Methods of qualitative data analysis

1
Methods of qualitative data analysis

• Module 1 Course
• Session 4

2
"What do we do with these disorderly piles of
data?" (Lewis 1991)

• Reducing the discursive data to manageable
  proportions?
• Distilling the meaning from the qualitative
  interviews?
• Requirement Being systematic!
• for exhaustiveness (nothing left out)
• for credibility (no bias)

3
We need methods for

• Organizing the interview text (manageability)
• Analysing the manifest meanings ('essence')
• Interpreting latent meanings (beneath the
  surface)
• No magical tool just hard work!

4
Transcribing interviews

• Different practices academic vs. commercial
  research
• Transcribe in toto all utterances, verbatim,
  indicate 'circumstances of the utterance
  (laughter, irony, gesture, facial)
• The best way to defeat faulty first impressions
• Go back to the tapes a more faithful rendering
  of what was said and how

5
We don't start from scratch!(as in radical
grounded theory)

• "The central task of interview analysis rests
  with the researcher, with the thematic questions
  he or she has asked from the start of the
  investigation and followed up through designing,
  interviewing, and transcribing." (Kvale, 187)
• Start with the themes of the interview guide!
• (Researching Audiences Box 9.2 and 9.4)

6
The pervasiveness of interpretation 1

• 'analysis' vs. 'interpretation' a positivist
  misconception interpretation is not an isolated
  stage! BUT
• Kvale pragmatically distinguishes surface
  'analytical' operations from deeper kinds of
  'interpretative' operations (188,193, 201)
• ANALYSIS text reduction, manifest meanings, what
  is directly said, more objective
• INTERPRETATION text expansion, deeper
  dimensions, meanings not apparent, more
  speculative

7
The pervasiveness of interpretation 2

• - also the lesson from the Chinese Boxes model
• The outer box The meaning construction process
  goes on making sense of/interpreting/decoding
  the transcript.
• Two dimensions to analyse
• 1. Language representing social reality, talking
  about the world
• 2. Language as patterns of interaction,
  turn-taking, co-production of meaning,
  negotiation of agreement.

8

9
5 approaches to interview analysis

• Meaning condensation one looks for the natural
  meaning units in the interview (e.g. a couple of
  paragraphs) and rephrases them as a 'theme'. (Ex.
  Kvale p.195)
• little interpretation involved - the analyst
  tries phenomenologically - to be the
  informants mouthpiece (Kvale p. 196).
• Corp.ad. study articulating the essence of an
  informant's experience of one ad

10

• Categorization the analyst codes aspects of the
  interview text into a small number of dimensions
  or categories "This experience is X or Y or Z".
  On the road to quantification.
• Corp.ad study This experience of ad A is
  category Symp/Agn/Cyn.
• Narrative analysis registers stories told during
  interviews plot, good versus bad, the point of
  the narrative. Or analyst creates a coherent
  story out of the many happenings reported
  throughout an interview
• Corp.ad study Not used.

11

• Interpretation analyst looks for latent
  meanings, beyond a structuring of the manifest
  meanings to deeper and more or less speculative
  interpretations.
• inferential meanings "profits cut dramatically"
  (Res.Aud. 168).
• everyday metaphors "add a bit of spice", "just
  myself again" (Res.Aud. 167)

12

• Ad Hoc meaning generation
• Using different approaches and techniques for
  meaning generation (e.g. the previous 4)
• "free interplay of techniques"
• No standard method
• "We believe that to use any one interpretive tool
  is insufficient to deal with the richness of the
  interview data. () we recommend the ad hoc
  procedure, as long as it remains systematically
  structured and does not become idiosyncratic and
  impressionistic" (Res.Aud. p. 166 )

13
Benefits of 'categorization'

• Potentially provides an overview over the
  meanings experienced by the informant (Kvale
  fig.11.3, p. 198).
• Corp.ad study 1. Constructing each informant as
  an 'aggregate' category of Sym/Agn/Cyn. 2.
  Constructing a map for positioning informants in
  landscape of intersecting Sym/Agn/Cyn
  territories.
• Provides a basis for determining typicality of
  each informant.
• Potentially enables social differentiation
  between informants (level of education).
• Potentially enables comparisons (DK vs. UK)

14
Limitations of 'categorization'

• Reduction of complexity
• Rendering ambivalences invisible

15
Control of analysis Credibility, quality,
craftmanship?

• Did other eyes see the same?
• Intercoder reliability/Multiple interpreters
• procedures for coding agreement.
• Interviewer reliability different interview
  styles
• Would other eyes see the same? (yours!)
• Documentation through procedural explicitness
  and extensive transcript excerpts.

16
What reliability is all about- when more than
one interpretation is possible

• "The chief point to be remembered with this type
  of research is not so much whether another
  position with respect to the data could be
  adopted (), but whether a reader, adopting the
  same viewpoint as articulated by the researcher,
  can also see what the researcher saw, whether or
  not he agrees with it. That is the key criterion
  for qualitative research."