Interviewing

Disclosure
the aim of an interview is to get people to open up

Idiographic
study of facts or processes
particular biographic, historical case study
small number appropriate
qualitative analysis


nomothetic
study of general laws
General tendencies of populations
Large numbers required
quantitative analysis

Design issues

schedule wording


Structured
Exactly the same for everyone
same wording


semi-structured
same set of themes and basic items for everyone
allow for follow up questions and prompts
more flexible


unstructured
General themes and issues you want to discuss
less comparable across cases

Semi-structured

Schedule wording

Introduction


set the agenda
Tell them length
discuss confidentiality

Principles

Do


Funnelling - from general to specific
helps gathering detail, checking possible alternate responses
From open to closed


prompts
can you say more about, what do you mean?
Repeat back to them (almost Rogerian style)


feign ignorance
I don't know much about x, can you tell me?

don't


introduce assumptions
"don't you think that..."


Complex words or jargon


double negatives

Pilot study


Try schedule out on one or two people before using

Rapport


An informal conversation-like ambience to enhance disclosure
informal turn taking


But not
Agenda
roles of interviewer and interviewee


relaxed, atmosphere of openness, gain trust (particularly about sensitive, confidential topics)

Conduct
give space (physical and conversational)
if constrained time, then need to keep loose control

Recording issues & Transcription & coding

Recording


Take notes in session
Notes from memory
Recording devices and microphones
video (captures non-verbal communication)

transcription
1 hour interview = 6 hours transcription (more if granular)


Granularity
include duration of pauses
include 'erm' or 'um'
word stressing e.g. no I don't
NVC


Meanings, the gist, or verbatim (recommended verbatim)

Verbatim includes 'erm' and pauses.
Pauses will indicate something
'correct' punctuation will give false impression

Coding


selective coding - identify relevant material
complete coding - line by line


Research question determines what you code
identify references to research question themes/topics


vary how much in a coded chunk
too small chunk, miss detail meaning

Analysis issues

method of analysis


top down
you already know what you're interested in and look for


bottom up (thematic analysis)
note features without judgement
stay close to 'data' (language, concepts, categories of text)


Something in between

quantitative analysis


if you have 100 interviewees, then chi square etc. possible
count coded chunks (e.g. operationalise 'selfishness' and 'help' get frequency)
Detail lost


Consider time, effort, reliability


e.g. 5 interviewees, no choice for stats

Ambivalent attitudes would be lost
if someone says both yes and no in an answer, frequency loses all of this.

Interviews


theoretical orientation


explore a topic
give person a voice (humanistic, idiographic perspective)
interest in pps own language, concepts and categories
interest in subjective experience

Interviews high in validity (authenticity)
flexibility to find out and probe what is meant


questionnaires high in reliability
how many people... x but not what x means

Criticism

reliance on self-report


true of questionnaires
may be interested in feelings/perceptions in own right
some topics `(e.g. mass emergencies) no observational data of behaviour

Unstandardised makes comparison difficult


if you need to see how people respond to same question, this is possible
perhaps use questionnaire

Time


analysis, transcription, coding... takes time
Keep few participants

Interviewer effects


gender, power, appearance of interviewer
responses of interviewer (actions)


Some answers inhibited, some encouraged

match interviewer and interviewee
train interviewers in skills

responses taken at face value 'window into mind"
social interaction, not neutral medium
what said in interview might be different
keep context of interviewer

Advantages in other designs


focus groups - adds dynamics, interactions = data, more difficult to structre


ethnography (participant observation)


Piloting for questionnaires