Fundamentals of methodology Lecture 02: Scientific Knowledgeuser.mendelu.cz/cenek/Fundamentals_of_Methodology/Lectures_and… · debate while recognizing that there are di ﬀerences

FRRMS MENDELU

2019

Fundamentals of methodology

Lecture 02: Scientific Knowledge

Reader

Created for FRRMS students as a part of a course reading

Includes (for full citations see the course syllabus):

Obligatory reading:

Halperin, Heath, p. 25-52

della Porta, Keating p. 19-39

Suggested reading:

Bryman, p. 19-43

Corbeta 9-29

Part IEpistemology and philosophy ofthe social sciences

2 How many approaches in the socialsciences? An epistemologicalintroduction

Donatella della Porta and Michael Keating

Paradigms in the social sciences

Partisans articulate their positions with passion and intensity, yet the nature of whatdivides them is hard to pin down. At times we hear of a stand-off between ‘qualitative’scholars, who make use of archival research, ethnology, textual criticism, and dis-course analysis; and ‘quantitative’ scholars, who deploy mathematics, game theory,and statistics. Scholars in the former tradition supposedly disdain the new, hyper-numerate, approaches to political science as opaque and overly abstract, while scholars of the latter stripe deride the ‘old’ ways of studying politics as impression isticand lacking in rigor. At other times the schism is portrayed as being about the properaspiration of the discipline – between those who believe that a scientific explanationof political life is possible, that we can derive something akin to physical laws ofhuman behavior, and those who believe it is not . . . at still other times the rivals areportrayed as ‘rational choice theorists,’ whose work is animated by the assumptionthat individuals are rational maximizers of self-interest (often economics, sometimesnot), and those who allow for a richer range of human motivations (Shapiro, Smithand Masoud 2004a: 1).

This quotation from the introduction to a recent volume on Problems andMethods in the Study of Politics addresses a core methodological issue for thesocial sciences in general: how many approaches/methods are available forstudents in the discipline? And what are the main cleavages along which theyare divided?

In The Structure of Scientific Revolutions, Thomas Kuhn (1962) suggestedthat mature scientific disciplines rely upon a paradigm that defines what tostudy (relevance of social phenomena), why to study (formulating explana-tory hypotheses) and how to study (through which methods). In normal timesthe presence of a paradigm, based upon previous acquisitions in a discipline,

19

allows for the accumulation of knowledge. In times of turbulence, scientificrevolutions produce changes of paradigm. An important element of a para-digm is that it is accepted by the whole community of scientists active in acertain discipline. According to Kuhn, in the 1960s the existence of a paradigmin the social sciences was an open question; in the 2000s, it remains so.

Some social scientists insist that there is only one approach (and thus oneparadigm) in the social sciences. King, Keohane and Verba (1994: 6) synthe-sized the ‘ideal to which any actual quantitative and qualitative research’should aim in the following definition of ‘scientific research’:

1 The goal is inference. Scientific research is designed to make descriptive or explana-tory inferences on the bases of empirical information about the world . . .

2 The procedures are public. Scientific research uses explicit, codified, and publicmethods to generate and analyse data whose reliability can therefore be assessed . . .

3 The conclusions are uncertain . . .4 The content is the method. . . . scientific research adheres to a set of rules of infer-

ence on which its validity depends.

Not all social scientists, however, share all these assumptions or evenbelieve in the possibility of a common definition of scientific research. Somethink that social science is pre-paradigmatic, still in search of a set of unify-ing principles and standards; others believe that it is post-paradigmatic,having shed a set of scientistic assumptions tied to a particular conception ofmodernity (the post-modern approach). Yet others believe that it is non-paradigmatic, in that there can never be one hegemonic approach and set ofstandards, but that the social world is to be understood in multiple ways, eachof which may be valid for specific purposes; or even that it is multiparadig-matic, with different paradigms either struggling against each other or ignor-ing each other.

Some social scientists are specifically concerned with this issue, specializingin the philosophy of social science and the theory of knowledge. Others takethe basic issues for granted and concentrate on empirical research. We agreethat not all social scientists need to be philosophers, and certainly most socialscience research would never get off the ground if we had first to resolve thefundamental questions about being and knowing. Nevertheless, somereflection on the foundations of knowledge is necessary as a preliminary to allresearch.

We argue that it is possible to encompass much of the field, not by impos-ing a single truth, but by setting certain standards of argumentation anddebate while recognizing that there are differences in approaches and types of

20 Donatella della Porta and Michael Keating

evidence. Although these do not inevitably constitute fundamentally differentworld-views, they are not necessarily all compatible. Researchers need to beaware of the various approaches, the differences among them, and the extentto which they can be combined.

Disputes over approaches are often presented in a rhetorical form basedupon a dualist opposition of two main approaches (usually positivistic versushumanist, or quantitative versus qualitative) (Cresswell 1994). Others followa more nuanced ‘two-plus-one’ approach, with two more extreme positionsand a more moderate version of one of them (as in Corbetta 2003). In what-follows, we have constructed some simplified ideal types of rival approachesin order to explore their inherent logic. Such devices are inescap able if weare to understand clearly the main issues at stake, although in practice socialscience research is more complex and different approaches are mixed invarious ways. We do not claim that any social scientists follow preciselythese formulations, but many of the issues discussed below provide relevantguidelines for the methodological choices we often have to make in ourresearch.

What can we know and how? Ontologies and epistemologies in the socialsciences

Usually, competing approaches in the social sciences are contrasted on (a)their ontological base, related to the existence of a real and objective world; (b)their epistemological base, related to the possibility of knowing this world andthe forms this knowledge would take; (c) their methodological base, referringto the technical instruments that are used in order to acquire that knowledge(Corbetta 2003: 12–13).

The ontological question is about what we study, that is, the object of inves-tigation. Disputes about the existence of a physical world go back to theancients. This is not the point at issue here, since few people now bother todispute the existence of physical objects.1 Rather, the question is how theworld fits together and how we make sense of it. The natural sciences are stillhome to arguments about how we identify natural phenomena, for examplewhether taxonomies of species really exist in nature or are the product ofscientific classification. For nominalists, categories only exist because we arbi-trarily create them. For realists,2 the categories are there to be discovered.Again, we should not overstate this point. There are certain categories that areunchallenged and others that everyone accepts are the product of convention.

21 How many approaches in the social sciences?

Almost everyone accepts a distinction between living forms and inert objects,and most accept a distinction between human beings and other animals. Onthe other hand, there was an argument in 2006 about the definition of a planetfollowing the discovery of objects in the solar system smaller than Pluto, whichhad been accepted as a planet for years. This was not an argument about facts(the existence or size of the new body), but a purely nominalist argumentabout definitions (Kratochwil, ch. 5, uses the same example).

Most disputes between nominalists and realists in the natural sciences areat the margins, where conventional categories and labels can be challenged onthe grounds that they are misleading or that they reify what should properlybe seen as concepts rather than objects. In the social sciences there are muchwider differences about the degree to which the world of social phenomena isreal and objective, endowed with an autonomous existence outside the humanmind and independent of the interpretation given to it by the subject(Corbetta 2003). For some, the only ‘real’ object is the individual person, withall other units being mere artefacts. This is the basis for ‘methodological indi-vidualism’ and for most, but not all, rational choice approaches.3 Most socialscientists, however, use larger categories such as class, gender or ethnicity, pro-voking disputes about the extent to which these are real objective distinctions,the product of our own categorization, or just concepts.4

Epistemology is about how we know things. It is a branch of philosophy thataddresses the question of the ‘nature, sources and limits of knowledge’ (Klein2005). Knowledge here is propositional knowledge – distinct from ‘belief ’ inthat it requires that we give reasons for saying that something is so and canpotentially convince others. Again, the question arises also in the natural sci-ences; but they have shared standards of evidence, argument and logic. This isnot so in the social sciences, with some social scientists calling for objective evi-dence akin to that of the natural sciences, while others insist that other formsof knowledge are possible. For example, a common device in positive socialscience is to contrast ‘myth’, as widely shared belief, with ‘reality’, revealed byempirical research; the task of the social scientist is to expose this falsehood anddiscard what is not empirically verifiable or falsifiable. Many anthropologists,however, would reject this way of proceeding, on the grounds that myths andbeliefs are data as valid as any other and that we have no business telling otherpeople (especially in other cultures) that their construction of the world iswrong, as opposed to merely different. Less radically, many social scientistswould agree that myths are important factors in themselves and their role insocial behaviour is independent of whether they are true or false. Of course,social science itself can be charged with existing on myths, for example the


myth of rationalized institutions that – according to neoinstitutional analysisof organizations – dominates in modern societies (Meyer and Rowan 1983:27). As in other domains, this modernist myth is challenged by other dis-courses stressing the post-modern character of contemporary societies.

Taking these two dimensions together, we can identify four broadapproaches (Table 2.1). Again, these should not be taken as hard categories (orfixed labels), but rather as positions on a spectrum from the most positivisticto the most humanistic.

The traditional approach in positivism (as represented in the work ofComte, Spencer and, according to some, Durkheim)5 is that social sciences arein many ways similar to other (physical) sciences. The world exists as an objec-tive entity, outside of the mind of the observer, and in principle it is knowablein its entirety. The task of the researcher is to describe and analyse this reality.Positivist approaches share the assumption that, in natural as in social sci-ences, the researcher can be separated from the object of his/her research andtherefore observe it in a neutral way and without affecting the observed object.As in the natural sciences, there are systematic rules and regularities govern-ing the object of study, which are also amenable to empirical research. In the


Table 2.1. How many ontologies and epistemologies in the social sciences?

Positivist Post-positivist Interpretivist Humanistic

Ontological issuesDoes social Objective; Objective, Objective and Subjective: reality exist? realism critical realism subjective as science of the

intrinsically spiritlinked

Is reality Yes, and easy Yes, but not Somewhat, but No; focus on knowable? to capture easy to capture not as separate from human

human subjectivity subjectivity

Epistemological issuesRelationship Dualism: scholar Knowledge is Aims at No objective between the and object are influenced by understanding knowledge is scholar and two separate the scholar; subjective possiblehis/her object things; inductive deductive knowledge

procedures procedures

Forms of Natural laws Probabilistic Contextual Empathetic knowledge (causal) law knowledge knowledge

words of Emil Durkheim (1982: 159), ‘Since the law of causality hasbeen verified in other domains of nature and has progressively extended itsauthority from the physical and chemical world to the biological world, andfrom the latter to the psychological world, one may justifiably grant that it islikewise true for the social world.’

In neo-positivism and then post-positivism, these assumptions are relaxed.Reality is still considered to be objective (external to human minds), but it isonly imperfectly knowable. The positivist trust in causal knowledge ismodified by the admission that some phenomena are not governed by causallaws but, at best, by probabilistic ones. This does not represent a sharpbreak with the natural sciences but follows modern scientific developments(Delanty 1999). If positivism closely resembles the traditional scientificmethod (or Newtonian physics) in its search for regularities, post-positivismis closer to modern scientific approaches, which accept a degree of uncer-tainty. Critical realist epistemology holds that there is a real material world butthat our knowledge of it is often socially conditioned and subject to challengeand reinterpretation.6 There are mechanisms governing human affairs thatmay be unobserved and unobservable, but these are not therefore to be dis-counted. Again, this is also true in the natural sciences, where theories haveoften been formulated and applied before the underlying causal mechanismshave been explicated.

Similar ideas are present in (social) constructionism (sometimes called con-structivism7). This approach does not, as is sometimes thought, argue that thephysical world itself is the product of the imagination of the social scientist;rather, it is he/she who puts order onto it. As Hacking (1999: 33) explains:‘Social constructionists tend to maintain that classifications are not deter-mined by how the world is but are convenient ways to represent it.’ Theoriesare not descriptions to be evaluated by their literal correspondence to somediscoverable reality, but partial ways of understanding the world, whichshould be compared with each other for their explanatory power (Kratochwil,ch. 5). The world is not just there to be discovered by empirical research;rather, knowledge is filtered through the theory the researcher adopts.

These ontologies and epistemologies shade into the interpretivist approach.Here, objective and subjective meanings are deeply intertwined. Thisapproach also stresses the limits of mechanical laws and emphasizes humanvolition. Since human beings are ‘meaningful’ actors, scholars must aim at dis-covering the meanings that motivate their actions rather than relying on uni-versal laws external to the actors. Subjective meaning is at the core of thisknowledge. It is therefore impossible to understand historical events or social


phenomena without looking at the perceptions individuals have of the worldoutside. Interpretation in various forms has long characterized the study ofhistory as a world of actors with imperfect knowledge and complex motiva-tions, themselves formed through complex cultural and social influences, butretaining a degree of free will and judgement.8

Historians also recognize that the interpretation is often dependent on thevalues and concerns of the historian him/herself and that reinterpretation ofthe past (revisionism) is often stimulated by the political agenda of thepresent. Such traditional forms of interpretation have been joined by a newerschool of interpretivism derived from post-modernist premises (Bevir andRhodes 2003). This school casts doubts on the epistemological constants ofmuch social science, which it sees as unduly influenced by modernist assump-tions about order, causation and progress (themselves in turn derived fromnineteenth-century natural science). Interpretation works at two levels. Theworld can be understood not as an objective reality, but as a series of inter-pretations that people within society give of their position; the social scientist,in turn, interprets these interpretations. In a further reflexive turn, social scientists’ interpretations feed back to the people through literature andmedia, influencing them yet again in what Giddens (1976) calls the ‘doublehermeneutic’. This is one reason why relationships that may have held in thepast might not hold in the future (Hay 2002).

The humanistic approaches shift the emphasis further towards the subjec-tive. In this perspective, what distinguishes human science from natural sciences is that human behaviour is always filtered by the subjective under-standings of external reality on the part of the people being studied and theresearcher him/herself. Social science is therefore, in the often-quoteddefinition proposed by Clifford Geertz (1973: 5), ‘Not an experimental sciencein search of laws but an interpretative science in search of meaning’. In themost radical versions of this approach, reality does not exist beyond the (rel-ative and partial) images the various actors have of it. Knowing the reality istherefore impossible, and scholars should focus on the meaning throughempathetic knowledge.

How many methodologies in the social sciences?

The methodological question refers to the instruments and techniques we useto acquire knowledge. At one level, this is independent of the ontological andepistemological questions just discussed, since there are multiple ways of


acquiring each type of knowledge. In practice, they tend to be linked, sincepositivistic social science lends itself naturally to ‘hard’ methods, seekingunambiguous data, concrete evidence and rules and regularities, while moreinterpretive approaches require ‘softer’ methods allowing for ambiguity andcontingency and recognizing the interplay between the researcher and theobject of research (but see below). All these differences are linked with thediffering final scope of the research.

In the positivist tradition, research aims at singling out causal explanations,on the assumption of a cause–effect relationship between variables (seeHéritier, ch. 4). Researchers aim at an explanation that is structural andcontext-free, allowing generalization and the discovery of universal laws ofbehaviour. Such laws may be discovered in two ways. The inductive approach,which is associated with pragmatism or behaviourism (Hay 2002), involvesderiving generalizations from specific observations in a large number of cases.Positivists in the more scientific tradition, however, would insist that one startwith a theory, which then generates hypotheses (an expected state of affairs)which are then subjected to the test of hard facts and only accepted if theysurvive the ordeal (see Héritier, ch. 4).9 This is the hypothetico-deductive(deductive-empirical) method,10 in which the study of social reality utilizesthe conceptual framework, techniques of observation and measurement,instruments of mathematical analysis and procedures of inference of thenatural sciences (Corbetta 2003: 13). Since it is rarely possible in the social sci-ences to conduct experiments, large datasets and statistical analyses are usedin order to identify and isolate causes and effects in a rigorous manner andarrive at a single explanation. This is not to say that positivists use only quan-titative methods; but where they use other (qualitative) methods, they followthe same logic of inference. The main aim is ‘identifying, assessing and elim-inating rival explanations’ (Collier, Brady and Seawright 2004a: 229).

By contrast, interpretive/qualitative research aims at understanding eventsby discovering the meanings human beings attribute to their behaviour andthe external world. The focus is not on discovering laws about causal rela-tionships between variables, but on understanding human nature, includingthe diversity of societies and cultures. More specifically, following Weber, thistype of social science aims at understanding (verstehen) the motivations thatlie behind human behaviour, a matter that cannot be reduced to anypredefined element but must be placed within a cultural perspective, whereculture denotes a web of shared meanings and values (see della Porta, ch. 11,and Keating, ch. 6). Theory is important, but is not always established priorto the research as in the deductive-empirical approach. In the form of


‘grounded theory’, it may be built up in the course of the research, but thenbe available for further research and the study of other cases. Cases are notbroken down into variables but considered as interdependent wholes; gener-alization is achieved by assigning cases to classes and approximating themto ideal types. Context is considered as most important since research onhuman activity must consider an individual’s situational self-interpretation(Flyvbjerg 2001: 47). Predictability is impossible since human beings changein time and space and, in the words of Bourdieu (1977: 109), ‘practice has alogic, which is not that of logic’. The outcome of the research then takes theform of specific explanations of cases, but also of refined concepts for theanalysis of future cases.

This type of research, like the positivist approach, seeks explanations forsocial outcomes but does not expect to derive these from universal rules.Rather, explanation comes from the interpretation of people’s motives fortheir actions. Ferejohn (2004: 146) clarifies this distinction by contrasting‘externalist’ and ‘internalist’ explanations:

Externalists explain action by pointing to its causes; internalists explain action byshowing it as justified or best from an agent’s perspective. Externalist explanations arepositivist and predictive; internalist explanations are normative or hermeneutic.Externalists tend to call themselves political scientists; internalists, political theorists.And, both externalists and internalists agree, if they agree on little else, that they areengaged in different enterprises.

Sometimes this difference is presented as a contrast between quantitative(positivist) and qualitative (interpretive) methods (Creswell 1994; Corbetta2003). This is a source of considerable confusion, conflating ontology andepistemology on the one hand with methods and methodology on the other.The quantitative method refers to sophisticated data analysis using largenumbers; there is certainly a stream in social science that is both positivist andquantitative in approach. Brady, Collier and Seawright (2004) describe a‘mainstream quantitative method’ as an approach based upon the use ofregression analysis and related techniques aiming at measuring causal infer-ence; but note that work in the positivist tradition also makes use of non-quantitative material such as case studies, paired comparisons, interviewrecords and even ethnographic approaches in field research and interpreta-tion. King, Keohane and Verba (1994), leading exponents of the positivistapproach, accept that qualitative methods may be used as a supplement toquantitative methods as long as they follow the same logic. The chapters inBrady and Collier (2004) argue that qualitative methods can tackle questions


that quantitative methods cannot encompass, but remain within the samepositivist epistemological framework. Even participant observation is oftenused within ‘theory-driven’ research designs (Lichterman 2002). Laitin (2003)likewise admits to the validity of narrative approaches but only as part of a tri-partite approach in conjunction with statistics and formal modelling. ForLaitin, narratives can provide plausibility tests for formal models, mecha-nisms that link dependent and independent variables, and ideas for searchingfor new specifications of variables that have yet to be modelled.

There is, however, another rather different, more specific meaning oftengiven to the term qualitative methods, linked to the interpretive approachderived from ethnography and anthropology and which has now arrived inother areas of the social sciences. As defined by Denzin and Lincoln (2000: 3):

Qualitative research is a situated activity that locates the observer in the world. It con-sists of a set of interpretive practices that make the world visible. These practices trans-form the world. They turn the world into a series of representations, including fieldnotes, interviews, conversations, photographs, recordings and memos to the self. Atthis level, qualitative research involves an interpretive, naturalistic approach to theworld. This means that qualitative researchers study things in their natural settings,attempting to make sense of, or to interpret, phenomena in terms of the meaningspeople bring to them.

Favoured methods for this are unstructured interviews, focus groups, textualanalysis and content analysis (see Bray, ch. 15). However, just as positivistsmay make use of interviews, case studies and even participant observation, sointerpretivists sometimes use quantitative techniques. Sophisticated com-puter software is available for analysis of the content of speech and texts toidentify key words, patterns of symbols, codes and references. This shows onceagain that we should not confuse issues of epistemology with those ofmethodology or research technique.

From methodology to method

It would therefore be a great simplification to say that there is a distinctionbetween quantitative and qualitative methods corresponding to the distinc-tion between positivist and interpretivist epistemologies. Methods are nomore than ways of acquiring data. Questions about methods do, however,come together with epistemology and theory in discussions about method -ology, which refers to the way in which methods are used. Here we face choices


pointing in the direction of more or less formally structured approaches and‘harder’ or ‘softer’ methods.11 To explore them, we first present a simplified setof choices to be made in research design and in method selection (see alsodella Porta, ch. 11).

The first choice is in the framing of the research question. Positivists willusually start with a hypothesis deductively derived from theory and previousknowledge. Typically, this will postulate some expected state of affairs orcausal relationship and be empirically falsifiable. By this we do not mean thatit is actually false, merely that the conditions under which it can be rejectedare specified. If it is not falsified, then it can be taken as true, not only for thecases in question but for all cases with the same characteristics. Interpretivists(or qualitative researchers in the restricted sense) work more inductively,build up the research question in the course of the research and are preparedto modify the design while the research is in progress. There is thus no cleartime distinction between the research design and its implementation, as theyare interlinked with continuous feedbacks. Positivists take care to opera-tionalize their concepts and hypotheses in scientific and general terms, whileinterpretivists let the concepts emerge from the work itself.

Another difference refers to the number of cases analysed, as well as the cri-teria for selecting them. Positivists will often choose a large number of casesto achieve the maximum generalizability and capture most sources of varia-tion. Alternatively, they will choose a small number of cases, but rigorouslyselect them in such a way that their differences can be specified precisely. InJ. S. Mill’s (1974) classic formulation, two cases should be chosen such thatthey share only one attribute in common, or so that they differ in only oneattribute. In this approach, numbers are not necessarily used, and cases can befew: the logic is, however, the approximation to a statistical type of analysis,with concerns with (statistical) representativity, validity and reliability. Non-quantitative techniques must thus follow the same logical structure and rulesfor scientific inference (King, Keohane and Verba 1994).12 Interpretivists, onthe other hand, will select cases on the basis of their inherent interest (forexample, paradigmatic cases), not because they are typical of a category butfor what they tell us about complex social processes.

Positivists usually employ the language of variables. That is, they are notinterested in cases as such, but in the properties of those cases that cause themto differ. Since they are concerned with general or universal laws, they want toknow what factors cause which outcomes in social life, for example what is thecausal relationship between economic growth and democratization. Thisrequires that they develop an operational definition of economic growth and


of democratization and ways of measuring them. These then become the vari-ables in the analysis, with economic growth as the ‘independent’ or causalvariable and democratization as the ‘dependent’ or caused variable. Of course,it is rarely the case that one independent variable will everywhere and alwaysproduce the same effects on the dependent variable, but this merely meansthat more variables need to be added so that, eventually, all variation isaccounted for. In the words of Przeworski and Teune (1970), the aim is ulti-mately to ‘eliminate proper names’ – that is, to account for social processes byreference to general rules without talking about individual cases, since thesewill all be accounted for within the general rules (Corbetta 2003). Context forthese social scientists merely consists of variables that have yet to be specifiedadequately (Laitin 2003).

Neo-positivist approaches have relaxed the assumption that knowledge iscontext-free and that the same relationships among variables will hold every-where and at all times. Instead, there is more emphasis on the particular andthe local, and on the way in which factors may combine in different circum-stances. To capture this contextual effect, researchers have increasinglyresorted to the idea of institutions as bearers of distinct patterns of incentivesand sanctions, and on the way that decisions taken at one time constrain whatcan be done later. These institutional factors may be expressed in the form ofvariables, but an important role is played by comparative study of a smallnumber of cases, where the variation is the institutional structure and its his-torical evolution (see Steinmo, ch. 7). Neo-positivists seek to express the effectof context in the form of institutional structures and try to avoid the conceptof culture as impossible to operationalize and inimical to general theorizing.Others, however, have moved from institutions into culture, providing abridge between interpretivist and positivist approaches (see Keating, ch. 6).

Interpretive analyses keep a holistic focus, emphasizing cases (which couldbe an individual, a community or other social collectivity) as complex entities(della Porta, ch. 11) and stressing the importance of context. Concepts are ori-entative and can be improved during the research. The presentation of thedata is usually in the form of thick narratives, with excerpts from texts (inter-views, documents and ethnographic notes) presented as illustration. Theassumption of mutual influence among the many factors at work in any casediscourages any attempt to reason about causes and effects or to generalize.Understanding reality implies ‘immersing ourselves in information about theactors in question, and using both empathy and imagination to constructcredible accounts of their senses of identity’ (Smith 2004: 43). In such anenterprise, methods generally labelled as qualitative – such as interpretative


textual analyses, ethnographic fieldwork, biographical studies or participantobservation – are key (see Bray, ch. 15).

Another difference is in the relationship of the researcher to the researchobject: how much participation is permissible in the situation to be observed?How much of a stranger should the researcher be? And how sympathetictowards the point of view of the object of his/her research? The positivist setsup a complete separation between the observer and what is observed, takingcare not to ‘contaminate’ the research by becoming part of it. S/he will preferstandardized questionnaires and interview schedules, anonymized surveys,rigorous coding of responses and, often, quantitative techniques. The inter-pretivist will tend, on the contrary, to immerse him/herself in the situation tobe studied, to empathize with the population and to see things from their per-spective. Anthropologists spend long periods in the field seeking to gainan inside knowledge. The sociology of intervention (as pioneered by AlainTouraine) involves the researchers working with social movements and theactivists they study in a common path, with the aim of helping the latter tointerpret the situation and engaging in mutual learning. In the most radicalunderstanding, all statements about the external world have such strong sub-jective elements that no shared observation can exist. The acknowledgementof the role of interactions between researchers and the object of the researchposes many ethical issues; among others, whom to accept as a sponsor, howmuch to reveal about the research to the interviewees, how to protect theirprivacy, how to compensate them for their collaboration, how to keep theminformed about the results of the research and how to avoid manipulation.

Another critical question that differentiates approaches concerns value-neutrality. In the positivist perspective, the researcher brings no normative,ideological or political perspectives to bear on the research. S/he is merelyseeking the unadorned truth. Critics would argue that this often conceals anormative agenda and indeed that the founding assumptions of positivismthemselves reflect a value choice.13 Positivists counter that, if this is the case,then all such normative tendencies should be declared in advance. Normativework as such is, according to this perspective, a separate endeavour, whichbelongs in the field of ethical philosophy. Interpretivists would tend not tomake such a sharp distinction between empirical and normative work; takento its fullest, this approach denies the distinction between facts and valuesaltogether. More moderate versions argue that most language and speech actshave both descriptive and normative elements within them, that conceptsthemselves usually have some normative content, and that the researchershould be aware of this. Recently, there have been conscious efforts to pull


together normative work derived from philosophy with empirical research(see Bauböck, ch. 3). While in one sense new, this also represents a return tothe classical era of social thought. Flyvbjerg (2001) has controversially sug-gested that, since the social sciences can never gain the explanatory power ofthe natural sciences because of the nature of the world, they should return tothis earlier age and seek to provide reflexive analysis and discussions of valuesand interests aimed at praxis, that is, to contribute to the realization of a bettersociety. This in turn has sparked some critical rejoinders (Laitin 2003).

Returning to our fourfold classification, and with the caveats already men-tioned, we can summarize some main methodological assumptions (Table 2.2).

How many ways to knowledge?

How exclusive must be our methodological choices? Should we leave space forepistemological anarchism, and trust exchanges with scholars working withinthe other ‘paradigm’? Even switching between the two? Or is the buildingof knowledge only possible within one paradigm? Is the combination ofapproaches/methods useful to overcome the limits of each methodology? Orwould it risk undermining the soundness of the empirical results?

Three approaches to these issues can be singled out in the social sciences:(a) Paradigmatic, exclusive approach. In the light of Kuhn’s conception of

the role of paradigm, some social scientists aim at a paradigmatic science,


Table 2.2. How many methodologies in the social sciences?

Positivist Post-positivist Interpretivist Humanistic

Which Empiricist, aiming Mainly empiricist, Relative focus Focus on values,methodology? at knowing the recognizing on meanings, meaning and

reality context context purposes

Which Imitating the Based upon Seeking Empathetic method/s? natural method approximations to meaning interactions

(experiments, the natural (textual between mathematical method analysis, researchers andmodels, (experiments, discourse object of statistical statistical analysis, analysis) researchanalysis) quantitative

interviews)

where only one paradigm is considered as the right one, combining theory,methods and standards together, usually in an inextricable mixture (Kuhn1962: 109). Those who see the social sciences as paradigmatic stress the impor-tance of converging on (or imposing) one single way to knowledge.

(b) Anarchist, hyper-pluralistic approach. At the other extreme, there is an‘inclusive’ position that combines scepticism about a ‘true’ knowledge withenthusiasm for experimentation with different paths to knowledge. Thosewho subscribe to this position to various extents support Feyerabend’s anar-chism and his belief that:

the world we want to explore is a largely unknown entity. We must therefore keep ouroptions open . . . Epistemological prescriptions may look splendid when comparedwith other epistemological prescriptions . . . but how can we guarantee that they arethe best way to discover, not just a few isolated ‘facts’, but also some deep-lying secretsof nature? (Feyerabend 1975: 20).

(c) The search for commensurable knowledge. Between those two extremes,there are positions that admit the differences in the paths to knowledge anddeny the existence of a ‘better one’, but still aim at rendering differences com-patible.

Within this third perspective – which we tend to follow in this volume – itis important to compare the advantages and disadvantages of each methodand methodology but also be aware that not all are compatible. Goals thatcannot be maximized at the same time include seeking precise communica-tion as opposed to fertility in the application of concepts, parsimonious expla-nations as opposed to thick descriptions, and generalizability as opposed tosimplicity (Collier, Brady and Seawright 2004a: 222). It may therefore be ne -cessary to trade off one advantage against another. This choice will be madeon the basis of the fundamental question the researcher is trying to answer –for example, whether he/she is trying to explain a particular case; to gainnomothetic knowledge (discovering general rules); or seeking ways to achievea better society. It depends on the preferences of the researcher, and on thesorts of data that are available, including reliable statistical data or detailedfield data requiring long immersion in the field.

The choice of approach is linked to another choice in social scienceresearch: whether to start with a theory, a method or a problem. Those aimingat a paradigmatic social science will often start with a theory, seeking to testit with a view to proving, disproving or modifying it and so contributingto universal knowledge. This is often tied to a particular methodology toallow studies to be reproduced and compared. Those interested in a specific


problem, on the other hand, will tend to look for the method and approachthat seems to offer more by way of understanding of the case. Exponents ofthe first approach are accused of studying methods for their own sake andchoosing only issues that are amenable to that method – summed up in theold adage that if the only tool you have is a hammer, every problem starts tolook like a nail (Green and Shapiro 1994; Shapiro 2004). Those who focus onproblems, in contrast, are accused of adding nothing to the writings of his -torians and journalists (Shapiro, Smith and Masoud 2004a).

Ways of combining knowledge can be characterized as synthesis, triangu-lation, multiple perspectives and cross-fertilization. Synthesis involvesmerging elements of different approaches into a single whole and can bedone at various levels. Synthesizing different epistemologies is virtuallyimpossible, since they rest on different assumptions about social reality andknowledge. Methodologies may be easier to synthesize since, as we have seen,they are not necessarily tied to specific epistemological assumptions.Techniques and methods are most easily combined since, as we have noted,many of them can be adapted to different research purposes. So comparativehistory and historical institutionalism have adopted and adapted techniquesfrom comparative politics, history and sociology to gain new insight intoprocesses of change.

Triangulation is about using different research methods to complement oneanother. Again, it is difficult to triangulate distinct epistemologies, easier withmethodologies and very common with methods. So positivists can incorpo-rate interviews and textual analysis into their research designs, although usingthese as hard data rather than in the manner of interpretivists. Case studies arefrequently used to complement large-N statistical analyses as ways of openingthe ‘black box’ of explanation (see Héritier, ch. 4). Survey research may becomplemented by ethnographic work, which explores the way in which ques-tions are understood and the meanings of the responses.

Multiple perspectives implies that a situation may have more than one inter-pretation according to how we view it. De Tocqueville (1999) wrote that in hislife he had met theorists who believed that events in the world owed everythingto general causes, and practical people who imagined that daily events andactions were those that moved the world – he added that both were mistaken.Allison’s (1971) study of the Cuban Missile Crisis examined the same eventsusing different frames to come up with different explanations.

It has been said that everyone is born either an Aristotelian or a Platonist(Hacking 1999: 84), yet hardly any social scientist now is a naïve empiricistwho believes that the world represents itself to us without interpretation.


Conversely, nobody in mainstream social science denies the existence of thephysical world or maintains that reality is entirely subjective and in our minds.This encourages a cross-fertilization in a large middle ground.

Concepts often arise in the social sciences by different tracks, derived fromslightly different starting points but ending in similar places. For example, theconcept of ‘framing’, widely used in policy analysis to indicate the differentways in which people will define and conceptualize a policy issue or problem,can be derived from an anti-positivist and interpretivist position (Fischer2003) but also from a positivist one. It has been used in social movementresearch since long before the so-called ‘cultural turn’ by scholars interested instrategic action by collective actors (such as David Snow), but also by othersmore interested in the micro-dynamics of cognition (such as WilliamGamson). In all cases, the idea is that situations can be interpreted differentlyand presented differently to evoke different reactions from the same set offacts. The differences are in exactly how much weight is given to the objectiveworld and how much to its interpretation. The concept of culture, much usedby interpretivists, is rejected by positivists and rational choice analysts butthen often brought back in as normative institutionalism or shared meaningsand understandings that underpin policy communities (see Keating, ch. 6).Context is central to ethnographic and interpretivist approaches, where it isdeeply textured and rich, but is also used in neoinstitutionalist analysis andeven features in the hardest regression analyses (where difficult whole cases areexpressed as dummy variables). New institutionalism has come into the socialsciences through several doors: in political science, where it is a response todecontextualized rational choice approaches; in sociology, where it draws onorganizational theory; and in economics, where it draws on economic sociol-ogy. The result is a set of concepts that are very similar but, because of theirdistinct origins and vocabulary, never quite identical.

There is also a large crossover in ways of developing and using theory. Asmentioned, grounded theory does not start with a deductively producedhypothesis but with experience; nevertheless, it does then go on to build upgeneral theory of wider applicability. It owes a lot to the American pragmatisttradition, with roots in a ‘realist’ ontology, but it has been extended and elab-orated in more interpretivist approaches. Meanwhile, in the United States,that same realist ontology has evolved into varieties of rational choiceapproaches, based supposedly on the solid foundation of the individualperson, but in practice using an ideal-type construct and models derived fromdeductive reasoning. Indeed, rational choice approaches themselves seem tobe compatible both with determinism (on the assumption that preferences are


knowable and outcomes predictable from individual self-maximization) andwith free will (in that the individual does choose). A great deal of social scienceproceeds by going back and forth between theory and cases, using the one todevelop and deepen understanding of the other.

Sometimes the cross-fertilization is explicitly acknowledged. In a contribu-tion to a volume significantly titled Rethinking Social Inquiry. Diverse Tools,Shared Standards, Collier, Seawright and Munck (2004) stress the importanceof good theories and empirical methods, but also appreciate the contributionof interpretive work to concept formation and fine-grained description. Manyof the classic works in sociology and political science have taken the form ofinterpretive case studies from which general theories have been developed byexample, replication and extension (Van Langenhove 2007). Examples areAlexis de Tocqueville’s De la démocracie en Amérique and L’ancien régime et larévolution, but also more recent historical sociology in the school ofBarrington Moore Jr. Qualitative analysis has also been used to highlightcausal effects by focusing on striking cases where the impact is clearest and thedetailed mechanisms can be examined. In this way, social scientists canproceed from correlation, where the same causes are associated with the sameeffects, to explanations of why and how.

Influences come not only from within the discipline but also from otherareas of science. Newtonian physics, with its search for laws and constants, hasbeen an inspiration for positivist social science, while its opponents havedrawn attention to the uncertainties underlying modern physics and the hugeepistemological assumptions among which scientists have to choose (such asthe existence of one or parallel universes). Evolutionary biology now providesinspiration for historical institutionalists (see Steinmo, ch. 7).14 Rationalchoice scholars are inspired by neoclassical economists, while institutionaleconomists learn from sociology. History long provided the model and toolsfor the study of politics in Britain, while law was its basis in many Europeancountries. After a period in which the social sciences insisted on their ownspecificity, many scholars are now turning back to history, while developmentsin legal scholarship (including law in context, critical legal theory and consti-tutionalism) are linking back to concerns in political science and sociology.Literature has helped inspire the ‘sociological imagination’ by portraying dra-matic situations that need to be explained and resolved and drawing attentionto the conflicts within the individual mind.

Cross-fertilization, however, is inhibited by the existence or closing up ofresearch communities, groups of scholars in regular contact and discussion,who may define their common interest by substantive topic, methodology, or


both (Sil 2004). These are reified and perpetuated by processes which them-selves are worthy of sociological analysis, including the existence of journalswedded to particular approaches, the orientation of individual departmentsor sections, patterns of graduate supervision and discipleship, routinizedassessment procedures, and routes to career advancement. When researchcommunities are defined both by substantive topic and by method, barriersmay be very high and knowledge remain limited to the problems each methodis best fit to tackle, secluded from external stimuli and challenges. On the otherhand, when barriers are more fluid, the problem emerges of the commensu-rability of different forms of knowledge, as well as ‘fuzzy’ and ill-defined stan-dards (Ruggie 1998). This makes it all the more important for researchers toknow the field and to be able to compare standards and arguments with thosefrom different communities. This is what Sil (2004) suggests under the labelof eclecticism, where problems of incommensurability are not absolute andcomparisons can be made across fields to the advantage of both empiricalknowledge and theoretical innovation.

Further problems are caused by the tendency for concepts or expressions tobecome fashionable and then stretched beyond their original or indeed anyuseful meaning. In recent years, for example, the use of the word ‘governance’has exploded. For some scholars, this is a specific phenomenon distinct fromgovernment and capable of operationalization, but for others it is used inter-changeably with government. Still others see it as less than government, refer-ring to a specific way of governing through networks, alongside traditionalinstitutional government. Others see it as a broader category of social regula-tion, of which government is a subcategory. Some see it as an alternative togovernment – that we are moving from a world of government to one of gov-ernance. ‘Construction’ or ‘social construction’ are similarly stretched to coveralmost everything (Hacking 1999) as, for a while, was the term ‘invention’.Discourse analysis is sometimes used as a specific methodology, with its ownontology (speech acts themselves) and its own techniques; at other times it isapplied to any technique that involves using texts and interviews. Sometimesthe blame for all this confusion lies with scholars thinking that they need toget inside the current paradigm in order to make their point; often it is merelya matter of publishers looking for a trendy title.

Of course, not everything is methodologically healthy, and the label of eclec-ticism should not be used to justify hybrids that violate, if not rules, at leastcodes of conduct of what we have presented here as main approaches. Althoughthe triangulation of various methods and methodologies within the sameresearch project often increases reliability and improves our understanding, the


different parts of the enterprise must respect internal coherence. If an ‘eclecticknowledge’ of qualitative and quantitative techniques enriches a researcher’scurriculum, human limits, together with the increasing sophistication of mostqualitative and quantitative techniques, impose some specialization. The following chapters offer differing approaches in ontology, epistemology andmethodology but also indicate points of commonality and overlap.

NOTES

11 This is either because they accept the material world, or because it is a question that cannotand need not be answered and is therefore futile to debate.

12 This is one of the terms in social science that has a multiplicity of meanings. In internationalrelations it has a rather different meaning from the one given here (see Kratochwil, ch. 5).

13 In fact, even the individualist solution, reducing the ontology to the individual human being,does not answer this question definitively, as one might argue that even the self-regardingrational individual is an artefact of social science methodology and not something thatoccurs naturally, since the original condition of human beings is the group. This is argued inAdam Ferguson’s (1966) Enlightenment classic, Essay on the History of Civil Society, of 1767.

14 A classic example of this is the case of gender. While nobody denies the existence of sexualdifferences, there is a big dispute over the category of gender, which includes a lot of otherattributes and roles which have been mapped onto sex differences.

15 Van Langenhove (2007) claims that late twentieth-century social scientists have often por-trayed the classical sociologists as more simplistically positivist than they really were.

16 Critical realism has been defined as ‘a philosophical view of science and/or theology whichasserts that our knowledge of the world refers to the-way-things-really-are, but in a partialfashion which will necessarily be revised as that knowledge develops’. ChristopherSouthgate, www.meta-library.net/.

17 For a discussion of the difference, see Hacking (1999: 47–9). He recommends leaving theterm ‘constructivism’ to the mathematicians.

18 This taps into a long-standing division in philosophy between determinists and thoseemphasizing free will. While for St Augustine and John Calvin, determinism was a matterof divine selection, for modern social scientists it is a matter either of genetic programming,social conditioning or a predictable response to institutional incentives. Believers in free willcannot by definition be certain about how another actor will behave, no matter what con-straints they are under.

19 In practice, social scientists often go back and forth between inductive and theory-drivenapproaches as they seek to frame their projects.

10 This is not to be confused with the pure deductive method, in which conclusions are derivedfrom premises by pure reasoning, with no empirical research involved. Héritier (ch.4)explains the link between induction and deduction in the positivist tradition.

11 These terms are not used in a value-laden way to suggest that one is better than the other.Hard methods correspond to the view that social science can be made to resemble the phys-ical sciences; soft methods to the view that social reality is more elusive.


12 For example, case studies can be accepted either to disconfirm a hypothesis (since it onlytakes one case to disprove a rule) or as a basis for formulating hypotheses for general testing.They are not valuable in themselves.

13 This is perhaps most obviously so in rational choice analysis, which claims a strictly posi-tivist basis but includes some strong assumptions and tends to lead to highly normative con-clusions.

14 This is not to say that the unity of the natural and social sciences can thereby be restored, asmany people insist that the specificity of the latter is that the objects of study are endowedwith consciousness and can act on their own volition.


Forms of Knowledge: Laws, Explanation, and Interpretation in the Study of the Social World

Chapter Summary

This chapter considers fundamental assumptions that researchers make about how we can know and develop knowledge about the social world, including assumptions about the nature of human behaviour and the methods appropriate to investigating and explaining that behaviour. The core concern is whether and how we can pursue a systematic and rigorous study of social phenomena in the way that scientists pursue study of the natural world. Without considering this issue, it is diffi cult to design or structure an approach to research into political phenomena, and to make any claim with respect to the fi ndings that result from that research.

This chapter focuses on three different answers to the question of how to ap-proach the study of social phenomena: those offered by positivism, scientifi c realism, and interpretivism. In exploring the differences among them and their implications for conducting political research, our discussion will engage with a number of ques-tions, including the following:

● What form(s) of knowledge should be the goal of political research?

● Should the social sciences strive to emulate natural science methods, or is understanding social phenomena something essentially different from explanation in the natural sciences?

● Can we study politics scientifi cally? What does it means to be ‘scientifi c?

● What distinguishes science from non-science?

Introduction

Every researcher must confront fundamental questions about the nature of knowledge and how we acquire it. Th ese questions are the focus of key debates in political research, and the subject of an ongoing inquiry into scientifi c practice, forms of knowledge, and the world of politics. What sort of knowledge can we gain about the social world? Is it the same sort of knowledge that scientists are, able to obtain about the natural world? Or are the forms of knowledge concerning the social world and the natural world necessarily diff erent? If they are diff erent, is it still possible to produce knowledge that is reliable and objective? What counts as legitimate knowledge of the social world? Th ese questions

PHILOSOPHY OF SOCIAL SCIENCE26

bear directly on research practice and, consequently, are of primary concern to those who seek to understand political processes and structures. Th e answer or answers you accept will determine the sort of research you pursue, the claims you make on the basis of that research, and your assessment of the fi ndings of the research produced by others in our fi eld.

We will consider three diff erent approaches to these questions: positivism, scientifi c real-ism, and interpretivism. Each approach diff ers from the others with respect to its ontologi-cal, epistemological, and methodological premises. Th ese diff erences are summarized in Box 2.3 in the concluding section of this chapter.

Th e terms ‘ontology’, ‘epistemology’, and ‘methodology’ relate to fundamental issues concerning research practice and knowledge. Ontology is concerned with ‘what is’: with assumptions about the nature of the social world and the basic elements that make up this world. Questions of ontology relevant to political research include whether the social world is fundamentally diff erent from the natural world; whether it is an objective reality that exists independently of us or is in important respects subjectively created. Episte-mology is concerned with what is knowable, with what we can know about social phe-nomena, and, consequently, what type or form of knowledge we should pursue and treat as legitimate knowledge about the social world. It is only when we have considered these ontological and epistemological questions that we can move to a consideration of meth-odological questions. Methodology is concerned with how we obtain knowledge, with the means and methods that can provide us with legitimate knowledge of the political world. Box 2.1 shows how these key issues concerning knowledge are related.

We begin this chapter with a discussion of the development of positivist thought and practice, including classical and logical positivism, Karl Popper’s critique of these, and the role of general laws and causation in social-scientifi c explanation. We then focus on two non-positivist positions: scientifi c realism and interpretivism. Th e three positions diff er from one another in many ways and, in particular, with respect to their view of how the assumptions, logic, and methods of science can be used by scholars to study human behav-iour. However, though each position has developed, in part, through a critique of the oth-ers, each of them produces useful forms of knowledge. Taken together, they have enabled us to broaden the range and type of questions that political research can eff ectively address.

BOX 2.1 Ontology X Epistemology X Methodology

Ontology Epistemology Methodology

What exists?

What is the nature of the

social world?

What sort of knowledge

of it is possible? How

can we know about it?

What strategies

can we use to gain

that knowledge?

FORMS OF KNOWLEDGE 27

Positivism

As a prelude to our discussion of positivism, it would be helpful to get a sense of its role in political research by briefl y considering behaviouralism and the ‘behavioural revolution’ in the fi eld of politics.

Behaviouralism is the term used for the application of positivism and empiricism to political research.1 What has been called the ‘behavioural revolution’ was concerned to pro-mote the systematic search for sound and reliable knowledge about politics based on a posi-tivist approach to knowledge. For behaviouralists, political research involves studying and explaining the observable behaviour of individuals or aggregates of individuals (parties, classes, interest groups, governments, social movements).

Behaviouralist research focuses on the question of what political actors do and why they do it. Until the mid-1970s, behaviouralist researchers emphasized an inductivist approach to research which, as we shall see, is associated with classical positivism. An inductive approach to social inquiry is one in which ‘knowledge is arrived at through the gathering of facts that provide the basis for laws’ (Bryman 2004: 11). Although behaviouralist research can employ both quantitative and qualitative data, during the 1950s and 1960s behaviouralist researchers tended to focus on questions that could be answered by gathering and studying data condu-cive to exact measurement, as for instance voting data or data from public-opinion polls and social surveys. Th is tendency generated the criticism that, by focusing on phenomena that lent themselves more easily to measurement, the fi eld had become preoccupied with tech-nique rather than substance, and was failing to address signifi cant problems.

Th ese concerns triggered a ‘post-behavioural revolution’. Despite its name, this ‘revolu-tion’ was not concerned to displace behaviouralism, but to ‘propel political science in new directions’ (Easton 1969: 1051). Some of these new directions moved the fi eld towards a further realization of positivist and behaviouralist goals, such as the trend towards ‘positive political theory’ or rational choice theory. Positive political theory assumes that rational self-interest, ‘as opposed to attitudes, which are the subject of study in much behavioral research’, provides the motivational foundation for behaviour; and that individual self-interested rational action combines to produce collective political outcomes (Amadae and Bueno de Mesquita 1999: 270). But while the post-behaviouralist revolution moved behavioural research forward, it also set in motion trends that moved the fi eld in non-positivist direc-tions, and encouraged the emergence of an array of theoretical approaches that represented a self-conscious rejection of behavioural and positivist assumptions. Normative theory, which we will consider in Chapters 3 and 6, witnessed a re-birth, and oft en self-consciously as a response to the infl uence of behaviouralist research. In addition, there emerged a set of approaches based on non-positivist assumptions and associated with ‘interpretivism’, including constructivism, feminism, post-modernism, and critical theory.

Th e behavioural revolution set in motion an important process of discussion and debate within political research about the methods and goals of the fi eld. It began a discussion on the desirability and possibility of attaining reliable, empirical, causal knowledge about political life. It promoted more methodologically self-conscious research; and, though much behavioural research originally focused on what might be characterized as a narrow range of questions, it also succeeded in broadening the research domain, as behavioural researchers, seeking insights


from the theories, research methods, and fi ndings of other disciplines, opened the way to greater interdisciplinarity in the fi eld. Behaviouralism established an emphasis on research based on empirical observation, testing involving systematic evidence, and falsifi able and causal explanation. By emphasizing the importance of research that is capable of replication by others, behaviouralism makes researchers more precise about what they want to know, what explanation they are advancing, and how they intend to demonstrate it.

We will gain a better understanding of this revolution, and of both its positivist and non-positivist legacy, as we explore the basic tenets and contours of positivist thought.

Positivism began as a movement to establish a sound basis for social-scientifi c inquiry. Th is is a fundamentally important issue in political research. Political researchers want to be able to off er credible answers to important questions, and they are concerned to ensure that the research practices and methods they employ enable them to do this. Positivism off ers a particular approach to resolving these issues. It maintains that it is possible to arrive at factual, reliable, and objective answers to questions about the social world by employing the methods used in the natural sciences. Depending on your point of view, this position may strike you as highly controversial or as plain common sense. A large number of researchers in our fi eld react to positivist thought in one or the other of these two ways. Consequently, it is likely that positivism will continue to occupy a central place in our fi eld, both in providing a foundation for research and in stimulating the articulation of alternative methodological positions.

Th e term ‘positivism’ was invented by the French philosopher Auguste Comte (1798–1857) to describe what he saw as the last of three phases in the development of society and its search for truth. It was Comte’s view that society had passed through a theological stage and then a metaphysical stage; and that now it had entered into a fi nal ‘positive’ stage in which the search for truth is characterized by the systematic collection of observed facts. Th e term ‘sociology’, which refers to the scientifi c study of the social world, was also his invention. Both terms expressed the same belief: that the social world could be explained using similar methods to those used to explain natural phenomena.

Th is view of social science methodology, in common with the other approaches to be discussed in this chapter, commits us to a number of ontological and epistemological claims. Th e nature and implications of these claims and their relationship to a positivist methodol-ogy will become clear as we identify and discuss the basic tenets of positivism.

We begin discussion of these tenets by fi rst considering the classical positivist tradition, and then focusing on the development of positivist thought through the movement of ‘logical posi-tivism’ and Karl Popper’s critique of logical positivist tenets. In discussing these developments, our purpose is not to provide an intellectual history of positivism: the ideas of classical positiv-ism were not superseded by those advanced by logical positivists; nor were those associated with logical positivism supplanted or displaced by the ideas of Karl Popper. In other words, the devel-opment of positivism over time did not always or usually lead to the wholesale rejection of previous ideas, but rather to an expansion of the array of positions associated with it.

Classical positivism

Th e fi rst tenet of positivism—one implied by our previous discussion—is naturalism. Natu-ralism is the idea that there are no fundamental diff erences between the natural and the social sciences. Note that this idea entails an ontological presupposition about the social world: if


there is no diff erence between the social and natural sciences, it must be because there is no fundamental diff erence between the social and natural worlds. Both claims provide positivism with a basis for building a larger edifi ce of thought concerning the nature and goals of social-scientifi c inquiry. As we shall see, positivism maintains that, since the social sciences are no diff erent from the natural sciences, they should have the same structure and logical character-istics as the natural sciences. We’ll return to this notion in a moment when we discuss the third tenet of positivism. But fi rst let’s consider a second tenet of positivism: empiricism.

Empiricism is a philosophical theory of knowledge which claims that what we know of the world is limited to what can be observed. Knowledge is only that which originates in sensory experience: there is no a priori knowledge, no knowledge of reality that is acquired prior to sense experience. So, an empiricist epistemology commits positivism to the view that social reality can only be known through what is observed and that knowledge of the social world is therefore limited to phenomena that can be observed by the senses. Positivists maintain that social science should be empirical, based on evidence that is visible in the world. Its goal should be to gain knowledge of social reality through concepts which apply to or derive from what is observable and measurable.

Additional tenets of positivism provide further elaboration of its position concerning the basis of knowledge and the form it takes. Consider a third tenet of positivism: that the goal of social science is to explain and predict social phenomena by means of laws. Th e German logician Carl Gustav Hempel (1905–1997) argued that if the discovery of laws is necessary in the physical or natural sciences, then laws must be necessary also in social science. If the social world is like the natural world, then, like the natural world, it also must be regular, systematic, and law-governed. Th ere are regularities in, and ultimately laws of, social and political processes; and we can explain social events and phenomena by means of law-like generalizations that have the same status as natural scientifi c laws.

Th e possibility of discovering laws in the social sciences is one of the key issues on which positivism and its critics divide. As we shall see, there is considerable debate concerning whether social laws exist. Some non-positivist approaches insist that there is a diff erence ‘“in kind” between the subject matter of natural and of social science, which precludes the use of laws in the explanation of human behavior and makes it impossible to establish social laws’ (McIntyre 1994: 131). We will be considering this view later in the chapter.

We have said that positivism holds that the social world is regular, systematic, and law- governed, like the natural world; that social phenomena can be explained and predicted by means of laws that have the same status as natural scientifi c laws; and that the purpose of social science, therefore, is to discover these laws. But how do we go about discovering laws of social life? Classical positivist thought maintains that laws can be discovered through systematic investigation of observable events and happenings, and by means of inductive reasoning. Induction is a means of reasoning that begins with specifi c observations and measures. It moves to an identifi cation of patterns and regularities and to the formulation of some tentative hypotheses that can be explored; and it ends by developing some general conclusions or theories. An inductive approach to social inquiry is, as we noted earlier, one in which ‘knowledge is arrived at through the gathering of facts that provide the basis for laws’ (Bryman 2004: 11). We will have more to say about induction and other means of rea-soning (i.e. ‘deduction’ and ‘abduction’) further on in our discussion of the development of positivist thought.


We are still discussing the third tenet of positivism: the view that explanation of social phenomena should proceed by the discovery of laws. But, for positivism, there is another key element in social science explanation: explanation must not only proceed with reference to law-like generalizations, it must also establish a cause–eff ect relationship between events in the world. Positivism sees the social world as comprising phenomena that are causally related to each other; consequently, to explain a social outcome we are required to show the factors or conditions that combined to bring it about or caused it to be more likely to occur in the circumstances.

Virtually all social research is concerned to discover causes. But there are diff erent con-ceptions of causation. Th e positivist conception of causation is an empiricist conception which was introduced by the Scottish philosopher, economist, and historian, David Hume (1711–1776). Most of us probably carry in our minds an idea of causation as a relation between two events, the cause and the eff ect, which expresses some type of ‘necessary con-nection’ between them. But Hume pointed out that we cannot directly perceive causal rela-tionships. He points out that ‘when we look about us towards external objects, and consider the operation of causes, we cannot in any instance discover a power, necessary connexion, or quality which binds the eff ect to the cause and renders one an infallible consequence of the other’ (1966: 51). Instead, we observe only the ‘constant conjunction’ of events; we observe only that one thing follows the other. Our experience of observing this ‘constant conjunction’ between events conveys to our minds a necessary relation between these events. So the causal conclusions we reach are based, not on ‘knowledge of causal mechanisms and the generative properties of things’, but only ‘on the observation of how a certain event is followed again and again by a certain other event’ (Ekström 1992: 108).

According to this conception, then, causation is constituted by facts about empirical regularities among observable variables. Th ere is no underlying power or necessity deriving from the laws of nature. All we can do is observe that one thing follows another with regular-ity; and, because of this observation, we develop a psychological expectation that Y will occur whenever X does. But we cannot know that X is the cause of Y by observing that X is constantly followed by Y. Consequently, in establishing the basis of causal explanations, positivists are concerned with observing empirical regularities rather than in discovering causal mechanisms. Th is is a subjective conception of causation: causation as a perceived regular association among variables. An objective conception of causality, one involving causal necessity or causal mechanisms is, according to positivism, metaphysical. Th is objec-tive conception of causation features prominently in the critique of classical positivism artic-ulated by logical positivism, a subject to which we will turn next. But before moving on, we need to briefl y note a fourth tenet of positivism: that it is possible to make a distinction between facts and values.

Positivism maintains that we can gain knowledge of the social world through applica-tion of the scientifi c methods used in the natural sciences. According to this fourth tenet of positivism, the pursuit of knowledge through these methods can be value-free or objective, because statements of fact (confi rmed by the senses) can be distinguished from normative statements. Science is concerned with the discovery of facts, whereas values relate to ethics or policy studies. Th e argument that it is possible to distinguish between facts and values, and to treat ‘facts’ as independent of the observer and of his or her val-ues, represents a key diff erence between positivists and adherents of alternative approaches. However, we will leave discussion of this issue for the time being, since we


will be exploring it in some detail in Chapter 3. Instead, we turn to a consideration of the further development of positivist thought as a result of logical positivism and Karl Pop-per’s critique of it.

Empiricism and logic as the basis of truth claims

Th e ideas of classical positivism were developed by a movement that adopted the name ‘logical positivism’, as well as by the highly infl uential critique of Karl Popper.

Logical positivism began in the early twentieth century as a movement within philos-ophy. Inspired by developments in twentieth-century logic and mathematics, its goal was to introduce logical reasoning and mathematics as sources of knowledge in addition to empir-icism. It advanced the idea that social inquiry should combine induction (based on empiri-cism) and deduction (in the form of logic) as methods of reasoning.

We have previously discussed induction as a means of discovering laws. Induction, you will recall, is a process of reasoning from particular facts to a general conclusion. As Figure 2.1 shows, in induction we begin with particular observations or cases and then develop generalizations about them. Deduction works the other way around. As Figure 2.2 shows, deduction moves from broader generalizations and theories to specifi c observations. We start, not with an observation, but either with a theory that has already been confi rmed or with a logical argument, and then we draw out the meaning or implications this has for explaining some particular case or phenomena.

To digress from our discussion of logical positivism for a moment, it should be noted that, in practice, researchers do not use solely one method or the other. Scientifi c inquiry typically involves a process of continuous interaction between theory and observation, in which the researcher moves from observation to theory (induction) and from theory back to observa-

Theory

Tentativehypothesis

Pattern

Observation

Figure 2.1 Induction

Source: Trochim 2006.

BOX 2.2 Induction, Deduction, Retroduction

Induction observation ® theory

Deduction theory ® observation

Retroduction observation «␣theory


tion (deduction). Box 2.2 illustrates how this process contrasts with and combines induction and deduction. Th e compiling of evidence (induction) leads the researcher to theory (deduc-tion); and once a hypothesis is formed, the researcher brings it ‘backward’ for readjustment or redefi nition. Th e term ‘retroduction’ describes this interaction of induction and deduc-tion in an evolving, dynamic process of discovery and hypothesis formation.

We have said that logical positivism introduced the idea that social inquiry should combine both induction and deduction. It also established ‘verifi cation’ (of statements or propositions) as the goal of social science research. Verifi cation was held to be the main criterion for establishing truth claims and a means of defi ning a clear line of division between science and metaphysics.

Both of these tenets of logical positivism became the target of a critique by Karl Popper (1902–94), a philosopher of science who also wrote extensively on social and political philosophy. Popper’s critique had a decisive impact on social-scientifi c thought. In fact, its infl uence was so great that logical positivism’s most important contribution to social science, it might be argued, is the role it played in having served as the focus of this critique. Th is does not diminish its con-tribution: in the quest to establish a sound basis for scientifi c inquiry, logical positivism raised important questions about the concepts and practices of science which continue to have rele-vance for social-scientifi c inquiry today. Moreover, while Popper was a critic of logical positiv-ism, there are also many affi nities between his views and those held by logical positivists.

Logical positivists had argued that both inductive and deductive methods of reasoning should be used to acquire knowledge of social phenomena. But Popper argued that induc-tion must be rejected entirely. Moreover, the argument he advanced for rejecting induction also provided grounds for rejecting verifi ability as a basis for establishing truth claims.

Popper elaborates these arguments in his book, Logik der Forschung, published in 1934, and later published in English under the title Th e Logic of Scientifi c Discovery (1959). Th e book addresses two issues. Th e fi rst is what David Hume calls ‘the problem of induction’. Th e problem is whether experience can provide the basis for gaining general theoretical knowl-edge. Since experience is particular, while knowledge is general or even universal, how do we achieve universal knowledge on the basis of particular experience? How can we reach gen-eral statements of scientifi c law on the basis of experiences that are necessarily limited and particular? Popper argues that no matter how many experiences we have of observing some-thing, this does not permit the deduction of a general statement of scientifi c knowledge.

Th e reasoning that leads him to this conclusion begins with David Hume’s argument about the limits of inductive reasoning. Hume argued that since we cannot observe the uni-verse at all times and in all places, but are only able to observe particulars, we are not justifi ed in deducing general laws based on inductive evidence. Popper’s now famous story of the

Hypothesis

Theory

Observation

Confirmation

Figure 2.2 Deduction

Source: Trochim 2006.


black swan illustrates what happens when we attempt to formulate laws based on observa-tion. Th e story is that, once upon a time, Europeans thought that all swans were white because, having found nothing but white swans for thousands of years, Europeans concluded on the basis of their experience that all swans were white. But one day Europeans went to New Zealand (as Popper had), and there they found black swans. What this story tells us is that no matter how many observations confi rm a theory, it only takes one counter-observation to falsify it: only one black swan is needed to repudiate the theory that all swans are white. And since it only takes a single unforeseen or seemingly improbable event to invalidate a gener-alization based on empirical observation, then empirical observation alone cannot generate ‘laws’. Popper therefore concludes that, rather than endeavouring to discover laws through induction, what scientists should be doing is testing theory deductively.

Popper’s critique of induction leads him to reject another tenet of logical positivism: the notion that scientists should seek to verify hypotheses. Popper argues that, since a single exception to the rule destroys inductively generated theory, then conclusive verifi cation of a hypothesis is not possible. So Popper proposes that we reverse the logical positivist assump-tion about verifi ability: he argues that rather than continually attempting to prove a theory, scientists should attempt to disprove it. Since we cannot verify a hypothesis , our aim should be to falsify it. We should formulate propositions in such a way as to enable them to be refuted. By doing this, it will be possible for us to show a theory to be wrong; and we can then introduce a new theory which better explains the phenomena. Th is, Popper argues, is how we achieve progress in science.

Th e notion of falsifi ability is the basis of Popper’s argument concerning the second issue he addresses in Th e Logic of Scientifi c Discovery: the problem of demarcation. Th is refers to the problem of determining how to diff erentiate science from non-science. It is a key problem in the philosophy of science and the subject of ongoing debate. For Popper, it is falsifi ability—and not verifi ability, as logical positivists argued—that defi nes the boundary between science and pseudo-science or metaphysics. Anything non-falsifi able is outside science.

Consider religions and ideologies in this regard. Religions and ideologies are logically consistent statements which provide a guide for understanding the world. But they cannot be proved false: potentially disconfi rming or anomalous facts do not prove them false, but are incorporated within them. A scientifi c theory, however, must state what evidence would disconfi rm it or prove it to be false. If you cannot think of anything that might disconfi rm a theory, then it is not a theory at all but a set of self-verifying statements—an ideology.

To sum up: in rejecting induction, Popper was rejecting the idea that observation pro-vides the basis for the formulation of scientifi c theories. Th eories cannot be derived from observation (induction), because at any time a single observation can disconfi rm the theory. Popper concludes that social inquiry must proceed deductively, through a process in which observations are not the basis of theories, but are derived from and used to ‘test’, or falsify, them. According to Popper’s notion of falsifi ability, we endeavour to falsify hypotheses. We reject those which are falsifi ed and we continue to test those that are not until they become so thoroughly tested that we can consider them to be ‘confi rmed’, though it remains possible that some day someone may falsify or signifi cantly modify them.

Two objections have been made to this formulation. Th e fi rst objection is to the distinc-tion which Popper seems to make between facts and theories. Popper seems to assume that the observations or facts that we pursue as a means of testing theories can be established


independently of the theory that they are meant to test. We consider the debate concerning this issue in some detail in Chapter 3. A second objection is that Popper’s notion of falsifi a-bilty is at odds with how scientists actually go about developing and testing theories. Do researchers seek to disprove or falsify their own theories? Do they discard their theories when they are confronted with disconfi rming evidence? We will consider Th omas Kuhn’s and Imre Lakatos’ answers to these questions, and the further evolution of Popper’s falsifi ca-tionist position that developed as a response to them, in Chapter 3.

Here, however, we have still to consider a further question: how do we use deductive reason-ing to discover laws of social life as a basis for explanation? Previously we have discussed the classical positivist approach to explanation: inductive reasoning based on systematic investiga-tion of observable events and happenings. As we have seen, logical positivists maintain that both induction, based on empiricism, and deduction in the form of logic could be used to discover laws. Popper argues that we can establish laws of social life as a basis for explanation only through deduction. But, what is the process through which deduction operates as a means of explaining social phenomena? For the answer to this question, we turn, again, to Carl Gustav Hempel.

Hempel maintains that explanation in the social and natural sciences is the same, not only because both involve the search for and discovery of law-like generalizations, but because the social and natural worlds are subject to laws in the same way (see Hempel 1994). Th e logic and function of laws, what Hempel calls ‘general laws’, are the same. In both the natural and social sciences, individual events can be subsumed within hypotheses about general laws of nature: what this means is that to explain some fact is to cite some law or laws and other conditions from which the fact can be deduced.

Hempel formalizes this defi nition of explanation in his deductive-nomological model.A deductive-nomological explanation is deductive because the phenomenon to be

explained (explanandum) is logically deducible from that which does the explaining (the explanans); and it is nomological because the explanans includes at least one law (‘nomos’ is the Greek word for law). According to this model, then, something is explained when it is shown to be a member of a more general class of things, when it is deduced from a general law or set of laws. A full explanation of an event requires that we give an account of how a phe-nomenon follows deductively from a well-confi rmed general law. For instance, ‘To explain fully why an actor votes (a “fact”) we must do more than just isolate the particular cause of this particular event (for example, the intensity of the voter’s concern about unemployment). We must subsume this act of participation under a “law” that explains why, under certain condi-tions, the voter had to vote: “persons with intense preferences for candidates or issues”, every-thing else being equal, will become “active in politics”’ (Milbrath 1965: 53; quoted in Schwartz 1984: 1123). Given the general law, the particular case in question was to be expected.

But, how do we confi rm a regularity or generalization that what we take to be a ‘law’ is, in fact a law? A regularity might be true, accurate, or supported by evidence; but it might be only ‘accidentally true’: true only as a result of circumstance or coincidence. Explaining how to distinguish law-like generalizations from those that are ‘accidental’ is one of the central problems in the philosophy of science. However, in general, we can say that a law expresses a necessary connection between properties, while an accidental generalization does not. If a necessary connection exists between its properties, then we should be able to test a law by its ability to predict events. If we predict something on the basis of a law and fi nd that the prediction was true, then the law can be said to be confi rmed.


Th is is what Carl Hempel proposes that we do in his hypothetico-deductive model of confi rmation. We confi rm that the generalization is a law (rather than an accidental gener-alization) by treating it as a hypothesis and then we test the hypothesis by deducing from the hypothesis a suffi cient number of explicit predictions of further phenomena that should be observable as a consequence of the hypothesis. Observations that run contrary to those predicted are taken as a conclusive falsifi cation of the hypothesis; observations which are in agreement with those predicted are taken as corroborating the hypothesis. It is then suppos-edly possible to compare the explanatory value of competing hypotheses by looking to see how well they are sustained by their predictions.

An example of what is regarded as a law or, at least, a law-like generalization in our fi eld is Duverger’s Law. Th e sociologist, Maurice Duverger, proposed that the plurality rule for selecting the winner of elections favours the two-party system. Duverger off ers two theoretical explanations for why a plurality rule election system tends to favour a two-party system. Th e fi rst is the ‘mechanical eff ect’ of under-representing losing parties; and the sec ond is a ‘psychological factor’: voters don’t want to waste their votes on losers (Riker 1982: 761). William Riker explains: ‘when the defi nition of winning forces candi-dates to maximize votes in order to win (as in plurality systems), they have strong motives to create a two-party system; but when the defi nition of winning does not require them to maximize votes (as in runoff and proportional systems), then this motive for two par-ties is absent’ (Riker 1982: 755).2

To sum up: the deductive-nomological model holds that an observed phenomenon is explained if it can be deduced from a law-like generalization. Th e hypothetico-deductive model confi rms that a generalization is a law by treating the generalization as a hypothe-sis, and testing it by its deductive consequences. To explain some fact is to cite a law or laws plus other relevant conditions from which the explanandum may be deduced (the deductive-nomological model of explanation). To confi rm a hypothesis is to deduce some observed phenomenon from the hypothesis plus other relevant known conditions (the hypothetico-deductive model of confi rmation).

We have traced the development of positivist thought through a consideration of the basic tenets of classical and logical positivism and the arguments advanced by Karl Popper. We turn now to approaches that emerged as a challenge to positivist thought and research.

Challenges to positivist approaches within the social sciences

Th ere are a number of approaches to social inquiry that challenge the positivist position and that articulate a fundamentally diff erent basis for inquiry. Here, we focus on two alternative positions—those represented by scientifi c realism and interpretivism.

Scientifi c realism

Scientifi c realism is concerned to elaborate a non-positivist version of science, one that its adherents claim is more scientifi c than positivism. Th eir message, as Ruth Lane puts it (Lane 1996: 373), is that ‘we don’t have to be positivists to be scientifi c!’


Scientifi c realism appears to be similar to positivism in some ways because it accepts a number of assumptions of positivism that other non-positivist approaches reject. For instance, scientifi c realism assumes, like positivism, that the social and natural worlds are essentially similar, and that the social and natural sciences are therefore fundamentally sim-ilar, as well. Th ese assumptions are based on another shared assumption: realism. Realism holds that the world exists independently of our knowledge of it, that reality has an inde-pendent existence (it exists independently of human beings and their perceptions), and that it impacts directly upon the human mind without any refl ection on the part of the human knower. We can therefore gain objective knowledge of the world because our knowledge of it is directly determined by an objective reality within the world.

So, positivism and scientifi c realism share some key assumptions. However, there is a key diff erence between the two approaches—and it is an important one! Let’s recap for a moment before stating this diff erence. Both approaches maintain that the subject matter of scientifi c research and scientifi c theory exists independently of our knowledge of it, that we can there-fore gain objective knowledge of it, and can treat ‘facts’ as independent of the observer and of his or her values. Now, where the two approaches diff er is that, while positivists maintain that reality consists of only that which we can directly observe, for scientifi c realists, reality consists of observable elements as well as observable ones.

You will recall that positivists assume that statements not based on observable data are metaphysical. Scientifi c realists break decisively with this assumption. Th ey assume that there are knowable, mind-independent facts, objects, or properties that cannot be directly observed but which are, nonetheless, real. Th ey argue that unobservable elements of social life, such as structural relations between social phenomena, are crucial to an understanding and explanation of what goes on in the world. Th ey point out that the central role of unob-servable elements in shaping outcomes is one of the features that makes the social world similar to the natural world; that this ontological conception of the social world is not met-aphysical, but more scientifi c, and more closely aligned with the tenets of the natural sci-ences, than the positivist conception.

Consequently, for scientifi c realists, the goal of scientifi c research is to describe and explain both observable and unobservable aspects of the world. But how do we know these unobservable elements exist? According to scientifi c realism, we know they exist because we can observe their consequences: unobservable elements of social life can be treated as ‘real’ if they produce observable eff ects. To posit the existence of unobservable entities to explain observable outcomes is consistent with well-established scientifi c prac-tice. We treat gravity and subatomic particles as real because, even though we cannot see them, we can see their eff ects. Similarly, there are many elements in social and political life that are not directly observable—social structures, capitalism, society—but they have observable eff ects; and because their eff ects are observable, researchers in our fi eld treat them as real.

Given these assumptions, it follows that, for scientifi c realists, scientifi c knowledge does not take the form solely of empirical regularities, and scientifi c research cannot be solely concerned with the goal of formulating law-like generalizations based on observations. To state this diff erently, if scientifi c realists reject the notion that only entities of which we have direct sensory experience are ‘real’, then they cannot depend on an epistemology that places emphasis on direct observation for pursuing knowledge of the social world.


It follows that scientifi c realists also cannot accept the empiricist (Humean) conception of causality that positivists employ. Recall that positivists treat causation as constituted by facts about empirical regularities among observable variables, and seek to establish causal rela-tionships by observing these regularities rather than by discovering causal mechanisms. Th ey treat the notion that causal mechanisms produce outcomes in social life as metaphysi-cal, since we are unable to have knowledge of causal mechanisms through direct observation only. But scientifi c realists assume that unobservable elements are part of reality and are knowable, and so they treat causal mechanisms and causal powers as ‘real’, as a legitimate object of scientifi c investigation, and as fundamental to explanations of social outcomes. For scientifi c realists, explaining social outcomes entails providing an account of the causal mechanism that brought about a given outcome; and with developing empirically justifi ed theories and hypotheses about causal mechanisms.

A causal mechanism can be defi ned as ‘a series of events governed by lawlike regularities that lead from the explanans to the explanandum’ (Little 1991: 15); or ‘the pathway or proc-ess by which an eff ect is produced or a purpose is accomplished’ (Gerring 2007: 178). Charles Tilly identifi es three sorts of mechanism. Environmental mechanisms are ‘externally gener-ated infl uences on conditions aff ecting social life’; cognitive mechanisms ‘operate through alterations of individual and collective perception’; and relational mechanisms ‘alter connec-tions among people, groups, and interpersonal networks’ (2001: 24). Michael Ross defi nes an environmental mechanism—the nature of a government’s resource base—to explain the apparent link between oil exports and authoritarian rule. He calls this mechanism a ‘rentier eff ect’: ‘when governments derive suffi cient revenues from the sale of oil, they are likely to tax their populations less heavily or not at all, and the public in turn will be less likely to demand accountability from—and representation in—their government’ (2001: 332). Explanations of a variety of political outcomes might link them to the eff ect of increases or decreases in the government’s resource base (see e.g. Chapter 6). Cognitive mechanisms have been identifi ed to explain ethnic confl ict as, for instance, changing conceptions of racial, ethnic, gender, religious, or class diff erences (e.g. Hoff mann 2006). Relational mechan isms, such as governmental absorption and destruction of previously autonomous patron–client networks, or bureaucratic containment of previously autonomous military forces, have been held to eff ect ‘the likelihood of civil war, the level of domestic violence, and even the prospect that a given state will engage in international war’ (Tilly 2001: 38). Robert Gilpin (1981) has argued that there is a tendency for a disjuncture to arise between the costs and benefi ts of hegemony, and that, when it does, the hegemonic state begins to decline. Gilpin identifi es a number of mechanisms that cause this disjuncture. One is the ‘law of the increasing costs of war’: military techniques tend to rise in cost, and the increasing cost of war produces a fi scal crisis within the hegemonic state. Another is the ‘law of expanding state expenditures’: private and public consumption grows faster than the GNP as a society becomes more affl uent (the rich indulge increasingly in lavish consumption; the poor begin to clamour for welfare).

Th ose who emphasize the importance of mechanisms in causation have diff erent views about the nature or types of social mechanisms that operate to produce social outcomes. Th ese are linked to diff erent assumptions about what we should treat as the basic unit of analysis in social inquiry. So, for instance, those who treat individuals as the basic unit of social analysis favour agent-based models, or individual-level mechanisms to explain


outcomes. Th is is characteristic of rational choice approaches, which assume that the instrumental rationality of individuals is the causal mechanism that produces social out-comes. Structural models, on the other hand, attempt to demonstrate that there are struc-tural or institutional mechanisms that cause social outcomes. In Chapter 4 we will be exploring the diff erences between the individualist and collectivist (or holist) ontologies on which these diff erent models are based. We will also discuss how what Peter Hedström and Richard Swedberg (1998) call ‘social mechanisms’ produce outcomes through macro–micro interactions and linkages.

We have been discussing the assumptions of scientifi c realism with regard to basic ques-tions concerning the nature of the social world, forms of knowledge, and the goals of social science. We have said that, for scientifi c realists, the goal of scientifi c research is to describe and explain both observable and unobservable aspects of the world. It still remains to say how scientifi c realists establish that claims regarding unobservable social phenomena are true.

Scientifi c realists argue that knowledge of unobservable elements of social life can be obtained through the development of theoretical constructs. But how can we know whether our theories about unobservable elements of social life are true? Th e answer is that we can accept as true the theory or hypothesis which, from among those that have been advanced to explain a phenomenon, off ers the best explanation. Th e ‘best’ explanation or hypothesis is the one that, based on various ‘rules of method’, explains a fact better than other available hypotheses. For instance, it may be ‘best’ because it has been tested and not refuted, while the others have not; because it accounts for more, or better meets the standard of explanation we accept for other phenomena. ‘If a theory is certifi ed by such rules of method, a scientist is rationally justifi ed in accepting the theory’ (Sankey 2008: 28). Scientifi c realism maintains that we can accept that a theory is true if there is rational justifi cation for accepting it to be true; and it is rational to accept as true the best available explanation of any fact.

Th is position is summed up in the phrase ‘Inference to the Best Explanation’. Scientifi c realists maintain that inference from some data to the ‘best explanation’ justifi es our accept-ance of a hypothesis as true. By inference we mean the reasoning involved in the process of drawing conclusions based on facts or logical premises; and, according to scientifi c realists, the kind of inference that justifi es our accepting a hypothesis as true emerges from a type of reasoning called ‘abduction’. Contemporary philosophers use ‘inference to the best explana-tion’ and ‘abduction’ interchangeably.

Abductive reasoning is prior to and distinct from induction and deduction. Abduction starts with a hunch that a set of seemingly unrelated facts are connected in some way. Th e hunch or hypothesis can then be affi rmed by induction or deduction. Abduction may be used to explain singular events rather than, as in inductive reasoning, to form generaliza-tions on the basis of a large number of token instances; and, unlike induction, it can employ both observables and unobservables to explain events. Abductive reasoning requires that we choose from among competing explanations the best available explanation: the one that best explains a particular event or phenomenon given all the available evidence. Th e abduction is provisional: new evidence may later undermine it. But it is reasonable for us to believe it if it is the best explanation we have. We may later fi nd out that the explanation is wrong and then it will no longer be reasonable for us to believe it; but it remains the case that it was not wrong or unreasonable for us to have believed it prior to our fi nding out that it was wrong. When a


detective infers that a murder was committed by a suspect, he does so because this hypoth-esis provides the best explanation for the murder; and it is the ‘best’ explanation because it fi ts better with the forensic evidence, and/or provides a better account of motive and oppor-tunity. Later information may reveal this explanation to be false; but this does not make it unreasonable for the detective to have made the original inference.

We now have introduced some of the key terms of reference in a continuing debate about the scientifi c status of unobservable elements of social life. Because political researchers continually refer to unobservables such as ‘society’ and ‘structures of power’ to explain polit-ical events and processes, we will be returning to this debate in later chapters. But, before moving on from discussion of scientifi c realism to a consideration of other non-positivist approaches, it is worth noting a related position that has emerged within the fi eld of politics: critical realism.

We have said that scientifi c realism, like positivism, assumes that there exists a reality separate from our description of it. Critical realism represents a move away from this posi-tion. As we have seen, scientifi c realism is committed to identifying the unobservable struc-tures that work to generate observable outcomes. Th is, as Roy Bhaskar points out, is critical in that it opens up the possibility or our being able to change our world (1998: 2). But critical realism also rejects the view, accepted by scientifi c realists and associated with what adher-ents of critical realism call ‘naïve realism’, that the external world is as it is perceived. Instead, it holds that perception is a function of the human mind, and that we can therefore only acquire knowledge of the external world by critically refl ecting on perception. While some political researchers see the terms ‘scientifi c realism’ and ‘critical realism’ as synonymous (see e.g. Brown 2007: 409), this would seem to be a position that moves us further in the direction of the interpretivist approaches that we will be discussing next. Some examples of how a critical realist position informs political research are in Chapter 13, where we discuss critical discourse analysis, a type of textual analysis inspired by, and to a large degree consist-ent with, a critical realist philosophy of science.

We have been discussing scientifi c realism, an approach to social inquiry that provides an alternative to positivism. As we have seen, its main diff erence with positivism is that it does not place emphasis on direct observation in pursuing knowledge of the world; rather, it assumes that reality consists of both observable and unobservable elements. We have also seen that there are some respects in which scientifi c realism and positivism are more similar than dissimilar. For instance, both agree that the world exists independently of our knowledge of it. Th is assumption has important implications for what we treat as legitimate knowledge and how we conduct research. We turn now to a set of approaches that break decisively with this assumption, and promote ontological and epistemological positions that stand in diametric opposition to those of positivism.

Interpretivism

Interpretivism maintains that the social world is fundamentally diff erent from the world of natural phenomena, and so we cannot understand it by employing the methods used to explain the natural world. It argues that it is impossible for us to gain knowledge of the social world by searching for objective regularities of behaviour that can be summed up in social-scientifi c laws analogous to the laws of physics, because the social world does not exist


independently of our interpretation of it. Th e social world is what we experience it to be: it is subjectively created. Th e task of social science, then, is fundamentally diff erent from that of natural science, because the objects of the social sciences are diff erent from those found in the natural world. Social phenomena are socially or discursively constructed; so we cannot explain and predict social phenomena by means of laws. Th e primary goal of social science must be to achieve an understanding of human behaviour through an interpretation of the meanings, beliefs, and ideas that give people reasons for acting.

Let’s consider the implications of this view for how we conduct political research.Recall our earlier discussion about behaviouralism. Behaviouralist research is positivist

and empiricist. Its concern is with the question of what political actors do and why they do it. It seeks to discover the causes of behavioural outcomes by understanding the motivations of political actors. It uses public-opinion polls and social surveys to learn about the beliefs, attitudes, and values that motivate behaviour; or rational choice theory to explain how indi-vidual self-interested rational action motivates behaviour. However if, as interpretivists contend, people act on the basis of the meanings they attach to their own and to others’ actions, then understanding human behaviour requires an understanding of these mean-ings. Consequently, social science must be concerned, not with discovering causes of social outcomes, but with piecing together an interpretation of the meanings of a social outcome or production.

Intepretivists seek to understand human behaviour through interpretation and inter-pretive theory. Th ese are forms of social science ‘that emphasize understanding the meaning that social behaviour has for actors’ (Gibbons 2006: 563). Th ese forms include a multiplicity of approaches, most notably hermeneutics, cultural anthropology, verstehen social theory, critical theory, and post-structuralism. In what follows, we will focus on hermeneutics as a means of highlighting the diff erences between interpretivist and posi-tivist approaches. We will then consider how approaches based on interpretivist and positivist assumptions analyse a specifi c area of political inquiry.

‘Hermeneutics’ originally referred to a method used to interpret theological and legal texts. In fact, ‘the literal English translation of the German word “hermeneutics” is interpre-tation’ (Gibbons 2006: 563). Today, ‘hermeneutics’ refers to theories and methods that are used in the interpretation of texts of all kinds. Th ese texts include not just written docu-ments, but any object or practice that can be treated as a text and which can, therefore, be the subject of interpretation. But can human beings and their actions be treated as a text and the subject of hermeneutical interpretation? Interpretivists argue that they can. Hermeneutics can be used to study behavioural outcomes because, if behaviour is a product of the mean-ings and intentions employed by social actors, then the social scientist endeavouring to understand that behaviour is involved in an interpretive exercise not unlike that engaged in by the translator of a text.

Th e philosopher Charles Taylor elaborates this argument in an infl uential essay entitled ‘Interpretation and the Sciences of Man’ (1994). Taylor explains that any fi eld of study can be the object of hermeneutics if it meets two requirements. First, it must contain an object or a fi eld of objects that is a text, or a ‘text-analogue’. Second, this text must be ‘confused, incomplete, cloudy’ or ‘seemingly contradictory’; that is, it must be in some way ‘unclear’. When these criteria are met, hermeneutical interpretation can be used ‘to bring to light an underlying coherence or sense’ with respect to the objects defi ned by the fi eld of study and,


in this way, enable us to understand them (Taylor 1994: 181). Does the study of politics meet these criteria? Yes. We can treat the behaviour we are concerned to understand—the actions of a government, or the behaviour of members of a group towards one another—as ‘texts’; and, since the motives and goals of this behaviour are oft en unclear or at odds with the pro-nouncements of the political actors involved, we can use interpretative methods in order to make sense of this behaviour.

Interpretivists argue that it is necessary not only to employ a hermeneutical approach, but also to reject empiricist scientifi c methods for studying human behaviour. Empiricist methods treat social reality as consisting only of what Charles Taylor calls ‘brute data’. By ‘brute data’, Taylor means ‘data whose validity cannot be questioned by off ering another interpretation or reading, data whose credibility cannot be founded or undermined by fur-ther reasoning’ (Taylor 1994: 184). Th ese data capture political behaviour involving actions that have an identifi able physical end state. When actors raise their hands at a meeting at the appropriate time we can give this action a ‘brute data’ description and say that the actors are ‘voting for the motion’. However, the action may have meanings for the actors that are not captured by the ‘brute data description’ of it. It may be the case that when an actor votes for a motion, she is also expressing loyalty to her party or defending the value of free speech (Taylor 1994: 190). But a ‘behavioural’ (or positive, or empiricist) political science deals only with brute data and their logical consequences, and avoids addressing the meaning of political behaviour.

As Taylor points out, brute data captures more than behaviour that has an identifi able end state: it also captures the subjective reality of individuals’ beliefs, attitudes, and values ‘as attested by their responses to certain forms of words, or in some cases, their overt non-verbal behaviour’ (1994: 198–9). But while these data capture subjective meanings, there are non-subjective (intersubjective and common) meanings constitutive of social reality that they cannot capture such as, for instance, inter-subjective meanings and common meanings. Inter-subjective meanings are meanings that do not exist only in the minds of agents but are rooted in and constitutive of social relations and practices such as paying taxes and voting. Common meanings involve recognition or consciousness of shared beliefs, aspirations, goals, values, and a common reference point for public life of a society. Common meanings are the basis of community, in that they are expressed by collective aspirations, actions, and feelings (1994: 197). Taylor argues that we need to study these non-subjective (inter subjective and common) meanings in order to comprehend politi-cal issues such as social cohesion, stability, disorder, and legitimacy. Moreover, they are crucial for ‘a science of comparative politics’: without them, we ‘interpret all other socie-ties in the categories of our own’ (1994: 200), rendering invisible important diff erences among societies and making comparison impossible.

To stay with this point for a moment longer, analyses based on positivist epistemological assumptions, like those off ered by rational choice theory, depend on an abstract description of human agency, one that pays little attention to diff erences across social, cultural, and his-torical settings. Rational choice theory seeks to explain social phenomena as the outcome of purposive rationality, and of material and structural factors exercising causal infl uence on individuals. Its concern is to show how a given outcome is the result of purposive choices by individuals within a given set of material and structural circumstances. As Daniel Little puts it: ‘Agents like these in structures like those, produce outcomes like these’ (Little 2009). But


interpretive approaches see individuals as unique, and human activities, actions, and social formations as unique historical expressions of human meaning and intention. Consequently, they are concerned, not with abstract descriptions of human agents, but with detailed interpretive work on specifi c cultures.

Th ese diff erences between positivist and interpretive approaches can be illustrated by reference to a key area of research in our fi eld: political participation. Positivist studies typi-cally equate political participation with voting. However, interpretivists would argue that a particular voter may ‘not understand voting as participation at all, in contrast, say, to party activism. (It is neither practically nor logically impossible that an actor could say, “No, I do not participate in politics, but I do vote”)’ (Schwartz 1984: 1118). But positivists tend to treat participatory acts as ‘brute facts’, as having an ‘objective’ ontological status: as existing ‘in some sense “in the world” separate from the theoretical stance of the observer or of the par-ticipant’ (Schwartz 1984: 1119).

Joel Schwartz points out that there is no objective point of view from which to describe and understand participation, that participation ‘is a “subjective” phenomenon much like “justice” and “virtue”’ (Schwartz 1984: 1119). Consequently, ‘any successful attempt to describe and explain participatory acts must begin, not by imposing the observer’s theoreti-cal framework onto the data, but rather with a sensitivity to the frameworks of the partici-pants themselves’ (Schwartz 1984: 1120). While positivist studies typically equate political participation with voter turnout, participation involves a variety of political acts. By impos-ing their own concept of ‘participation’, researchers are prevented ‘from seeing the plural forms that participation in fact takes in the world. Whether acts (of an American voter or demonstrator, a French revolutionary, a Muslim revolutionary, a Solidarity member, and so on) count as acts of participation depends on those actors’ subjective understanding of what they are doing’ (Schwartz 1984: 1117).

In sum, interpretivism maintains that all social action is framed by a meaningful social world. To understand, explain, or predict patterns of human behaviour, we must fi rst under-stand the meanings concrete agents attribute to their environment (social and natural); the values and goals they possess; the choices they perceive; and the way they interpret other individuals’ social action. Social science is, therefore, fundamentally diff erent from natural science, and it is the importance of meaning that distinguishes social science from natural science. Humans act because of what things mean, so an understanding of human behaviour requires that we develop an understanding of meanings and intentions employed by social actors.

Many researchers have pointed to the tendency to cast positivism and interpretivism approaches as ‘two grand traditions in social science epistemology’ and to exaggerate the diff erences between them (Pollins 2007: 93). Positivism and interpretivism have diff erent ontological and epistemological commitments, but they don’t necessarily represent oppos-ing or competing traditions. Researchers working in both traditions generally follow the same methodological conventions, and so can understand what those working within the other tradition are doing. Researchers depend upon diff erent assumptions, and may be interested in and test diff erent questions. But while they may ‘be tackling diff erent kinds of questions’, ‘practical investigation of these questions oft en leads them to similar methodo-logical tasks’ (Finnemore and Sikkink 2001: 395). Ted Hopf argues that there is, in fact, ‘a


certain methodological unity’ between the two traditions. Th e methodological conven-tions they share include the following:

a. clear diff erentiation of premises from conclusions;b. acknowledgement that sampling strategies matter;c. recognition that some standards of validation must be established for the sources of evidence

used;d. diff erentiation of causes from correlations;e. recognition that the spectre of spuriousness haunts all correlations;f. acceptance of deductive logic;g. belief in the need for the contestability of fi ndings (2007: 56).

Th ese shared methodological conventions may, in fact, be seen as refl ecting a common research practice founded in the hypothetico-deductive method. As Brian Pollins puts it: some researchers ‘assess whether the information they have gathered fi ts with the interpreta-tion they have posited’, and others ‘consider the fi t of competing interpretations with the facts they have gathered’, but ‘in either case they are practicing the hypothetico-deductive method’ (Pollins 2007: 100). In fact, according to Dagfi nn Føllesdal, the hermeneutic method that we discussed in our consideration of interpretivism, above, ‘is actually the hypothetico-deductive method’ applied to materials that are ‘meaningful’, i.e. material that expresses an actor’s beliefs and values' (1994: 233). Interpretation-hypotheses can be judged by deducing consequences from them and confronting them with data, such as, for instance, a given text and related works bearing on it.

So interpretivists and positivists do not necessarily use diff erent approaches to gathering relevant evidence. However, they do diff er in their conception of what constitutes explana-tion (recall our discussion, above, about political participation). Th ey also diff er in their understanding of evidence.

The differences between positivist and interpretivist conceptions of both explana-tion and evidence might be described by defining a distinction between ‘external’ and ‘internal’ explanation and evidence. External explanations are associated with positiv-ist research: they tend to work via correlations or deductions on the basis of ascribed reasons, and so need not concern themselves with actors’ understandings of the world. Interpretive explanations, on the other hand, are ‘internal’ in the sense of their being concerned with the world of meanings inhabited by the actor (Hampsher-Monk and Hindmoor 2009: 48). The distinction can be applied to different types of evidence, as well: ‘external evidence’ consists of empirical evidence about the behaviour, and the effects of the behaviour, of particular actors; while ‘internal’ or interpretive evidence consists of evidence about the beliefs of actors whose actions comprise the phenomena to be explained.

To highlight these distinctions, let’s compare the analysis off ered by a specifi c positivist approach (rational choice theory) and a specifi c interpretivist approach (constructivism) with respect to a particular area of political inquiry: the eruption of ethnic confl ict within the former Yugoslavia.


The analysis of ethnic confl ict: a positivist (rational choice) and interpretivist (constructivist) approach

A number of rational choice and constructivist explanations have been off ered for why eth-nic confl icts erupted in Yugoslavia in the 1990s. Both types of explanation have been con-cerned to off er an alternative to the cultural, ‘ancient hatreds’ explanation for the war between the Serbs and Croats. As many people have noted, the vast majority of Serbs and Croats lived together peacefully until the spring of 1991, when Croatia declared its inde-pendence. Th ere is no evidence to suggest the existence of deep and widespread hatred in relations among Serbs and Croats during the sixty-year history of Yugoslavia. Moreover, even if evidence could be found for the persistence of ‘ancient hatreds’, they still cannot explain a key ‘puzzle’: why relations among communities that had been living together peacefully became polarized so quickly, before fi nally dissolving into savage violence.

Rational choice theory is the study of strategic political interactions, of how people (agents or players) determine strategies in diff erent situations. It explains outcomes as the result of rational choices made by individuals within a given set of material and structural circum-stances. It shows that, given a particular set of circumstances, the strategic interactions of agents will produce predictable, law-like outcomes. Much of the analysis of strategic political interactions focuses on how individual actors make decisions in game-like situations. A ‘game’ is any situation in which a fi xed set of agents or players, with a fi xed set of strategies available to them, compete against one another, and receive a payoff as a result of the strate-gies they and their fellow actors choose. Th e assumption is that all players know all possible outcomes and have preferences regarding these possible outcomes based on the amount of value or utility they derive from each of them. All players behave rationally; and they make rational decisions about what strategy to pursue based on a calculation of the costs and ben-efi ts of diff erent strategies for achieving their preferred outcome.

In a series of articles (1994, 1995, 1998), James Fearon used rational-choice assumptions as a basis for exploring the causes and conditions of ethnic confl ict and war.

In order to explain the ‘puzzle’ of the rapid polarization of Serbs and Croats in Croatia in 1991, James Fearon develops a game-theoretic model: a model of how groups of people interact when confronted with a situation of uncertainty, based on assumptions of game theory. Fearon points out that, just months before the eruption of war between Serbs and Croats in Croatia, journalists had reported that ‘most people seemed to have had no use for or interest in the exclusivist arguments pushed by the minority of extremists’ (1998: 114). ‘With the exception of a relatively small number of extremists . . . Serbs and Croats in the mixed population areas recognised that war would be costly and viewed it as unnecessary.’ But in spring 1991, ‘Serbs and Croats who had resisted the extremists appeals fi nally opted for division and war’ (Fearon 1998: 115).

Fearon argued that, following the collapse of the Soviet Union, ethnic confl ict erupted in Yugoslavia as a result of a ‘commitment problem’. Th e problem arises when ‘two political communities fi nd themselves without a third party that can guarantee agreements between them’ (1995: 2). If, in a new state, ethnic minorities don’t believe that the state can guarantee that the ethnic majorities will not infringe on their rights, they will prefer to fi ght for succes-sion while the state is still weak. Th is, he argues, helps to explain the rapid polarization of ethnic groups in Croatia following the declaration of Croatian independence in June 1991.


Explanation consistent with the hypothetico-deductive model here consists of (1) a set of initial determining conditions (circumstances pertaining at particular times and places); and (2) a general law or laws which connect these conditions to the type of events to be explained (‘hypotheses’ which are capable of being confi rmed or disconfi rmed by suitable empirical fi ndings). First, Fearon specifi es the set of initial conditions. Th e commitment problem, Fearon tells us, arises whenever three conditions hold: (1) the groups interact in anarchy, without a third party able to guarantee and enforce agreements between them; (2) one of the groups anticipates that its ability to secede or otherwise withdraw from joint arrangements will decline in the near future; and (3) for this group, fi ghting in the present is preferable to the worst political outcome it could face if it chooses continued interaction (1995: 10). Second, on the basis of a game-theoretic model he develops of the commitment problem, Fearon generates hypotheses about what makes ethnic war more or less likely. Th e key mechanisms include (1) the expected change in size in the relative military power between groups that would result from formation of a new state; (2) the relative size of the ethnic minority; (3) whether majority and minority groups’ costs for fi ghting are low; and (4) whether institutions can be created that give minority groups political power that is at least proportional to their numbers.

Fearon then applies the model to the war in Croatia in 1991–2. When Croatia declared its independence from Yugoslavia, minority Serbs living in Croatia faced the prospect of being in a state with no credible guarantees on their political status, or economic and even physical security. ‘If the commitment problem model does capture something of what was going on in Croatia, then we might expect to fi nd evidence of Croatian leaders trying to work out guarantees with Serb leaders’ (1998: 119). Th is, he fi nds, occurs on numerous occasions. Croatian President Tudjman met with the leader of the Serbs in Croatia, Jovan Raskovic, ‘to discuss the issue of commitment to guarantees on the Serbs’ status and “cultural autonomy”’ (1998: 119). But despite Tudjman’s eff orts to construct a credible set of guarantees for the Serb minority, his eff orts to solve the commitment problem were ultimately unsuccessful.With the ‘prospect of entering the new state of Croatia with no credible guarantees on their political status, or economic or even physical security’, the prospect of a war then appeared better to the Serbs than the prospect of fi ghting later, by which time the Croatian state would have grown stronger (Fearon 1998: 116).

Th e evidence, then, consists in showing that there is a ‘fi t’ between the deductions of the theory and the observed behavioural outcomes; that the outcome is consistent with the theoretical predictions. Fearon also endeavours to demonstrate that his explanation off ers a better ‘fi t’ with the facts than other explanations do. Finally, he argues that the basic com-mitment problem that he describes ‘appears either to lurk or to have caused interethnic violence’ in other cases, as well: in Azerbaijan, Georgia, Moldova, Ukraine, Estonia, Zimba-bwe, South Africa, and Northern Ireland (1995: 21).

Let’s consider how an interpretivist approach, constructivism, explains the rapid ethnic polarization that occurred in the former Yugoslavia in the 1990s.

Constructivism is an approach that has had an important infl uence on political inquiry. Consistent with interpretivist assumptions, constructivism maintains that reality does not exist as something independent of us and is not, therefore, merely discovered by us: it is socially, and actively, constructed. Constructivists assume that social phenomena are social constructs in the sense that their shape and form is imbued with social values, norms, and assumptions, rather


than being the product of purely individual thought or meaning. We live in ‘a world of our mak-ing’, as Nicolas Onuf (1989) has put it. Actors are not totally free to choose their circumstances, but make choices in the process of interacting with others and, as a result, bring historically, culturally, and politically distinct ‘realities’ into being. In this respect, the world of politics is a social construction rather than something that exists independently of human meaning and action. States and other social institutions take specifi c historical, cultural, and political forms that are a product of human interaction in the social world.

In contrast with positivist approaches which emphasize a single objective reality, the idea of social construction suggests diff erence across context. It is not only the material environment, but also the cultural and institutional environment that provides incentives and disincentives for behaviour. Society is more than just the site of strategic interaction to pursue pre-defi ned interests in a rational, utility-maximizing manner. It is a constitutive realm, an environment that forms and infl uences the identities and interests of actors and makes them who they are. Moreover, social interaction also infl uences the identity of actors. Th e properties of actors are not intrinsic to them: they are socially contingent, they depend on social interaction: bargain-ing/negotiating, arguing, communicating in general. Both the identities and interests of actors are constituted (formed, infl uenced) through interaction and by the institutionalized norms, values, and ideas of society. Since the interests and identities of actors are not given—but result from social interaction—they cannot be abstracted from the social conditions which produce them; and they are subject to change as a result of political processes.

Consistent with these assumptions, constructivism, like rational-choice approaches, rejects explanations of nationalist and ethnic phenomena as the outcome of essential cultural identities. But, unlike rational-choice approaches, it sees nationalist and ethnic confl ict as a phenomenon that has assumed a variety of forms across space and time; and it emphasizes the role of identities that are multiple and fl uid and politically malleable: infl uenced by sur-rounding structures and ‘constructed’ for political purposes.

Murat Somer (2002, 2001) addresses the same ‘puzzle’ as Fearon: the rapid ethnic polari-zation that occurred in Yugoslavia in the 1990s. Somer emphasizes the signifi cance of public discourses in forming individuals’ ethnic identities, and in suppressing and reviving domi-nant perceptions of ethnic identities. He argues that ethnic confl ict is a result of processes of ethnic identity construction in the public arena that construct a divisive image of identities. Public ethnic activities and expressions are immediately observed and they immediately aff ect the decisions of others.

Ethnic polarization changes the dominant images of ethnic categories in society through cascades of individual reactions. A cascading process changes behaviour and attitudes and, once begun, is very diffi cult to stop. In Yugoslavia, ethnic polarization in public discourses was engineered by ethnic entrepreneurs who constructed and promoted a divisive image of ethnic identities as mutually exclusive and incompatible with belonging to the same nation. Th is triggered a ‘cascade process’, which resulted in the creation of a critical mass of opinion around a new image of ethnic identities. People who secretly held this divisive view, as well as people who now felt compelled to support it, jumped on the bandwagon. Hence, the divi-sive image became the norm, and it became ‘inappropriate, even blasphemous, to defend interethnic mixing and brotherhood’ (Somer 2001: 128).

Somer draws a distinction between public and private ethnic polarization and highlights the way people publicly ‘falsify’ their private beliefs. During the communist era, state policies


had ‘aimed at eradicating the public expression of the divisive image of ethnic relations in the country; but they had insuffi ciently encouraged its elimination in private’. Consequently, the public discourse in Yugoslavia had exerted pressure for ‘downward falsifi cation’ to discour-age people from openly expressing their ethnic prejudices. Th is ‘downward preference falsi-fi cation concealed, to most observers, the private importance of the divisive image’; consequently, ‘even analysts who had a fair idea about the private signifi cance of the divisive image were surprised by the severity of polarization’ in the 1990s (2001: 136). During the 1990s, ‘the dominant public discourse emphasizing unity and brotherhood’ turned into one that emphasized ‘radical ethnonationalism’ (2001: 136). But this ‘public polarization far exceeded private polarization’ (2001: 143). Consequently, there was widespread ‘upward ethnic preference falsifi cation’, the exaggeration of public support for the divisive image, as the new nationalist regime exerted pressure for people—including liberal and moderately tolerant individuals—to think and act in an ethnically intolerant manner.

Somer uses survey research which indicates decreases in self-identifi cation with the overarching Yugoslav identity between 1981 and 1991. Th e respondents were anonymous, so these surveys were able to capture changes in people’s private preferences. During the 1980s there was a striking upsurge in ‘the public expression of the divisive image’ (Somer 2001: 143). But, ‘in 1989, when public polarization had reached an advanced state, anony-mous surveys continued to reveal that interethnic tolerance levels were high by global standards’. So, ‘while the public discourse was becoming increasingly more divisive and less tolerant of interethnic diff erences, private attitudes remained quite tolerant of interethnic diff erences’. In fact, ‘the highest levels of tolerance were found in Bosnia, the site of the most violent crimes’ (Somer 2001:144). ‘Desertion and call-up evasion were very common during the civil war when public support for the divisive image was at its peak’ (Somer 2001: 144).

Let’s sum up by comparing the two approaches to understanding ethnic polarization in Yugoslavia. Both highlight the signifi cance of social and political institutions in forming individuals’ ethnic identities. But Fearon, like other rational-choice theorists, tends to stress the structural and constraining features of institutions. Somer, on the other hand, empha-sizes their social and cognitive aspects. Both are constructivist in the sense that they see changes in an actor’s identity constructions as likely to occur in moments of crisis and dilemma. But Fearon, consistent with rational choice approaches, emphasizes the role of strategic calculation in identity construction, while Somer emphasizes cognitive features, such as norms of behaviour and inter-subjective understandings (though these don’t neces-sarily operate to the exclusion of the calculative element stressed in the rational choice explanations off ered by Fearon and others).

Let’s consider how the two analyses illustrate the distinction between ‘internal’ and ‘exter-nal’ explanation and evidence that we previously discussed. Th e analyses that Fearon and Somer off er are consistent with the assumptions, respectively, of rational choice and con-structivist approaches regarding actors’ interests: rational-choice theories assume that agents act on the basis of fi xed interests and preferences; constructivists assume that interests can only develop from the image an actor holds of himself and of others, that identities are the source of interests (and, therefore, the basis of action) (e.g. Wendt 1994; Ringmar 1996).

Fearon models external ‘behaviour’, and then seeks evidence by way of deductive fi t with that model. Empirical evidence consists of statements and activities of Croatian leaders


that indicate a concern for providing the Serb minority with a commitment to guarantee their status and cultural autonomy. But there is no direct evidence of the existence of a commitment problem among the Serb population, of strategic behaviour on the part of individuals or groups, or of the relationship between belief and action. Direct evidence concerning whether the individual or group choices which led to ethnic polarization were made for the reasons stated in his model might be impractical or impossible to obtain. Instead, the ‘test’ of the model involves (1) deducing the factors that, in the given circum-stances, might be expected to lead groups to resort to violence; and (2) observing the out-comes. Where we observe that circumstances favour the behaviours that the model tells us are most likely to occur, we can infer that there is a line of cause and eff ect that relates those circumstances and the outcome. Somer identifi es a mechanism, the ‘cascade’ process, which links popular beliefs, public political discourse, and relations across groups. He then combines survey data with detailed examination of the historical events to provide evidence of the changing nature of public and private views and suggest the relationship between them.

Both analyses have implications for politics and policy: which of the explanations for ethnic confl ict we choose to favour has implications for how people relate to each other and how governments act. Th e choice of which set of assumptions will provide the best starting point for your own research on a specifi c political issue or problem is one which you will need to carefully consider.

Conclusions

This chapter has begun our consideration of some fundamental ontological, epistemological, and methodological problems posed by social inquiry. These problems mostly branch out from one central question: are the methods of the social sciences essentially the same as, or essentially different from, those of the natural sciences? We have reviewed the basic tenets of three different answers to this question: positivism, scientifi c realism, and interpretivism. How these answers differ is presented in Box 2.3.

All of these defi ne a position with respect to how we study and conduct research in the social sciences. As Box 2.3 shows, all are based on fundamentally different assumptions about how we can know and develop knowledge about the social world; and all of them remain important perspectives for contemporary social research. The question of whether and how we can pursue a systematic and rigorous study of social phenomena in the way that scientists pursue study of the natural world and, more generally, philosophical presuppositions about ‘reality’ implicit in social science research, bears on how we design or structure an approach to research into political phenomena, and the claims we can make with respect to the fi ndings that result from that research.

At the heart of the debate among these perspectives is the question of what sort of knowledge we can gain about social phenomena. This question is also central to the controversy that we take up in Chapter 3, which is the debate about whether the knowledge produced through the study of the social world is or can be ‘objective’; and in Chapter 4, where we consider the question of what is the ‘social’.

Reality— is constantly being defi ned for us—by political scientists, by historians, by politicians in their speeches, by media analysts in their news reports. The ability to identify the underlying structure of assumptions or the implicit theory which shapes a given account of reality, whether presented by scholars, politicians, or journalists, allows us to become more active analysts of contemporary politics, rather than depending on the analysis of others.


Questions

● What is involved in providing an explanation of social phenomena? How is explanation distinct from and related to interpretation?

● What place does the concept of ‘law’ have in social-scientifi c explanation?

● What does ‘causality’ mean? How, according to different conceptions of causality, do we establish that something causes something else?

● How is describing, interpreting, and explaining human action different from describing, interpreting, and explaining non-human events?

● Should the primary goal of social science be to provide law-like explanations, capable of supporting predictions? Or are law-like explanations impossible, or unnecessary, within social science?

BOX 2.3 Positivism, Scientifi c Realism, and Interpretivism Compared

Positivism Scientifi c realism Interpretivism

Ontology: What

is the nature of

the social world?

No different from the

natural world: an objec-

tive reality that exists

independently of our

knowledge of it

(a ‘naturalist’ ontology).

No different from

the natural world: an

objective reality that

exists independently of

our knowledge of it (a

‘naturalist’ ontology).

Fundamentally different

from the natural world.

Reality is not mind-

independent: it is

subjectively created.

Epistemology: What sort of

knowledge of the

social world is

possible?

Scientifi c knowledge of

the social world is limited

to what can be observed.

We can explain and

predict social phenomena

by discovering empirical

regularities, formulating

law-like generalizations,

and establishing causal

relationships.

Scientifi c knowledge is

not limited to what can be

observed but also includes

theoretical entities (unob-

servable elements of social

life). We can explain and

predict social phenomena

based on theories about

these entities and certify

the truth of these theories

by employing various

rules of method.

Scientifi c knowledge

can be gained through

interpreting the meanings

which give people reasons

for acting. We can, in this

way, understand human

behaviour; but we cannot

explain or predict it on

the basis of law-like

generalizations and

establishing the existence

of causal relationships.

Causality: what

do we mean by

‘causes’?

A refl ection of the way

we think about the world;

established by discovering

observable regularities.

A refl ection of reality;

established by discovering

unobservable underlying

generative mechanisms.

We cannot seek

causes, but only uncover

meanings that provide the

reasons for action.

Methodology:

How can we gain

knowledge of the

social world?

Through direct

observation.

Through direct observa-

tion and logic applied

to both observable and

unobservable structures.

Interpretive theory and

textual strategies. The social world is like a text

and has to be interpreted

to discover hidden

meanings and subtexts.


Guide to Further Reading

March, D. and P. Furlong (2002), ‘A Skin not a Sweater: Ontology and Epistemology in Political Science’, in D. Marsh and G. Stoker (eds), Theory and Methods in Political Science, 2nd edition (Basingstoke: Palgrave), 17–41.

The authors discuss positivism, interpretivism, constructivism, and also realism, as different approaches to ontology and epistemology in political science, and illustrate their differences with case studies.

Martin, M. and L. C. McIntyre (eds) (1994), Readings in the Philosophy of Social Science (New York: MIT Press).

This volume brings together a collection of important texts on the disputed role of general laws in social-scientifi c explanation (Part II), and on interpretation and meaning (Part III).

Gibbons, M. T. (2006), Hermeneutics, Political Inquiry, and Practical Reason: An Evolving Challenge to Political Science’. American Political Science Review 100: 4 (November), 563–71.

Lane, R. (1996), ‘Positivism, Scientifi c Realism and Political Science’. Journal of Theoretical Politics 8 (3): 361–82.

This article explores the implications of scientifi c realist principles for political science, political research, and political theory, providing examples of a scientifi c realist approach in studies utilizing a variety of theoretical approaches, including rational choice, new institutionalism, and comparative politics.

Little, D. (1991), Varieties of Social Explanation: An Introduction to the Philosophy of Social Science (Boulder, Colorado: Westview Press), chapter 2 (‘Causal Analysis’), pp. 13–38.

Russo, F. (2009), Causality and Causal Modelling in the Social Sciences (New York: Springer).This book offers an overview of debates, and it provides a valuable analysis of reasoning about causation by looking at the causal arguments advanced in specifi c social science studies.

Sankey, H. (2008), Scientifi c Realism and the Rationality of Science (Aldershot: Ashgate), chapter 1.This chapter provides a clear exposition of the doctrines of scientifi c realism, which distinguishes between core and optional doctrines; and the principal arguments that have been advanced for scientifi c realism.

Amadae, S. M. and B. Bueno de Mesquita (1999), ‘The Rochester School: The Origins of Positive Political Theory’, Annual Review of Political Science 2: 269–95.

Benoit, K. (2007), ‘Electoral Laws as Political Consequences: Explaining the Origins and Change of Electoral Institutions’. Annual Review of Political Science 10 (June): 363–90.

Bhaskar, Roy (1998), The Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences, 3rd edition (Hemel Hempstead: Harvester Wheatsheaf).

Brown, Chris (2007), ‘Situating Critical Realism’, Millennium: Journal of International Studies 35(2) (March): 409–16.

Bryman, A. (2004), Social Research Methods (Oxford: Oxford University Press).

Colomer, J. P. (2005), ‘It’s Parties that Choose Electoral Systems (or, Duverger’s Laws Upside Down)’, Political Studies 53: 1–21.

Easton, D. (1969), ‘The New Revolution in Political Science’, American Political Science Review 63(4) (December): 1051–61.

Ekström, M. (1992), ‘Causal Explanation of Social Action: The Contribution of Max Weber and of Critical Realism to a Generative View of Causal Explanation in Social Science’, Acta Sociologica: Journal of the Scandinavian Sociological Association 35: 107–22.

References


Fearon, D. James (1994), ‘Ethnic War as a Commitment Problem’, paper presented at the annual meeting of the American Political Science Association, New York, http://www.stanford.edu/~jfearon/papers/ethcprob.pdf.

———(1995), ‘Rationalist Explanations for War’, International Organization 49(3): 379–414.

———(1998). ‘Commitment Problems and the Spread of Ethnic Confl ict’, in David Lake and Donald Rothchild (eds), The International Spread of Ethnic Confl ict (Princeton, NJ: Princeton University Press), 107–26.

Finnemore, M. and K. Sikkink (2001), ‘Taking Stock: The Constructivist Research Program in International Relations and Comparative Politics’, Annual Review of Political Science 4: 391–416.

Føllesdal, D. (1994), ‘Hermeneutics and the Hypothetico-Deductive Method’, in Michael Martin and Lee C. McIntyre (eds), Readings in the Philosophy of Social Science (Cambridge, MA: MIT Press).

Gerring, J. (2007), ‘Review Article: The Mechanismic Worldview: Thinking Inside the Box’. British Journal of Political Science 38: 161–79.

Gibbons, M. T. (2006), ‘Hermeneutics, Political Inquiry, and Practical Reason: An Evolving Challenge to Political Science’, American Political Science Review 100: 4 (November), 563–71.

Gilpin, R. (1981), War and Change in World Politics (Cambridge: Cambridge University Press).

Grix, J. (2002), The Foundations of Research (Basingstoke: Palgrave/Macmillan).

Hampsher-Monk, I. and A. Hindmoor (2009), ‘Rational Choice and Interpretive Evidence: Caught between a Rock and a Hard Place?’ Political Studies 58(1): 47–65.

Hedström, P.and R. Swedborg, eds (1998), Social Mechanisms: An Analysis Approach to Social Theory. (Cambridge : Cambridge University Press).

Hempel, C. G. (1994), ‘The Function of General Laws in History’, in Michael Martin and Lee C. McIntyre (eds), Readings in the Philosophy of Social Science (Cambridge, MA: MIT Press), 43–54.

Hoffmann, M. J. (2006), ‘Social (De)Construction: The Failure of a Multinational State’, in Jennifer Anne Sterling-Folker (ed.), Making Sense of International Relations Theory (Boulder, CO: Lynne Reiner), 123–38.

Hopf, T. (2007), ‘The Limits of Interpreting Evidence’, in R. N. Lelbow and M. I. Lichbach (eds), Theory and Evidence in Comparative Politics and International Relations (Basingstoke: Palgrave/Macmillan), 55–84.

Hume, D. (1966), Enquiries Concerning the Human Understanding (Oxford: Clarendon Press).

Lane, R. (1996), ‘Positivism, Scientifi c Realism and Political Science’, Journal of Theoretical Politics 8(3): 361–82.

Little, D. (1991), Varieties of Social Explanation: An Introduction to the Philosophy of Social Science (Boulder, CO: Westview Press).

———(2009), ‘McIntyre and Taylor on the Human Sciences’. 16 July. http://understandingsociety.blogspot.com/2009/07/macintyre-and-taylor-on-human-sciences.html.

May, T (2003), Social Research: Issues, Methods And Process, 3rd edition (Buckingham: Open University Press).

McIntyre, L. C. (1994), ‘Complexity and Social Scientifi c Laws’, in Michael Martin and Lee C. McIntyre (eds), Readings in the Philosophy of Social Science (Cambridge, MA: MIT Press), 131–44.

Onuf, N. (1989), World of Our Making: Rules and Rule in Social Theory and International Relations (Columbia, SC: University of South Carolina Press).

Pollins, B. (2007), ‘Beyond Logical Positivism: Reframing King, Keohane, and Verba’, in R. N. Lelbow and M. I. Lichbach (eds), Theory and Evidence in Comparative Politics and International Relations (Basingstoke: Palgrave/Macmillan), 87–106.

Popper, Karl, (1959), The Logic of Scientifi c Discovery (London: Hutchinson of London).

Riker, W. H. (1982), ‘The Two-Party System and Duverger’s Law: An Essay on the History of Political Science’, American Political Science Review 76 (December): 753–66.

Ringmar, E. (1996), Identity, Interest and Action (Cambridge: Cambridge University Press).

Ross, M. (2001), ‘Does Oil Hinder Democracy?’ World Politics 53: 325–61.

Sankey, N. (2008), Scientifi c Realism and the Rationality of Science (Aldershot : Ashgate).

Schwartz, J. D. (1984), ‘Participation and Multisubjective Understanding: An Interpretivist Approach to the Study of Political Participation’, Journal of Politics 46: 1117–41.

http://www.stanford.edu/~jfearon/papers/ethcprob.pdf

http://www.stanford.edu/~jfearon/papers/ethcprob.pdf

http://understandingsociety.blogspot.com/2009/07/macintyre-and-taylor-on-human-sciences.html




Somer, M. (2001), ‘Cascades of Ethnic Polarization: Lessons from Yugoslavia’, The ANNALS of the American Academy of Political and Social Science 573(1): 127–51.

———(2002), ‘Insincere Public Discourse, Trust, and Implications for Democratic Transition: The Yugoslav Meltdown Revisited’, Journal for Institutional Innovation, Development and Transition 6 (December): 92–112.

Taylor, C. (1994), ‘Interpretation and the Sciences of Man’, in Michael Martin and Lee C. McIntyre (eds), Readings in the Philosophy of Social

Science (Cambridge, MA: MIT Press), 181–212.

Tilly, C. (2001), ‘Mechanisms in Political Processes’, Annual Review of Political Science 4: 21–41.

Trochim, W. M. (2006), The Research Methods Knowledge Base, 2nd edition. Available at: http://www.socialresearch methods.net/kb/.

Wendt, A. (1994), ‘Collective Identity Formation and the International State’, American Political Science Review 88(2) (June): 384–96.

Endnotes1. Note the difference between the terms ‘behaviourism’ and behaviouralism’: ‘behaviourism’ is a school of

psychology which studies observable behaviour, rather than ‘unobservable’ behaviour such as mental processes and intentions, and emphasizes experimentation and causal analysis. ‘Behaviouralism’ is the term adopted by political scientists. The key tenet of behaviouralism is that only observable behaviour may be studied.

2. In recent years some researchers have modifi ed Duverger’s Law by suggesting that ‘it is the number of parties that can explain the choice of electoral systems, rather than the other way round’ (Colomer 2005: 1; see also Benoit 2007).

http://www.socialresearchmethods.net/kb/

http://www.socialresearchmethods.net/kb/

Social research strategies 19

IntroductionThis book is about social research. It attempts to equip people who have some knowledge of the social sciences with an appreciation of how social research should be conducted and what it entails. The latter project involves situating social research in the context of sociology, which in turn means attending to the question of its role in the overall enterprise of the discipline. It would be much easier to ‘cut to the chase’ and explore the nature of methods of social research and provide advice on how best to choose between and implement them. After all, many people might expect a book with the title of the present one to be concerned mainly with the ways in which the different methods in the social researcher’s arsenal can be employed.

But the practice of social research does not exist in a bubble, hermetically sealed off from the social sciences and the various intellectual allegiances that their practi-tioners hold. Two points are of particular relevance here.

First, methods of social research are closely tied to dif-ferent visions of how social reality should be studied. Methods are not simply neutral tools: they are linked

with the ways in which social scientists envision the con-nection between different viewpoints about the nature of social reality and how it should be examined. However, it is possible to overstate this point. While methods are not neutral, they are not entirely suffused with intellectual inclinations either. Secondly, there is the question of how research methods and practice connect with the wider social scientifi c enterprise. Research data are invariably collected in relation to something. The ‘some-thing’ may be a burning social problem or, more usually, a theory.

This is not to suggest that research is entirely dictated by theoretical concerns. One sometimes fi nds simple ‘fact-fi nding’ exercises published. Fenton et al. (1998) conducted a quantitative content analysis of social re-search reported in the British mass media. They exam-ined national and regional newspapers, television and radio, and also magazines. They admit that one of the main reasons for conducting the research was to establish the amount and types of research that are represented. Sometimes, such exercises are motivated by

Chapter guide

The chief aim of this chapter is to show that a variety of considerations enter into the process of doing social research. The distinction that is commonly drawn among writers on and practitioners of social research between quantitative research and qualitative research is explored in relation to these considerations. This chapter explores:

• the nature of the relationship between theory and research, in particular whether theory guides research (known as a deductive approach) or whether theory is an outcome of research (known as an inductive approach);

• epistemological issues—that is, ones to do with what is regarded as appropriate knowledge about the social world; one of the most crucial aspects is the question of whether or not a natural science model of the research process is suitable for the study of the social world;

• ontological issues—that is, ones to do with whether the social world is regarded as something external to social actors or as something that people are in the process of fashioning;

• the ways in which these issues relate to the widely used distinction in the social sciences between two types of research strategy: quantitative and qualitative research; there is also a preliminary discussion, which will be followed up in Chapter 27, that suggests that, while quantitative and qualitative research represent different approaches to social research, we should be wary of driving a wedge between them;

• the ways in which values and practical issues also impinge on the social research process.

9780199588053_C02.indd 19 10/20/11 9:59 AM

Social research strategies20

a concern about a pressing social problem. McKeganey and Barnard (1996) conducted qualitative research involving observation and interviews with prostitutes and their clients in Glasgow. One factor that seems to have prompted this research was the concern about the role of prostitutes in spreading HIV infection (McKeganey and Barnard 1996: 3). Another scenario occurs when research is done on a topic when a specifi c opportunity arises. The interest of Westergaard et al. (1989) in the effects of redundancy seems to have been profoundly motivated by the opportunity that arose when a Sheffi eld steel company, which was close to their institutional base at the University of Sheffi eld, made a large number of people redundant. The fi rm’s management approached the authors a year after the redundancies to conduct research on what had happened to the individuals who had been made redundant. The authors conducted social survey research using a structured interview approach on most of those made redundant. Of course, the authors were infl uenced by theories about and previous research on unemployment, but the specifi c impetus for

the research on the effects of redundancy was not planned. Yet another stimulus for research can arise out of personal experiences. Lofl and and Lofl and (1995) note that many research publications emerge out of the researcher’s personal biography, such as Zukin’s (1982) interest in loft living arising out of her living in a loft in New York City. Another example is O’Reilly’s (2000) in-vestigation of British expatriates living on the Costa del Sol in Spain, which stemmed from her and her partner’s dream of moving to the area themselves, which in fact they eventually did. Certainly, my own interest in Disney theme parks can be traced back to a visit to Disney World in Florida in 1991 (Bryman 1995, 1999), while my inter-est in the representation of social science research in the mass media (Fenton et al. 1998) can almost certainly be attributed to a diffi cult encounter with the press reported in Haslam and Bryman (1994).

By and large, however, research data achieve signifi -cance in sociology when viewed in relation to theoretical concerns. This raises the issue of the nature of the rela-tionship between theory and research.

Student experiencePersonal experience as a basis for research interestsFor her research, Isabella Robbins was interested in the ways in which mothers frame decisions regarding vaccinations for their children. This topic had a particular signifi cance for her. She writes:

As the mother of three children I have encountered some tough decisions regarding responsibility towards my children. Reading sociology, as a mature student, gave me the tools to help understand my world and to contextualize some of the dilemmas I had faced. In particular, I had experienced a diffi cult decision regarding the vaccination status of my children.

To read more about Isabella’s research experiences, go to the Online Resource Centre that accompanies this book at: www.oxfordtextbooks.co.uk/orc/brymansrm4e/

Theory and research

Characterizing the nature of the link between theory and research is by no means a straightforward matter. There are several issues at stake here, but two stand out in par-ticular. First, there is the question of what form of theory one is talking about. Secondly, there is the matter of whether data are collected to test or to build theories.

Theory is important to the social researcher because it provides a backcloth and rationale for the research that is being conducted. It also provides a framework within which social phenomena can be understood and the research fi ndings can be interpreted.

9780199588053_C02.indd 20 10/20/11 9:59 AM

www.oxfordtextbooks.co.uk/orc/brymansrm4e/


What type of theory?

The term ‘theory’ is used in a variety of different ways, but its most common meaning is as an explanation of observed regularities—for example, why sufferers of schizophrenia are more likely to come from working-class than middle-class backgrounds, or why work alien-ation varies by technology. But such theories tend not to be the stuff of courses in sociological theory, which typically focus much more on theories with a higher level of abstraction. Examples of such theories include structural-functionalism, symbolic interactionism, crit-ical theory, poststructuralism, structuration theory, and so on. What we see here is a distinction between theories of the former type, which are often called theories of the middle range (Merton 1967), and grand theories, which operate at a more abstract and general level. According to Merton, grand theories offer few indications to re-searchers as to how they might guide or infl uence the collection of empirical evidence. So, if someone wanted to test a theory or to draw an inference from it that could be tested, the level of abstractness is likely to be so great

that the researcher would fi nd it diffi cult to make the necessary links with the real world. There is a paradox here, of course. Even highly abstract ideas, such as Parsons’s notions of ‘pattern variables’ and ‘functional requisites’, must have some connection with an external reality, in that they are likely to have been generated out of Parsons’s reading of research or his refl ections upon that reality or others’ writings on it. However, the level of abstractness of the theorizing is so great as to make it dif-fi cult for them to be deployed in research. For research purposes, then, Merton argues that grand theories are of limited use in connection with social research, although, as the example in Research in focus 2.1 suggests, an abstract concept like social capital (Bourdieu 1984) can have some pay-off in research terms. Instead, middle-range theories are ‘intermediate to general theories of social systems which are too remote from particular classes of social behavior, organization and change to account for what is observed and to those detailed orderly descriptions of particulars that are not generalized at all’ (Merton 1967: 39).

Research in focus 2.1Grand theory and social researchButler and Robson (2001) used Bourdieu’s concept of social capital as a means of understanding gentrifi cation of areas of London. While the term ‘social capital’ has acquired an everyday usage, Butler and Robson follow Bourdieu’s theoretical use of it, which draws attention to the social connectedness and the interpersonal resources that those with social capital can draw on to pursue their goals. While the term has attracted the interest of social policy researchers and others concerned with social exclusion, its use in relation to the middle class has been less prominent, according to Butler and Robson. Bourdieu’s treatment implies that those with social capital cultivate signifi cant social connections and then draw upon those connections as resources for their goals. Butler and Robson conducted semi-structured interviews with ‘gentrifi ers’ in each of three inner London areas. Responding to a tendency to view gentrifi cation in rather unitary terms, the authors selected the three areas to examine what they refer to as the ‘variability’ of the process. To that end, the areas were selected to refl ect variation in two factors: the length of time over which gentrifi cation had been occurring and the middle-class groupings to which each of the areas appealed. The selection of areas in terms of these criteria was aided by census data. Of the three areas, Telegraph Hill was the strongest in terms of social capital. According to the authors, this is revealed in ‘its higher levels of voluntary co-operation and sense of geographically focused unity’ (Butler and Robson 2001: 2159). It is the recourse to these networks of sociality that accounts for the successful gentrifi cation of Telegraph Hill. Battersea, one of the other two areas, entails a contrasting impetus for gentrifi cation in Bourdieu’s terms. Here, economic capital was more signifi cant for gentrifi cation than the social capital that was important in Telegraph Hill. The role of economic capital in Battersea can be seen in the ‘competitive access to an increasingly desirable and expensive stock of housing and an exclusive circuit of schooling centred on private provision’ (Butler and Robson 2001: 2159). In the former, it is sociality that provides the motor for gentrifi cation, whereas in Battersea gentrifi cation is driven by market forces and is only partially infl uenced by patterns of social connectedness. This study is an interesting example of the way in which a relatively high-level theoretical notion—social capital and its kindred concept of economic capital—associated with a social theorist can be employed to illuminate research questions concerning the dynamics of modern urban living.

9780199588053_C02.indd 21 10/20/11 9:59 AM


By and large, then, it is not grand theory that typically guides social research. Middle-range theories are much more likely to be the focus of empirical enquiry. In fact, Merton formulated the idea as a means of bridging what he saw as a growing gulf between theory (in the sense of grand theory) and empirical fi ndings. This is not to say that there were no middle-range theories before he wrote: there defi nitely were, but what Merton did was to seek to clarify what is meant by ‘theory’ when social scientists write about the relationship between theory and research.

Middle-range theories, unlike grand ones, operate in a limited domain, whether it is juvenile delinquency, racial prejudice, educational attainment, or the labour process

Even the grand/middle-range distinction does not en-tirely clarify the issues involved in asking the deceptively simple question of ‘what is theory?’ This is because the term ‘theory’ is frequently used in a manner that means little more than the background literature in an area of social enquiry. To a certain extent, this point can be taken to apply to fact-fi nding exercises such as those referred to above. The analysis of the representation of social research in the media by Fenton et al. (1998) was under-taken against a background of similar analyses in the USA and of studies of the representation of natural sci-ence research in the media in several different countries. In many cases, the relevant background literature relat-ing to a topic fuels the focus of an article or book and thereby acts as the equivalent of a theory, as with the research referred to in Research in focus 2.3. The litera-ture in a certain domain acts as the spur to an enquiry. The literature acts as an impetus in a number of ways: the researcher may seek to resolve an inconsistency between different fi ndings or between different interpretations of fi ndings; the researcher may have spotted a neglected

(see Research in focus 2.2). They vary somewhat in their range of application. For example, labelling theory repre-sents a middle-range theory in the sociology of deviance. Its exponents sought to understand deviance in terms of the causes and effects of the societal reaction to devi-ation. It was held to be applicable to a variety of different forms of deviance, including crime and mental illness. By contrast, Cloward and Ohlin’s (1960) differential associ-ation theory was formulated specifi cally in connection with juvenile delinquency, and in subsequent years this tended to be its focus. Middle-range theories, then, fall somewhere between grand theories and empirical fi nd-ings. They represent attempts to understand and explain a limited aspect of social life.

aspect of a topic; certain ideas may not previously have been tested a great deal; the researcher may feel that existing approaches being used for research on a topic are defi cient, and so provides an alternative approach; and so on.

Social scientists are sometimes prone to being somewhat dismissive of research that has no obvious con nections with theory—in either the grand or the middle-range senses of the term. Such research is often dismissed as naive empiricism (see Key concept 2.1). It would be harsh, not to say inaccurate, to brand as naive empiricism the numerous studies in which the publications-as-theory strategy is employed, simply because their authors have not been preoccupied with theory. Such research is con-ditioned by and directed towards research questions that arise out of an interrogation of the literature. The data collection and analysis are subsequently geared to the illumination or resolution of the research issue or problem that has been identifi ed at the outset. The literature acts as a proxy for theory. In many instances, theory is latent or implicit in the literature.

Research in focus 2.2Labour process theory: a middle-range theoryIn the sociology of work, labour process theory can be regarded as a middle-range theory. The publication of Labor and Monopoly Capital (Braverman 1974) inaugurated a stream of thinking and research around the idea of the labour process and in particular on the degree to which there has been an inexorable trend towards increasing control over the manual worker and the deskilling of manual labour. A conference volume of much of this work was published as Labour Process Theory (Knights and Willmott 1990). P. Thompson (1989) describes the theory as having four elements: the principle that the labour process entails the extraction of surplus value; the need for capitalist enterprises constantly to transform production processes; the quest for control over labour; and the essential confl ict between capital and labour. Labour process theory has been the focus of considerable empirical research (e.g. Knights et al. 1985).

9780199588053_C02.indd 22 10/20/11 9:59 AM


Indeed, research that appears to have the character-istics of the fact-fi nding exercise should not be prematurely dismissed as naive empiricism either. McKeganey and Barnard’s (1996) research on prostitutes and their clients is a case in point. On the face of it, even if one strips away the concern with HIV infection, the research could be construed as naive empiricism and perhaps of a rather prurient kind. However, this again would be a harsh and probably inaccurate judgement. For example, the authors relate their research fi ndings to the literature

reporting other investigations of prostitutes in a number of different countries. They also illuminate their fi ndings by drawing on ideas that are very much part of the soci-ologist’s conceptual tool kit. One example is Goffman’s (1963) notion of ‘stigma’ and the way in which the stig-matized individual seeks to manage a spoiled identity; another is Hochschild’s (1983) concept of ‘emotional labour’, a term she coined to denote the way in which air-line fl ight attendants need to express positive emotions as part of the requirements for their jobs. In doing so,

Research in focus 2.3Background literature as theory: emotional labour and hairstylistsOne component of R. S. Cohen’s (2010) mixed methods study of hairstylists’ relationships with their clients was a postal questionnaire survey of all salons and barbers’ shops in a northern city in England. Of the 328 enterprises contacted, 40 per cent replied to the questionnaire. The goal of the research was to examine how far the giving of emotional favours was affected by the nature of the relationship with the client in terms of whether the worker was an owner or a paid employee. Her survey data show that owners are more likely to stay late for clients and to try to fi nd a space for them between clients who have been booked in. Hochschild’s (1983) book, in which she fi rst coined the term ‘emotional labour’, and the many studies that have taken up this concept form the starting point of Cohen’s research. The signifi cance of this work is evident from Cohen’s two opening sentences:

Since Hochschild (1983) fi rst suggested that interactive service workers carry out emotional labour in the course of their work, this proposition has become widely accepted. However the relationship of emotional labour, and client–worker social interactions more generally, to the structural relations of employment has received surprisingly little attention . . . (R. S. Cohen 2010: 197)

Thus, the literature on emotional labour forms the background to the study and the main impetus for the interpretation of the fi ndings, some of which are gleaned from qualitative data deriving from semi-structured interviews with some owners and employees. For the latter, interactions with clients are much more likely to take the form of what Hochschild (1983) called ‘surface acting’, a superfi cial form of emotional labour and emotional engagement with the client.

Key concept 2.1 What is empiricism?The term ‘empiricism’ is used in a number of different ways, but two stand out. First, it is used to denote a general approach to the study of reality that suggests that only knowledge gained through experience and the senses is acceptable. In other words, this position means that ideas must be subjected to the rigours of testing before they can be considered knowledge. The second meaning of the term is related to this and refers to a belief that the accumulation of ‘facts’ is a legitimate goal in its own right. It is this second meaning that is sometimes referred to as ‘naive empiricism’.

9780199588053_C02.indd 23 10/20/11 9:59 AM


they contrive a demeanour of friendliness when deal-ing with passengers, some of whom may be extremely diffi cult (see also Research in focus 2.3).

It is not possible to tell from McKeganey and Barnard’s (1996) report whether the concepts of stigma and emo-tional labour infl uenced their data collection. However, raising this question invites consideration of another question: in so far as any piece of research is linked to theory, what was the role of that theory? Up to this point, I have tended to write as though theory is something that guides and infl uences the collection and analysis of data. In other words, research is done in order to answer questions posed by theoretical considerations. But an alternative position is to view theory as something that occurs after the collection and analysis of some or all of the data associated with a project. We begin to see here the signifi cance of a second factor in consider-ing the relationship between theory and research—whether we are referring to deductive or inductive theory.

Deductive and inductive theoryDeductive theory represents the commonest view of the nature of the relationship between theory and social research. The researcher, on the basis of what is known about in a particular domain and of theoretical consider-ations in relation to that domain, deduces a hypothesis (or hypotheses) that must then be subjected to empirical scrutiny. Embedded within the hypothesis will be con-cepts that will need to be translated into researchable entities. The social scientist must both skillfully deduce a hypothesis and then translate it into operational terms. This means that the social scientist needs to specify how data can be collected in relation to the concepts that make up the hypothesis.

This view of the role of theory in relation to research is very much the kind of role that Merton had in mind in connection with middle-range theory, which, he argued, ‘is principally used in sociology to guide empirical inquiry’ (Merton 1967: 39). Theory and the hypothesis deduced from it come fi rst and drive the process of gath-ering data (see Research in focus 2.4 for an example of a deductive approach to the relationship between theory and data). The sequence can be depicted as one in which the steps outlined in Figure 2.1 take place.

The last step involves a movement that is in the oppo-site direction from deduction—it involves induction, as the researcher infers the implications of his or her fi nd-ings for the theory that prompted the whole exercise. The fi ndings are fed back into the stock of theory and the

research fi ndings associated with a certain domain of enquiry. This can be seen in the case of the fi nal refl ec-tions of Butler and Robson’s (2001—see Research in focus 2.1) study of gentrifi cation in three areas of London when they write:

Figure 2.1Figure 2.1The process of deduction

1. Theory

3. Data collection

4. Findings

5. Hypotheses confirmed or rejected

6. Revision of theory

2. Hypothesis

Each of the three groups has played on its strengths, where it has them. Gentrifi cation, given this, cannot in any sense be considered to be a unitary phenomenon, but needs to be examined in each case according to its own logic and outcomes. The concept of social capital, when used as an integrated part of an extended conceptual framework for the apprehension of all forms of middle-class capital relations, can thus play an important part in discriminating between differing types of social phenomena. (Butler and Robson 2001: 2160)

9780199588053_C02.indd 24 10/20/11 9:59 AM


In these fi nal refl ections they show how their fi ndings and the interpretations of those fi ndings can be fed back into both the stock of knowledge concerning gentrifi ca-tion in cities and, in the third of the three sentences, the concept of social capital and its uses.

However, while this element of inductiveness un-doubtedly exists in the approach outlined, it is typically deemed to be predominantly deductive in orientation. Moreover, it is important to bear in mind that, when this deductive approach, which is usually associated with quantitative research, is put into operation, it often does not follow the sequence outlined in its pure form. As pre-viously noted, ‘theory’ may be little more than the litera-ture on a certain topic in the form of the accumulated knowledge gleaned from books and articles. Also, even when theory or theories can be discerned, explicit hypo-theses are not always deduced from them in the way that Kelley and De Graaf (1997) did in Research in focus 2.4. A further point to bear in mind is that the deductive

process appears very linear—one step follows the other in a clear, logical sequence. However, there are many instances where this is not the case: a researcher’s view of the theory or literature may have changed as a result of the analysis of the collected data; new theoretical ideas or fi ndings may be published by others before the researcher has generated his or her fi ndings; or, the relev-ance of a set of data for a theory may become apparent after the data have been collected.

This may all seem rather surprising and confusing. There is a certain logic to the idea of developing theories and then testing them. In everyday contexts, we com-monly think of theories as things that are quite illumin-ating but that need to be tested before they can be considered valid or useful. In point of fact, however, while the process of deduction outlined in Figure 2.1 does undoubtedly occur, it is better considered as a general orientation to the link between theory and research. As a general orientation, its broad contours may

Research in focus 2.4A deductive studyKelley and De Graaf (1997) show that a number of studies have examined the factors that have an impact upon individuals’ religious beliefs, such as parents, schools, and friends, but they also argue that there are good grounds for thinking that the nation into which one is born will be an important cross-cultural factor. These refl ections constitute what they refer to as the ‘theory’ that guided their research and from which the following hypothesis was derived: ‘People born into religious nations will, in proportion to the orthodoxy of their fellow-citizens, acquire more orthodox beliefs than otherwise similar people born into secular nations’ (Kelley and De Graaf 1997: 641). There are two central concepts in this hypothesis that would need to be measured: national religiosity (whether it is religious or secular) and individual religious orthodoxy. The authors hypothesized further that the religious orientation of the individual’s family (whether devout or secular) would affect the nature of the relationship between national religiosity and religious orthodoxy.

To test the hypotheses, a secondary analysis of data deriving from survey research based on large samples from fi fteen nations was conducted. UK readers will be interested to know that the British and Northern Irish (and Irish Republic) data were derived from the British Social Attitudes survey for 1991 (Jowell et al. 1992). Religious orthodoxy was measured by four survey questions concerned with religious belief. The questions asked about (1) whether the person believed in God, (2) his or her past beliefs about God, (3) how close the individual felt to God, and (4) whether he or she felt that God cares about everyone. To measure national religiosity, the fi fteen nations were classifi ed into one of fi ve categories ascending from secular to religious. The classifi cation was undertaken according to ‘an unweighted average of parental church attendance . . . and religious belief in the nation as a whole’ (Kelley and De Graaf 1997: 647). Family religious orientation was measured on a scale of fi ve levels of parental church attendance. The hypotheses were broadly confi rmed and the authors conclude that the ‘religious environment of a nation has a major impact on the beliefs of its citizens’ (Kelley and De Graaf 1997: 654). Some of the implications of the fi ndings for theories about international differences in religiosity are then outlined.

This study demonstrates the process whereby hypotheses are deduced from existing theory and these then guide the process of data collection so that they can be tested.

9780199588053_C02.indd 25 10/20/11 9:59 AM


frequently be discernible in social research, but it is also the case that we often fi nd departures from it. However, in some research no attempt is made to follow the sequence outlined in Figure 2.1. Some researchers prefer an ap-proach to the relationship between theory and research that is primarily inductive. With an inductive stance, theory is the outcome of research. In other words, the process of induction involves drawing generalizable inferences out of observations. Figure 2.2 attempts to cap-ture the essence of the difference between inductivism and deductivism.

However, just as deduction entails an element of induction, the inductive process is likely to entail a modi-cum of deduction. Once the phase of theoretical refl ection on a set of data has been carried out, the researcher may want to collect further data in order to establish the con-ditions in which a theory will and will not hold. Such a general strategy is often called iterative: it involves a weaving back and forth between data and theory. It is particularly evident in grounded theory, which will be examined in Chapter 24, but in the meantime the basic point is to note that induction represents an alternative strategy for linking theory and research, although it contains a deductive element too.

Figure 2.2Figure 2.2Deductive and inductive approaches tothe relationship between theory andresearch

Inductive approach

Deductive approach

Observations/Findings

Observations/Findings

Theory

Theory

Research in focus 2.5An inductive studyCharmaz (1991, 1997) has been concerned to examine a number of aspects of the experiences of people with chronic illness. One phase of her research entailed the examination specifi cally of men with such a condition. In one of her reports (Charmaz 1997), she discusses the results of her research into twenty men suffering from chronic illness. The bulk of her data derives from semi-structured interviews. In order to bring out the distinctiveness of men’s responses, she compared the fi ndings relating to men with a parallel study of women with chronic illness. She argues that a key component of men’s responses is that of a strategy of preserving self. Although the experience of chronic illness invariably necessitates a change of lifestyle that itself occasions a change in personal identity, the men sought to preserve their sense of self by drawing on ‘essential qualities, attributes, and identities of [the] past self’ (Charmaz 1997: 49). By contrast, women were less reliant in their strategies of preserving self on the recapturing of past identities. She relates her theoretical refl ections of her data to her male respondents’ notions of masculine identity. Her emphasis on the idea of preserving self allows her to assess the factors that lie behind whether a man with chronic illness will ‘reconstruct a positive identity or sink into depression’ (Charmaz 1997: 57). If they were unable to have access to actions that would allow their sense of past self to be extended into the future (for example, through work), the probability of their sinking into depression was enhanced.

In this study, the inductive nature of the relationship between theory and research can be seen in the way that Charmaz’s theoretical ideas (such as the notion of ‘preserving self’) derive from her data rather than being formed before she had collected her data.

9780199588053_C02.indd 26 10/20/11 9:59 AM


However, as with ‘theory’ in connection with the deductive approach to the relationship between theory and research, we have to be cautious about the use of the term in the context of the inductive strategy too. While some researchers undoubtedly develop theories, equally it is necessary to be aware that very often what one ends up with can be little more than empirical gen-eralizations of the kind Merton (1967) wrote about. Research in focus 2.5 is an example of research that can be classifi ed as inductive in the sense that it develops a theory out of interview data deriving from men suffering from chronic illness concerning what determines suc-cessful coping mechanisms for males affl icted with such a condition. In fact, the analytic strategy adopted by the author (Charmaz 1997) was grounded theory, and it is certainly the case that many of the most prominent examples of inductive research derive from this tradition (see the other chapters in Strauss and Corbin 1997b, from which Charmaz’s example was taken).

Charmaz’s (1997) research is an interesting illustra-tion of an inductive approach. Two points are particu-larly worth noting about it. First, as previously noted, it uses a grounded theory approach to the analysis of data and to the generation of theory. This approach, which was fi rst outlined by Glaser and Strauss (1967), is often regarded as especially strong in terms of generating theories out of data. This contrasts with the nature of

many supposedly inductive studies, which generate inter-esting and illuminating fi ndings but whose theoretical signifi cance is not entirely clear. They provide insightful empirical generalizations, but little theory. Secondly, in much the same way that the deductive strategy is associ-ated with a quantitative research approach, an inductive strategy of linking data and theory is typically associated with a qualitative research approach. It is not a coinci-dence that Charmaz’s (1997) research referred to in Research in focus 2.5 is based on in-depth, semi-structured interviews that produced qualitative data in the form of respondents’ detailed answers to her questions. However, as will be shown below, this characterization of the in-ductive strategy as associated with qualitative research is not entirely straightforward: not only does much qualita-tive research not generate theory, but also theory is often used at the very least as a background to qualitative investigations.

It is useful to think of the relationship between theory and research in terms of deductive and inductive strat-egies. However, as the previous discussion has implied, the issues are not as clear-cut as they are sometimes presented. To a large extent, deductive and inductive strategies are possibly better thought of as tendencies rather than as a hard-and-fast distinction. But these are not the only issues that impinge on the conduct of social research.

Epistemological considerations

An epistemological issue concerns the question of what is (or should be) regarded as acceptable knowledge in a discipline. A particularly central issue in this context is the question of whether the social world can and should be studied according to the same principles, procedures, and ethos as the natural sciences. The position that affi rms the importance of imitating the natural sciences is invariably associated with an epistemological position known as positivism (see Key concept 2.2).

A natural science epistemology: positivismThe doctrine of positivism is extremely diffi cult to pin down and therefore to outline in a precise manner, be-cause it is used in a number of different ways by authors. For some writers, it is a descriptive category—one that

describes a philosophical position that can be discerned in research—though there are still disagreements about what it comprises; for others, it is a pejorative term used to describe crude and often superfi cial data collection.

It is possible to see in the fi ve principles in Key con-cept 2.2 a link with some of the points that have already been raised about the relationship between theory and research. For example, positivism entails elements of both a deductive approach (principle 2) and an inductive strategy (principle 3). Also, a fairly sharp distinction is drawn between theory and research. The role of research is to test theories and to provide material for the develop-ment of laws. But either of these connections between theory and research carries with it the implication that it is possible to collect observations in a manner that is not infl uenced by pre-existing theories. Moreover, theoret-ical terms that are not directly amenable to observation

9780199588053_C02.indd 27 10/20/11 9:59 AM


Key concept 2.2 What is positivism?Positivism is an epistemological position that advocates the application of the methods of the natural sciences to the study of social reality and beyond. But the term stretches beyond this principle, though the constituent elements vary between authors. However, positivism is also taken to entail the following principles:

1. Only phenomena and hence knowledge confi rmed by the senses can genuinely be warranted as knowledge (the principle of phenomenalism).

2. The purpose of theory is to generate hypotheses that can be tested and that will thereby allow explanations of laws to be assessed (the principle of deductivism).

3. Knowledge is arrived at through the gathering of facts that provide the basis for laws (the principle of inductivism).

4. Science must (and presumably can) be conducted in a way that is value free (that is, objective).

5. There is a clear distinction between scientifi c statements and normative statements and a belief that the former are the true domain of the scientist. This last principle is implied by the fi rst because the truth or otherwise of normative statements cannot be confi rmed by the senses.

are not considered genuinely scientifi c; they must be sus-ceptible to the rigours of observation. All this carries with it the implication of greater epistemological status being given to observation than to theory.

It should be noted that it is a mistake to treat positivism as synonymous with science and the scientifi c. In fact, philosophers of science and of the social sciences differ quite sharply over how best to characterize scientifi c practice, and since the early 1960s there has been a drift away from viewing it in positivist terms. Thus, when writers complain about the limitations of positivism, it is not entirely clear whether they mean the philosophical term or a scientifi c approach more generally. Realism (in particular, critical realism), for example, is another philo-sophical position that purports to provide an account of the nature of scientifi c practice (see Key concept 2.3).

The crux of the epistemological considerations that form the central thrust of this section is the rejection by some writers and traditions of the application of the canons of the natural sciences to the study of social reality. A diffi culty here is that it is not easy to disentangle the natural science model from positivism as the butt of their criticisms. In other words, it is not always clear whether they are inveighing against the application of a general natural scientifi c approach or of positivism in particular. There is a long-standing debate about the appropriate-ness of the natural science model for the study of society, but, since the account that is offered of that model tends to have largely positivist overtones, it would seem that it is positivism that is the focus of attention rather than

other accounts of scientifi c practice (such as critical realism—see Key concept 2.3).

InterpretivismInterpretivism is a term given to a contrasting epistem-ology to positivism (see Key concept 2.4). The term subsumes the views of writers who have been critical of the application of the scientifi c model to the study of the social world and who have been infl uenced by different intellectual traditions, which are outlined below. They share a view that the subject matter of the social sciences —people and their institutions—is fundamentally differ-ent from that of the natural sciences. The study of the social world therefore requires a different logic of re-search procedure, one that refl ects the distinctiveness of humans as against the natural order. Von Wright (1971) has depicted the epistemological clash as being between positivism and hermeneutics (a term that is drawn from theology and that, when imported into the social sci-ences, is concerned with the theory and method of the interpretation of human action). This clash refl ects a divi-sion between an emphasis on the explanation of human behaviour that is the chief ingredient of the positivist approach to the social sciences and the understanding of human behaviour. The latter is concerned with the empathic understanding of human action rather than with the forces that are deemed to act on it. This contrast refl ects long-standing debates that precede the emergence of the modern social sciences but fi nd their expression in

9780199588053_C02.indd 28 10/20/11 9:59 AM


such notions as the advocacy by Max Weber (1864–1920) of an approach referred to in his native German as Verstehen (which means understanding). Weber (1947: 88) described sociology as a ‘science which attempts the

interpretive understanding of social action in order to arrive at a causal explanation of its course and effects’. Weber’s defi nition seems to embrace both explanation and understanding here, but the crucial point is that the

Key concept 2.3 What is realism?Realism shares two features with positivism: a belief that the natural and the social sciences can and should apply the same kinds of approach to the collection of data and to explanation, and a commitment to the view that there is an external reality to which scientists direct their attention (in other words, there is a reality that is separate from our descriptions of it). There are two major forms of realism:

• Empirical realism simply asserts that, through the use of appropriate methods, reality can be understood. This version of realism is sometimes referred to as naive realism to refl ect the fact that it is often assumed by realists that there is a perfect (or at least very close) correspondence between reality and the term used to describe it. As such, it ‘fails to recognise that there are enduring structures and generative mechanisms underlying and producing observable phenomena and events’ and is therefore ‘superfi cial’ (Bhaskar 1989: 2). This is perhaps the most common meaning of the term. When writers employ the term ‘realism’ in a general way, it is invariably this meaning to which they are referring.

• Critical realism is a specifi c form of realism whose manifesto is to recognize the reality of the natural order and the events and discourses of the social world and holds that ‘we will only be able to understand—and so change—the social world if we identify the structures at work that generate those events and discourses. . . . These structures are not spontaneously apparent in the observable pattern of events; they can only be identifi ed through the practical and theoretical work of the social sciences’ (Bhaskar 1989: 2).

Critical realism implies two things. First, it implies that, whereas positivists take the view that the scientist’s conceptualization of reality actually directly refl ects that reality, realists argue that the scientist’s conceptualization is simply a way of knowing that reality. As Bhaskar (1975: 250) has put it: ‘Science, then, is the systematic attempt to express in thought the structures and ways of acting of things that exist and act independently of thought.’ Critical realists acknowledge and accept that the categories they employ to understand reality are likely to be provisional. Thus, unlike naive realists, critical realists recognize that there is a distinction between the objects that are the focus of their enquiries and the terms they use to describe, account for, and understand them. Secondly, by implication, critical realists unlike positivists are perfectly content to admit into their explanations theoretical terms that are not directly amenable to observation. As a result, hypothetical entities that account for regularities in the natural or social orders (the ‘generative mechanisms’ to which Bhaskar refers) are perfectly admissible for realists, but not for positivists. Generative mechanisms entail the entities and processes that are constitutive of the phenomenon of interest. For critical realists, it is acceptable that generative mechanisms are not directly observable, since they can be admitted into theoretical accounts on the grounds that their effects are observable. Also crucial to a critical realist understanding is the identifi cation of the context that interacts with the generative mechanism to produce an observed regularity in the social world. An appreciation of context is crucial to critical realist explanations because it serves to shed light on the conditions that promote or impede the operation of the causal mechanism. What makes critical realism critical is that the identifi cation of generative mechanisms offers the prospect of introducing changes that can transform the status quo. A further point to note about critical realism is that the form of reasoning involved in the identifi cation of generative causal mechanisms is neither inductive nor deductive. It is referred to by Blaikie (2004) as retroductive reasoning, which entails making an inference about the causal mechanism that lies behind and is responsible for regularities that are observed in the social world. Research in focus 26.1 provides an example of research using a critical realist approach. This example can be read profi tably at this stage even though it is in a much later chapter.

9780199588053_C02.indd 29 10/20/11 9:59 AM


task of ‘causal explanation’ is undertaken with reference to the ‘interpretive understanding of social action’ rather than to external forces that have no meaning for those involved in that social action.

One of the main intellectual traditions that has been responsible for the anti-positivist position has been phenomenology, a philosophy that is concerned with the question of how individuals make sense of the world around them and how in particular the philosopher should bracket out preconceptions in his or her grasp of that world. The initial application of phenomenological ideas to the social sciences is attributed to the work of Alfred Schutz (1899–1959), whose work did not come to the notice of most English-speaking social scientists until the translation from German of his major writings in the 1960s, some twenty or more years after they had been written. His work was profoundly infl uenced by Weber’s concept of Verstehen, as well as by phenomenological philosophers, like Husserl. Schutz’s position is well cap-tured in the following passage, which has been quoted on numerous occasions:

Two points are particularly noteworthy in this quotation. First, it asserts that there is a fundamental difference between the subject matter of the natural sciences and the social sciences and that an epistemology is required that will refl ect and capitalize upon that difference. The fundamental difference resides in the fact that social reality has a meaning for human beings and therefore human action is meaningful—that is, it has a meaning for them and they act on the basis of the meanings that they attribute to their acts and to the acts of others. This leads to the second point—namely, that it is the job of the social scientist to gain access to people’s ‘common-sense thinking’ and hence to interpret their actions and their social world from their point of view. It is this particular feature that social scientists claiming allegiance to phe-nomenology have typically emphasized. In the words of the authors of a research methods text whose approach is described as phenomenological: ‘The phenomenologist views human behavior . . . as a product of how people interpret the world. . . . In order to grasp the meanings of a person’s behavior, the phenomenologist attempts to see things from that person’s point of view’ (Bogdan and Taylor 1975: 13–14; emphasis in original).

In this exposition of Verstehen and phenomenology, it has been necessary to skate over some complex issues. In particular, Weber’s examination of Verstehen is far more complex than the above commentary suggests, because the empathetic understanding that seems to be implied above was not the way in which he applied it (Bauman 1978), while the question of what is and is not a genu-inely phenomenological approach to the social sciences is a matter of some dispute (Heap and Roth 1973). How-ever, the similarity in the writings of the hermeneutic–phenomenological tradition and of the Verstehen approach, with their emphasis upon social action as being meaning-ful to actors and therefore needing to be interpreted from their point of view, coupled with the rejection of positiv-ism, contributed to a stream of thought often referred to as interpretivism (e.g. J. A. Hughes 1990).

Key concept 2.4 What is interpretivism?Interpretivism is a term that usually denotes an alternative to the positivist orthodoxy that has held sway for decades. It is predicated upon the view that a strategy is required that respects the differences between people and the objects of the natural sciences and therefore requires the social scientist to grasp the subjective meaning of social action. Its intellectual heritage includes: Weber’s notion of Verstehen; the hermeneutic–phenomenological tradition; and symbolic interactionism.

The world of nature as explored by the natural scientist does not ‘mean’ anything to molecules, atoms and electrons. But the observational fi eld of the social scientist—social reality—has a specifi c meaning and relevance structure for the beings living, acting, and thinking within it. By a series of common-sense constructs they have pre-selected and pre-interpreted this world which they experience as the reality of their daily lives. It is these thought objects of theirs which determine their behaviour by motivating it. The thought objects constructed by the social scientist, in order to grasp this social reality, have to be founded upon the thought objects constructed by the common-sense thinking of men [and women!], living their daily life within the social world. (Schutz 1962: 59)

9780199588053_C02.indd 30 10/20/11 9:59 AM


Verstehen and the hermeneutic–phenomenological tradition do not exhaust the intellectual infl uences on interpretivism. The theoretical tradition in sociology known as symbolic interactionism has also been regarded by many writers as a further infl uence. Again, the case is not clear-cut. The implications for empirical research of the ideas of the founders of symbolic interactionism, in particular George Herbert Mead (1863–1931), whose discussion of the way in which our notion of self emerges through an appreciation of how others see us, have been hotly debated. There was a school of research, known as the Iowa school, that drew heavily on Mead’s con-cepts and ideas, but proceeded in a direction that most people would prefer to depict as largely positivist in tone (Meltzer et al. 1975). Moreover, some writers have argued that Mead’s approach is far more consistent with a natural science approach than has typically been recognized (McPhail and Rexroat 1979). However, the general tendency has been to view symbolic inter-actionism as occupying similar intellectual space to the hermeneutic–phenomenological tradition and so as broadly interpretative in approach. This tendency is largely the product of the writings of Herbert Blumer, a student of Mead’s who acted as his mentor’s spokesman and interpreter, and his followers (Hammersley 1989; R. Collins 1994). Not only did Blumer coin the term symbolic interaction; he also provided a gloss on Mead’s writings that has decidedly interpretative overtones. Symbolic interactionists argue that interaction takes place in such a way that the individual is continually interpreting the symbolic meaning of his or her environ-ment (which includes the actions of others) and acts on the basis of this imputed meaning. In research terms, according to Blumer (1962: 188), ‘the position of symbolic interaction requires the student to catch the process of interpretation through which [actors] construct their actions’, a statement that brings out clearly his views of the research implications of symbolic interactionism and of Mead’s thought.

It should be appreciated that the parallelism between symbolic interactionism and the hermeneutic–phenomenological tradition should not be exaggerated. The two are united in their antipathy for positivism and have in common an interpretative stance. However, symbolic interactionism is, at least in the hands of Blumer and the many writers and researchers who have followed in his wake, a type of social theory that has dis-tinctive epistemological implications; the hermeneutic–phenomenological tradition, by contrast, is best thought of as a general epistemological approach in its own right. Blumer may have been infl uenced by the hermeneutic–

phenomenological tradition, but there is no concrete evidence of this. There are other intellectual currents that have affi nities with the interpretative stance, such as the working-through of the ramifi cations of the works of the philosopher Ludwig Wittgenstein (Winch 1958), but the hermeneutic–phenomenological, Verstehen, and symbolic interactionist traditions can be considered major infl uences.

Taking an interpretative stance can mean that the researcher may come up with surprising fi ndings, or at least fi ndings that appear surprising if a largely external stance is taken—that is, a position from outside the par-ticular social context being studied. Research in focus 2.6 provides an interesting example of this possibility.

Of course, as the example in Research in focus 2.6 suggests, when the social scientist adopts an interpreta-tive stance, he or she is not simply laying bare how members of a social group interpret the world around them. The social scientist will almost certainly be aiming to place the interpretations that have been elicited into a social scientifi c frame. There is a double interpretation going on: the researcher is providing an interpretation of others’ interpretations. Indeed, there is a third level of interpretation going on, because the researcher’s inter-pretations have to be further interpreted in terms of the concepts, theories, and literature of a discipline. Thus, taking the example in Research in focus 2.6, Foster’s (1995) suggestion that Riverside is not perceived as a high crime area by residents is her interpretation of her subjects’ interpretations. She then had the additional job of placing her interesting fi ndings into a social scien-tifi c frame, which she accomplished by relating them to existing concepts and discussions in criminology of such things as informal social control, neighbourhood watch schemes, and the role of housing as a possible cause of criminal activity.

The aim of this section has been to outline how epi-stemological considerations—especially those relating to the question of whether a natural science approach, and in particular a positivist one, can supply legitimate knowledge of the social world—are related to research practice. There is a link with the earlier section in that a deductive approach to the relationship between theory and research is typically associated with a positivist posi-tion. Key concept 2.2 does try to suggest that inductivism is also a feature of positivism (third principle), but, in the working-through of its implementation in the practice of social research, it is the deductive element (second prin-ciple) that tends to be emphasized. Similarly, the third level of interpretation that a researcher engaged in inter-pretative research must bring into operation is very much

9780199588053_C02.indd 31 10/20/11 9:59 AM


part of the kind of inductive strategy described in the previous section. However, while such interconnections between epistemological issues and research practice exist, it is important not to overstate them, since they represent tendencies rather than defi nitive points of

correspondence. Thus, particular epistemological prin-ciples and research practices do not necessarily go hand in hand in a neat unambiguous manner. This point will be made again on several occasions and will be a special focus of Chapter 26.

Research in focus 2.6Interpretivism in practiceFoster (1995) conducted ethnographic research using participant observation and semi-structured interviews in a housing estate in East London, referred to as Riverside. The estate had a high level of crime, as indicated by offi cial statistics on crime. However, she found that residents did not perceive the estate to be a high crime area. This perception could be attributed to a number of factors, but a particularly important reason was the existence of ‘informal social control’. People expected a certain level of crime, but felt fairly secure because informal social control allowed levels of crime to be contained. Informal social control comprised a number of different aspects. One aspect was that neighbours often looked out for each other. In the words of one of Foster’s interviewees: ‘If I hear a bang or shouting I go out. If there’s aggravation I come in and ring the police. I don’t stand for it.’ Another aspect of informal social control was that people often felt secure because they knew each other. Another respondent said: ‘I don’t feel nervous . . . because people do generally know each other. We keep an eye on each other’s properties . . . I feel quite safe because you know your neighbours and you know they’re there . . . they look out for you’ (Foster 1995: 575).

Ontological considerations

Questions of social ontology are concerned with the nature of social entities. The central point of orientation here is the question of whether social entities can and should be considered objective entities that have a reality external to social actors, or whether they can and should be considered social constructions built up from the per-ceptions and actions of social actors. These positions are frequently referred to respectively as objectivism and constructionism. Their differences can be illustrated by reference to two of the most common and central terms in social science—organization and culture.

ObjectivismObjectivism is an ontological position that implies that social phenomena confront us as external facts that are beyond our reach or infl uence (see Key concept 2.5).

We can discuss organization or an organization as a tangible object. It has rules and regulations. It adopts standardized procedures for getting things done. People are appointed to different jobs within a division of

labour. There is a hierarchy. It has a mission statement. And so on. The degree to which these features exist from organization to organization is variable, but in thinking in these terms we are tending to the view that an organ-ization has a reality that is external to the individuals who inhabit it. Moreover, the organization represents a social order in that it exerts pressure on individuals to conform to the requirements of the organization. People learn and apply the rules and regulations. They follow the standardized procedures. They do the jobs to which they are appointed. People tell them what to do and they tell others what to do. They learn and apply the values in the mission statement. If they do not do these things, they may be reprimanded or even fi red. The organization is therefore a constraining force that acts on and inhibits its members.

The same can be said of culture. Cultures and subcul-tures can be viewed as repositories of widely shared values and customs into which people are socialized so that they can function as good citizens or as full partici-pants. Cultures and subcultures constrain us because we

9780199588053_C02.indd 32 10/20/11 9:59 AM


internalize their beliefs and values. In the case of both organization and culture, the social entity in question comes across as something external to the actor and as having an almost tangible reality of its own. It has the characteristics of an object and hence of having an objec-tive reality. To a very large extent, these are the ‘classic’ ways of conceptualizing organization and culture.

ConstructionismHowever, we can consider an alternative ontological position—constructionism (Key concept 2.6). This posi-tion challenges the suggestion that categories such as organization and culture are pre-given and therefore confront social actors as external realities that they have no role in fashioning.

Let us take organization fi rst. Strauss et al. (1973), drawing on insights from symbolic interactionism, car-

ried out research in a psychiatric hospital and proposed that it was best conceptualized as a ‘negotiated order’. Instead of taking the view that order in organizations is a pre-existing characteristic, they argue that it is worked at. Rules were far less extensive and less rigorously im-posed than might be supposed from the classic account of organization. Indeed, Strauss et al. (1973: 308) prefer to refer to them as ‘much less like commands, and much more like general understandings’. Precisely because relatively little of the spheres of action of doctors, nurses, and other personnel was prescribed, the social order of the hospital was an outcome of agreed-upon patterns of action that were themselves the products of negotiations between the different parties involved. The social order is in a constant state of change because the hospital is ‘a place where numerous agreements are continually being terminated or forgotten, but also as continually being established, renewed, reviewed, revoked, revised. . . . In

Key concept 2.5What is objectivism?Objectivism is an ontological position that asserts that social phenomena and their meanings have an existence that is independent of social actors. It implies that social phenomena and the categories that we use in everyday discourse have an existence that is independent or separate from actors.

Key concept 2.6What is constructionism?Constructionism is an ontological position (often also referred to as constructivism) that asserts that social phenomena and their meanings are continually being accomplished by social actors. It implies that social phenomena and categories are not only produced through social interaction but that they are in a constant state of revision. In recent years, the term has also come to include the notion that researchers’ own accounts of the social world are constructions. In other words, the researcher always presents a specifi c version of social reality, rather than one that can be regarded as defi nitive. Knowledge is viewed as indeterminate, a position redolent of postmodernism (see Key concept 17.1, which further examines this viewpoint). This sense of constructionism is usually allied to the ontological version of the term. In other words, these are linked meanings. Both meanings are antithetical to objectivism (see Key concept 2.5), but the second meaning is also antithetical to realism (see Key concept 2.3). The fi rst meaning might be thought of usefully as constructionism in relation to the social world; the second as constructionism in relation to the nature of knowledge of the social world (and indeed the natural world).

Increasingly, the notion of constructionism in relation to the nature of knowledge of the social world is being incorporated into notions of constructionism, but in this book I will be using the term in relation to the fi rst meaning, whereby constructionism is presented as an ontological position in relating to social objects and categories—that is, one that views them as socially constructed.

9780199588053_C02.indd 33 10/20/11 9:59 AM


any pragmatic sense, this is the hospital at the moment: this is its social order’ (Strauss et al. 1973: 316–17). The authors argue that a preoccupation with the formal prop-erties of organizations (rules, organizational charts, regu-lations, roles) tends to neglect the degree to which order in organizations has to be accomplished in everyday interaction, though this is not to say that the formal properties have no element of constraint on individual action.

Much the same kind of point can be made about the idea of culture. Instead of seeing culture as an external reality that acts on and constrains people, it can be taken to be an emergent reality in a continuous state of construction and reconstruction. Becker (1982: 521), for example, has suggested that ‘people create culture continuously. . . . No set of cultural understandings . . . provides a perfectly applicable solution to any problem people have to solve in the course of their day, and they therefore must remake those solutions, adapt their under-standings to the new situation in the light of what is different about it.’ Like Strauss et al., Becker recognizes that the constructionist position cannot be pushed to the extreme: it is necessary to appreciate that culture has a reality that ‘persists and antedates the participation of particular people’ and shapes their perspectives, but it is not an inert objective reality that possesses only a sense of constraint: it acts as a point of reference but is always in the process of being formed.

Neither the work of Strauss et al. nor that of Becker pushes the constructionist argument to the extreme. Each admits to the pre-existence of their objects of inter-est (organization and culture respectively). However, in each case we see an intellectual predilection for stressing the active role of individuals in the social construction of social reality. Not all writers adopting a constructionist position are similarly prepared to acknowledge the exis-tence or at least importance of an objective reality. Walsh (1972: 19), for example, has written that ‘we cannot take for granted, as the natural scientist does, the availability of a preconstituted world of phenomena for investiga-tion’ and must instead ‘examine the processes by which the social world is constructed’. Constructionism essen-tially invites the researcher to consider the ways in which social reality is an ongoing accomplishment of social actors rather than something external to them and that totally constrains them.

Constructionism also suggests that the categories that people employ in helping them to understand the natural and social world are in fact social products. The categories do not have built-in essences; instead, their meaning is constructed in and through interaction. Thus, a category

like ‘masculinity’ might be treated as a social construc-tion. This notion implies that, rather than being treated as a distinct inert entity, masculinity is construed as something whose meaning is built up during interaction. That meaning is likely to be a highly ephemeral one, in that it will vary by both time and place. This kind of stance frequently displays a concern with the language that is employed to present categories in particular ways. It suggests that the social world and its categories are not external to us, but are built up and constituted in and through interaction. This tendency can be seen particu-larly in discourse analysis, which is examined in Chap-ter 22. As Potter (1996: 98) observes: ‘The world . . . is constituted in one way or another as people talk it, write it and argue it.’ This sense of constructionism is highly antithetical to realism (see Key concept 2.3). Constructionism frequently results in an interest in the representation of social phenomena. Research in focus 2.7 provides an illustration of this idea in relation to the representation of the breast cancer epidemic in the USA.

Constructionism is also frequently used as a term that refl ects the indeterminacy of our knowledge of the social world (see Key concept 2.6 and the idea of construction-ism in relation to the nature of knowledge of the social world). However, in this book, I will be using the term in connection with the notion that social phenomena and categories are social constructions.

Relationship to social researchQuestions of social ontology cannot be divorced from issues concerning the conduct of social research. Onto-logical assumptions and commitments will feed into the ways in which research questions are formulated and research is carried out. If a research question is for-mulated in such a way as to suggest that organizations and cultures are objective social entities that act on indi-viduals, the researcher is likely to emphasize the formal properties of organizations or the beliefs and values of members of the culture. Alternatively, if the researcher formulates a research question so that the tenuousness of organization and culture as objective categories is stressed, it is likely that an emphasis will be placed on the active involvement of people in reality construction. In either case, it might be supposed that different approaches to the design of research and the collection of data will be required. Later in the book, Research in focus 20.8 provides an illustration of a study with a strong commitment to a constructionist ontology and its implications for the research process.

9780199588053_C02.indd 34 10/20/11 9:59 AM


Research in focus 2.7Constructionism in actionLantz and Booth (1998) have shown that breast cancer can be treated as a social construction. They note that US data show a rise in the incidence of the disease since the early 1980s, which has led to the depiction of the trend as an epidemic. The authors examined a variety of popular magazines using qualitative content analysis (see Key concept 13.1 for a brief description of this m ethod). They note that many of the articles draw attention to the lifestyles of modern women, such as delaying fi rst births, diet and alcohol consumption, and having careers. The authors argue that the articles

ascribe blame to individual behaviors by listing a wide array of individual risk factors (many of which are not behaviors of ‘traditional’ women), and then offering prudent prescriptions for prevention. Women are portrayed as victims of an insidious disease, but also as victims of their own behaviors, many of which are related to the control of their own fertility. . . . These articles suggest that nontraditional women experience pathological repercussions within their bodies and, in turn, may be responsible for our current epidemic of breast cancer. (Lantz and Booth 1998: 915–16)

This article suggests that, as a social category, the breast cancer epidemic is being represented in popular magazines in a particular way—one that blames the victims and the lifestyles of modern women in particular. This is in spite of the fact that fewer than 20 per cent of cases of breast cancer are in women under the age of 50. Lantz and Booth’s study is fairly representative of a constructionist ontology in suggesting that the epidemic is not simply being construed as a social fact but is being ascribed a particular meaning (one that blames the victims of the disease). In this way, the representation of the disease in popular magazines forms an important element in its social construction.

Research strategy: quantitative and qualitative research

Many writers on methodological issues fi nd it helpful to distinguish between quantitative research and quali-tative research. The status of the distinction is ambigu-ous, because it is almost simultaneously regarded by some writers as a fundamental contrast and by others as no longer useful or even simply as ‘false’ (Layder 1993: 110). However, there is little evidence to suggest that the use of the distinction is abating and even consider-able evidence of its continued, even growing, cur-rency. The quantitative/qualitative distinction will be employed a great deal in this book, because it repres-ents a useful means of classifying different methods of social research and because it is a helpful umbrella for a range of issues concerned with the practice of social research.

On the face of it, there would seem to be little to the quantitative/qualitative distinction other than the fact that quantitative researchers employ measurement and qualitative researchers do not. It is certainly the case

that there is a predisposition among researchers along these lines, but many writers have suggested that the differences are deeper than the superfi cial issue of the presence or absence of quantifi cation. For many writers, quantitative and qualitative research differ with respect to their epistemological foundations and in other re-spects too. Indeed, if we take the areas that have been the focus of the previous three sections—the connection between theory and research, epistemological consider-ations, and ontological considerations—quantitative and qualitative research can be taken to form two distinctive clusters of research strategy. By a research strategy, I simply mean a general orientation to the conduct of social research. Table 2.1 outlines the differences between quantitative and qualitative research in terms of the three areas.

Thus, quantitative research can be construed as a research strategy that emphasizes quantifi cation in the collection and analysis of data and that

9780199588053_C02.indd 35 10/20/11 9:59 AM


• entails a deductive approach to the relationship between theory and research, in which the accent is placed on the testing of theories;

• has incorporated the practices and norms of the natural scientifi c model and of positivism in particular; and

• embodies a view of social reality as an external, objec-tive reality.

By contrast, qualitative research can be construed as a research strategy that usually emphasizes words rather than quantifi cation in the collection and analysis of data and that

• predominantly emphasizes an inductive approach to the relationship between theory and research, in which the emphasis is placed on the generation of theories;

• has rejected the practices and norms of the natural sci-entifi c model and of positivism in particular in prefer-ence for an emphasis on the ways in which individuals interpret their social world; and

• embodies a view of social reality as a constantly shift-ing emergent property of individuals’ creation.

There is, in fact, considerably more to the quantitative/qualitative distinction than this contrast. In Chapters 7 and 17 the nature of quantitative and then qualitative research respectively will be outlined in much greater detail, while in Chapters 26 and 27 the contrasting features will be further explored. In particular, a number of distinguishing features fl ow from the commitment of the quantitative research strategy to a positivist epistem-ology and from the rejection of that epistemology by practitioners of the qualitative research strategy. In other words, the three contrasts in Table 2.1 are basic, though fundamental, ones.

However, the interconnections between the differ-ent features of quantitative and qualitative research are

not as straightforward as Table 2.1 and the previous paragraph imply. While it is useful to contrast the two research strategies, it is necessary to be careful about hammering a wedge between them. It may seem per-verse to introduce a basic set of distinctions and then suggest that they are problematic. A recurring theme of this book is that discussing the nature of social research is just as complex as conducting research in the real world. You may discover general tendencies, but they are precisely that—tendencies. In reality, the picture becomes more complicated the more you delve.

For example, it is common to describe qualitative research as concerned with the generation rather than the testing of theories. However, there are examples of studies in which qualitative research has been employed to test rather than to generate theories. For example, Adler and Adler (1985) were concerned to explore the issue of whether participation in athletics in higher edu-cation in the USA is associated with higher or lower levels of academic achievement, an issue on which the existing literature was inconsistent. This is an illustration of the use of the existing literature on a topic being employed as a kind of proxy for theory. The fi rst author was a partici-pant observer for four years of a basketball programme in a university, and both authors carried out ‘intensive, taped interviews’ with players. The authors’ fi ndings do lead them to conclude that athletic participation is likely to result in lower academic achievement. This occurs because the programme participants gradually drift from idealistic goals about their academic careers, and a variety of factors lead them to become increasingly detached from academic work. For example, one student is quoted as saying: ‘If I was a student like most other students I could do well, but when you play the calibre of ball we do, you just can’t be an above-average student. What I strive for now is just to be an average student. . . . You just can’t fi nd the time to do all the reading’ (Adler and Adler 1985: 247). This study shows how, although

Table 2.1Fundamental differences between quantitative and qualitative research strategies

Quantitative Qualitative

Principal orientation to the role of theory in relation to research

Deductive; testing of theory Inductive; generation of theory

Epistemological orientation Natural science model, in particular positivism Interpretivism

Ontological orientation Objectivism Constructionism

9780199588053_C02.indd 36 10/20/11 9:59 AM


qualitative research is typically associated with generat-ing theories, it can also be employed for testing them.

Moreover, it is striking that, although the Adler and Adler study is broadly interpretivist in epistemological orientation, with its emphasis on how college athletes view their social situation, the fi ndings have objectivist, rather than constructionist, overtones. For example, when the authors describe the students’ academic performance as ‘determined less by demographic characteristics and high school experiences than by the structure of their college experiences’ (Adler and Adler 1985: 249), they are positing a social world that is ‘out there’ and that has a formal, objective quality. It is an example of qualitative research in the sense that there is no quantifi cation or very little of it, but it does not have all the other features outlined in Table 2.1. Similarly, the previously mentioned study by Westergaard et al. (1989) of the effects of redundancy was a quantitative study in the sense of being concerned to measure a wide variety of concepts, but exhibited little evidence of a concern to test theor-ies of unemployment or of a stressful life event like redundancy. Instead, its conclusions revolve around seek-

ing to understand how those made redundant responded to the experience in terms of such things as their job-search methods, their inclination to fi nd jobs, and their political attitudes. As such, it has interpretivist overtones in spite of being an exercise in quantitative research.

The point that is being made in this section is that quantitative and qualitative research represent different research strategies and that each carries with it striking differences in terms of the role of theory, epistemological issues, and ontological concerns. However, the distinc-tion is not a hard-and-fast one: studies that have the broad characteristics of one research strategy may have a characteristic of the other. I will say more about the common features in quantitative and qualitative research in Chap ter 26. Not only this, but many writers argue that the two can be combined within an overall research project, and Chapter 27 examines precisely this possibil-ity. In Chapter 27, I will examine what is increasingly referred to as mixed methods research. This term is widely used nowadays to refer to research that combines methods associated with both quantitative and qualita-tive research.

Research in focus 2.8Mixed methods research—an exampleIn 2001, Britain was profoundly affected by the Foot and Mouth Disease (FMD), which had a big impact on people’s movements. Poortinga et al. (2004) were interested in how far the public trusted the information the government was supplying and how it perceived the risks associated with the disease. Such issues were of interest in part because the researchers felt that the ways in which the public responds to a crisis was an important topic, but also because the issues connect with the infl uence in recent years of the notion of the ‘risk society’ (Beck 1992), which has attracted a good deal of sociological attention. At the height of the disease during 2–5 April 2001, the researchers conducted a survey by administering a self-completion questionnaire (see Chapter 10) to samples in two contrasting areas: Bude in Cornwall and Norwich in Norfolk. These two areas were chosen because they were very differently affected by FMD. The questionnaire covered the following areas: level of agreement with statements about the outbreak of the disease (for example, ‘My main concerns about FMD are to do with the possible impacts on human health’); perceptions of who was to blame; level of agreement with statements about the government’s handling of FMD; degrees of trust in various sources of information about the disease; and personal information, such as any connection with the farming or tourist industries. In addition, a qualitative research method—focus groups (see Chapter 21)—was employed. In May and June 2001, these groups were convened and members of the groups were asked about the same kinds of issues covered in the questionnaire. Focus group participants were chosen from among those who had indicated in their questionnaire replies that they were willing to be involved in a focus group discussion. Three focus group discussions took place. While the questionnaire data were able to demonstrate the variation in such things as trust in various information sources, the focus groups revealed ‘valuable additional information, especially on the reasons, rationalizations and arguments behind people’s understanding of the FMD issue’ (Poortinga et al. 2004: 86). As a result, the researchers were able to arrive at a more complete account of the FMD crisis than could have been obtained by either a quantitative or a qualitative research approach alone. This and other possible advantages of mixed methods research will be explored further in Chapter 27.

9780199588053_C02.indd 37 10/20/11 9:59 AM


In Research in focus 2.8 and 2.9, I present examples of mixed methods studies. I am presenting them here partly to provide some early insights into the possibility of doing mixed methods research, but also to show how a wedge need not and should not be driven between

quantitative and qualitative research. By contrasting the two approaches, it is easy to see them as incompatible. As the examples in Research in focus 2.8 and 2.9 show, they can be fruitfully combined within a single project. This point will be amplifi ed throughout Chapter 27.

Research in focus 2.9Mixed methods research—an exampleThis second example of mixed methods research is probably one of the biggest studies in the UK using the approach—the Cultural Capital and Social Exclusion (CCSE) project. Like the research referred to in Research in Focus 2.1, the CCSE project was profoundly infl uenced by Bourdieu and in particular by his infl uential research on cultural capital and its role in the reproduction of social divisions (Bourdieu 1984). While the CCSE project was inspired by Bourdieu’s research, at the same time the researchers had some reservations about the methodological approach taken, the theoretical approach, and its relevance beyond the period in which the research was conducted and its milieu (France). The research was designed around three research questions:

• ‘What is the nature of cultural capital in Britain? What kinds of social exclusion are generated by the differential distribution of cultural capital across class positions?’

• ‘What are the relationships between economic capital, social capital and cultural capital, in particular how is cultural capital related to other forms of capital?’

• ‘What role does cultural capital play in relation to existing patterns of social exclusion? How can a closer knowledge of this assist in developing cultural policies designed to offset the effects of social exclusion?’(www.open.ac.uk/socialsciences/cultural-capital-and-social-exclusion/research-questions.php, emphasis removed (accessed 13 August 2010))

Each of these three research questions was broken down into several subquestions. In order to address these research questions, the authors employed three main research methods:

1. Twenty-fi ve focus groups, with each group being made up of a distinctive group of members, for example, Pakistani middle class, supervisors, self-employed.

2. A structured interview survey of a large representative sample of 1,781 respondents within the UK.

3. Semi-structured interviews with 44 individuals from 30 households. The interviewees were sampled from the survey on the basis of socio-demographic and cultural capital characteristics. The interviewers also took notes about the households. In addition, 11 interviews were conducted with ‘elite’ individuals, because it was felt that these were not suffi ciently present in the sample.

Thus, the CCSE project comprised two qualitative research methods (focus groups and semi-structured interviewing) and one quantitative method (a structured interview survey). The mixed methods aspect of this research fulfi lled several roles for the researchers. For example, although the focus groups yielded fi ndings that could be linked to the survey ones, they were also used to inform the design of the survey questions. There will be further reference to the utility of the mixed methods approach in Chapter 27, while the components of the CCSE project will be referred to in the interim chapters.

Sources: Silva and Wright (2008); Bennett et al. (2009); Silva et al. (2009); www.open.ac.uk/socialsciences/cultural-capital-and-social-exclusion/project-summary.php (accessed 13 August 2010).

9780199588053_C02.indd 38 10/20/11 9:59 AM

www.open.ac.uk/socialsciences/cultural-capital-and-social-exclusion/research-questions.php

www.open.ac.uk/socialsciences/cultural-capital-and-social-exclusion/project-summary.php

www.open.ac.uk/socialsciences/cultural-capital-and-social-exclusion/project-summary.php


We are beginning to get a picture now that social re-search is infl uenced by a variety of factors. Figure 2.3 summarizes the infl uences that have been examined so far, but has added two more—the impact of values and of practical considerations.

ValuesValues refl ect either the personal beliefs or the feelings of a researcher. On the face of it, we would expect that social scientists should be value free and objective in their research. After all, one might want to argue that research that simply refl ected the personal biases of its practitioners could not be considered valid and scien-tifi c because it was bound up with the subjectivities of its practitioners. Such a view is held with less and less frequency among social scientists nowadays. Émile Durkheim (1858–1917) wrote that one of the corollaries of his injunction to treat social facts as things was that all ‘preconceptions must be eradicated’ (Durkheim 1938: 31). Since values are a form of preconception, his exhor-tation was at least implicitly to do with suppressing them when conducting research. His position is unlikely to be regarded as credible nowadays, because there is a grow-ing recognition that it is not feasible to keep the values that a researcher holds totally in check. These can intrude at any or all of a number of points in the process of social research:

• choice of research area;

• formulation of research question;

• choice of method;

• formulation of research design and data-collection techniques;

• implementation of data collection;

• analysis of data;

• interpretation of data;

• conclusions.

There are, therefore, numerous points at which bias and the intrusion of values can occur. Values can materialize at any point during the course of research. The researcher may develop an affection or sympathy, which was not necessarily present at the outset of an investigation, for the people being studied. It is quite common, for example, for researchers working within a qualitative research strategy, and in particular when they use participant observation or very intensive interviewing, to develop a close affi nity with the people whom they study to the extent that they fi nd it diffi cult to disentangle their stance as social scientists from their subjects’ perspective. This possibility may be exacerbated by the tendency that Becker (1967) identifi ed for sociologists in particular to be very sympathetic to underdog groups. Equally, social scientists may be repelled by the people they study. The social anthropologist Colin Turnbull (1973) reports the results of his research into an African tribe known as the Ik. Turnbull was appalled by what he witnessed: a love-less (and for him unlovable) tribe that left its young and very old to die. While Turnbull was able to point to the conditions that had led to this state of affairs, he was very honest in his disgust for what he witnessed, particularly during the period of his initial sojourn among the tribe. However, that very disgust is a product of Western values about the family, and it is likely, as he acknowledged, that these will have infl uenced his perception of what he witnessed.

Another position in relation to the whole question of values and bias is to recognize and acknowledge that research cannot be value free but to ensure that there is no untrammelled incursion of values in the research pro-cess and to be self-refl ective and so exhibit refl exivity(see Key concept 17.5) about the part played by such factors. As Turnbull (1973: 13) put it at the beginning of his book on the Ik: ‘the reader is entitled to know some-thing of the aims, expectations, hopes and attitudes that the writer brought to the fi eld with him, for these will surely infl uence not only how he sees things but even

Infl uences on the conduct of social research

gu e .3Figure 2.3Infl uences on social research

Theory Practical considerations

Social research

Epistemology

Values Ontology

9780199588053_C02.indd 39 10/20/11 9:59 AM


what he sees.’ Researchers are increasingly prepared to forewarn readers of their biases and assumptions and how these may have infl uenced the subsequent fi ndings. There has been a growth since the mid-1970s of collec-tions of inside reports of what doing a piece of research was really like, as against the generalities presented in social research methods textbooks (like this one!). These collections frequently function as ‘confessions’, an ele-ment of which is often the writer’s preparedness to be open about his or her personal biases.

The signifi cance of feminism in relation to values goes further than this, however. In particular, several feminist social researchers around the early 1980s proposed that the principles and practices associated with quantitative research were incompatible with feminist research on women. For writers like Oakley (1981), quantitative research was bound up with male values of control that can be seen in the general orientation of the research strategy—control of the research subject/respondent and control of the research context and situation. Moreover, the research process was seen as one-way traffi c, in which researchers extract information from the people being studied and give little, or more usually nothing, in return. For many feminists, such a strategy bordered on exploitation and was incompatible with feminism’s values of sisterhood and non-hierarchical relationships between women. The antipathy towards quantitative research resulted in a preference for qualita-

Still another approach is to argue for consciously value-laden research. This is a position taken by some feminist writers who have argued that only research on women that is intended for women will be consistent with the wider political needs of women. Mies (1993: 68) has argued that in feminist research the ‘postulate of value free research, of neutrality and indifference towards the research objects, has to be replaced by conscious par-tiality, which is achieved through partial identifi cation with the research objects’ (emphases in original).

tive research among feminists. Not only was qualitative research seen as more consistent with the values of fem-inism; it was seen as more adaptable to those values. Thus, feminist qualitative research came to be associated with an approach in which the investigator eschewed a value-neutral approach and engaged with the people being studied as people and not simply as respondents to research instruments. The stance of feminism in relation to both quantitative and qualitative approaches demon-strates the ways in which values have implications for the process of social investigation. In more recent years, there has been a softening of the attitudes of feminists towards quantitative research. Several writers have acknowledged a viable and acceptable role for quanti-tative research, particularly when it is employed in conjunction with qualitative research (Jayaratne and Stewart 1991; Oakley 1998). This issue will be picked up in Chapters 17, 26, and 27.

Student experienceThe infl uence of feminism on research questionsSarah Hanson is very clear about the infl uence of feminism on her research and on her research questions in particular.

My research project focused on the representation of women through the front covers of fi ve women’s magazines, combining the application of feminist theory with the decoding practices of content analysis. Throughout the project I wanted to understand the nature of women’s magazines, the infl uences they have on women’s sense of self and identity and the role the magazines play. I asked: do women’s magazines support or destroy women’s identity and do they encourage self-respect or self-scrutiny? I wanted to combine theory with fact, focusing on the meanings behind the presentation of images and text.

Similarly, for her research on non-governmental organizations (NGOs) and sex workers in Thailand, Erin Sanders wrote that she ‘employed a feminist methodology—and as such attempted to engage with my research participants, particularly the sex workers, as a “friend” rather than as a “researcher” ’. She also writes:

I chose to use a feminist methodology because I wanted to eliminate the power imbalance in the research relationship. As there are a number of power issues with a ‘White’, ‘Western’ woman interviewing ‘Non-White’, ‘Non-Western’ sex workers, I had hoped a feminist methodology . . . would help redress some of the power issues.

To read more about Sarah and Erin’s research experiences, go to the Online Resource Centre that accompanies this book at: www.oxfordtextbooks.co.uk/orc/brymansrm4e/

9780199588053_C02.indd 40 10/20/11 9:59 AM



There are, then, different positions that can be taken up in relation to values and value freedom. Far fewer writers overtly subscribe to the position that the prin-ciple of objectivity can be put into practice than in the past. Quantitative researchers sometimes seem to be writing in a way that suggests an aura of objectivity (Mies 1993), but we simply do not know how far they subscribe to such a position. There is a greater awareness today of the limits to objectivity, so that some of the highly confi dent, not to say naive, pronouncements on the subject, like Durkheim’s, have fallen into disfavour. A further way in which values are relevant to the con-duct of social research is through adherence to ethical principles or standards. This issue will be followed up in Chapter 6.

Practical considerationsNor should we neglect the importance and signifi cance of practical issues in decisions about how social research should be carried out. There are a number of different dimensions to this issue. For one thing, choices of re-search strategy, design, or method have to be dovetailed with the specifi c research question being investigated. If we are interested in teasing out the relative importance of a number of different causes of a social phenomenon, it is quite likely that a quantitative strategy will fi t our needs, because, as will be shown in Chapter 7, the assess-ment of cause is one of its keynotes. Alternatively, if we are interested in the world views of members of a certain social group, a qualitative research strategy that is sensi-tive to how participants interpret their social world may be the direction to choose. If a researcher is interested in a topic on which no or virtually no research has been done in the past, the quantitative strategy may be diffi -cult to employ, because there is little prior literature

from which to draw leads. A more exploratory stance may be preferable, and, in this connection, qualitative research may serve the researcher’s needs better, since it is typically associated with the generation rather than the testing of theory (see Table 2.1) and with a rela-tively unstructured approach to the research process (see Chapter 17). Another dimension may have to do with the nature of the topic and of the people being investigated. For example, if the researcher needs to engage with indi-viduals or groups involved in illicit activities, such as gang violence (Patrick 1973), drug dealing (P. A. Adler 1985), or the murky underworld of organs-trading (Scheper-Hughes 2004), it is unlikely that a social survey would gain the confi dence of the subjects involved or achieve the necessary rapport. In fact, the idea of con-ducting survey research in such contexts or on such respondents looks rather ridiculous. It is not surprising, therefore, that researchers in these areas have tended to use a qualitative strategy where there is an opportunity to gain the confi dence of the subjects of the investiga-tion or even in some cases not reveal their identity as researchers, albeit with ethical dilemmas of the kind dis-cussed in Chapter 6. By contrast, it does not seem likely that the hypothesis in the research described in Research in focus 2.4 could have been tested with a qualitative method like participant observation.

While practical considerations may seem rather mun-dane and uninteresting compared with the lofty realm inhabited by the philosophical debates surrounding such discussions about epistemology and ontology, they are important ones. All social research is a coming-together of the ideal and the feasible. Because of this, there will be many circumstances in which the nature of the topic or of the subjects of an investigation and the constraints on a researcher loom large in decisions about how best to proceed.

Student experienceA practical consideration in the choice of research methodOne of the factors that infl uenced Rebecca Barnes’s choice of the semi-structured interview for her study of violence in women’s same-sex intimate relationships was that she felt that the topic is a highly sensitive area and that she therefore needed to be able to observe her interviewees’ emotional responses.

I felt that, given the sensitivity of the research topic, semi-structured, in-depth interviews would be most appropriate. This gave me the opportunity to elicit women’s accounts of abuse in a setting where I was able to observe their emotional responses to the interview and endeavour to minimize any distress or other negative feelings that might result from participating in the research.

To read more about Rebecca’s research experiences, go to the Online Resource Centre that accompanies this book at: www.oxfordtextbooks.co.uk/orc/brymansrm4e/

9780199588053_C02.indd 41 10/20/11 9:59 AM



Key points

● Quantitative and qualitative research constitute different approaches to social investigation and carry with them important epistemological and ontological considerations.

● Theory can be depicted as something that precedes research (as in quantitative research) or as something that emerges out of it (as in qualitative research).

● Epistemological considerations loom large in considerations of research strategy. To a large extent, these revolve around the desirability of employing a natural science model (and in particular positivism) versus interpretivism.

● Ontological considerations, concerning objectivism versus constructionism, also constitute important dimensions of the quantitative/qualitative contrast.

● Values may impinge on the research process at different times.

● Practical considerations in decisions about research methods are also important factors.

● Feminist researchers have tended to prefer a qualitative approach, though there is some evidence of a change of viewpoint in this regard.

Questions for review

Theory and research

● If you had to conduct some social research now, what would the topic be and what factors would have infl uenced your choice? How important was addressing theory in your consideration?

● Outline, using examples of your own, the difference between grand and middle-range theory.

● What are the differences between inductive and deductive theory and why is the distinction important?

Epistemological considerations

● What is meant by each of the following terms: positivism; realism; and interpretivism? Why is it important to understand each of them?

● What are the implications of epistemological considerations for research practice?

Ontological considerations

● What are the main differences between epistemological and ontological considerations?

● What is meant by objectivism and constructionism?

● Which theoretical ideas have been particularly instrumental in the growth of interest in qualitative research?

Research strategy: quantitative and qualitative research

● Outline the main differences between quantitative and qualitative research in terms of: the relationship between theory and data; epistemological considerations; and ontological considerations.

● To what extent is quantitative research solely concerned with testing theories and qualitative research with generating theories?

Infl uences on the conduct of social research

● What are some of the main infl uences on social research?

9780199588053_C02.indd 42 10/20/11 9:59 AM


Online Resource Centrewww.oxfordtextbooks.co.uk/orc/brymansrm4e/

Visit the Online Resource Centre that accompanies this book to enrich your understanding of social research strategies. Consult web links, test yourself using multiple choice questions, and gain further guidance and inspiration from the Student Researcher’s Toolkit.

9780199588053_C02.indd 43 10/20/11 9:59 AM


This chapter illustrates the philosophicalbases of the two basic approaches to socialresearch which gave rise to the families ofquantitative and qualitative techniques. Wewill begin with the concept of paradigm – thatis, the perspective that inspires and directs agiven science. Then we shall examine the his-torical roots and the guiding principles of thepositivist and the interpretive paradigms. Thechapter ends with a few reflections concern-ing currents trends in social research.

1. KUHN AND THE PARADIGMSOF SCIENCES

The notion of ‘paradigm’ has ancient originsin the history of philosophical thought. It wasutilized both by Plato (to mean ‘model’) andby Aristotle (to mean ‘example’). In the socialsciences its use has been inflated and con-fused by multiple and different meanings:these range from a synonym for theory toan internal subdivision of a theory, from asystem of ideas of a pre-scientific nature toa school of thought, from an exemplaryresearch procedure to the equivalent of

method. It seems useful therefore briefly toreview the meaning given to the concept ofthe paradigm by the scholar who, in the1960s, brought it once again to the attention ofphilosophers and sociologists of science.We are referring to Thomas Kuhn and hiscelebrated essay The Structure of ScientificRevolutions (1962).

Reflecting on the historical development ofthe sciences, Kuhn refuted the traditionalunderstanding of the sciences as a cumulativeand linear progression of new acquisitions.According to the traditional conception, singleinventions and discoveries would be added tothe previous body of knowledge in the samemanner as bricks are placed one on top ofanother in the construction of a building.According to Kuhn, however, while this is theprocess of science in ‘normal’ times, there arealso ‘revolutionary’ moments, in which thecontinuity with the past is broken and a newconstruction is begun, just as – to take up thebuilding metaphor again – from time to time,an old brick building is blown up to make roomfor a structurally different one, for example askyscraper made of glass and aluminium.

Kuhn illustrates his argument with a richcollection of examples from the natural

1 Paradigms of Social Research

1. Kuhn and the paradigms of sciences 92. Three basic questions 113. Positivism 134. Neopositivism and postpositivism 175. Interpretivism 206. A final note: radicalization, criticism and new tendencies 25

3069-CH01.QXD 3/6/03 7:27 PM Page 9

sciences (especially from physics). Forinstance, he cites the development of opticalphysics, which is currently interpreted inquantum terms; according to this view, lightis made up of photons, which display some ofthe features of waves and some of the proper-ties of particles. Kuhn points out that, beforequantum theory was developed by Planck,Einstein and others, light was believed to bea transversal wave movement. This lattertheory was developed at the beginning ofthe nineteenth century. Still earlier, in theseventeenth century, the dominant view wasthat of Newtonian optics, according to whichlight was made up of material corpuscles.

The shift from one theoretical perspectiveto another is so pervasive and has such radi-cal consequences for the discipline concernedthat Kuhn does not hesitate to use the term‘scientific revolution’. What changes in agiven discipline after one of these revolu-tions? It produces ‘a shift in the problemsavailable for scientific scrutiny and in thestandards by which the profession deter-mined what it should count as an admissibleproblem or as a legitimate problem-solution’(1962: 6). A reorientation in the disciplineoccurs that consists of ‘a displacement of theconceptual network through which scientistsview the world’ (1962: 102). This ‘conceptualnetwork’ is what Kuhn calls a ‘paradigm’,

and it is this aspect of his theorising, ratherthan his analysis of the developmental processin science, that interests us here.

Without a paradigm a science lacks orienta-tions and criteria of choice: all problems, allmethods, all techniques are equally legiti-mate. By contrast, the paradigm constitutes aguide: ‘Paradigms’ – recalls Kuhn – ‘providescientists not only with a map but also withsome of the directions essential for map-making. In learning a paradigm the scientistacquires theory, methods, and standardstogether, usually in an inextricable mixture’(1962: 109).

Kuhn defines normal science as those phasesin a scientific discipline during which a givenparadigm, amply agreed to by the scientificcommunity, predominates. During this phase,as long as the operating paradigm is notreplaced by another in a ‘revolutionary’manner, a scientific discipline does indeeddevelop in that linear and cumulative waythat has been attributed to the whole of scien-tific development. ‘No part of the aim ofnormal science is to call forth new sort ofphenomena … Instead, normal-scientificresearch is directed to the articulation of thosephenomena and theories that the paradigmalready supplies’ (Kuhn, 1962: 24).

Numerous examples of scientific para-digms are to be found in the history of the

10 Socia l Research

What does Thomas Kuhn mean by ‘paradigm’? He means a theoreticalperspective:

• accepted by the community of scientists of a given discipline • founded on the previous acquisitions of that discipline• that directs research through:

! the specification and choice of what to study ! the formulation of hypotheses to explain the phenomenon observed ! the identification of the most suitable empirical research techniques.

BOX 1.1 PARADIGM

3069-CH01.QXD 3/6/03 7:27 PM Page 10

natural sciences. Going back to our previousexample, we can speak of corpuscular, wave,and quantum paradigms in optical physics.Likewise, as examples of alternative para-digms that have succeeded one another intime, we can quote Newtonian and Einsteinianmechanics, Ptolemaic and Copernican cosmo-logy, and so on.

To what extent can we speak of paradigmsin the social sciences? Kuhn notes that theparadigm is a characteristic feature of the‘mature’ sciences. Before the corpusculartheory of light was introduced by Newton, nocommon paradigm existed among scientistsin this sector; instead, various schools andsub-schools opposed and competed with oneanother, each with its own theory and point ofview. Consequently, concludes Kuhn, ‘The netresult of their activity was something lessthan science’ (1962: 13). In this perspective,because the social sciences lack a single para-digm broadly shared by the scientific commu-nity, they are in a pre-paradigmatic state,except perhaps for economics (according toKuhn, ‘economists agree on what economicsis’, while ‘it remains an open question whatparts of social science have yet acquired suchparadigm at all’ (1962: 14).

What has been said with regard to thesocial sciences also holds for sociology.Indeed, it is difficult to identify a paradigmthat has been agreed upon, even for limitedperiods, by the community of sociologists.Nevertheless, there exists another interpreta-tion of the thinking of Kuhn, which hasbeen proposed in an attempt to apply hiscategories to sociology. This interpretationredefines the concept of the paradigm, main-taining all the elements of the original defini-tion (theoretical perspective that defines therelevance of social phenomena, puts forwardinterpretative hypotheses and orients thetechniques of empirical research) except one:that the paradigm is agreed upon by themembers of the scientific community. Thispaves the way for the presence of multipleparadigms inside a given discipline; thus,instead of being a pre-paradigmatic discipline,sociology becomes a multi-paradigmatic one.

This is the interpretation of Friedrichs (1970)who, after highlighting the paradigm inspiredby Parsons’ structural-functionalism, sees inthe Marxist dialectic approach the second par-adigm of sociology, in which the concepts ofsystem and consensus that are central to func-tionalism are replaced by that of conflict.

This interpretation of the concept of theparadigm in terms of an overall theoreticalperspective which does not exclude other per-spectives but rather is in open competitionwith them, is certainly the most widespreadinterpretation and corresponds to the currentuse of the term in the social sciences.Nevertheless, this less rigorous interpretation,which adapts Kuhn’s original category to thestatus of the social sciences, must not be trivi-alized by equating a paradigm with a theoryor a school of thought. Indeed, fundamentalto the concept of the paradigm is its pre-theoretical and, in the final analysis, meta-physical character of a ‘guiding vision’, ‘a viewof the world’, which shapes and organizesboth theoretical reflection and empiricalresearch and, as such, precedes both.

In this interpretation, the concept of theparadigm seems useful in analysing thevarious basic frames of reference that havebeen put forward, and which are still beingevaluated in the field of social researchmethodology.

2. THREE BASIC QUESTIONS

Having defined and circumscribed the con-cept of a paradigm and briefly discussed itsapplication to the social sciences, we will nowabandon the slippery terrain of the paradigmsof sociological theory (one paradigm? two para-digms? a hundred paradigms?) for more solidground: the methodology of social research. Wewill not, however, go deeply into the complexepistemological problems of how many andwhich philosophical frameworks guideempirical research in the social sciences.Instead, we will confine ourselves to a historical

Parad igms o f Soc ia l Research 11

3069-CH01.QXD 3/6/03 7:27 PM Page 11

review by briefly describing the fundamentalperspectives that have been proposed andbecome accepted during the evolution of thediscipline. Since this is a book on socialresearch techniques, it seems natural andproper to begin by raising the question of thefounding paradigms of social research, fromwhich the first operative procedures emerged,and which subsequently guided the develop-ment of empirical research. Indeed, as hasbeen said, one of the functions of a paradigmis to establish acceptable research methodsand techniques in a discipline. As Hugheswrites:

Every research tool or procedure is inextrica-bly embedded in commitments to particularversions of the world and ways of knowingthat world made by researchers using them.To use a questionnaire, an attitude scale ofbehavior, take the role of a participantobserver, select a random sample .. . is to beinvolved in conceptions of the world whichallow these instruments to be used forthe purposes conceived. No technique ormethod of investigation . . . is self valida-ting: its effectiveness, its very status asa research instrument .. . is dependent, ulti-mately, on philosophical justification.(Hughes, 1980: 13)

Within the philosophical perspectives thatgenerated and have accompanied the growthof social research, can we identify visions thatare sufficiently general, cohesive and opera-tive to be characterized as paradigms? Itseems so. Indeed, there is broad agreementamong scholars that two general frames ofreference have historically oriented socialresearch since its inception: the ‘empiricist’vision and the ‘humanist’ vision. Variouslabels have been used, including ‘objectivism’and ‘subjectivism’; here, we will utilize thecanonical term ‘positivism’ and the less con-solidated ‘interpretivism’. As we will soonsee, these are two organic and stronglyopposed visions of social reality and how itshould be understood; and they have gener-ated two coherent and highly differentiatedblocks of research techniques. Before describing

these techniques, however, it is essential toexplore their philosophical origins, since onlyby doing so can we achieve a full understand-ing of them.

In order to adequately compare the twoabove-mentioned paradigms, we will attemptto understand how they respond to the fun-damental interrogatives facing social research(and scientific research in general). These canbe traced back to three basic questions: Does(social) reality exist? Is it knowable? How canwe acquire knowledge about it? In otherwords: Essence, Knowledge and Method.1

The ontological question2 This is the questionof ‘what’. It regards the nature and formof social reality. It asks if the world of socialphenomena is a real and objective worldendowed with an autonomous existence out-side the human mind and independent fromthe interpretation given to it by the subject. Itasks, therefore, if social phenomena are‘things in their own right’ or ‘representationsof things’. The problem is linked to the moregeneral philosophical question of the exis-tence of things and of the external world.Indeed, the existence of an idea in the mindtells us nothing about the existence of theobject in reality, just as a painting of a unicorndoes not prove the existence of unicorns.

The epistemological question3 This is thequestion of the relationship between the‘who’ and the ‘what’ (and the outcome of thisrelationship). It regards the knowability ofsocial reality and, above all, focuses on therelationship between the observer and thereality observed. Clearly, the answer to thisquestion depends on the answer to the previ-ous ontological question. If the social worldexists in its own right, independently fromhuman action, the aspiration to reach it andunderstand it in a detached, objective way,without fear of altering it during the course ofthe cognitive process, will be legitimate.Closely connected with the answer given tothe epistemological question are the formsknowledge can take: these range from deter-ministic ‘natural laws’ dominated by the

12 Socia l Research

3069-CH01.QXD 3/6/03 7:27 PM Page 12

categories of cause and effect, to less cogent(probabilistic) laws, to various kinds of gener-alizations (e.g. Weberian ideal types), to theexclusion of generalizations (only specific andcontingent knowledge being admissible).

The methodological question4 This is thequestion of ‘how’ (how can social reality bestudied?). It therefore regards the technicalinstruments of the cognitive process. Here,too, the answers depend closely on theanswers to the previous questions. A vision ofsocial reality as an external object that is notinfluenced by the cognitive research proce-dures of the scientist will accept manipulativetechniques (e.g. experimentation, the controlof variables, etc.) more readily than a perspec-tive that underlines the existence of inter-active processes between the scholar and theobject studied.

The three questions are therefore inter-related, not only because the answers to eachare greatly influenced by the answers to theother two, but also because it is sometimesdifficult to distinguish the boundariesbetween them (though, for the purpose of ourexposition, we will try to do so). Indeed, it isdifficult to separate conceptions of the natureof social reality from reflections on whether(and how) it may be understood and, in turn,to separate these from the techniques that canbe used to understand it. Then again, theseinterrelations are implicit in the very defini-tion of the scientific paradigm which, as wehave seen, is both a theoretical perspectiveand a guide to research procedures.

3. POSITIVISM

Table 1.1 shows a synopsis of the differentparadigms with regard to the fundamentalquestions introduced above. First of all, it willbe noted that two versions of positivism arepresented: the original nineteenth-centuryversion, to which even the most tenaciousempiricists no longer subscribe, and itstwentieth-century reformulation, which was

constructed to address the manifest limits ofthe original version. The original positivistparadigm is presented both for historicalreasons – since it was the vision that accom-panied the birth of the social sciences and, inparticular, the birth of sociology – andbecause the character of the other two para-digms can be better understood by examiningthe criticisms levelled against it.

Sociology was born under the auspicesof positivist thought. In the middle of thenineteenth century, when the investigation ofsocial phenomena was evolving into a subjectof scientific study, the paradigm of the naturalsciences reigned supreme. Inevitably, the newdiscipline took this paradigm as its model.Indeed, the founders of the discipline,Auguste Comte and Herbert Spencer amongthem, shared a naïve faith in the methods ofnatural science. The positivist paradigm is nomore than this: the study of social reality utiliz-ing the conceptual framework, the techniques ofobservation and measurement, the instruments ofmathematical analysis, and the procedures of infer-ence of the natural sciences.

Let us look more closely at the distinctiveelements of this definition. The conceptualframework: the categories of ‘natural law’,cause and effect, empirical verification, expla-nation, etc. The techniques of observation andmeasurement: the use of quantitative variables,even for qualitative phenomena; measurementprocedures applied to ideological orientation,mental abilities and psychological states (atti-tude measurement, intelligence tests, etc.)Mathematical analysis: the use of statistics,mathematical models, etc. The procedures ofinference: the inductive process, wherebyhypotheses regarding the unknown are formedon the basis of what is known and specificobservations give rise to general laws; the useof theory to predict outcomes; extrapolationfrom the sample to the whole population.

According to Comte, the prophet ofnineteenth-century sociological positivism, theacquisition of the positivist viewpoint consti-tuted, in all sciences, the end-point of a trendthat had previously passed through theologicaland metaphysical stages. Such development


3069-CH01.QXD 3/6/03 7:27 PM Page 13

did not occur at the same time in all disci-plines; it first took place in the inorganicsciences, such as astronomy, physics andchemistry, followed by the organic sciences,such as biology. It was therefore natural, inthe progression from simple to complex mate-rial, that the positivist approach should beapplied to the most complex material of all:

society. Thus, a new science would emerge:sociology, the positive science of society.According to this view, science is universal,and scientific method is unique. The socialsciences do not differ from the naturalsciences, and the positivist way of thinkingthat brought such great advances in the fieldsof astronomy, physics and biology is destined

14 Socia l Research

Table 1.1 Characteristics of the basic paradigms of social research

Constructivism: the knowableworld is that of meaningsattributed by individuals.Relativism (multiplerealities): theseconstructed realities vary inform and content amongindividuals, groups, andcultures

Non-dualism; non-objectivity.Researcher and object ofstudy are not separate, butinterdependent

Interpretive science in searchof meaning

Goal: comprehensionGeneralizations: opportunity

structures; ideal types

Empathetic interactionbetween scholar and objectstudied

InterpretationObserver-observed interaction

Inuction (knowledge emergesfrom the reality studied)

Qualitative techniques.

Analysis ‘by cases’

Ontology

Epistemology

Methodology

Naïve realism: socialreality is ‘real’ andknowable (as if itwere a ‘thing’)

Dualism-objectivity

True results

Experimental sciencein search of laws

Goal: explanationGeneralizations:

‘natural’ immutablelaws

Experimental-manipulative

ObservationObserver-observed

detachmentMostly induction

Quantitative techniques

Analysis ‘by variables’

Critical realism: socialreality is ‘real’ butknowable only in animperfect andprobabilistic manner

Modified dualism-objectivity

Results probabilisticallytrue

Experimental sciencein search of lawsMultiplicity of theoriesfor the same fact

Goal: explanationGeneralizations:

provisional laws,open to revision

Modified experimental-manipulative

ObservationObserver-observed

detachmentMostly deduction

(disproof ofhypotheses)

Quantitative techniqueswith some qualitative

Analysis ‘by variables’Source: Partially adapted from Guba and Lincoln (1994: 109).

Positivism Postpositivism Interpretivism

3069-CH01.QXD 3/6/03 7:27 PM Page 14

to triumph even when its focus shifts fromnatural objects to social objects, such asreligion, politics and work.

The first attempt to apply this overall theo-retical perspective to empirical research wasmade by Durkheim. Indeed, as Durkheimpointed out:

Up to now sociology has dealt more or lessexclusively not with things, but with concepts.It is true that Comte proclaimed that socialphenomena are natural facts subject to naturallaws. In so doing he implicitly recognizedthere are only things. Yet when, leavingbehind these general philosophical state-ments, he tries to apply his principle anddeduce from it the science it contained, it isideas which he too takes as the object of study.(Durkheim, 1895: 63)

By contrast, Durkheim actually tried to trans-late the positivist principles of thought intoempirical procedures; he was the first ‘socialscientist’, the first true positivist sociologist.His empirical procedure is founded on thetheory of ‘social fact’. In his Rules ofSociological Method, he states at the outset that‘the first and most basic rule is to considersocial facts as things’ (1895: 60). For Durkheim,social facts are:

Ways of acting, thinking and feeling whichpossess the remarkable property of existing out-side of the consciousness of the individual .. .When I perform the duties as a .. . husband or acitizen . . . I carry out the commitments I haveentered into, I fulfil obligations which aredefined in by law and custom and which areexternal to myself and my actions. Even whenthey conform to my sentiments and when Ifeel their reality within me, that reality doesnot cease to be objective, for it is not I whohave prescribed these duties; I have receivedthem through education . . . Similarly thebeliever has discovered from birth, readyfashioned, the beliefs and practices of his reli-gious life; if they existed before he did, itfollows that they exist outside him .. . (Likewise,for as far as) the system of signs that I employto express my thoughts, the monetary systemI use to pay my debts . . . the practices I follow

in my profession, etc., all function indepen-dently from the use I make of them.(Durkheim, 1895: 50–51)

These social facts, even if they are not materialentities, nonetheless have the same propertiesas the ‘things’ of the natural world, and fromthis derive two consequences. On the onehand, social facts are not subject to human will;they are things that offer resistance to humanintervention; they condition and limit it. Onthe other hand, just like the phenomena of thenatural world, they function according to theirown rules. They possess a deterministic struc-ture that can be discovered through scientificresearch. Thus, notwithstanding their differentobjects, the natural world and the social worldshare a substantial methodological unity (theycan both be studied through the same inves-tigative logic and the same method, hence thename ‘social physics’ attributed to the studyof society).

The first assertion is, therefore, that socialreality exists outside the individual. Thesecond is that this social reality is objectivelyunderstandable, and the third that it can bestudied by means of the same methods as thenatural sciences. As Durkheim states, ‘Ourrule implies no metaphysical conception, nospeculation about the innermost depth ofbeing. What it demands is that the sociologistshould assume the state of mind of physicists,chemists or in physiologists, when they ven-ture into an as yet unexplored area of theirscientific field’ (1895: 37). And again: ‘Our mainobjective is to extend the scope of scientificrationalism to cover human behaviour . . .What has been termed our positivism ismerely a consequence of this rationalism.’(Durkheim, 1895: 33)

Let us now look at how this understandingis acquired. Positivism is fundamentallyinductive, where induction means ‘movingfrom the particular to the general’5 the processby which generalizations or universal laws arederived from empirical observation, from theidentification of regularities and recurrencesin the fraction of reality that is empiricallystudied. Implicit in inductive procedures is


3069-CH01.QXD 3/6/03 7:27 PM Page 15

the assumption of order and uniformity innature, that universal organizing principlesexist. The task of the scientist is, of course, todiscover these. This vision has long domi-nated the natural sciences and has even beenidentified with the scientific method. Inassuming that social life, like all other pheno-mena, is subject to immutable natural laws,the positivist conception of society fully

adopts this vision. According to Durkheim,the social scientist is an explorer ‘Consciousthat he is penetrating into the unknown. Hemust feel himself in the presence of factsgoverned by laws as unsuspected as those oflife before the science of biology was evolved.He must hold himself ready to make discov-eries which will surprise and disconcert him.’(1895: 37)

16 Socia l Research

Ontology: naïve realism This position stems from everything that has beensaid regarding the ‘codification’ of social reality, and can be succinctlyexpressed by two propositions: (a) there exists an objective social reality thatis external to human beings, whether they are studying or performing socialacts; (b) this reality is knowable in its true essence.6

Epistemology: dualist and objectivist; natural law The assertion that knowledgeis attainable is based on two assumptions: (a) that the scholar and the objectstudied are independent entities (dualism); (b) that the scholar can study theobject without influencing it or being influenced by it (objectivity). Investigationis carried out as if through a ‘one-way mirror’. Knowledge assumes the formof ‘laws’ based on the categories of cause and effect. These laws are part ofan external reality that is independent of the observer (‘natural laws’); thescientist’s task is to ‘discover them’. There is no fear that the researcher’svalues might distort her reading of social reality, or vice versa. This position,which excludes values in favour of facts, necessarily derives from the visionof social fact as given and unmodifiable.

Methodology: experimental and manipulative The methods and techniques ofpositivist research – like its basic conception – draw heavily on the classicalempiricist approach to the natural sciences. Two features of the experimentalmethod are taken up: (a) its use of inductive procedures, whereby generalformulations are derived from particular observations; and (b) its mathematicalformulation which, though not always attainable, is the final goal of the positivistscientist. The ideal technique remains – even though its applicability to socialreality is limited – that of experiment, founded on manipulation and control ofthe variables involved and the detachment of the observer from what isobserved.

BOX 1.2 ANSWERS GIVEN BYPOSITIVISM TO THE THREE BASIC QUESTIONS

3069-CH01.QXD 3/6/03 7:27 PM Page 16

Finally, with regard to the ‘form’ of thisknowledge, there is no doubt that these lawsof nature will eventually be identified, formu-lated, demonstrated and ‘proved’; in theirmost complete form, they are laws that linkcause and effect:

Since the law of causality has been verified inthe other domains of nature and has progres-sively extended its authority from the phys-ical and chemical world to the biologicalworld, and from the latter to the psychologicalworld, one may justifiably grant that it is like-wise true for the social world. Today it is pos-sible to add that the research undertaken onthe basis of this postulate tends to confirmthis. (Durkheim, 1895: 159).

In the positivist paradigm, the elements thatwe have called ‘naïve faith’ in the methods ofthe natural sciences are all too evident.Underlying the various manifestations ofpositivism there is always, in fact, a sort ofenthusiasm for ‘positive’ scientific knowledge,whereby the ‘scientific method’ is viewed asthe only valid means of achieving true know-ledge in all fields of human endeavour.

4. NEOPOSITIVISM ANDPOSTPOSITIVISM

Throughout the twentieth century, the posi-tivist approach was continually revised andadjusted in attempts to overcome its intrinsiclimits. The reassuring clarity and linearity ofnineteenth-century positivism gave way to atwentieth-century version that was muchmore complex and detailed and, in somerespects, contradictory and unclear. However,some basic assumptions were maintained,such as ontological realism (‘the world existsindependently of our awareness of it’) and thepre-eminent role of empirical observation inunderstanding this world. We will not enterinto the details of this development, or thevarious phases of its history; rather, we willmention only ‘neopositivism’, the term used

to denote the approach that dominated inthe period from the 1930s to the 1960s, and‘postpositivism’, which is used to identify itsfurther evolution from the end of the 1960sonwards.7 We will therefore outline the prin-cipal shifts in perspective that occurred – overtime and with differing degrees of intensity –with respect to the positivist orthodoxypresented in the previous section.

One of the first revisions of nineteenth-century positivism was made by the schoolknown as logical positivism, which gave riseto neopositivism. The movement formedaround the discussions of a group of scholarsof different disciplinary origins who, in thesecond half of the 1920s, constituted theso-called ‘Vienna Circle’. Among its principalexponents were the philosophers Schlick andCarnap, the mathematician Hahn, the econo-mist Neurath, and the physicist Frank. A fewyears later, a group of like-minded thinkers(Reichenbach, Herzberg, Lewin, Hempel andothers) was formed in Berlin. In the wake ofNazi persecution, some notable representa-tives of this school emigrated to the UnitedStates, where the affinity between their viewsand American pragmatism contributed con-siderably to the spread of neopositivistthought. This influenced other disciplines,including sociology, which had been develop-ing a very rich tradition of empirical researchin the United States throughout the 1930s.

The new point of view assigned a centralrole to the criticism of science and redefinedthe task of philosophy, which was to abandonits broad theorization in order to undertakecritical analysis of the theories elaboratedwithin single disciplines (Schlick hoped to seea time when there would be no more books onphilosophy, but all books would be writtenin a ‘philosophical way’). This led to therejection of the ‘great questions’ and of allmetaphysical issues that could not be demon-strated (‘pseudo-problems’), and which weretherefore branded as meaningless. Instead,the utmost attention was devoted to method-ological problems in every science, to thelogical analysis of their language and their


3069-CH01.QXD 3/6/03 7:27 PM Page 17

theoretical output, to the criticism of theirassumptions, and – not least – to the proce-dures by which conceptual elaboration couldbe empirically verified.

From what has been said, it is evident thatepistemological questions are central to thismovement of thought, and the influence ithad on the methodology of the sciences,including the social sciences, is comprehensi-ble. It must be remembered that one of thepostulates of neopositivism is the widespreadconviction that the meaning of a statementderives from its empirical verifiability. Theformula ‘the meaning of a proposition is themethod of its verification’ neatly summarizesthis point of view.

What did this conception of science andscientific knowledge mean for social research?What were the repercussions on operationalprocedures and research techniques? Themain consequence was the development of acompletely new way of speaking about socialreality, using a language borrowed frommathematics and statistics. Paul F. Lazarsfeld,the principal exponent of neopositivist empiri-cal methodology in sociology, called this thelanguage of variables. Every social object,beginning with the individual, was analyti-cally defined on the basis of a range of attri-butes and properties (‘variables’), and wasreduced to these; and social phenomena wereanalysed in terms of relationships amongvariables. The variable, with its neutral char-acter and objectivity, thus became the protag-onist of social analysis; there was no longerany need to recompose the original object orindividual as a whole again. In this way socialresearch became ‘depersonalized’, and thelanguage of variables, with the measurementof concepts, the distinction between depen-dent and independent variables, the quantifi-cation of their interrelations and theformulation of causal models, provided aformal instrument that allowed social scien-tists to go beyond ‘the notoriously vagueeveryday language (in a process of) clarifica-tion and purification of discourse (that is)

very important for the social scientist; . . . wemust sort out this knowledge and organize itin some manageable form; we must reformu-late common sense statements so that theycan be subjected to empirical test’ (Lazarsfeldand Rosenberg, 1955: 2,11). In this way, allsocial phenomena could be surveyed, meas-ured, correlated, elaborated and formalizedand the theories either confirmed or dis-proved in an objective manner withoutambiguity.

But nothing would ever be the same again.The twentieth-century conception of sciencewas by now far removed from the solid cer-tainties of nineteenth-century positivism, inwhich a ‘mechanical’ conception of realitydominated, together with a reassuring beliefin immutable laws and faith in the irresistibleprogress of science. This new philosophic-scientific atmosphere arose first of all out ofdevelopments in the natural sciences and, inparticular, in physics, during the early yearsof the new century. Quantum mechanics,Einstein’s special and general theories of rela-tivity, Heisenberg’s principle of uncertainty –to cite only a few of the cornerstones of thenew physics – introduced elements of proba-bility and uncertainty to crucial areas suchas the concept of causal law, the objectivity ofthe external world, and even the classicalcategories of space and time.

Theories were no longer expressed in termsof deterministic laws, but of probability. Thecrucial moment in this change was the shiftfrom classical physics (Newtonian approach)to quantum physics. According to quantummechanics, there are processes in elementaryphysics – so-called quantum jumps – that arenot analyzable in terms of traditional causalmechanisms, but are absolutely unpredictablesingle facts governed by probabilistic laws.Scientific theories would no longer explainsocial phenomena through models characteri-zed by logical necessity, and deterministiclaws were replaced by probabilistic lawsthat implied the existence of haphazard ele-ments and the presence of disturbances and

18 Socia l Research

3069-CH01.QXD 3/6/03 7:27 PM Page 18

fluctuations. If this notion of probabilisticindeterminism was valid for the naturalworld, then it would be even more valid forthe social world, the world of language,thought, and human interaction.

An element introduced into scientificmethodology by this evolution of positivismis the concept of falsification, which wastaken up as a criterion for the empirical vali-dation of a theory or a theoretical hypothesis.This states that a theory cannot be positivelyconfirmed by data, and that empirical valida-tion can take place only in the negative,through the ‘non-confutation’ of the theory bythe data – that is to say, by demonstrating thatthe data do not contradict the hypothesis and,therefore, that the theory and the data aremerely compatible. Positive proof is imposs-ible, since the same data could be compatiblewith other theoretical hypotheses.

This position gives rise to a sense of the pro-visional nature of any theoretical statement,since it is never definitively proven and alwaysexposed to the axe of possible disproof. AsPopper writes, the idol of certainty crumbles:‘The old scientific ideal of episteme – ofabsolutely certain, demonstrable knowledge –has proved to be an idol. The demand forscientific objectivity makes it inevitable thatevery scientific statement must remain tenta-tive for ever’ (1934, English translation 1992:280). Man cannot know but only conjecture.This point is also illustrated by a statementattributed to Einstein: ‘to the degree that ourpropositions are certain, they say nothingabout reality; to the degree that they saysomething, they are uncertain’.

Lastly, and this brings us to the most recentdevelopment of the postpositivist approach, ithas become a widespread conviction thatempirical observation, the very perception ofreality, is not an objective picture, but istheory-laden,8 in the sense that even the simplerecording of reality depends on the researcher’sframe of mind, and on social and cultural con-ditioning. In other words, despite theassumption that reality exists independently

from the cognitive and perceptive activity ofhumans, the act of understanding remainsconditioned by the social circumstances andthe theoretical framework in which it takesplace. The thesis of the theory-laden nature ofempirical observations – that is to say, theclaim that no clear distinction exists betweentheoretical concepts and observed data –brings down the last positivist certainty: thatof the objectivity of the data collected and ofthe neutrality and inter-subjectivity of thelanguage of observation.

It must be said, nonetheless, that thisprocess of moving away from the originalpositivist orthodoxy, first through neoposi-tivism and then postpositivism, did not meanthat the empiricist spirit was abandoned.Modern positivism, when its states that laws(both natural and social) are probabilistic andopen to revision, when it affirms the conjec-tural nature of scientific knowledge and in theend, the theoretical conditioning of the obser-vation itself, has come a long way from thenaïve interpretation of the deterministic lawsof the original positivism. It has lost its cer-tainties, but does not repudiate its empiricistfoundations. The new positivism redefinesthe initial presuppositions and the objectivesof social research; but the empirical approach,though much amended and reinterpreted,still utilizes the original observationallanguage, which was founded on the corner-stones of operationalization, quantificationand generalization. And, since we are dealingwith research techniques, it is this point thatinterests us here. The operational procedures,the ways of collecting data, the measurementoperations and the statistical analyses havenot fundamentally changed. Conclusionsare more cautious, but the (quantitative) techni-ques utilized in reaching them are still the same.

At this point, we will conclude our briefexcursus on the developments of the posi-tivist paradigm by filling out the column inTable 1.1 regarding the positions of modernpostpositivism on the three fundamentalquestions.


3069-CH01.QXD 3/6/03 7:27 PM Page 19

5. INTERPRETIVISM

5.1 BeginningsTwo versions of the positivist paradigm havebeen presented: the initial nineteenth-centuryperspective and its critical revision, carriedout in the 1930s and again in the 1970s. Theparadigm presented in this section underwent

an almost symmetrical development. If wewished to stress the analogy between the twoparadigms, we would introduce the initialvision of ‘interpretive sociology’, which owedboth its methodological elaboration and itsfirst attempts at empirical research, at thebeginning of the twentieth century, to MaxWeber (his role was symmetrical to thatplayed by Durkheim in positivism). This

20 Socia l Research

Ontology: critical realism As in the case of positivism, the existence of areality external to human beings is assumed; but – contrary to what is upheldin the positivist paradigm – this reality is only imperfectly knowable, bothbecause of the inevitable imperfection of human knowledge and because ofthe very nature of its laws, which are probabilistic. This point of view has alsobeen called ‘critical realism’: realism, in that it assumes that cause-effectrelationships exist in reality outside the human mind; critical, in that itunderlines the view that the scientist must always be prepared to questionevery scientific acquisition.

Epistemology: modified dualism-objectivity; middle range, probabilistic andconjectural laws With regard to the question of the relationship between thescholar and the object studied, dualism, in the sense of separation and non-interference between the two realities, is no longer sustained. It is recognizedthat the subject conducting the study may exert a disturbing effect on theobject of study, and that a reaction effect may ensue. The objectivity of knowledgeremains the ideal goal and the reference criterion, but this can only beachieved approximately. In the cognitive process, deductive procedures areemphasized, through the mechanism of falsifying hypotheses. The intentremains that of formulating generalizations in the form of laws, even if limitedin scope, probabilistic and provisional.

Methodology: modified experimental-manipulative The operational phases ofresearch remain fundamentally inspired by a substantial detachment betweenthe researcher and the object studied (experiments, manipulation of variables,quantitative interviews, statistical analysis, etc.). Nevertheless, qualitativemethods are admitted. The scientific community is important as it criticallyanalyses new hypotheses, and can confirm results by means of new experiments(repeated results are more likely to be true).

BOX 1.3 ANSWERS GIVEN BY NEO- AND POST-POSITIVISM TO THE THREE BASIC QUESTIONS

3069-CH01.QXD 3/6/03 7:27 PM Page 20

would then be followed by the 1960s reinter-pretation of the original approach, above allin American sociology. This, in turn, gave riseto the various lines of thought found insymbolic interactionism, phenomenologicalsociology and ethnomethodology, which, inspite of their differences, are unified by acommon emphasis on individual interaction.

However, we prefer not to proceed in thismanner, since there is no discontinuitybetween the original Weberian vision andsubsequent developments, as there was in theshift from nineteenth to twentieth-centurypositivism. Instead, we will put these twohistorical blocks of approaches to socialresearch together under the same heading andutilize the general term ‘interpretivism’ for allthe theoretical visions in which reality is notsimply to be observed, but rather ‘interpreted’.

How did this new vision of social sciencearise? While positivism originated innineteenth-century French and English cultures(we need mention only Auguste Comte, JohnStuart Mill and Herbert Spencer) and owed itssociological development chiefly to the Frenchculture (we are, of course, referring toDurkheim), its most radical and organiccriticism emerged in the context of Germanhistoricism.

In general, the German philosopherWilhelm Dilthey is credited with the first criti-cal attack on Comtean scientism in the nameof the autonomy of the human sciences – inthe sense that they are non-homologous to thenatural sciences. In his Introduction to theHuman Sciences (1883), Dilthey draws afamous distinction between ‘sciences ofnature’ and ‘sciences of the spirit’, basingthe difference between them precisely on therelationship that is established between theresearcher and the reality studied. Indeed, inthe natural sciences the object studied consistsof a reality that is external to the researcherand remains so during the course of thestudy; thus, knowledge takes the form ofexplanation (cause-effect laws, etc.). In thehuman sciences, by contrast, since there is nosuch detachment between the observer andwhat is observed, knowledge can be obtained

only through a totally different process, thatof comprehension (Verstehen). According toDilthey, we explain nature, whereas we under-stand the life of the mind.

5.2 Max Weber: objectivity andorientation towards individualityBut it is only with Max Weber that this newperspective enters fully into the field of soci-ology. Indeed, Dilthey had spoken genericallyof ‘sciences of the spirit’, among which hesingled out historiography. Weber broughtthe concept of Verstehen into sociology, andrevised Dilthey’s original position. Whileadopting the principle of Verstehen, Weber didnot want to fall into subjectivist individualismor psychologism; he wanted to preserve theobjectivity of social science both in terms of itsbeing independent of value judgements, andin terms of the possibility of formulating state-ments of a general nature, even when an‘orientation towards individuality’ is adopted.

Regarding the first point, throughout hislife Weber reiterated the need for the histori-cal and social sciences to be free from anyvalue judgement whatsoever. However, hisawareness of the problem (sharpened by hisintense involvement in politics and, later, bythe ethical questions arising from the immi-nent threat of world war) exceeded his abilityto provide an unequivocal answer. None-theless, he never abandoned his convictionthat the historical and social sciences must bevalue-free. ‘The distinction between knowl-edge and judgement – that is to say, betweenfulfilling the scientific responsibility of seeingfactual reality and the fulfilling the practicalresponsibility of defending one’s own ideals –this is the principle to which we must adheremost firmly’ (Weber, 1904).

While value judgements must be kept outof the historical and social sciences, valueswill, according to Weber, inevitably influencethe choice of the objects of study, thus takingon a guiding role for the researcher. Even ifthey play no role in forming judgements,values are still involved in what could becalled a ‘selective function’; they serve to decideupon a field of research in which the study


3069-CH01.QXD 3/6/03 7:27 PM Page 21

proceeds in an objective manner in order toreach causal explanations of phenomena.

Freedom from values was therefore thefirst condition for objectivity in the socialsciences. The terms of the second condition,understood as the ability to produce state-ments which would be to some extent general,remained to be defined. According to Weber,the social sciences are to be distinguishedfrom the natural sciences not on the basis oftheir object (as in Dilthey’s contraposition ofhuman sciences with the sciences of thespirit), nor because their goal is to study socialphenomena in their individuality, since thesocial sciences also aim at formulating gener-alizations; rather, the distinction lies in their‘orientation towards individuality’.

This orientation is primarily one of method.For Weber the method is that of ‘Verstehen’.However, in defining what he means by this,Weber rejects any form of psychologism.Verstehen is neither psychological perspicacitynor sudden illumination; it is the rationalcomprehension of the motivations underlyingbehaviour. It is not intuition, but ‘interpreta-tion’: understanding the purpose of the actionand grasping the intentional element inhuman behaviour. The ability to identify withothers, which is inherent in Verstehen, is alsochannelled towards rational interpretation:putting oneself into the other person’s posi-tion so as to ‘understand’. This involvesunderstanding the motivations of actions, thesubjective meaning that individuals attributeto their own behaviour: because every action,even the most apparently illogical, has itsown inner rationality, its own interior ‘sense’.As Boudon writes:

For Weber, to understand an individual actionis to acquire sufficient means of obtaininginformation to understand the motives behindit. In his view, observers understand the actionof an observed subject as soon as they canconclude that in the same situation it is quiteprobable that they too would act in the sameway. . . . As can be seen, understanding in theWeberian sense implies the ability of theobserver to put him or herself in the actor’s place,but does not in any way imply that actor’s

subjectivity is immediately transparent. . . .Indeed, the Weberian notion of comprehensiondesignates a procedure which is very close towhat textbooks of logic call ‘ampliative induc-tion’ and which consists of reconstructingmotives not directly accessible by cross-checking facts. (Boudon, 1984: 31, 51)

How can this orientation towards individual-ity yield objectivity? If we start with the indi-vidual and the subjective sense of his action,how can we attain objective knowledge thathas general characteristics? Here we are facedwith the second condition for objectivity inthe historical and social sciences.

The answer is provided by the Weberianconcept of the ideal type. For Weber, idealtypes are forms of social action that are seento recur in human behaviour, the typical uni-formity of behaviour constituted through anabstractive process which, after isolatingsome elements within the multiplicity ofempirical fact, proceeds to coordinate theminto a coherent picture that is free from con-tradiction. The ideal type, then, is an abstrac-tion that comes from empirically observedregularities.

The Weberian ideal type impinges upon allfields of social science and can be found at dif-ferent levels of generality, ranging from thesingle individual to society as a whole. Weberexemplified ideal types with reference tosocial structures (for example capitalism),institutions (e.g. bureaucracy, church and sect,forms of power) and individual behaviour(e.g. rational behaviour).

These ‘ideal types’, writes Weber, are not tobe ‘confused with reality . . . they were con-structed in an ideal heuristic manner’ (Weber,1922a); they are ‘ideal’ in that they are mentalconstructs; they carry out a ‘heuristic’ func-tion in that they direct knowledge. Theyare empty shells, ‘fictions lacking life’ asSchutz has described them; they have no con-crete counterpart in reality, but are theoreticalmodels that help the researcher to inter-pret reality. For example, probably none ofthe three ideal types of power Weberdistinguishes – charismatic power, traditionalpower, and rational-legal power – has ever

22 Socia l Research

3069-CH01.QXD 3/6/03 7:27 PM Page 22

existed in its pure form. The ideal type is aclear, coherent, rational, unambiguous con-struct. Reality, however, is much more com-plex, contradictory and disorderly. No form ofcharismatic power that has ever existed hasbeen wholly and exclusively charismatic;though globally identifiable with thisWeberian ‘type’, the actual form will doubt-less contain elements of the other two formsof power.

The regularities that the researcher pursuesand identifies in order to interpret social real-ity are not ‘laws’ in the positivist sense. ForWeber, ‘the number and type of causes thathave determined any individual event what-ever, are in fact, always infinite . . . and thecausal question, when treating the individual-ity of a phenomenon is not a question of lawsbut rather a question of concrete causal con-nections . . . the possibility of a selection withinthe infinity of determining elements’ (Weber1922b). Instead of laws, then, we have causalconnections, or rather, to use Boudon’s expres-sions, mere possibilities or opportunity structures(‘If A, then most frequently B’, Boudon, 1984:75). It is therefore impossible to establish thefactors that determine a certain social event orindividual behaviour, but one can trace theconditions that make it possible.

Thus, in contraposition to the causal laws ofthe positivist approach, which are general anddeterministic (though less so in the more prob-abilistic neopositivist interpretation), we havestatements and connections characterized byspecificity and probability.

5.3 Further developmentsWeber has been discussed at some lengthbecause the work of the great German sociolo-gist anticipated practically all the themes thatwould be subsequently developed in the richvein of sociological theory and research thatgave rise to approaches such as phenomeno-logical sociology (Husserl and Schutz), sym-bolic interactionism (Mead and Blumer) andethnomethodology (Garfinkel and Cicourel),which became established in American socio-logy from the 1960s onwards. All theseperspectives share fundamental characteristics

of the Weberian approach: a strong anti-deterministic conviction; opposition to allphilosophies of history and all forms of evolu-tionism; the fundamental ‘ontological’ differ-ence between natural sciences and socialsciences, and the irreducibility of the latter tothe former’s methods of research; and thecriticism of any attempt to explain humanaction by starting from social systems and theconditioning factors within them. Finally, allof these approaches share – this time in posi-tive terms – a strong conviction that ‘indivi-dual action endowed with meaning’ must beseen as the core of every social phenomenonand of the sociologist’s work.

Weber, however, did not push his method-ological approach to extreme consequences.While he elaborated these concepts in hismethodological writings, in his theoreticalreflections and empirical research he con-stantly operated on a macrosociologicallevel, adopting the perspective of compara-tive history, in an effort to understandmacrostructural phenomena such as theeconomy, the state, power, religion, and thebureaucracy. By contrast, the movement thatarose in the United States in the 1960s devel-oped the Weberian perspective in its naturaldirection, that is, in a ‘micro’ perspective. Ifsociety is built on the interpretations ofindividuals, and if it is their interactionthat creates structures, then it is the inter-action of individuals that one must study inorder to understand society. This convictionopened up a completely new area of socio-logical research, the study of everyday life,which had formerly been disregarded asnon-scientific.

It is clear that the interpretivist paradigmdiffers radically from the positivist frame ofreference. The ‘subjectivist’ view is first of alla reaction to the ‘objectivist’ positivist posi-tion. By treating social reality and humanaction as something that could be studiedobjectively, the positivist approach over-looked the individual dimension: all thoseaspects that distinguish the world of humanbeings from the world of things. The veryelements that disturbed the ‘scientific’


3069-CH01.QXD 3/6/03 7:27 PM Page 23

research of the positivist approach and weretherefore excluded – individual, motivationsand intentions, values, free will, in short,the subjective dimension that cannot beperceived by quantitative tools – become the

primary object of interpretive research. It ison this fundamental difference between theobjects studied that the interpretive point ofview bases its alleged superiority over thepositivist approach. The convinced supporter

24 Socia l Research

Ontology: constructivism and relativism (multiple realities) ‘Constructivism’: theknowable world is that of the meanings attributed by individuals. The radicalconstructivist position virtually excludes the existence of an objective world(each individual produces his own reality). The moderate position does not askwhether a reality external to individual constructions exists, since it claimsthat only the latter can be known. ‘Relativism’: these meanings, or mentalconstructions, vary among individuals; and even when they are not strictlyindividual in that they are shared by the individuals within a group, they vary amongcultures. A universal social reality valid for all persons, an absolute reality, doesnot exist; rather, there are multiple realities in that there are multiple anddifferent perspectives from which people perceive and interpret social facts.

Epistemology: non-dualism and non-objectivity; ideal types, possibilities, opportunitystructures The separation between the researcher and the object of studytends to disappear, just like that between ontology and epistemology. In contrastto the positivist vision, social research is defined as ‘not an experimentalscience in search of law, but an interpretive one in search of meaning’ (Geertz,1973: 5), in which the central categories are those of value, meaning andpurpose. In pursuing its objective, which is to understand individual behaviour,social science can utilize abstractions and generalizations: ideal types andpossibilities or opportunity structures.

Methodology: empathetic interaction between the researcher and the object ofstudy The interaction between the researcher and the object of study duringthe empirical phase of research is no longer judged negatively but constitutes,instead, the basis of the cognitive process. If the aim is to understand themeanings that subjects attribute to their own actions, the research techniquescannot be anything but qualitative and subjective, meaning that they will varyfrom case to case depending on the form taken by the interaction between theresearcher and the object studied. Knowledge is obtained through a processof induction; it is ‘discovered in reality’ by the researcher who approaches itwithout prejudices or preconceived theories.

BOX 1.4 ANSWERS GIVEN BY INTERPRETIVISMTO THE THREE BASIC QUESTIONS

3069-CH01.QXD 3/6/03 7:27 PM Page 24

of the interpretive paradigm affirms not onlythe autonomy and diversity of the historicaland social sciences from the natural sciences,but also their superiority, since only anapproach that adopts the principle ofVerstehen can achieve that understandingfrom the inside which is the basis of theknowledge of behaviour and of the socialworld.

These fundamental differences inevitablyimply different techniques and research pro-cedures. And it is this aspect that most inter-ests us here. Indeed, if the essence of humanlife differs from that of the natural world, thenit should be studied by means of differentmethods from those of the positivistapproach. The subjectivist position cannotadopt ‘the language of variables’. It cannotadopt it in the phase of empirical observationon account of the centrality of intentional andsubjective components which, by definition,escape objective quantification and can beseized only through empathy. It cannot adoptit during the phase of data analysis because itcannot imagine analysing human behaviourin terms of the interaction of separate compo-nents (variables), as the human being is awhole that cannot be reduced to the sum of itsparts. The subjectivist position has thereforedeveloped its own research procedures, itsown observation techniques and its ownways of analysing empirical reality, whichform the body of so-called ‘qualitativeresearch’. This will be discussed in greaterdetail later. For now, we will conclude ourpresentation of the interpretive paradigm bysummarizing this approach according to thescheme shown in Table 1.1.

6. A FINAL NOTE: RADICALIZATION,CRITICISM AND NEW TENDENCIES

In the previous sections we have described –with reference to their fundamental conceptsand their founding fathers – the two paradigms

which have guided social research andshaped its strategies and techniques since itsinception. We will now mention the criticismslevelled at these two approaches and a fewinstances of their radicalization.

For what concerns the positivist paradigm,we have seen that great attention wasfocused, especially in the period of neoposi-tivism, on formulating and developing empiri-cal procedures. The radicalization of thistrend gave rise to a sort of anti-speculativeempiricism in which ‘the method’, and subse-quently ‘the data’, reigned supreme; the taskof the social scientist was no longer to formu-late theories and then to test them empirically,but to collect and describe data underthe naïve illusion that ‘the data speak forthemselves’.

This was a process of progressive reduction(hence the accusation of ‘reductionism’) thatwent through various phases. First, theboundaries of theoretical exploration wereshrunk; questions of verification, or confirma-tion of hypotheses (ars probandi), werestressed at the expense of discovery (ars inve-niendi). Subsequently, attention was shiftedfrom the content to the method. This empha-sis on empirical validation meant thatquestions which could not be translatedimmediately and simply into empiricallyverifiable procedures were excluded from the-oretical considerations. Theoretical complexitywas therefore gradually reduced to banality.Finally, attention was shifted from the methodto the data, from the operationalization ofconcepts to the practical problems of collec-tion and analysis of data (perhaps even statis-tically sophisticated) – data which by nowwere bereft of theoretical and methodologicalbackground. As Luciano Gallino points out,‘The immediate results of the research werewhat the critics of sociological neopositivismmight have expected: a huge mass of data,meticulously recorded, measured and classi-fied, but uncoordinated, lacking significantconnections, and unable to yield adequateknowledge of the object to which they nomi-nally refer’ (Gallino, 1978: 457).


3069-CH01.QXD 3/6/03 7:27 PM Page 25

Interpretivism was no less exposed tocriticism. It was not so much Weber’s originalmodel as its subsequent interpretations thatcame under fire; as we have seen, these tookto the extreme the original concept of ‘orien-tation towards the individual’. Weber himselfstrove to go beyond subjectivity. He did notrule out the possibility of reaching forms ofcognitive generalization (ideal types); more-over, a considerable number of his metho-dological treatise deal with his attempt toreconcile causality and comprehension. Inaddition, although he started out by focusingon the individual, he did not neglect the greatsystemic issues or the institutional dimensionof society.

By contrast, the new schools of sociologicalthought that developed from the 1960sonwards accentuated the subjective characterof Weber’s original model and shifted theirattention to the world of everyday life and tointer-subject interaction. Again, this occurredthrough a process of reduction, though in thiscase it was the breadth of reflection that wasreduced, while in the case of neopositivismthe reduction was in the depth of reflection.This shift gave even greater impetus to thetwo basic criticisms levelled at the interpre-tive paradigm.

The first of these holds that extreme subjec-tivity rules out the very existence of science,and of social science in particular. If humanaction always has a unique dimension or ifreality is merely a subjective construction,then generalizations above the individuallevel cannot be made and knowledge cannotbe objective. Moreover, the objectivity ofknowledge is also denied by the very mecha-nism through which knowledge is pursued,since this involves the non-separation of theresearcher from the object studied. In addition,the fact that the researcher cannot transcendthe object studied also excludes the possibilityof inter-subject verification, which is a funda-mental principle of science (that is to say, thatanother researcher can obtain the same resultby elaborating the same or other data).

Second, the interpretive approach – againon account of its focus on the individual – is

accused of ignoring those objects that shouldstand at the centre of sociological reflection:institutions. Thus, it allegedly neglects aspectsof society which, though stemming from indi-vidual interaction, have become independentof individuals and their choices. This samebasic criticism is also levelled at phenomeno-logical sociology, ethnomethodology andsymbolic interactionism, which are accused oflimiting their interests to interaction andinterpersonal relationships, in that they areunwilling or unable to address problems thattranscend the minutiae of everyday life.

Up to now we’ve discussed these issuesagainst the backdrop of the major currents ofsociological thought, on which the disciplineof sociology was founded, which have shapedits research techniques and dominated socialenquiry from its very beginnings up to themid-1970s. The last quarter of the twentiethcentury has challenged the preceding historyof social research. The 1960s – featuring thecivil rights movement, student protests, racialconflicts in urban settings, struggles againstpoverty and inequality, and the rise of femin-ism – were an extremely lively period inWestern societies. Sociological theory andresearch played a central role and achieved agreat degree of popularity in such a context,and sociology seemed to uncover a new ‘mis-sion’ in its reflections on that decade’s socialchanges. There emerged new theoretical per-spectives, such as the neo-Marxian and neo-Weberian approaches, critical theory and othernew radical perspectives which openly con-tested the comfortable alliance betweenneopositivism and functionalism that had pre-viously dominated social theory and research.

In those same years these new macro-perspectives were accompanied by noveldevelopments in the field of so-called ‘micro-sociology’, an umbrella term groupingdifferent schools of thought and theoreti-cal outlooks, that resembled each other intheir interest for the ‘minor’ facts of everydaylife, micro-interactions among individuals,interpersonal dynamics (rather than greathistorical transformations and major socialprocesses).

26 Socia l Research

3069-CH01.QXD 3/6/03 7:27 PM Page 26

This abandonment of comprehensivetheoretical perspectives and wide-rangingexplanations eventually led to a generalizedcritique of any theoretical explanation andquestioned sociology’s status as a science.This tendency has assumed particularly radicaltraits in recent years (in the 1990s, roughlyspeaking) in a heterogeneous (and sometimesconfusing) intellectual movement that hasbeen labelled ‘post-modernism’.

In extremely simplified terms, one coulddefine this movement in terms of what it chal-lenges: modernism, i.e. the consequences ofthe Enlightenment, including the critical useof reason over humanity, nature, and society,and confidence in science, based on order,rationality, simplicity of scientific explana-tions and the cumulative nature of knowl-edge. ‘Post’-modernism transcends (anddisputes) modernity’s achievements, with acritique which can be summed up in fourpoints: (a) rejection of general theories, whichstands accused of totalitarianism, culturalimperialism, negation and repression ofdifferences among societies in order to perpet-uate the hegemonic goals of Western culture;promotion of multiple theoretical approachesand languages; defence of the fragmentaryand non-unitary nature of scientific explana-tion; (b) rejection of rationality, linearity, andscientific knowledge’s simplicity; praise forparadoxes, contradictions, opacity, alternativeand incompatible multi-faceted outlooks;(c) exaltation of differences, multiplicity of localand contextual truths, rejection of the cumula-tive nature of science; and (d) exaltation of the‘Other’, differences, minorities; identificationwith the oppressed, assumption of ‘power’ asan explanatory category at the basis of allsocial relationships and structures.

This overview of recent tendencies andpotential paradigms in contemporary socialscience is too simple and brief, but we will notfurther develop the issue. Our primary interestis to describe the basic social science paradigmswhich have influenced and shaped empiricalresearch strategies, methods and techniques.The new perspectives which have emergedin the last quarter-century have not had

revolutionary effects on social researchtechniques, except for promoting the fulllegitimacy and actual use of qualitativeresearch techniques (but without innovatingthem in any appreciable way).

SUMMARY

1. Any ‘mature’ science is accompanied by,in any given moment in history, its ownparadigm. Each science’s paradigm is its‘guiding vision’, a theoretical perspectiveaccepted by the community of scientiststhat directs research effort by specifyingwhat to study and formulating hypothe-ses to explain observed phenomena.

2. In the social sciences, the two paradigmsthat have historically oriented socialresearch since its inception have been‘positivism’ and ‘interpretivism’. In orderto compare them, we have attempted tounderstand how they deal with the threefundamental questions facing socialresearch: the ontological question (doessocial reality exist?); the epistemologicalquestion (is it knowable?); and the episte-mological question (how can we acquireknowledge about it?).

3. The positivist paradigm started to takeroot in social research in the second half ofthe nineteenth century, due to the greatsuccess achieved by the natural sciences.Positivism applied to social researchmaintained that social reality should bestudied through the same investigativelogic and the same method of the naturalsciences, hence the name ‘social physics’attributed to the study of society.

4. Over the twentieth century the originalpositivist outlook has been adaptedto overcome its intrinsic limits. Accordingto the neopositivist and postpositivistparadigm, social theories are no longerexpressed in terms of deterministic laws,but of probability. Any theoretical state-ment has a provisional nature, is neverdefinitively proven and always remains


3069-CH01.QXD 3/6/03 7:27 PM Page 27

exposed to possible disproof. Moreover,the research community grew increasinglyconvinced that any empirical observationis not an objective depiction, but is rathertheory-laden, in the sense that even thesimple recording of reality depends onthe mental framework employed by theresearcher. This revised form of posi-tivism, however, does not repudiate itsempiricist foundations nor its faith inquantification and generalization; andit promoted a further development ofquantitative empirical research methods,the so-called ‘language of variables’, alanguage borrowed from mathematicsand statistics.

5. According to interpretivism, there exists afundamental ‘epistemological’ differencebetween social and natural sciences. Thisperspective holds that social reality cannotsimply be observed, but rather needs to be‘interpreted’. In the natural sciences theobject of study consists of a reality that isexternal to the researcher and remains soduring the course of research; thus,knowledge takes the form of explanation.In the human sciences there is no suchdetachment between the observer andwhat is observed; and knowledge can beobtained only through a totally differentprocess, that of comprehension (Verstehen).These fundamental differences inevitablyimply different techniques and researchprocedures. The subjectivist positioncannot adopt the ‘language of variables’and has therefore developed its ownobservation techniques and its own waysof analysing empirical reality, which formthe body of so-called ‘qualitative research’.

FURTHER READING

A useful collection of essays that explore thetheoretical perspectives that have shapedsocial research methods is the reader editedby G. Ritzer and B. Smart, Handbook of SocialTheory (Sage, 2001, pp. 552). The issues

addressed in this chapter are further exam-ined in M. Gane, Durkheim’s Project for aSociological Science; P. Halfpenny, Positi-vism in Twentieth Century; S. Whimster, MaxWeber: Work and Interpretation; K.L. Sanstrom,D.D. Martin and G.A. Fine, Symbolic Inter-actionism at the End of the Century; S. Crook,Social Theory and the Postmodern.

An introductory discussion about the paradig-matic divisions between quantitative and qual-itative research traditions is given in the firstchapter of A. Tashakkori and C. Teddlie, MixedMethodology: Combining Qualitative andQuantitative Approaches (Sage, 1998, pp. 185).A more comprehensive guide to the differentanswers given to fundamental social researchdilemmas by classical and contemporaryschools of thought can be found in N. Blaikie,Approaches to Social Inquiry (Polity Press,1993, pp. 238).

An attempt to place current approaches to quali-tative research in a theoretical perspectivecan be found in an essay by Y.S. Lincoln andE.G. Guba, Paradigmatic Controversies, Con-tradictions, and Emerging Confluences, inDenzin and Lincoln (2000). Another, moredetailed attempt, is the book by J.F. Gubriumand J.M. Holstein, The New Language ofQualitative Method (Sage, 1997, pp. 244): theauthors identify four ‘idioms’ (naturalism,social constructionism, emotionalism, post-modernism) which inspire recent qualitativeresearch. A discussion of current trends insocial research from a quantitative standpointcan be found in J.H. Goldthorpe, On Sociology:Numbers, Narratives, and the Integration ofResearch and Theory (Oxford University Press,2000, pp. 337).

NOTES

1. The treatment illustrated in the followingpages borrows heavily from Guba and Lincoln(1994), which deals with the topics moreextensively.

2. Ontology: that part of philosophy that stud-ies the essence of ‘being’; from the Greekóntos (to be, being) and lógos (discourse,reflection).

28 Socia l Research

3069-CH01.QXD 3/6/03 7:27 PM Page 28

3. Epistemology: reflection on scientific knowl-edge, from the Greek epistéme (certainknowledge).

4. Methodology: from the Greek méthodos(pathway to, method). The methodological ques-tion has to do with ‘methods’ of social research,meaning an organic body of techniques. It couldalso be called (perhaps more correctly) ‘techno-logical question’, in that it focuses on techniques;this term has been avoided as it has taken on adifferent meaning in the common language.

5. Stuart Mill states that induction is ‘thatoperation of the mind by which we infer what weknow to be true in a particular case or cases, will

be true in all cases which resemble the former incertain assessable respects’ (Mill, 1843: 288).

6. Some epistemological questions (regardingthe knowability of reality) are introduced into ourdiscussion of the ontological issue (the essenceof reality) in order to facilitate understanding forthe reader new to these concepts. Moreover, aswill be seen in the section on the interpretiveparadigm, the two issues are inseparable.

7. The criticisms of neopositivism that gave riseto what is now called postpositivism are generallyattributed to Kuhn, Lakatos and Feyerabend.

8. The expression comes from Hanson(1958).


3069-CH01.QXD 3/6/03 7:27 PM Page 29

Documents

Fundamentals of methodology Lecture 02: Scientific Knowledgeuser.mendelu.cz/cenek/Fundamentals_of_Methodology/Lectures_and… · debate while recognizing that there are di ﬀerences