Martyn Minns 3rd November, 1996.



ψ

University of the west of england

A thematic analysis of a set of interview data and a critical discussion of the analysis and interpretation

Student Number : 02976321

Award : MSc Qualitative Methods in Psychology (Level M)

Module : Analysis and Interpretation of Interview Data (USP402HM)

Module Leader : Dr Kate Gleeson

Date : 25th August 2003

Word Count : 1999

A thematic analysis of a set of

interview data and a critical discussion of the analysis and interpretation

Thematic analysis is historically a conventional practice in qualitative research which involves searching through data to identify recurrent issues (Creswell, 1994, pp.153-155; Dey, 1993; Hayes, 2000, pp.173-179; Holliday, 2002, pp.98-122; Holloway, 1997, p.152; Miles & Huberman, 1994; Ritchie et al., 2003, pp.219-262; Smith, et al., 1995, pp.9-26; Patton, 2002, pp.452-471). A theme (or construct) is a cluster of linked categories conveying similar meanings. I performed an inductive thematic analysis whereby themes emerged from data collected by another researcher’s semi-structured interview. The exploratory power of this qualitative method was enhanced since I lacked previous knowledge about the interview and the research agenda. However, revealing my personal values and assumptions is crucial since qualitative research involves being a central channel for analysis (see reflexive statement, appendix-i; Alvesson & Sköldberg, 2000; Creswell, 1994; Kopala & Suzuki, 1999). In order to begin analysis a researcher must have at least some conceptual resource to guide the processes of representation or interpretation, otherwise no sense at all could be made of any data.

My analytic procedures (appendix-ii) which drew upon closely related techniques from interpretative phenomenological analysis - IPA (Smith & Osborn, 2003; Willig, 2001, pp.50-69) and grounded theory (Richardson, 1996, pp.75-101; Strauss & Corbin, 1998). However, I need to clearly state my epistemological position as qualitative analyses vary considerably and often reflect pragmatic conditions rather than philosophic principles. There is no simple distinction between qualitative (naturalistic, contextual, idealist) and quantitative (experimental, positivist, realist) methodologies (Hammersley, 1996; Creswell, 1994). Since researchers move back and forth between abstract concepts and data all research involves processes of induction and deduction (Robson, 1993), especially thematic analysis whereby induction creates themes and deduction authenticates them. My epistemological approach aligned with inductive-realism. It was inductive because themes emerged from the data and realist because my task as the analyst was to describe themes which exist independently from my own interpretation. This differs to the positivist-realist approach involving theory-led thematic analysis which tests hypotheses about textual data; comparable with analytic induction which seeks to identify deterministic laws embedded within the essential nature of phenomena. Research should clearly differentiate itself as primarily being explorative, generating theoretical ideas or testing existing hypotheses. Because this exploratory analysis remained at the level of description rather than formulating any interpretations or explanations it is closer to empiricism than constructionism.

I manually performed the analysis using basic data management facilities of a computer for cutting, pasting and ordering information. This enabled easy transferral of original information into this report in order to make my operations visible and assist reader evaluation. Although textual investigations are aided by computer assisted qualitative data analysis software (CAQDAS) these are largely architectural and the actual process of description, or interpretation, remains a human endeavour (Dey, 1993; Willig, 2001, p.151; Kelle, 1995; Miles & Huberman, 1994; Weitzman, 2000). Most CAQDAS offer the capacity for rapid searching and retrieval of data linked with the analyst’s annotations or may even display conceptual hierarchies (e.g. Nudist, Codatext, Nvivo, Atlas, Qualpro, Ethnograph; Scolari, 2003). They allow the researchers to store volumes of information in one place without accumulating masses of written material.

The audiotaped interview data had already been transcribed into text by the original researcher. However, it needed formatting for my own analysis and assigning line numbers to identify individual databits (appendix-ii, stage 1). Eleven areas of interest arose from my initial reading of the text which familiarised me with the interview’s material (appendix-ii, stage 2.i). The data was then systematically reduced during open coding which fragmented it into separate units for close examination and evaluation for similarities or differences (appendix-i; appendix-ii, stage 2.ii; Strauss & Corbin, 1998, pp.101-121). At this stage the participants information was summarised using line-by-line coding (or indexing) which allowed individual bits of data to be closely identified and scrutinised. This microanalytic process functioned to restrict my own interpretations by preventing me from becoming immersed in the respondent’s worldview (Smith, et al., 1995, p.37). Unlike performing an analysis on data collected for one’s own research this meant covering the whole data to avoid rejecting any potentially irrelevant material. Whilst intensely focused upon this process I made brief notes for the main issues without letting these dominate the further decontextualisation of the data. This process of reduction is a vital part of performing qualitative analysis so the source of themes remain rooted in the data, preventing the participant’s reality disappearing by simply overlaying themes from the researcher’s perspective (Ritchie et al., 2003, pp.237-248). The IPA technique helped simultaneous deconstruction of both the data and the dynamics between the interviewer and the participant. Data reduction established two hundred and twenty four databits which I manually assigned to sixteen categories (segmenting) by cutting and pasting on the computer (appendix-iii). This is a creative yet rigorous process which fully demands the researcher’s analytic abilities so that each databit is given equal treatment. At this stage the sub-categories are loosely defined and mundane category labels are recommended for general reference to empirical instances, preferably using the participant’s own terms to ‘sensitize concepts’ (Blumer, 1969, cited in Smith, 2003, p.85). Breaking the data down into small fragments means databits can potentially relate to more than one category. However, this analysis involved a natural assimilation of databits into sub-categories which was likely to be the result of such microanalytic line-by-line fragmentation.

Themes began to emerge when databits relating to similar topics were placed together in categories, whilst simultaneously moving between these categories and the data (appendix-ii, stage 3; appendix-iii). This stage involved searching for positive or negative comparisons between categories and databits to form clusters of initial proto-themes. Re-ordering these clusters helped freely develop and modify the emerging themes. I found that keeping proto-theme definitions as simple as possible at this stage assisted the categorisation process but was very complicated because it involved being fully immersed in the data. There had to be a balance between grouping information and not creating too large a proto-theme to avoid collapsing categories to early. Assigning identifiers to categories (i.e. C1, C2, etc.) and using line numbers helped track the analysis because the names and definitions of emergent themes had a tendency to transform throughout the analysis. Hence the need for a coherent coding system.

The next stage (appendix-ii, stage 4) involved the first process of trawling back through the data to examine how information was assigned to each proto-theme and to evaluate their embryonic meanings. A provisional name and flexible definition was created for each proto-theme. Whilst themes evolve by the continual process of assembling and reassembling categories it is important to have control over how separate databits relate to one or more of the categories. Up to this point themes were primarily formed through inductive procedures but the next stage (appendix-ii, stage 5) needed to verify their development via deductive procedures. This involved axial coding which takes themes individually and searches through the original text for related databits to check for deviant incidents or data inappropriately assigned to a particular theme. This is closely associated with theory-led thematic analysis and involves changing the epistemological position of the analyst (Strauss & Corbin, 1998, pp.123-142). This recontextualisation is critical to the analytic process because human perception is selective and may easily overlook relevant information or contradictory incidents, particularly as themes evolve through very dynamic processes (Tesch, 1990). Finally, ‘analyst constructed’ typologies were produced by assessing all the material relating to each theme (Patton, 2002, p.458). Themes are brought alive with illustrations of the participant’s actual words in order to tangibly portray their meanings to the reader (appendix-ii, stages 6 & 7; appendix-iv). The resulting themes could be used for researcher triangulation to verify the authenticity of the original researcher’s analyses (Bannister, et al., 1994, p.146).

Thematic analysis, IPA and grounded theory are methodologically similar analytic frameworks but the manner in which themes, concepts and categories are epistemologically managed varies considerably between these approaches (Spencer, et al., 2003; Ritchie, et al., 2003). They all attempt to represent a view of reality via systematically working through text to identify topics that are progressively integrated into higher order themes via processes of decontextualisation and recontextualisation (Charmaz, 1995; Tesch, 1990; Willig, 1990). Their procedures are more conceptually demanding than content analysis which employs a simpler empirical epistemology (Wilson, 1995). However, content analysis and thematic analysis are often confused. The former highlights criteria for the reliability and validity of the frequency of a theme or databit’s occurrence, using predefined mutually exclusive and collectively exhaustive categories which are statistically generalisable to outside factors such as gender (Bryman, 2001, pp.180-193). IPA is generally applied to understanding individual cases whereas grounded theory was developed to understand social processes. Both of these methods have the ability to draw upon numerous sources of data (Pidgeon, 1996). However, there are several versions of grounded theory, namely objective/realist and social constructionist (Charmaz, 2000). Thematic analysis does not extend as far as grounded theory which generates hypotheses from data sets and concerns issues of discovery. Thematic analysis and IPA are epistemologically similar because, in general, researchers respectively produce descriptions and interpretations of individual subjective experiences. The major difference occurs between their initial stages of coding since IPA uses multiple re-readings to make wide-ranging and unfocused notes which are quite different to the open coding strategy of thematic analysis and grounded theory.

Thematic analysis, IPA and grounded theory are ontologically similar because they treat language transparently by simply taking for granted the meanings expressed in the data’s vocabulary. This radically contrasts with discourse analysis (DA) and conversational analysis (CA) which promote the constant deconstruction of accounts. DA and CA are part of the linguistic tradition whereas thematic analysis grew out of the sociological tradition which treats text as a window into human experience (Ryan & Bernard, 2000). The language-based analytic methods of DA and CA focus upon the construction and structure of talk, text and interaction whereby meanings are conceptualised as being obscured by rhetorical devices, linguistic styles or the interactive performance (Spencer, et al., 2003). Consequently the system of transcription for DA and CA (depending on which method is used) include phonetic, prosodic, paralinguistic and extralinguistic features of text and is exceedingly more complex and time consuming than that required for thematic analysis. DA especially treats text as having action orientations designed to achieve various socially interactive functions like questioning, agreement or justification (Cole, 1995). Empirical thematic analysis is not as explanatory and thus can not comment upon the contextual function of themes. So generally, IPA and DA are easily designed for interpretive and explanatory analyses whereas thematic analysis and CA remain purely descriptive and are more suitable for exploratory investigations, from which point research questions and hypotheses can be considered.

This reflects how qualitative methods can work from separate or integrated epistemological perspectives and that this diversity means a particular study’s position needs identifying before we can evaluate it (Willig, p.149). The mechanics of categorising data via qualitative thematic analysis involves an vigorous process which spirals between the data, the analysis and the final account, as opposed the distinctive linear processes in most quantitative research (Dey, 1993, pp.30-54). A well structured system for effective data management is essential for the continual iterative movement between the original data and any emergent themes throughout and after the analysis (Bryman, 2001; Fielding, 1993; Wisker, 2001). Qualitative analysis requires a blend of systematic searching, creativity, insightfulness and diligent detection for the development of themes, involving a dialogue between analyst and data (Ritchie & Lewis, 2003). Several qualitative researchers may discover different themes from a generic text, and often disagree with another’s classification of methods. Consequently, it is crucial that analyses are transparently communicated for others to enter into discussion, particularly if this involves research colleagues, supervisors or financers. The ultimate purpose of conducting qualitative research is to recognise how the existence of themes and the organisation of data are connected with a particular topic or the development of a line of reasoning. Thematic analyses therefore shape, adjust or reform these objectives (Holliday, 2002, 98-122).

References

Alvesson, M., & Sköldberg, K. (2000) Reflexive methodology: New vistas for qualitative research. London: Sage.

Bannister, P., Burman, E., Parker, I., Taylor, M. & Tindall, C. (eds) (1994) Qualitative methods in psychology: A research guide. Buckingham: Open University Press.

Cole, A. (1995) Discourse analysis (chapter in) Breakwell, G., Hammond, S., & Fife-Schaw, C. (eds) (1995) Research methods in psychology. 243-258. London: Sage.

Breakwell, G., Hammond, S., & Fife-Schaw, C. (eds) (1995) Research methods in psychology. London: Sage.

Bryman, A. (2001) Social research methods. Oxford: Oxford University Press.

Charmaz, K. (1995) Grounded theory (chapter in) Smith, J., Harre, R. & Van Langenhove, L. (eds) (1995) Rethinking methods in psychology. 27-49. London: Sage.

Charmaz, K. (2002) Grounded theory: Objectivist and constructivist methods (chapter in) Denzin, N. & Lincoln, Y. (eds) (2000) Handbook of qualitative research (2nd edn). 509-535. London: Sage.

Creswell, J. (1994) Research design: Qualitative and quantitative approaches. London: Sage.

Dey, I. (1993) Qualitative data analysis: A user-friendly guide for social scientists. London: Routledge.

Denzin, N. & Lincoln, Y. (eds) (2000) Handbook of qualitative research (2nd edn). London: Sage.

Fielding, J. (1993) Coding and managing data (chapter in) Gilbert, N. (ed) (1993) Researching social life. 218-238. London: Sage.

Hammersley, M. (1996) The relationship between qualitative and quantitative research: Paradigm loyalty versus methodological eclecticism (chapter in) Richardson, J. (ed) (1996) Handbook of qualitative research methods: For psychology and the social sciences. 159-174. Leicester: BPS.

Hayes, N. (2000) Doing psychological research. Buckingham: Open University Press.

Holliday, A. (2002) Doing and writing qualitative research. London: Sage.

Holloway, I. (1997) Basic concepts for qualitative research. Oxford: Blackwell.

Kelle, U. (ed) (1995) Computer-aided qualitative data analysis: Theory, methods and practice. London: Sage.

Kopala, M. & Suzuki, L. (1999) Using qualitative methods in psychology. London: Sage.

Miles, M. & Huberman, A. (1994) Qualitative data analysis: An expanded sourcebook (2nd edn). London: Sage.

Patton, M. (2002) Qualitative research and evaluation methods (3rd edn). London: Sage.

Pidgeon, N. (1996) Grounded theory: Theoretical background (chapter in) Richardson, J. (ed) (1996) Handbook of qualitative research methods: For psychology and the social sciences. 75-85. Leicester: BPS.

Pilgrim, D. & Rogers, A. (1993) A sociology of mental health and illness. Buckingham: Open University Press.

Richardson, J. (ed) (1996) Handbook of qualitative research methods: For psychology and the social sciences. Leicester: BPS.

Ritchie, J. & Lewis, J. (eds) (2003) Qualitative research practice: A guide for social scientist students and researchers. London: Sage.

Ritchie, J., Spencer, L. & O’Connor, W. (2003) Carrying out qualitative analysis (chapter in) Ritchie, J. & Lewis, J. (eds) (2003) Qualitative research practice: A guide for social scientist students and researchers. 219-262. London: Sage.

Robson, C. (1993) Real world research: A resource for social scientists and practitioner-researchers. Oxford: Blackwell.

Ryan, G. & Bernard, H. (2000) Data management and analysis methods (chapter in) Denzin, N. & Lincoln, Y. (eds) (2000) Handbook of qualitative research (2nd edn). 769-802. London: Sage.

Scolari, (2003) Sage publications software. Internet: scolari.co.uk.

Smith, J. (ed) (2003) Qualitative psychology: A practical guide to research methods. London: Sage.

Smith, J., & Osborn, M. (2003) Interpretative phenomenological analysis (chapter in) Smith, J. (ed) (2003) Qualitative psychology: A practical guide to research methods. 51-80. London: Sage.

Smith, J., Harre, R. & Van Langenhove, L. (eds) (1995) Rethinking methods in psychology. London: Sage.

Spencer, L., Ritchie, J. & O’Connor, W. (2003) Analysis: Practices, principles and processes (chapter in) Ritchie, J. & Lewis, J. (eds) (2003) Qualitative research practice: A guide for social scientist students and researchers. 199-218. London: Sage.

Strauss, A. & Corbin, J. (1998) Basics of qualitative research: Techniques for developing grounded theory (2nd edn). London: Sage.

Tesch, R. (1990) Qualitative research: Analysis types and software tools. London: Falmer.

Weitzman, E. (2000) Software and qualitative research (chapter in) Denzin, N. & Lincoln, Y. (eds) (2000) Handbook of qualitative research (2nd edn). 803-820. London: Sage.

Willig, C. (2001) Introducing qualitative research in psychology: Adventures in theory and method. Buckingham: Open University Press.

Wilson, M. (1995) Structuring qualitative data: Multidimensional scalogram analysis (chapter in) Breakwell, G., Hammond, S., & Fife-Schaw, C. (eds) (1995) Research methods in psychology. 259-274. London: Sage.

Wisker, G. (2001) The postgraduate research handbook: Succeed with your MA, MPhil, EdD and PhD. Basingstoke: Palgrave.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download