Software_and_Fieldwork_Friese_in_Hobbs-Ch19.pdf

Part 9Fieldwork, Science and Technology

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 307

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 308

The aim of the chapter is to give you anintroduction to Computer-Aided QualitativeData Analysis Software (CAQDAS) and howit can be utilised in fieldwork. To set thestage, I review the almost 20 years longhistory of qualitative computing and providesome anecdotes based on my personal expe-rience with the field. Then I provide someexamples on how qualitative computing soft-ware can be put into use while in the field orwhen returning from the field. With this asbackground information, we look at the link-age between software and method. I present ageneral model of qualitative data analysis andexplain how the various steps can be translatesinto computer functions and a computer-aided analysis. This gives a first impressionwhich computer function might be impor-tant and necessary for which purpose. In thesection ‘What to look for when deciding on apackage’ the central functions of CAQDASand their variations are explained. The aim isto sharpen the view and to enable the readerto make a better informed decision when itcomes to matching personal and projectrequirements with a particular package.The lastpart of the chapter is devoted to a description

of four popular programs (in alphabeticalorder): ATLAS.ti 5, The Ethnograph v 5.08,MAXqda, and QSR Nud*ist 6. The softwaredescriptions are structured around the centralfunctions so that a comparison becomespossible.

LOOKING BACK AT ALMOST 20 YEARSOF QUALITATIVE COMPUTING

The first programme developed for thepurpose of supporting qualitative data analy-sis was The Ethnograph (http://www.qualis-research.com). The programme was firstlaunched in 1985. Thus, we can look backat a history of nearly 20 years of computer-supported qualitative data analysis. At thebeginning of the 1990s, a number of addi-tional software packages were launchedlike Nud.ist 3, MAX, HyperResearch andATLAS.ti. At this time, all programmes wereDOS-based and offered basic code andretrieve functions. Nevertheless, they werenot all the same. The developers or designersoften were researchers who needed com-puter support for a particular project. As a

19Software and Fieldwork

S U S A N N E F R I E S E

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 309

HANDBOOK OF FIELDWORK

consequence, the requirements posed by thedata at hand, the research questions and thechosen methodological approach guidedthe development and the design of the soft-ware. The study motivating the developmentof N4, for example, included large amountsof open-ended questions from a survey. Thisnecessitated a tool that allowed for auto-matic data processing on the basis of com-mand files. The first version of MAX wasoriginally also developed to support theanalysis of open-ended questions in surveyquestionnaires. However, the features ofNud.ist 3 and MAX were not the same as themethodological approach differed. MAXwas designed to support the methodologicalapproach of case-oriented quantificationbased on the works of Max Weber and AlfredSchütz (Kuckartz, 1995). The developmentof ATLAS.ti was guided by project needs aswell as by a combination of various meth-ods, that is, phenomenology, hermeneuticsand Grounded Theory (Böhm et al., 1992).This heritage is still obvious today in certainfeatures like the ‘code family’ or in menulabels like ‘open coding’. Thus, there is a storyto tell about all programmes and their devel-opment. Let me finish with an anecdote thatled to the development of the pioneer of allprogrammes, The Ethnograph.

John Seidel, a sociologist by training,developed The Ethnograph while he wasworking on this PhD thesis. At the time, per-sonal computers were not as widespread astoday. Software for statistical analyses ran onlarge mainframe computers and in order toobtain results, it was necessary to type in anappropriate syntax to tell the computer whatto do. This was not the same as programmingsoftware but very similar to it. John workedas an assistant in the research lab supportingstatistical applications, but for his PhD hecollected qualitative data. His raw data con-sisted of many pages of transcripts. All

readers who have conducted a qualitativedata analysis by hand know what this entails.Piles of paper need to be ordered and sorted,cut into smaller units, pasted on to differentpapers according to themes, sorted and orderedagain, and so on. John, while in the midst ofmaking sense of his data, however, lived notalone. There were also two cats in the house.They loved to stroll around the stacks of paper,though not always respecting that there was aparticular order to them that should not bemessed up. Before a catastrophe could occur(at least from the viewpoint of a PhD student),John utilized his skills gained from workingwith mainframe computers to literally movehis piles of paper from the floor into the com-puter. This resulted in the development ofThe Ethnograph 1.0. Fellow students andcolleagues were amazed to see that comput-ers could also be used to support qualitativedata analysis. When demonstrating his soft-ware and creating an output of segmentssorted by selected code words, they gatheredaround the dot-matrix printer to celebratethe wonders of computer technology. Basedon popular demand, John continued todevelop the programme that was originallyintended to only serve his PhD research.Version 2 was made available to colleaguesand friends, and version 3 was the first com-mercial version released in 1985.

Since 1985 development has not stopped.In 1995, Prein, Kelle and Bird provided anoverview of twelve CAQDAS programmes;Weitzman and Miles (1995) described tenprogrammes. Today most of them still exist,joined by a number of new packages. All pro-grammes have been developed further andtoday most of them are no longer DOS butWindows-based programmes, supplementedby a few native Macintosh programmes.1 Thedistribution of these programmes varies. Someare not spread much beyond country bound-aries like Kwalitan, which is mainly used in

310

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 310

The Netherlands, or AQUAD, which ismainly used in Germany. Others, such asQSR NVivo and N6, MAXqda or ATLAS.ti,have a worldwide user group. These pro-grammes offer sophisticated features andup-to-date technology. On the downside, thelearning curve is steeper and users needto invest more time to become skilled athandling these packages.

With increasing development of CAQ-DAS, some diversification has taken place.The main features of CAQDAS in the earlydays of qualitative computing were the codeand retrieve functions. Today a number ofthe programs have become quite sophisti-cated and offer a variety of other features (see‘Software Description’, page xxx below).There are still software packages that offerjust basic code and retrieve functions, someof them are distributed as freeware. Forcertain groups of users such packages areentirely adequate, offering all the functional-ity they need.

The diversification of CAQDAS has somepositive but also some negative side-effects.The focus on different features above andbeyond basic code and retrieve functionshave made it easier to advise potential userswhich package to choose. For example,there are only a few programmes that sup-port the analysis of photographs, audioand video data (e.g., HyperResearch andATLAS.ti). Some programmes offer featuresor add-on modules to combine qualitativeand quantitative content analysis, such asMAXqda or C-I-SAID. Others, like NVivo,are strong in supporting analyses based ona combination of codes, data attributes andcross-tabulation. On the negative side, itbecomes more and more difficult for oneperson, even though an expert in CAQDAS,to stay on top of the development and toknow all available programs in detail.Comparative reviews will therefore become

scarcer in the future or are limited to fewerprograms.

USER ADAPTATION OF CAQDAS INTHE TWENTY-FIRST CENTURY

Gathering around a dot-matrix printer andcelebrating a computer printout may soundridiculous to (at least some) computer userstoday. But even in 2004, there are qualitativeresearchers who refuse to use computer pro-grams to support their research. Such a refusalis adequate in some cases because not all qual-itative methods call for a code and retrieveapproach (e.g., sequential analytic approachesor objective hermeneutic). In other cases, thenegative responses to software is related to thebelief that the researcher would lose the close-ness to the data, or that software would auto-matically code and analyse the data and thustake away the analysis from the human inter-preter, or that software would lead to shallowanalyses. These are prejudices based on anincomplete understanding of what qualitativecomputing is all about.2 CAQDAS does notanalyse data. Moreover, the software is a toolthat (only) supports the process of qualitativedata analysis. Computers are generally verygood at finding things like strings of charactersor coded data segments in a large variety ofcombinations. But it is still the researcher whoneeds to tell the computer which data segmenthas which meaning by way of coding. Let’slook once more at the acronym that wasintroduced at the beginning of this chapter:CAQDAS. It stands for Computer-AidedQualitative Data Analysis Software. It is asomewhat lengthy acronym as compared to‘QDA Software’, a short form that can also befound in the literature. The latter stands forQualitative Data Analysis software and may beresponsible for some of the misunderstand-ings and misperceptions related to CAQDAS.3

SOFTWARE AND FIELDWORK 311

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 311


This explanation is at times also used as anargument against using software according tothe following logic: if the computer doesn’tdo the coding, then what is it good for? Andwithout ‘test driving’ a CAQDAS package,they judge software as not something to beused in a qualitative research context andreturn to their manual methods of usingcolour pencils and filing cabinets. To mymind, they forego an opportunity to improvethe validity of their research. Software, usedappropriately, offers the possibility to verifyor falsify ideas, hypotheses, theoretical con-structs or models at any stage of the researchprocess because the data can easily beaccessed, they can effortlessly be grouped andregrouped, compared and contrasted. It is noproblem to rename, modify or merge codes ifone gains more and more insights into thedata. One’s thoughts about the data are likelyto be different three or six months into theanalysis as compared to the very early stagesand modification of codes and concepts is aninert part of qualitative data analysis. But thisis not the only advantage. With the aid ofcomputers, this process can also be easily doc-umented. The steps of analysis can be tracedand the entire process does not remain insidea black box, as often is the case.

Thus, even if you don’t have cats or kids orother intervening forces that can mess upyour carefully sorted piles of transcripts orfield notes, software offers clear advantagesover manual techniques. It frees the user fromthose tasks that a machine can do much moreeffectively, such as modifying code words andcoded segments, retrieving data based onvarious criteria, searching for words, integrat-ing all material at one place, attaching notesand finding them again, counting the numbersof coded incidences, offering overviews atvarious stages of a project,and so on.When usingsoftware, it becomes much easier to analysedata systematically and to ask questions that

otherwise would not be asked because themanual tasks involved would be too time-consuming. Even large volumes of data, anddata of various media types can be structuredand integrated very quickly with the aid ofsoftware. In addition, a carefully conductedcomputer-aided qualitative data analysis alsoincreases the validity of research results.Especially when one has reached the concep-tual stage of an analysis, it is easy to ‘forget’ theraw data behind the concepts. In case of amanual analysis, it is quite laborious to getback into the data. In a software-supportedanalysis, the raw data are only a few mouseclicks away and it is much easier to remindoneself and to verify or falsify one’s develop-ing theoretical thoughts about the data.

THE USE OF SOFTWARE IN FIELDWORK

As the title of this chapter implies, CAQDAScan not only be used after data have alreadybeen collected, but also during the data col-lection process. This does not apply to allstudies, but there are a number of researchsituations where it can make sense to utilisea CAQDAS package from the very begin-ning. An example would be a research pro-ject where a variety of data types are collectedover an extended period of time, as the casein ethnographic studies, for instance. Fieldnotes can be directly typed into a CAQDASpackage, arising interpretative thoughts intomemos. Pictures can be assigned, organisedinto groups according to specific criteria,such as, dates and themes. Comments anddescriptions can immediately be added todevelop a picture archive. The same can bedone with short sequences of video footage.Interviews may be recorded on a laptop andimmediately archived into the growing data-base inside the CAQDAS package. At thisstage, no detailed analysis is yet taking place,

312

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 312

but valuable thoughts that might otherwisebe forgotten can immediately be writtendown in the form of comments and memosand attached to the data.

This requires some technical know-howand adequate technical equipment, especiallyif one works with multimedia data. In recentyears, though, multimedia technology hasbecome much more accessible and affordablealso for the ‘lay’ user. Most laptops alreadyinclude the possibility to record and to savethe outcome in the form of a digital file; theuse of digital cameras and the skill to down-load pictures on to a computer has entered themainstream, and with the right software youalso no longer have to be an expert to downloadand work on video footage.

CAQDAS offers the possibility to integrateall relevant data materials into one place tocreate a growing database already during theprocess of data collection. This can, of course,also be done without collecting multimediadata. Instead of writing field notes and com-ments into a word-processing package or anote book, they can be typed directly into aneditor provided by almost all CAQDAS pack-ages. A further advantage is that in addition tothe raw data, comments and memos can beincluded.

Another under-explored usage of CAQ-DAS is its support in terms of data represen-tation, as has already been suggested byCoffey, Holbrook and Atkinson (1996). Atpresent, only ATLAS.ti offers an HTML andXML output option.4 But others may followsuit. This allows for projects to be publishedindependently of the software either asWeb hypermedia presentations or distrib-uted on CD-ROMs or DVDs (see Pink,2001). Examples of hypermedia presenta-tions can be found at the following sites:http://www.lboro.ac.uk/departments/ss/visualising_ethnography/index.html and http://lucy.ukc.ac.uk/Stirling/index.html.

As far as the usage of image, audio andvideo in such presentation is concerned,a number of new ethical issues need to beconsidered. This might be a reason why onetoday rarely finds hypermedia presentations.But it may also be a matter of timing relatedto advances in computer technology (price)and user friendliness. It only has recentlybecome feasible to even think of ‘home made’multi- and hypermedia productions withouthaving to involve experts.

SOFTWARE AND METHODS

A common question of novices is whetherparticular software packages are suitable fora specific methodological approach such asgrounded theory. The answer to this questionis simple: one can use any of the softwarepackages for any approach that involves cod-ing. It is a myth that some packages can orcannot be used in conjunction with a partic-ular methodological approach. Related tothis myth is the implicit assumption that amethodological approach somehow magi-cally emerges just because one enters andcodes data in a CAQDAS package. The disen-chanting truth is that learning a methodolog-ical approach and a software that supportsthe analysis process are two distinct, even ifrelated, processes. You should have at leastsome basic knowledge about qualitative dataanalysis techniques before you embark on acomputer-aided analysis. The purpose ofCAQDAS is not to manipulate or direct youranalysis; its purpose is simply to support it.The user should be the one manipulating thesoftware to make it perform those tasksthat best suit the selected methodologicalapproach. This sounds logical, but everydayreality often looks different. Therefore, below,a very basic model of qualitative data analy-sis is presented, followed by an explanation of


19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 313


how it can be translated into a computer-supported analysis.

According to Seidel (1998), the generalidea behind an analysis method that is basedon a code-and-retrieve approach is that dataare broken up, separated or disassembledinto pieces, parts, elements or units. The pur-pose of it is to end up with manageable piecesof data sorted by themes or concepts. Thisgreatly facilitates comparing and contrastingand thinking about the data. The researchercan then sort and sift the pieces, searching forexample for patterns, classes, sequences,types or processes with the aim to assembleor reconstruct the data in a meaningful andcomprehensible fashion. This process involvesthree basic steps: noticing things, collectingthings and thinking about things (see Figure19.1 below).

These three basic steps can be found inalmost any qualitative data analysis approach.When reading your data, you start noticingthings. You are likely to mark those things inthe margins and add a few notes. Over time,you find other instances of the same note-worthy things in your data and you start col-lecting those things under the same name.You begin to create code words. Over time

you build up a coding system. The nextlogical step is to think about those things youhave noticed and collected. For this to bepossible you need to find and retrieve theseinstances in your data. In this process theunderlying structure of your data willbecome more and more obvious. You will beable to see sequences, patterns, hierarchies,and wholes that had been hidden before inthe mass of your data. While you are thinkingabout your data, you may want to go back tothe original not-fragmented text, you maywant to re-code some passages, add new codewords or get rid of some old ones. The fol-lowing section shows how this process can betranslated into a computer-aided analysis.

THE COMPUTER-AIDED QUALITATIVEDATA ANALYSIS PROCESS

Noticing things – creating a projectand reading your data files

If you use a software package for analysingyour qualitative data, then the preparation ofone or more data files will be your first taskduring or after collecting the data. Text datacan be prepared within any of the commonlyused word processors, such as MS Word,WordPerfect, etc. Depending on the softwarepackage, you need to observe certain format-ting rules. Some packages still only allow youto work with text-only data, other packagesalso accept Word or Rich Text files.

After you have prepared your data files, thenext step is to create a project and to importyour data files into the program. Now you canembark on the task of reading your data files.

Collecting things – coding data files

When reading through your data, soonyou will start noticing multiple instances of

314

Notice things

Thinkaboutthings

Collectthings

Qualitative data analysis

Figure 19.1 The three basic steps of qualitativedata analysis (Seidel, 1998: 2)

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 314

certain occurrences and you will want to markthem with a code word. In order to do so, youhighlight the data segment in question with amouse and assign one or more code wordsto it. Some software packages restrict thenumber of codes you can assign to one datasegment, others have no limitation. But evenif a restriction applies, the restrictions aregenerous enough so that you probably won’tnotice that there is one. All packages allowoverlapping or nested segments as it is com-mon when working with qualitative datathat meaning is ambiguous and not alwaysconfined to whole sentences or paragraphs.Therefore being able to apply overlappingcodes is essential and one of the manyadvantages of using a software package.Nonetheless, coding the data material anddeveloping a suitable coding system is alaborious task also when analysing the datawith the aid of software. As mentionedabove, computers cannot think and there-fore this task has to be undertaken withmuch care. This said, software packages arealso superior to manual techniques duringthis part of the analysis. With the aid of soft-ware, it is much easier to modify coded datasegments, to rename codes globally orlocally, or to reverse coding. Notes, includingfirst interpretations or clarification, can beattached to data segments and one does notend up with post-it notes or scribbled hand-writing all over the place. With the help ofthe retrieval functions the software pack-ages provide, it is also very easy to find thesenotes again.

Thinking about things – searchingdata files

Once a coding system has been developedand the data have been coded, data segmentscan be searched, retrieved and displayed in anumber of ways. A search operation can be

very simple, that is only being based onsingle code word. It is, however, also possibleto build more complicated search requests,which consist of multiple code words linkedby Boolean, proximity, or contextual opera-tors. Such requests can also be combinedwith variables like gender, age, education orother data attributes.

The retrieval of data segments makes iteasier to see things in your data and to thinkabout them. It helps the researcher to recog-nise relationships that previously wentunnoticed because they were disguised bytoo much noise in the data. Some packagesallow the visualisation of the establishedrelationships in form of networks or graphicmodels.

WHAT TO LOOK FOR WHENDECIDING ON A PACKAGE

Prerequisites

As mentioned above, when wanting to useCAQDAS, it is essential to work with a methodthat is based on a code-and-retrieve approach.A clear distinction needs to be made betweenquantitative content analysis programs, such asSonar Professional, Folio VIEWS, etc., and thehere described CAQDAS packages. The focusof CAQDAS is on thick descriptions, verstehenand context-rich analysis, and not on thequantification of qualitative data. Thereforethe first questions that needs to be answeredbefore deciding on a software package is whatkind of method should be applied to collectand analyse data. A well-known method is thegrounded theory approach originally devel-oped by Glaser und Strauss (1967). Otherapproaches are biographical life historyresearch, case studies, phenomenology orethnography. Within each of these methodsvarious ways of data collection and data


19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 315


analysis are described (see Creswell, 1998, foran overview). After you have decided thatCAQDAS is suitable for the kind of analysisthat you intend to conduct, looking at the fol-lowing program features can help you to selecta package that fits your needs.5

Data entry

With regard to data entry, it is worthwhileinvestigating the following questions: Whatkind of data do you want to analyse and isthis a format supported by the software?Some programmes can only handle text doc-uments; others also support the analysis ofgraphic image, photographs, audio andvideo data. If you only want to analyse textdocuments, it may be worth investigatingthe fit of programs that require documentsin text-only format versus those that allow toimport Rich Text or Word documents. Thelearning curve for the latter is often steeperas these programs generally offer more fea-tures and are more complex. For some pur-poses, a simple program might also besufficient. Another issue to look at is the pos-sibility to edit the data material once it isimported into the package and coded.

Coding

The issues to look for here are: What is thesmallest chunk of text that can be coded?How are coded data segments and codewords displayed? Can I use codes that arenot linked to any data segment (e.g., for thepurpose of structuring code lists or whendeveloping models and theory)? Is auto-matic coding possible?

Memos and comments

Analysing qualitative data means writing alot. A great deal of analysis happens while

you write notes in the form of commentsand memos. Therefore, it is vital to look atthe features that the various software pack-ages provide in order to support this process.What kind of notes can you write? Is there adistinction between writing code definitions(= a short comment) and memos, or is thesame technical solution offered independentof content? Look for what kinds of objectcomments and memos can be written for andwhere? How can you track that you haveattached or inserted a memo or commentsomewhere? Is it made visible and how?Further analytic support is provided whenyou can link memos and comments to otherobjects. Is this possible?

Data retrieval

The search functions and features are essen-tial as it is here where the computer powercan be fully utilised. Investigate what can besearched and how search results are dis-played? Can you search for words, strings oftext or text patterns? Can the finds be codedautomatically? The purpose of the text searchand auto coding is to allow you to ‘dive’ intoyour data and to give you a quick feel for whatis there and what to look for. Another appli-cation is to code structured information likethe responses to entire questions, or responsesin focus-group transcripts.

Searching for coded data segments is asignificant part of your analysis after havingcoded the data. Therefore you need to lookout for the ways this is facilitated by the soft-ware package. How conveniently can codeddata passages be accessed? Can they be dis-played in context? What kinds of searchoperator are available? They allow you to askquestions such as: Show me all data passageswhere Code_A and Code_B overlap/where Bfollows A/all passages coded with child codes(lower order) of A, etc. Another important

316

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 316

search function is the ability to restrictsearches to particular documents codes, ordata attributes. This allows you to ask ques-tions such as: Show me all data passageswhere Code_A and Code_B overlap, butonly in interviews with female respondents,age 21–30.

Output and display options

You may not necessarily always want to workin front of the computer. You may miss goingthrough stacks of real paper, and at somepoint in time you want to be able to transferyour analytic work to a word-processing pro-gram to write the final report. For all of this,you need output functions that facilitate thetransfer from the CAQDAS package to aprinter or another software application.Therefore, you should pay attention to theoutput destinations offered by the software.Can output be displayed in an editor, saved toa disk, and printed on paper? Is it saved intext-only or Rich Text format. Some packagesalso provide a quantitative overview of yourdata that can be outputted in an Excel-compatible format. Other issues to look forare: Can you select only parts of a result forreviewing, editing or saving? Is there an easyway to access the context of search results? Isit easy to identify where retrieved data pas-sages come from, how specific is the sourcetag? In addition, a number of packages pro-vide export functions to interface with otherprograms, such as SPSS or mapping software.Depending of your chosen method andanalysis aims, these might also be features tolook out for.

Other features

Some users prefer to organise their codingsystem in a hierarchical manner. Most pro-grams allow for this, but all in different ways.

Thus, if you are interested in a hierarchicalrepresentation, pay attention to how this isrealised and what best suits your needs andworking style.

Team work is also supported in differentways by different packages. When working inteams, it is important that a program offersa merge function to combine work done byvarious team members. Another issue to payattention to is whether it is possible todistinguish work done by different teammembers.

Further decision criteria

There are simpler, and thus easier-to-learnpackages, and more complex programs witha steeper learning curve that may best belearned when attending a professional train-ing session. Therefore, you should ask your-self what kind of computer user you are. Ifyou think of yourself as a novice or not veryexperienced user, insecure in issues relatedto data management, a simpler programmay be a better choice for you. This may alsobe the case, if you only need some basicfunctions and do not need the advancedoptions provided by the more sophisticatedpackages.

A further decision criterion should bewhether the software is only to be used for asingle project, or also in the future for otherprojects and possibly by other users. If it is tobe only chosen for a specific project, then aprogram that fits the specific requirements ofthis project is needed. If one is buying for thelong term, it may be better to choose a packagethat offers additional features, even if they maynot be required at present.

Last but not least …

Rely on your intuition when deciding on aprogram. All companies offer a demo version


19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 317


and tutorial that can be downloaded fromtheir respective websites. Look at the demoversions and go step-by-step through theprovided tutorials and sample project(s). Thetime invested will be worthwhile. A programmay provide all the features that you need,but you may feel uncomfortable with theuser interface and the ways the data are han-dled. This is something you can only find outby gaining some hands-on experience your-self. Reading this chapter or the programdescriptions on the company websites cannever substitute your personal experience.

SOFTWARE DESCRIPTION

The functions considered in the programdescription below are: data entry, coding, textsearch and auto coding, and commentsmemos, data retrieval, mapping and teamsupport. If applicable, special features thatonly apply to a specific program, and thusdistinguish it from others, are described atthe end of each section.6

ATLAS.TI 5

Data entry

Four documents media types are supported:text, graphic, audio and video data. Text doc-uments can be imported as text-only docu-ments, as Rich Text files (RTF), or as Worddocuments. If documents are not yet con-verted to Rich Text, ATLAS.ti 5 providesautomatic conversion of different file for-mats to RTF. As documents are assigned toa project and not imported, they do notbecome part of the project file and thus filesize does not play a role. ATLAS.ti canhandle large data sets, and large video filesalso do not create a problem. The overview

(which can be found at www.qualisresearch.com/QDA. htm) lists the multimedia fileformats that are accepted by ATLAS.ti.

Documents can be grouped into so-calledfamilies. This is a way to assign variables bycreating, for example, a female family, a malefamily, families containing interviews fromrespondents between the ages of 21 and 30,between the ages 31 and 40, and so on. Thesecan later be used to restrict or filter searchresults.

Documents that have been assigned toATLAS.ti can be edited at all stages of theproject.

Coding

Coding is an interactive process and theresults are immediately visible on the screen.Code words are displayed in the margins andthe length of the coded data passage(=quote) is marked by a bracket. The small-est text unit that can be coded is one charac-ter; the smallest unit in audio files is amillisecond, in video files a frame. Graphicalquotes are rectangular sections of the origi-nal image.

There are no restrictions with regard tothe number of levels that can be coded or thelength of a code word. All code words inATLAS.ti are interactive. If you click on acode word once, the text passage codedwith this code word is highlighted. If youdouble-click on a code word, the codeword’s comment is displayed. The modifica-tion of already coded data segments is quickand easy.

Many drag and drop operations are avail-able to facilitate coding, the modification ofcoded data passages, and the unlinking andexchanging of code words. For novices, anoption is provided to deactivate the manydrag and drop operations to make it easier toconcentrate on the basics first.

318

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 318

A code word list is automatically createdand updated while you code your data.Behind each code word, two numbers aredisplayed: the first number tells you howoften a code word is used, indicating the fre-quency or groundedness of a code word; thesecond number informs you about the linksthat exist to the code word, thus indicatingthe density of a code word. A code manageroffers additional functions to manage yourcode word list conveniently. Codes can begrouped in so-called code families and datasegments can be retrieved based on theirfamily membership. Furthermore, codes canbe ordered hierarchically with the aid of thenetwork tools (see ‘Mapping’, below).

Figure 19.2 below shows the ATLAS.ti 5user interface. A coded document is loaded.In addition to codes, hyperlinks andmemos are also shown in the margin. This

can be recognised by the different symbolsused. The code ‘Horror %1’ is highlightedin context and the code definition has beenactivated via a double click. The CodeManager is open, as is commonly the casewhen coding a document.

Text search and auto coding

You can search all primary documentsusing a procedure that is similar to whatyou might know from word-processingprograms. To find instances that areanchored at the beginning or end of a lineor sentence, certain number patterns, etc.GREP expressions can be used. Search pat-terns that are potentially used more thanonce, can be saved. The text search functiononly searches through primary data, butnot through other texts, such as comments


Figure 19.2 The ATLAS.ti 5 user interface with open Code Manager and activatedcode and code definition

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 319


or memos. In order to search through alltexts, the so-called Object Crawler can beused. It finds all first instances of an enteredsearch term in all specified object types.

Text search can be combined with autocoding. This means that you can search for astring of characters in the text and code thehits, the surrounding word, sentence orparagraph automatically with a selectedcode word. This process can run fully auto-matically, or you can control it by decidingeach time a match is found whether or notyou want to code the instance.

Memos and comments

For virtually every object in ATLAS.ti youcan write a memo or a comment.Comments are directly attached to theobject they relate to and cannot be handledindependently. A comment can be writtenfor the entire project, each primary docu-ment, quote, and code word. Commentedobjects are indicated by a tilde sign (~) inthe respective object lists. Memos are inde-pendent objects and can be attached toquotes, codes and other memos. Whenattached to quotes, this is indicated in themargin next to the coded segment (seeFigure 19.2). When linked to codes or othermemos, this can be made visible in the net-work editor (see below). As a way to facili-tate memo management, memo types canbe defined and used for sorting and filter-ing purposes.

Both memos and comments can be for-matted as Rich Text, just like primary docu-ments, including embedded objects.

Data retrieval

ATLAS.ti offers you a wide variety of optionsto search for coded data segments – all in

all there are 14 operators, including Boolean,proximity and semantic operators. Hits listscan be cleaned up or filtered according tospecified criteria, such as only those quotesfrom interviews with female respondents.All finds can be displayed in context orreviewed in an editor, saved to disk orprinted. The output contains clear sourcetags, but unfortunately these cannot easilybe removed to display a generic view, ifdesired. Each query can be saved in the formof a super code. A super code contains thequery and each time you click on a super code,the query is executed again. This is a useful fea-ture when using more complex queries thatyou want to re-run or re-use at a later time inyour analysis. Super codes can also be turnedinto regular codes, presenting a snapshot of thepresent stage of the analysis.

Mapping

A distinct feature of ATLAS.ti is the way thevarious objects of an ATLAS.ti project can bevisualised and linked to each other in so-called network views. Unlike in other pro-grams that offer a mapping or modellingfeature, connections between codes and con-nections between quotes can be named. Theprogram offers seven relation types, butthese can be edited or new ones can bedefined. Objects can be linked visually in annetwork editor, but also via menu functions.

The place of objects in a network editor isnot static. In other words, all objects can bemoved around freely without having to obey adefault structure. From a network view youcan access your primary data directly and dis-play them in context. Figure 19.3 shows anATLAS.ti network view in auto-colour mode.Auto-colouring indicates the groundednessand density of a code. The more frequently acode was applied, the redder it is coloured; the

320

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 320

higher the number of connections to othercodes, the bluer it is coloured. Frequency anddensity are also displayed as numbers behindthe code words (the display of nodes and links,and the auto-colour mode are optional fea-tures). Also shown in Figure 19.2 are a linkedmemo, the context menu for a code node, andthe type of relation between two connectedcodes (i.e., is part of, contradicts). A commenthas been written for the contradicts relation, asindicated by the tilde sign (~).

Network views can be saved as graphic filesand edited in a graphic program. They canalso be copied to the clipboard and directlyinserted into Word files or PowerPointpresentations.

You can create a hierarchical view of yourcoding system by linking codes either in net-work views, by using the link menu optionor per drag and drop in the Code Manager.The hierarchy of codes can then be displayedin the form of a sideways tree, similar to thedisplay of folders and files in WindowsExplorer (Figure 19.4). However, codes

cannot be manipulated in the tree view, onlyvia the above mentioned functions.

Team support

ATLAS.ti supports team work. All users canbe given a user name and if logged in, all newentries are ‘stamped’ with the name of thelogged-in user. Thus, all work done on a pro-ject, such as creating code words, memos,etc., carries the name of the current user(=author). In this way it is possible to tracethe analytic steps of the various teammembers. Documents can be analysed simul-taneously when working on a shared net-work drive, given that each user works on


Figure 19.4 Hierarchical display of codesin ATLAS.ti

Figure 19.3 An ATLAS.ti network view

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 321


his/her own project file. When documentsare edited, the modifications are propagatedto all users of the same document and theaffected document will be synchronised.All affected coded data segments will beadjusted. When splitting a project into sub-parts, these can be put together using theATLAS.ti merge function. The ATLAS.timerge function offers a variety of strategiesto either add or merge the various objecttypes, such as documents, quotes, codes,memos, networks or families.

Other features

Hypertext

ATLAS.ti offers the possibility to link quoteswithin and across documents, thus creatinga hypertext structure this way. The originalsequential document is de-linearised, bro-ken down into pieces, which are then recon-nected, making it possible to traverse fromone piece of data to another piece of dataregardless of their original positions. This isa unique feature of ATLAS.ti.

Additional output andexport options

Projects can be saved as HTML documentand published on a website. In addition, aXML export option is available. This offersthe possibility to represent data in a formthat does not restrict their usability to thecontext and output options provided byATLAS.ti. With the aid of style sheets, userscan then create their own reports, beyond theoptions ATLAS.ti offers.

A further option is the export the entirecoding system form of a SPSS syntax file.When run by SPSS, a data matrix is createdcontaining all codes as variables and allcoded segments (quotes) as cases.

Word frequencies

The ATLAS.ti Word Cruncher creates a listof words for the selected or all primary doc-ument. This feature is useful for a simplequantitative content analysis. A stop list canbe used to exclude certain words, charactersor patterns from the frequency count.Results can directly be opened in Excel.

Summary

ATLAS.ti 5 is one of the more complexCAQDAS packages, offering a large numberof tools and options for analysing qualitativedata. In order to learn the software, you needto set aside some time or, if possible, go on atraining session. However, if you invest thetime, you have made a decision for a productthat incorporates up-to-date Windows tech-nology, offering you powerful tools toanalyse your data. Some of the highlights arethat you can assign Word or Rich Text docu-ments, image, sound and video files. Theinteractive display and handling of codesand coded text passages as well as the hyper-text and network functions are unique inATLAS.ti. You may not utilise all functionsin a first project when still learning the pro-gram, and not all tools may always beexploited in all projects, but with this pack-age you certainly have software that can beextended for future use.

MAXQDA

Data entry

In MAXqda you can only analyse text docu-ments in Rich Text format. Documentscan be assigned to certain text groups, suchas articles, documents, interviews, etc., and

322

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 322

organised into sets, such as interviews withfemale or males, documents from 1950 to1960, etc. Text groups and sets serve in laterstages of the analysis as filter devices toretrieve coded segments that match certainspecified criteria, such as only responsesfrom female interviewees working as teach-ers in primary schools.

Documents that have been importedinto MAXqda can be edited at all stages of aproject. Certain precautions have to be takeninto account when working in teams.

Coding

The smallest unit of text you can code is onecharacter. Text is selected by highlighting theappropriate passage with the mouse. Allcode words are represented in form of asideways tree hierarchy, similar to the kindof tree structure in the Windows Explorer(Figure 19.5).

The structure of the tree can be manipu-lated by dragging and dropping codes toother places in the tree. Symbols next to thecode words indicate whether a memo has beenwritten for a code. Code frequencies arelisted at the right of the code system window.

There is no limit with regard to the numberof code words you can use or the number ofnests and overlaps that result from the codingprocess. The code word tree can be up to tenlevels deep. Modifications of the codingsystem is possible by copying or moving allcoded text at a code word to another codeword. The code words of individual coded textpassages cannot be modified. If this shouldbecome necessary, you first need to deletethem and assign them again.

A coded text in MAXqda can be seen inFigure 19.6.

In the left margin of the displayed text,MAXqda offers three types of information.In the first column of the left margin, yousee the codes displayed against the text. The


Figure 19.5 Code system in MAXqda

Figure 19.6 Coded text and activated memo in MAXqda

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 323


coded segments are referenced by a straightline with a little square box in the middle.When moving the cursor to this box, thecode word for this segment is displayed.After attaching a memo to a text passage,a yellow post-it symbol is shown in thesecond column. When moving with the cur-sor over this symbol, the memo text is dis-played. The third column lists paragraphnumbers.

Memos

Memos can be attached to documents andcodes. Each memo is date and time stamped.Optionally, an author and code words canbe associated. This information can be usedto filter, search and sort the memos. A fur-ther option is to define memo types byselecting different memo icons. Memos arein Rich Text format.


MAXqda offers you a number of option tosearch the text. You can search in all or onlyselected documents or document passages,in retrieved text passages or in memos.Entered search strings can be combinedusing the logic operators AND and OR.The results are displayed in a table, whichcan be sorted by headers. When clicking ona result in the table, the match is high-lighted in context. Search results can alsobe saved and exported as Rich Text file.Rather than just exporting the found searchstring, additional context can be exportedas well.

The result table also offers a function tocode the search results with a selected codeword. One can either code just the hits orselect x number of paragraphs surroundingthe found search string.

Data retrieval

The retrieval of coded text segments followsa simple logic, which is particular toMAXqda. The retrieval logic is based onactivating codes and documents. If onewants to retrieve all coded segments ofCode_A from interviews 1 and 5, one needsto activate Code_A and interviews 1 and 5.The coded segments are immediately dis-played in a separate window. A retrieved seg-ment is clearly indicated by a source tag andwhen selected is displayed in context. Theresults of a search can be saved under a newcode word.

It is also possible to conduct some morecomplex retrievals using Boolean and prox-imity operators. A total of ten logic operators,such as OR, intersection, if inside, if outside,followed by, etc., are available. In addition,you can use variables as filters in a searches.For instance, you could ask for all occurrencesof Code_A where it overlaps with Code_B inall the responses of men, age category 5, livingin Munich. Variables can be managed intables and exported or imported as text(tab-delimited) files.

Team support

MAXqda offers team support in two ways.Team members can either work with the samemaster version of a project (option A), or thedifferent members of the team divide theirwork (option B). When choosing option A,team members can never work simultane-ously on a project; only one person at a timecan code a text and then the entire project hasto be given to the other team members. Whenchoosing option B, team members can worksimultaneously on a project, but only on dif-ferent parts. When a team member has codeda particular text or part of a text, this text can

324

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 324

be transferred to other team members. Thesame applies to memos.

Another option is to merge two projects.New texts and all memos will be inserted,codes and variables will be merged.

Other features

Scoring relevance

Each coded segment can be given a weightbetween 0 and 100. This can be used as anindicator of how relevant or how typical acode is for a certain category. Depending onthe kind of analysis, the weights can betreated as fuzzy variables. When creating atable for all codings or for the codings ata particular code, the relevance scorces(weights) are included.

Quantitative content analysis functions

If you are interested in quantitative contentanalysis, MAXqda offers the add-on moduleMAXdictio. This module allows you toanalyse vocabulary and to create dictionaries.You can examine the vocabulary used in thetext to find out which words can be found inwhich text passages or the full text. For exam-ple, among the offered functions are the cre-ation of word frequency lists of the wholetext, of marked passages, in-text groups or in-text sets, the creation of an index of selectedwords of one or more texts, the building-upof word-based dictionaries, the export andimport of dictionaries from MS Office pro-grams like Excel, or the further processing ofthe results in SPSS and Excel. All content ana-lytic functions can be integrated into thefunctionality of MAXqda. Thus, if you areinterested in a combination of quantitativeand qualitative content analysis, the combina-tion of MAXqda with the add-on modulemight be the right choice for you.

Summary

MAXqda convinces due to its simple logicand structure. It is likely to appeal to userswho prefer to think of their coding system asa hierarchical construct consisting of higherand lower order codes. Conceptual-level rela-tionships cannot be mapped within the soft-ware. Nonetheless, the software provides allnecessary features to perform a qualitativedata analysis and, by keeping it simple, thelearning curve is not as steep as for othersoftware packages.

The add-on module MAXdictio gives thesoftware a unique profile, allowing for thecombination of qualitative and quantitativecontent analysis methods.

THE ETHNOGRAPH V5.08

Data entry

The Ethnograph supports the analysis oftext data. Before importing the data, it isnecessary to format them in a particular way.An Ethnograph data file has a 40-characterline and hanging paragraph indents. All ofyour data must be saved as ASCII text. Theprogram provides an editor that facilitatesthe creation and re-formatting of data filesinto format required by The Ethnograph.The special formatting rules allow the pro-gram to identify speaker or section turns.Thus, there is no need to code speaker orother sections and variables can be directlyattached to them.

Coding

It is possible to code your data interactively.This means that you can view and readthe text of your data files on screen whileyou are coding. For each coded segment the


19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 325


boundaries are marked in the margin andthe code word appears one line above thecoded text segment. Code words can be con-veniently selected from either the CodeBook or the Tree View. Both are automati-cally created and updated when coding.

The smallest size of a text segment youcan code is one line; the maximum length ofcode words is ten characters. This is a neces-sary restriction, otherwise it would beimpossible to display all of your code wordswithin the text (see Figure 19.7). Longer def-initions of up to 500 characters (roughly halfa page) can be written into the Code Book.Each coded segment can be defined by up to12 code words and these code words can benested or overlapped up to seven levels deep.

Within the Coding Window simple textsearch and code search function are available.The coding system can be structured hierar-chically by assigning parent and child codes in

the Code Book, or by dragging and droppingcodes in the Tree View. Parent codes arehigher-level and child codes lower-levelcodes. A code can be both a parent code ofa subgroup of codes and a child code ofa super-ordinate code.

A hierarchical presentation of the codingsystem is possible in the Tree View, whichdisplays the codes in the form of a side-waystree. You can either view just the Tree or boththe Code Book and the Tree side by side (seeFigure 19.8 below). By collapsing or openingthe various branches of the Tree View, it ispossible to focus one’s view on just a partic-ular area of the coding system.

Memos

Three types of memos are available: textmemos, project memos and file memos. Youcan attach up to 26 memos to each line in a

326

Figure 19.7 Coding window in The Ethnograph

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 326

data file, up to 1,000 memos to a project andup to 1,000 memos to each data file. Eachmemo can be up to 32 pages in length and isdate and time stamped. In addition, the ini-tials of the author can be entered, memo cat-egories can be defined and up to three codewords can be associated with a code word.The memo categories, code words, linenumbers, date and time of creation/modifi-cation, author’s name and memo title can beused as sorting devices.

All memos can be accessed from a varietyof places in the project. Text memos forexample are identified by the letter M, whichis displayed next to the first line number ofthe segment the memo is attached to (seeFigure 19.7). A double click on the M opensthe memo window.

Data retrieval

The Ethnograph search tool nicely integratesthe various search options and filters thatcan be used in a search. You can search forsegments coded by a single code word, or forsegments coded by multiple code words. In

the latter case, Boolean and proximityoperators can be used to combine a string ofup to five code words. Multiple code wordsearches can be saved and called up again,reused or edited for later searches.

Figure 19.8 shows the selection window forcodes in a single code search. Codes can eitherbe selected from the Code Book or the TreeView. The parent/child structure of the codingsystem allows you to use the equivalent ofa semantic DOWN operator (i.e., SCIENCEwith kids).

In addition to codes, speaker or sectionidentifiers can be used in a search; variablesand file codes can be used to restrict searchesto certain documents or attributes. An iden-tifier search retrieves all instances of a certainspeaker or of text passages marked with a spe-cific tag. Identifiers are defined in the processof transcribing the data (see data entry above).A common use of identifiers is to mark speakerturns, for example the name of the intervieweeor the names of participants in a focus group.Those identifies are referred to as SpeakerIdentifiers. If your data consists of documentslike letters or newspaper articles, you can


Figure 19.8 Single code search showing the Code Book and the Tree View

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 327


identify them by date and source. These typesof identifiers are called Section Identifiers.When employed as part of a search, bothSpeaker and Section Identifiers can be used asif they were code words.

For each identifier, variables such asgender, age, occupation, and so on, can beentered. In a search, this information can beutilised by retrieving text passages coded bysingle or multiple code words in combina-tion with selected variables. Thus, you can,for instance, ask for all comments abouta particular topic that can be attributedto only male interviewees between the ages of25 and 30 with an income of at least $50,000per year. Variables in The Ethnograph allowfine-grained searches in that you cannot onlychoose categories but enter continuousnumeric values.

File codes also serve as variables in that theypermit you to attach specific attributes to anentire document. File codes, like variables, canbe combined by AND, OR and NOT opera-tors. The difference between file codes andvariables attached to identifiers is that you canassign up to 16,000 file codes to a data file butonly 40 variables to an identifier. Thus, theprogram offers more than enough options torestrict and filter searchers. Search results canbe reviewed on screen, printed or saved todisk, and displayed in three different ways:

• In the form of coded text segments (withor without the surrounding context, linenumbers, boundaries, code words ormemos).

• In the form of summary reports.Summary reports give you a count of allhits. An additional option is to display allof the start and stop lines of the codedsegments that have been found.

• As frequency counts. Frequency outputslist the percentages of all hits within andacross data files.

Before saving or printing search results,segments of interest can be marked and theoutput can be restricted to only the markedsegments. Text output is saved in ASCII textformat, all quantitative output is saved in theform of tables and can be imported to a sta-tistical package for further analysis.

The entire search results or only selectedsegments can be saved as a new data file,which is automatically imported into theproject and available for further analysis andcoding.

Other features

Splitting and combining files

It is possible to split data files and to combinedata files. When you combine data files, youhave the option of combining only selectedparts of the data files, for example lines 1 to 500from data file X and lines 350 to 1,300 fromdata file Y. Inserted sub-headers inform you ofthe original source. Splitting and combiningdata files can be done on the basis of numberedor coded versions of your data files. Thisprocess does not affect the coding.You can saveand print numbered, coded or generic versionsof your data files.

Summary

The Ethnograph is one of the simpler pro-grams to analyse qualitative data and it isprobably the easiest of all programs to learn.The procedures are straightforward and bydesign they are strictly separated, so that onedoes not get lost or confused by multipleoverlapping windows. However, experiencedWindows users might expect more in termsof user friendliness. The DOS roots of theprogram are still noticeable. File and foldernames that are longer than eight characters

328

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 328

are not accepted or are abbreviated. Data,code, search and memo procedures cannotbe run side by side. For instance, one cannotadd additional files to a project whilethe coding or search window is open. Onecannot search while the coding or memowindow is still open, etc. This may not beexperienced as a problem and some mayeven appreciate the clear separation of func-tions. The Ethnograph certainly offers verygood search functions in combination witheasy-to-handle variables. The option towork with section and Speaker Identifiers isa unique feature of The Ethnograph. Otherprograms offer similar but less elegant solu-tions. All in all, The Ethnograph is a pro-gramme that offers good value for money;it is the least expensive of all programsdiscussed here.

QSR N6

Data entry

QSR N6 can be used to analyse text data. Therequired format for importing data into N6is plain text. The smallest segment that canbe coded is a text unit, which can be set tolines, sentences or paragraphs. A paragraphis defined as two successive hard returns.This needs to be taken into account whentranscribing or formatting the data. Newdata can be appended to existing documentsand already imported documents can beedited. This includes inserting or removingtext units, sub-headers and annotations.

Coding

Coding can be done interactively while youread the text on screen or, if you prefer, on ahard copy of your data. Coding done onpaper can be transferred to the computer by

using text unit numbers as reference points.All coded segments are stored in a so-callednode. Coded segments are not indicatedinside the text or in the margin area. Theycan only be made visible by specificallyrequesting a report that includes codingstripes for selected code words (= nodes).

Nodes can be understood as containersthat have a node address and node title.There are different types of node, those hold-ing coded data segments, results of text ornode searchers, document annotations or thecontent of the clipboard when cutting orcopying node contents. Nodes containingcoded text passages can either be handled as‘free’ node or can be integrated into the indextree. The index tree offers the possibility tostructure and organise your codes hierarchi-cally. All nodes are displayed in the NodeExplorer (see Figure 19.9). A descriptionfor each node can be entered into the rightwindow pane of the node explorer.


Figure 19.9 QSR N6 Node Explorerdisplaying free nodes and index tree

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 329


The Node Explorer offers a sideways repre-sentation of the coding system. Within theNode Explorer you can cut, delete and mergenodes, you can browse the text that is coded ata node, you can change node addresses, writeand review memos or modify the coding at anode. In addition, the Node Explorer is a veryhandy device when it comes to searching.From there you have direct access to theresults of the search operations (see SearchFunctions). To efficiently build up a hierar-chical index system, certain rules should befollowed. For a introduction to creating ahierarchical category scheme, see Richardsand Richards (1995).

The index tree can also be displayed graph-ically in the form of an upside-down tree. Thetop of the index tree is the root and fromthere all the branches ramify. An overview ofthe entire tree is provided. However, the usercan only access two levels at a time (see Figure19.10). Nodes can be manipulated in eitherthe side-ways or the graphical display.

It is possible to create the tree before youstart coding your data. In fact, it is very com-mon to set up certain parts of the index treeat the start of a project. This part of the tree isgenerally referred to as the base data tree. Itcan be used for coding factual informationsuch as gender, age, race, occupation, or timeseries. Since this is a fairly routine task, thisprocess can be automated by using commandfiles. If the factual information is available inthe form of a spread sheet, you can use the

Import Table function for generating the basedata tree and to code your data. This featureof the program allows you to link quantitativeinformation that you have about your datawith the qualitative-rich content. Documentbase or case data and coding at a particularnode can be exported as plain text or tab-separated tables.

Memos and annotations

In addition to writing a description for eachdocument and note, you can attach onememo to each document and node. Memosare indicated with an exclamation mark (!)before the object they belong to and their con-tent in plain text format can be cut and pastedsomewhere else. Memos can be retrieved viathe context menu or when creating a report.

Annotations are short notes that can beattached as a text-unit to any text-unit in adocument. All annotations are coded auto-matically at the Document Annotation nodein the node system. In order to distinguishannotations from the rest of the text in yourdocuments, they are put into <<brackets>>.


In N6 you can search for words or phrasesin either the documents or the coded textsegments. The search can be restricted by anumber of criteria. In addition, GREP expres-sions can be used for pattern searches. Youcan keep all finds or review the finds first anddecide which ones you want to keep. Theresulting hits are automatically coded at anew node in the Text Search area of the nodesystem.

Data retrieval

The search engine in N6 is a very powerful tool.It offers 17 different ways of asking questions

330

Figure 19.10 Tree display in N6

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 330

about your data, including Boolean, context,sequence, proximity, matrix and vectorsearches. The results of a search are saved tothe node clipboard. In addition, a node con-taining the search results is automatically cre-ated and saved in the Node Search area of thenode systems. Results of matrix or vectorsearches can be exported as plain text ortab-separated tables.

Node searches can be restricted to selecteddocuments and nodes. As in N6, the textcoded at the base data nodes is a form ofdefining data attributes or variables. This is away to combine codes with variables, thusbeing able to ask questions such as: Give meall text related to attitude x, but only fromfemale respondents of age group 1.

Having the results of a search saved at anode, you can ask further questions aboutspecific parts of your data. This way itbecomes possible to build up more and morelayers of analysis and to go deeper and deeperinto your data.

For utilizing the search functions in N6 inthe most efficiently manner, you are advisedto build up the index system following therules of hierarchical category schemes.

Team support

QSR offers a free companion tool as an add-on to N6 to merge projects. This tool allowsyou to compare and contrast two differentprojects or to build a central project frommultiple sites. It is also possible to use themerge tool across different platforms (i.e.,Macintosh and PC platforms).

Other features

Command files

Many processes in N6 can be automated byusing command files. Command files containwritten instructions that can be executed by

N6 to carry our different processes, suchas importing documents, setting up the basedata tree, adding and deleting coding, copy-ing and pasting parts of the tree, deletingnodes, displaying the tree, searching text andnodes, setting restrictions, or making andprinting reports. N6 makes the use of com-mand files fairly simple by providing anumber of templates. Thus, command files canbe written without knowledge of the underly-ing syntax.

Model-building – exporting the index tree

N6 does not offer a model-building functionitself, but the index tree can be exported toeither Decision Explorer or Inspiration. Thisallows you to break away from the hierarchicalstructure enforced by N6 and to freely playaround with the nodes, thus to enter the phaseof model-building.While you lose direct accessto your raw data, you can still view and editmemos.

Summary

QSR N6 is a good choice for those who preferto organise their coding system in a hierarchi-cally manner, or who have data that lendthemselves to such a structure. Even thoughN6 (as compared to previous versions) offersthe inclusion of free nodes into the indexsystem, the program would lose much of itspower if you do not choose to utilise the tree.Comparable to the The Ethnograph, N6 hasrecognisable DOS roots. File and foldernames can be no longer than eight charactersor have to be abbreviated. All imported docu-ments need to be saved as plain text; outputformats are also plain or tab-separated texts.Unique to the QSR products (i.e., N6 andNVivo) are the matrix and vector searchesand the possibility to display and export thesearch results as tables. Another distinctivefeature is that all text and node searches are


19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 331


automatically saved at a node and can easilybe subjected to further analysis.

You may teach yourself the programe withthe help of the provided tutorial, but a pro-fessional training session is especially usefulto start on the right foot in terms of settingup the index system.

NOTES

1 On the following website you find a list of free andcommercial CAQDAS packages: http://caqdas.soc.surrey.ac.uk/links1.htm

2 Interesting to read in the context of arguing for oragainst computer technology in the qualitative data analy-sis process is the book chapter by John Seidel: ‘Methodsand Madness in the Application of Computer Technologyto Qualitative Data Analysis’ (Seided, 1991).

3 The acronym CAQDAS was developed by the direc-tors of the CAQDAS networking project at the Universityof Surrey, Guildford, UK. http://caqdas.soc.surrey.ac.uk/

4 see ‘ATLAS.ti goes XML’ at http://www.atlasti.de/xml/.5 Further advice in deciding between software packages

can also be found on the website of the CAQDAS networkingproject: http://caqdas.soc.surrey.ac.uk/.

6 Other descriptions and software overviews can befound at: Friese, S. (2004) Qualitative research and con-sulting: http://www.quarc.de/software_overview_table.pdf

Alexa and Zuell (1999) Commonalities, differences andlimitations of text analysis software: the results of a review:http://www.gesis.org/Publikationen/Berichte/ZUMA_Arbeitsberichte/99/99_06abs.htm

REFERENCES

Alexa, M. and Zuell, C. (1999). Commonalities, dif-ferences and limitations of text analysis software:The results of a review, ZUMA-ArbeitsberichtNo. 99/06. http://www.gesis.org/Publikationen/Berichte/ZUMA_Arbeitsberichte/99/99_06abs.htm(last accessed 4 July 2004).

Böhm, A., Legewie, H. and Muhr, T. (1992). KursusTextinterpretation. Globalauswertung undGrounded Theory. Unpublished manuscript,Technical University Berlin, Germany.

Coffey, A., Holbrook, B. and Atkinson, P. (1996)Qualitative data analysis: technologies and

representations, Sociological Research Online, vol. 1,no. 1, http://www.socresonline.org.uk/socresonline/1/1/4.html (last accessed 4 July 2004).

Creswell, J.W. (1998). Qualitative Inquiry and ResearchDesign: Choosing among Five Traditions. London:Sage.

Friese, S. (2004). Qualitative research and consult-ing, http://www.quarc.de/software_overview_’

Glaser, B.G. and Strauss, A.L. (1967). The Discoveryof Grounded Theory: Strategies for QualitativeResearch. Chicago: Aldine.

Kuckartz, U. (1995). Case-oriented quantification.In U. Kelle (ed.), Computer-Aided QualitativeData Analysis. London: Sage, pp. 158–166.

Pink, S. (2001). Doing Visual Ethnography. London:Sage.

Prein, G., Kelle, U. and Bird, K. (1995). An overviewof software, In: U. Kelle (ed.), Computer-AidedQualitative Data Analysis. London: Sage,pp. 190–210.

Richards, T. and Richards, L. (1995). Using hierar-chical categories in qualitative data analysis. InU. Kelle (ed.), Computer-Aided Qualitative DataAnalysis: London: Sage, pp. 80–95.

Seidel, J. (1991). Methods and madness in the appli-cation of computer technology to qualitative dataanalysis. In N. Fielding and R. Lee (eds), UsingComputers in Qualitative Research. London: Sage,part two.

Seidel, J. (1998). Qualitative Data Analysis. Availableat: http://www.qualisresearch.com (last accessed 4July 2004).

Weitzman, E.A. and Miles, M.B. (1995). A SoftwareSourcebook: Computer Programs for QualitativeData Analysis. London: Sage.

SOFTWARE

ATLAS.ti http:://www.atlasti.comMAXqdahttp://www.maxqda.comThe Ethnograph http://www.qualis

research.comQSR N6 http://www.qsr.com.au

332

19-Hobbs-3326-19.qxd 10/5/2005 8:38 PM Page 332