Download - Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

Transcript
Page 1: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

Qualitative Sociology, Vol. 24, No. 3, 2001

Integrating Technology to Improve the Efficiencyof Qualitative Data Analysis—A Note on Methods

Calvin Smith and Patricia M. Short

Qualitative data analysis (QDA) is often a time-consuming and laborious processusually involving the management of large quantities of textual data. Recently de-veloped computer programs offer great advances in the efficiency of the processesof QDA. In this paper we report on an innovative use of a combination of extantcomputer software technologies to further enhance and simplify QDA. Used inappropriate circumstances, we believe that this innovation greatly enhances thespeed with which theoretical and descriptive ideas can be abstracted from rich,complex, and chaotic qualitative data.

KEY WORDS: qualitative data analysis; computers.

INTRODUCTION

Technological advances have led to enhancements in the efficiency of qual-itative data analysis (QDA). Computer programs such asThe EthnographandNUD?IST have made it possible for qualitative data analysts to manage largevolumes of textual data. Such programs offer an immense improvement in theefficiency and ease with which QDA can be done and they continue to be im-proved in scope and function. Yet, until very recently, preparing documents foranalysis in these programs involved transcription to text files. Notwithstandingrecent advances in software design, qualitative data analysis can still be very time-consuming and expensive; interviews or field notes still are usually transcribedand then “coded” in order to reduce data to an organised and coherent collectionof ideas.

Correspondence should be directed to Calvin Smith, Teaching and Educational Development Insti-tute, University of Queensland, Queensland, Australia, 4072, and Patricia M. Short, Department ofSociology, Anthropology and Archaeology, University of Queensland, Queensland, Australia, 4072;e-mail: [email protected].

401

C© 2001 Human Sciences Press, Inc.

Page 2: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

402 Smith and Short

We offer in this paper an advance in efficiency that does not involve verbatimtranscription of recorded data and draws on a combination of extant technologiesto facilitate faster, less expensive QDA.

We acknowledge that recent developments in QDA software enable re-searchers to code directly from audio or sound files (e.g.,C-I-SAID) or to create“proxy” documents linked to (nontranscribed) sources in different formats—mediafiles, photographic images, etc. (e.g., the latest QSR product,NVivo.Still, whileC-I-SAID, for example, provides a sophisticated method for directly linking cod-ing charts and video/audio sources, the lexical coding system provided for appearssomewhat limited as a tool for genuine inductive analysis. The method we describewill be more useful for researchers who are interested in progressively building,changing, and documenting a coding system. Also, our method will assist thoseusing earlier or similar versions of the QSR software (NUD?IST IVand earlier) andwho wish to maximize and balance efficiency, affordability, convenience, and rigorin qualitative inquiry, especially in contexts where rapid assessment and analysisis necessary.

The essence of the innovation we describe here is the production and codingof a simplified text file that represents the sequence and length of segments ofrecorded data (on audio- or videotape or digital recording device) passing througha playback machine. We call this text file the “counter-run” file. It is the counter-runfile that is introduced into a QDA program such asNUD?ISTor The Ethnographrather than a full transcription of recorded data. Theoretical ideas are cataloguedagainst the counter-run text that serves as an index representing the location of thedata giving rise to those ideas in the original data record.

This approach to QDA recommends itself highly for the routine work of QDA.It facilitates efficient analysis of qualitative data while still enabling the analyst torefer to the detail of the raw data at all stages of analysis. It is especially suitablewhere the goals of analysis are analytic induction, grounded theory building, andso on (Glaser and Strauss 1967; Lofland 1971; Strauss 1987; Strauss and Corbin1990), especially where coarse-grained coding is all that is required or is a usefulpreliminary (e.g., Short 1996). In contexts where rapid assessment of a field or issueis necessary (e.g., for social impact assessment, action research, or social planningcontexts) or research is strictly time-limited (e.g., for undergraduate or some short-term postgraduate study programs), this technique is most appropriate. It affordsspeedy analysis, rigorous documentation of analytic procedures to demonstratevalidity and verify findings, and swift retrieval of data for transcription, reportingand publication.

It should be noted, however, that the approach to QDA described here willnot always be appropriate. For instance, it does reduce the efficiency with whichan analyst can search for related “strings” of text in order to conduct detailedanalysis of the nuances of expression and meaning of particular ideas. And itwould not be appropriate for conversation analysts or linguists who must record

Page 3: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

Integrating Technology to Improve the Efficiency of Qualitative Data Analysis 403

and document every minute semantic and syntactic detail in order to do theirwork.

This said, we describe the technique in some detail below and then makefurther comment on some aspects of method.

THE SIMPLE VERSION—TAPE-RECORDED DATA

Most tape machines have a counter that ticks over as the tape plays throughthe machine. The innovation proposed here simply involves setting up a counter-run file that represents chunks of time (and therefore of the talk or images thattranspired within them) as ranges of digits that appear in the counter window ofthe tape or video (e.g., 0–10, 11–20, etc.).

Since it is a text file, it is the counter-run file that is read into programs suchasThe Ethnographor NUD?ISTand gets coded with theoretical ideas that come tomind as one listens to the original data source. The codings are mapped against thetext units of the counter-run. The text units are the digits representing the numberin the window of the tape counter and therefore representing various points alongthe tape. This allows researchers to code as they listen to the recorded data withoutfirst transcribing it, effecting a vast saving of time, effort, and cost. The samemethod can be applied, of course, to video data where the video machine has abuilt-in tape counter.

Table I illustrates the method with the third column showing the text of thecounter-run file.

By conducting a preliminary test run, the analyst can determine the appropri-ate scale for the counter-run to accommodate speed of speech or the through-flowof useful, theoretically relevant bits of information. If the scale is over-coarse, toomany ideas may be coded onto one band on the counter-run scale, and retriev-ing exemplary segments for later analysis and reporting will be less reliable andconvenient. If it is too fine, there will be unnecessary “gaps” in the coding of thecounter-run.

During coding, the analyst can edit the counter-run text to include notes onthe segment or snippets of text to signify, at a glance, the flow of the text orparticular expressions/images that might be returned to for further analysis. It ispossible to produce an abstract of the interview in this way, and this may providesome additional advantages over full transcription or no transcription for bothresearchers and participants.1

1For instance, Duncan (1997) suggests that an abstract and audiotape copy of interviews may allowmore effective review of interviews by participants because the abstract is a smaller, more manageabledocument providing a succinct guide or index to the content of a taped interview, thus allowingparticipants to focus more carefully on the parts of the interview that are most important for them toreview.

Page 4: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

404 Smith and Short

Table I. Example of Coding Against Counter-Run

Transcribed Text (taken from an interviewwith a househusband—see Smith 1998) Tape counter Counter-run Coding

Q: Do you have an arrangement with U, 189 189 Financialyour wife, where she pays you an 199 arrangementsamount each week or anything like that 200which is just yours to do with 201 201 Planning thewhatever you want to? 202 transition

A: No. We talked about something like 203that in the initial stages but we 204never really followed it through. 205

Q: Why did you decide not to go that road 206 206and what road did you go down? 207

A: Well I think it was a decision by 208default rather than a conscious 209decision—we tended to just have a 210joint account and as I was saying to 211 211 Responsibilityyou before I always tended to do the 212 for shopping;shopping, I still do the shopping so 213 power overyou just draw it out. So there’s money 214 expenditurethere—I mean we sat down and budgeted: 215there’s x-amount for you to spend each 216 216 Budgeting;week—and you just don’t pay any 217 standard ofattention to that (laughter). I guess 218 livingwe’ve never really had to worry about 219counting every penny (mm hmm), things 220like that so we’ve been fortunate that 221 221way—so long as there’s enough there to 222meet our bills and commitments we 223don’t have much problem. 224

225

INCORPORATING OTHER TECHNOLOGIES IN THE INNOVATION

In order to make counter-run files, one can simply type and save a “template”file that can be reproduced for each transcript being analyzed, incorporated into,and coded in programs such asNUD?ISTor The Ethnograph.Alternatively, onecan utilize a feature of spreadsheets that makes the construction of a counter-runfile very simple and efficient. Counter-run files to suit any scaling of a counter-runcan easily be made this way.

The way to do this is quite simple. In a Microsoft Excel spreadsheet (version97 or later):

1. Put in two numbers, one under the other, in a column (say 0 and 10);2. highlight both;3. pointing at the bottom right-hand corner of the highlighted block, mouse-

click and drag down the column (while holding the mouse buttondown).

Page 5: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

Integrating Technology to Improve the Efficiency of Qualitative Data Analysis 405

The package will fill in the cells, incrementing each by the difference betweenthe values in the first two that you typed in. That is, if these numbers are “0” and“10,” the program will fill the next cells with “20,” “30,” “40,” etc., incrementingeach time by 10.

By doing this, the column can be copied and pasted directly into a wordprocessor file or text editor (including the text editor inNUD?IST) or, more easily,in the latest version ofNUD?IST, N5, imported directly from the clipboard. Thedocument can then be prepared in the usual way for importing into a program suchasThe EthnographandNUD?ISTfor analysis. You could copy it straight into such apackage, but you may like to prepare it for use first in a word processor or text editor.This is useful if you wish to add information such as interviewee’s pseudonym,details of the data source, a fieldwork date, and any technical information, such astape speed2 at the head of the document.

MORE SOPHISTICATED VERSIONS—DIGITAL DATA STORAGE

Storing the data digitally and accessing it via software such asSound EditPro (Macintosh) orSound Forge(PC) for audio, orAvid Cinemafor video, is nowpossible and is navigationally more reliable than using analog tape counters. Digitalrepresentations of the data also allow for the calibration of the “counter-run” to befurther fine-tuned.

To use digitized sound software, you may need to first convert the audio orvideo data into digital format; this may require special connectors and software tomanage the connection between a video or tape machine and the computer. Youalso need considerable storage capacity to store these forms of data digitally. Thechief benefit is the capacity to navigate easily around the data source by pointingand dragging the computer’s cursor or mouse pointer over a graphic representationof the data (e.g., an oscilloscopic trace of audio data).

In deciding the scale of counter-run intervals in these applications, the issueof how much real time is represented by each point in the scale of the countercan be considered. In the case of the audio or video software, the amount ofreal time that is represented by each chunk of the counter usually can be varied.Sometimes it may be appropriate to use five-second intervals for each point on thecounter scale, sometimes more, sometimes less. A very fast speaker, for instance,will conceivably need to be mapped against a fine time-scaling with fewer scale-points in each range of the counter-run than might otherwise be used. The differentcombinations of the time-scale and counter-run intervals, and the characteristics of

2You may need to vary the speed at which the tape replays the recorded data (e.g., for slow or fastspeakers or to increase audibility). If so, you would need to keep a record of the speed at which youreplayed and coded each recording, so that you could return reliably to the same location on the tapefrom later transcription of a representative segment.

Page 6: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

406 Smith and Short

Diagram 1. Flow Chart of Use of Different Technologies.

the data to which each seems to be best suited, are matters of personal preferenceand fitness for the task at hand.

SUMMARY AND DISCUSSION

Diagram 1 shows the steps involved in the process of integrating computer-based technologies to improve the efficiency of QDA.

A Methodological Note

We acknowledge that there is a degree of debate in qualitative analysis litera-ture about the advantages and disadvantages of transcription and nontranscriptionmethods of interpretation and analysis (DeVault 1990; Duncan 1997).

Page 7: Integrating Technology to Improve the Efficiency of Qualitative Data Analysis—A Note on Methods

P1: VENDOR/FLF/GOQ P2: GCQ/GEE QC:

Qualitative Sociology [quso] PH052-341002 January 1, 1904 7:46 Style file version Nov. 19th, 1999

Integrating Technology to Improve the Efficiency of Qualitative Data Analysis 407

Some analysts may prefer to work (to code) from a transcript of the databecause it imparts a certain intensity to the work and closeness to the data. Wetake the view that the coding process is made more meaningful and accuratebecause multiple “listenings” and “viewings” of the data bring the researcher evencloser to it than a transcript does. Listening to and viewing the data allows usto retain paralinguistic cues (body language and the like) that supplement theverbal “message.” Using the counter-run method described here maintains a directlink with the interview as “an interaction grounded in talk” (DeVault 1990) andfacilitates the analyst’s interpreting the meaning-making aural dimensions of talksuch as silences (Duncan 1997; Opie 1995; Poland and Pederson 1998), timing andpacing, pitch, tone, and volume. Thus, our method also avoids some of the pitfallsof “editing” first-person narratives (especially nonstandard patterns of speech) atearly stages of analysis (Duncan 1997; Blauner 1987). A textual representation ofthe verbal always elides these complex and rich cues and so transcripts of data area poorer second cousin to the original than are audio and video recordings.

Facilitating the reliable and accurate return to the segments of original datathat gave rise to theoretical notions eases the researcher’s task of demonstrating toan audience of colleagues that what is proposed is a reasonable interpretation of thedata. Segments of data to be reported can be transcribed verbatim and publishedin support of findings, in the usual manner. What has been avoided by adoptingthis approach is the transcription of the entire body of data, thus saving much timeand labor and, therefore, cost.

REFERENCES

Blauner, B. (1987). Problems of editing first-person sociology.Qualitative Sociology, 10, 46–64.DeVault, M. L. (1990). Talking and listening from women’s standpoint: Feminist strategies for inter-

viewing and analysis.Social Problems, 37, 96–116.Duncan, J. (1997). To transcribe or not to transcribe? That is the question.Education Research and

Perspectives, 24, 1–13.Glaser, B., & Strauss, A. (1967).The discovery of grounded theory.Chicago: Aldine Publishing Com-

pany.Lofland, J. (1971).Analyzing social settings.Belmont, CA: Wadsworth.Opie, A. (1995).Beyond good intentions: Support work with older people.Wellington, New Zealand:

Institute of Policy Studies.Poland, B., & Pederson, A. (1998). Reading between the lines: Interpreting silences in qualitative

research.Qualitative Inquiry, 4, 293–313.Short, P. M. (1996). No-one to turn to: Estrangement, need and kinship economies. Paper presented

to the 5th Australian Families Research Conference, Brisbane, Queensland, November. Also at<http://www.aifs.org.au> and onAustralian Family and Society Abstracts (FAMILY)database.

Smith, C. D. (1998). Men don’t do this kind of thing: A case study of the social isolation of househus-bands.Men and Masculinities, 1, 138–172.

Strauss, A. (1987).Qualitative analysis for social scientists.Cambridge: Cambridge University Press.Strauss, A., & Corbin, J. (1990).Basics of qualitative research: Grounded theory procedures and

techniques.Newbury Park: Sage.