Upload
louise-franklin
View
216
Download
0
Embed Size (px)
Citation preview
The Comparative MethodThe Comparative MethodLecture 2
Why and how to compare countries;
description – classification – hypothesis-testing – prediction
problems of comparisons
ObjectivesObjectives
to know and understand the basics of the comparative method; its applications; its scope and limitations
Research traditions in comparative Research traditions in comparative politicspolitics
Early example of CP:Early example of CP:
Aristotle’s Aristotle’s Politics Politics (350 BC): compared (350 BC): compared constitutions of Greek city statesconstitutions of Greek city states
Growth in the Discipline Post WW2:Growth in the Discipline Post WW2: Doubling of Independent statesDoubling of Independent states south-east Asia, Middle East, north Africa and Africa south-east Asia, Middle East, north Africa and Africa
south of the Saharasouth of the Sahara WE since 1990: more than a dozen new cases for WE since 1990: more than a dozen new cases for
comparing liberal-democracy comparing liberal-democracy
SUBJECT and METHOD
Comparative Politics as a Subject
Examines domestic politics and government within numerous countries, whereas international politics looks at relations between different countries.
Comparative Politics as a Method
Comparative political analysis Various ways of analysis
Trends in the Comparative DisciplineTrends in the Comparative Discipline
Jean Blondel Jean Blondel Comparative GovernmentComparative Government
Three main phases in the study of CP:Three main phases in the study of CP:
1.1. Constitutionalist phase (Aristotle – Constitutionalist phase (Aristotle – 1900s)1900s)
2.2. Behaviouralist phase (1940s-1960s)Behaviouralist phase (1940s-1960s)
3.3. Institutionalist phase (1970s-)Institutionalist phase (1970s-)
Why compare?Why compare?
What does a comparative approach bring to What does a comparative approach bring to the study of politics?the study of politics?
1.1. KnowledgeKnowledge
2.2. ClassificationClassification
3.3. Formulate and test hypothesesFormulate and test hypotheses
4.4. Generalisations and predictionsGeneralisations and predictions
Hague and Harrop 2001; Landman 2003Hague and Harrop 2001; Landman 2003
1. Knowledge1. Knowledge
the simplest and the best reasonthe simplest and the best reason
In 1925, Munro described the purpose as In 1925, Munro described the purpose as aiding’ the comprehension of daily news aiding’ the comprehension of daily news from abroad’. from abroad’.
Background information about foreign Background information about foreign governments not only helps to interpret governments not only helps to interpret new developments, it also enables us to new developments, it also enables us to view our own country in a fresh light. view our own country in a fresh light.
2. Classifications2. Classifications
The classification of executives into The classification of executives into presidential and parliamentary types, presidential and parliamentary types, allows to look at the origins and effects of allows to look at the origins and effects of each. each.
Comparative method allows to observe Comparative method allows to observe variations of a concept or modelvariations of a concept or model
Without a classification of governments, Without a classification of governments, we have nothing to explainwe have nothing to explain
3. Formulate and test hypotheses3. Formulate and test hypotheses
We can develop and scrutinize questions as: We can develop and scrutinize questions as: Do plurality electoral systems always produce a two-Do plurality electoral systems always produce a two-
party system? party system? Do revolutions only occur after a country has suffered Do revolutions only occur after a country has suffered
defeat in war? defeat in war?
Without comparison we would lack general Without comparison we would lack general knowledge of politics and therefore the ability to knowledge of politics and therefore the ability to explain particular observations. explain particular observations.
‘‘You cannot be scientific if you are not You cannot be scientific if you are not comparing’. comparing’. The American political scientist James ColemanThe American political scientist James Coleman
4. Generalisations can generate predictions4. Generalisations can generate predictions
Example: Example: If we find that the plurality method of election always If we find that the plurality method of election always
produces a two-party system, we can predict that produces a two-party system, we can predict that countries which switch to this formula will probably countries which switch to this formula will probably witness a fall in the number of parties represented in witness a fall in the number of parties represented in their parliamentstheir parliaments..
Studying one case leads to studying several Studying one case leads to studying several cases, upon which a theory can be builtcases, upon which a theory can be built
A theory allows to explain singular cases againA theory allows to explain singular cases again
Additional benefitsAdditional benefits
To help us address the To help us address the counterfactuals – ‘what if…?’counterfactuals – ‘what if…?’
to avoid ethnocentrismto avoid ethnocentrism
Mackie and Marsh, 1995; Dogan and Pelasy, 1990Mackie and Marsh, 1995; Dogan and Pelasy, 1990
The difficulties of comparisonThe difficulties of comparison
a) Conceptual stretchinga) Conceptual stretching
b) interdependenceb) interdependence
c) too many variables too few casesc) too many variables too few cases
d) selection biasd) selection bias
Cases, units of observation, Cases, units of observation, variables, and observationsvariables, and observations
Cases: The countries that we study Cases: The countries that we study (France, Nicaragua, Egypt)(France, Nicaragua, Egypt)Units of observation: The `things’ that we Units of observation: The `things’ that we study (trade unions)study (trade unions)Variables: The features of the `things’ that Variables: The features of the `things’ that can vary (legal status, membership can vary (legal status, membership figures)figures)Observations: The data points (CGT in Observations: The data points (CGT in France 1982)France 1982)
a) Conceptual stretchinga) Conceptual stretching
Countries must be compared against a common Countries must be compared against a common concept but the meaning of that concept may concept but the meaning of that concept may itself vary: itself vary: The connotation of ‘national pride’ differs considerably The connotation of ‘national pride’ differs considerably
between, say, Germany and, Greecebetween, say, Germany and, Greece
Analyzing political behavior across countries Analyzing political behavior across countries depends on the conventions of the country depends on the conventions of the country concernedconcerned E.g. an EP voting against his/her own governmentE.g. an EP voting against his/her own government
Use more abstract conceptsUse more abstract concepts
b) Interdependenceb) Interdependence
Countries do not develop separately from each Countries do not develop separately from each otherotherThey copy, compete with, influence and They copy, compete with, influence and (sometimes) invade each other in a constant (sometimes) invade each other in a constant process of interactionprocess of interaction The spread of Napoleon’s The spread of Napoleon’s Code CivilCode Civil The impact of industrializationThe impact of industrialization The impact of the European Union (Europeanization)The impact of the European Union (Europeanization)
Galton’sGalton’s problem: the difficulty of testing whether problem: the difficulty of testing whether similarities between nations are caused by similarities between nations are caused by diffusion across countries or alternatively by diffusion across countries or alternatively by parallel but independent developmentparallel but independent development
c) Too many variables, too few countriesc) Too many variables, too few countries
a major problem for scholars:a major problem for scholars:The small-N problem (not enough cases at The small-N problem (not enough cases at hand)hand)Only a handful of cases in WE politicsOnly a handful of cases in WE politicsHow can we isolate one factor to test our How can we isolate one factor to test our hypothesis?hypothesis? E.g. Why do we find the strongest communist parties
in France and Italy? Possible answers: strong catholic church OR late
inclusion of working class into political process OR Both
For more valid explanations more cases would For more valid explanations more cases would be necessary, but are simply not therebe necessary, but are simply not there
c) Too many variables, too few countriesc) Too many variables, too few countries
Often a problem even if the number of cases Often a problem even if the number of cases exceeds the number of variables:exceeds the number of variables:
Variables must indeed vary over countriesVariables must indeed vary over countries
Does PR lead to multi-party systems?Does PR lead to multi-party systems?
Cannot be tested if all countries have PR and Cannot be tested if all countries have PR and a MPS, no matter how many countries are a MPS, no matter how many countries are comparedcompared
d) Selection biasd) Selection biaswhen the choice of what to study, or even how to study when the choice of what to study, or even how to study it, produces unrepresentative resultsit, produces unrepresentative resultswhen generalizations cover only a small number of when generalizations cover only a small number of countriescountriesoften an unintended consequence of a process of case often an unintended consequence of a process of case selection that is arbitrary but not truly random:selection that is arbitrary but not truly random:
E.g. choosing countries which speak the same language, or E.g. choosing countries which speak the same language, or which we have personal relations/experiences with, which we have personal relations/experiences with,
The result is that findings of comparative politics are The result is that findings of comparative politics are often weightedoften weighted
toward consolidated developed democracies (actually a rare toward consolidated developed democracies (actually a rare form of polity in the expanse of human history) form of polity in the expanse of human history)
Large, powerful countries.Large, powerful countries.
Only covering a large number of countries reduces the Only covering a large number of countries reduces the selection bias risk. selection bias risk. If the study covers all countries, selection bias If the study covers all countries, selection bias disappearsdisappears
Further Pitfalls of ComparingFurther Pitfalls of Comparing
Pitfalls:Pitfalls:
1.1. Description is not Comparison (Macridis, 1955)Description is not Comparison (Macridis, 1955)
2.2. The persistence of ‘Cultural Idiosyncrasies’ The persistence of ‘Cultural Idiosyncrasies’ (Mayer, 1972; Ragin, 1987)(Mayer, 1972; Ragin, 1987)
3.3. Trade-off between number of cases and level Trade-off between number of cases and level of detailof detail
4.4. Ecological and individual fallaciesEcological and individual fallacies
Ecological FallacyEcological Fallacy
Support for the Extreme Right in the Support for the Extreme Right in the German General Election of 1990 German General Election of 1990 concentrated in areas with high proportion concentrated in areas with high proportion of foreignersof foreigners
Conclusion: Foreigners supported the Conclusion: Foreigners supported the Extreme Right – WRONGExtreme Right – WRONG
Lesson: Do not use macro level data for Lesson: Do not use macro level data for inferences about micro level behaviourinferences about micro level behaviour
Individual FallacyIndividual Fallacy
Do not use micro data to make statements about Do not use micro data to make statements about the macro levelthe macro levelUnless macro features are analytical, i.e. simply Unless macro features are analytical, i.e. simply aggregate measuresaggregate measuresDoes a prevalence of individual authoritarian Does a prevalence of individual authoritarian attitudes render a attitudes render a societysociety authoritarian? authoritarian? Yes if `society’ is defined as distribution of individual Yes if `society’ is defined as distribution of individual
features (cf. Almond/Verba)features (cf. Almond/Verba) No if `society’ includes institutions, works of art etc.No if `society’ includes institutions, works of art etc.
Whenever and wherever "1", "2", and "3" are present revolution will occur – a comparative (general) statement.
Case study: Causes of a Revolution
Hypothesis:
"Revolution is caused by the combination of three factors:
1. High income inequality,2. conflict within the governing group, 3. defeat in war."
Method of AgreementMethod of Agreement
Case 1
abc Revolutiondef
abcghi
Revolution
Case 2
Method of DifferenceMethod of Difference
Case 1
abc Revolutiondefg
Case 2
--- No Revolutionde f g
Strategies for ComparisonStrategies for Comparison
For large number of cases, use regression or For large number of cases, use regression or other statistical techniquesother statistical techniquesFor small number of cases For small number of cases Method of differences leads to Most Similar Systems Method of differences leads to Most Similar Systems
Design (MSSD, popular in area studies)Design (MSSD, popular in area studies) Method of agreement leads to Most Different Systems Method of agreement leads to Most Different Systems
Design (MDSD)Design (MDSD)
Both methods focus on one key explanatory Both methods focus on one key explanatory variable, others constant or variedvariable, others constant or variedUseful, but cannot overcome the basic problem Useful, but cannot overcome the basic problem of small n (third variables)of small n (third variables)
MSSD vs. MDSDMSSD vs. MDSD
Compare how?Compare how?
Case studies:Case studies:Representative casesRepresentative cases –the study of a typical, –the study of a typical, standard example of a wider category.standard example of a wider category.Prototypical casesPrototypical cases – a topic is chosen not – a topic is chosen not because it is representative but because it is because it is representative but because it is expected to become soexpected to become soDeviant caseDeviant case –to cast light on the exceptional –to cast light on the exceptional and the untypical; can provide the variation and the untypical; can provide the variation without which well-founded explanation is without which well-founded explanation is impossibleimpossibleCrucial caseCrucial case – seeks initial support for a theory – seeks initial support for a theory by testing it in favorable conditions.by testing it in favorable conditions.
ConclusionConclusion
The comparative method is both a subject The comparative method is both a subject and a methodand a methodIt allows the researcher to gain knowledge It allows the researcher to gain knowledge on other countries/systems, to provide on other countries/systems, to provide classifications, to test hypotheses and to classifications, to test hypotheses and to make predictionsmake predictionsThe advantage of the method is surplus The advantage of the method is surplus knowledge compared to single case knowledge compared to single case studiesstudies