12.01.2015 Views

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

VALIDITY AND RELIABILITY IN INTERVIEWS 151<br />

minimize this. Silverman (1993) suggests that it<br />

is important for each interviewee to understand<br />

the question in the same way. He suggests that<br />

the reliability of interviews can be enhanced by:<br />

careful piloting of interview schedules; training of<br />

interviewers; inter-rater reliability in the coding<br />

of responses; and the extended use of closed<br />

questions.<br />

On the other hand, Silverman (1993) argues for<br />

the importance of open-ended interviews, as this<br />

enables respondents to demonstrate their unique<br />

way of lo<strong>ok</strong>ing at the world – their definition of<br />

the situation. It recognizes that what is a suitable<br />

sequence of questions for one respondent might be<br />

less suitable for another, and open-ended questions<br />

enable important but unanticipated issues to be<br />

raised.<br />

Oppenheim (1992: 96–7) suggests several<br />

causes of bias in interviewing:<br />

biased sampling (sometimes created by<br />

the researcher not adhering to sampling<br />

instructions)<br />

poor rapport between interviewer and interviewee<br />

changes to question wording (e.g. in attitudinal<br />

and factual questions)<br />

poor prompting and biased probing<br />

poor use and management of support materials<br />

(e.g. show cards)<br />

alterations to the sequence of questions<br />

inconsistent coding of responses<br />

selective or interpreted recording of data/<br />

transcripts<br />

poor handling of difficult interviews.<br />

One can add to this the issue of ‘acquiescence’<br />

(Breakwell 2000: 254), the tendency that<br />

respondents may have to say ‘yes’, regardless of<br />

the question or, indeed, regardless of what they<br />

really feel or think.<br />

There is also the issue of leading questions. A<br />

leading question is one which makes assumptions<br />

about interviewees or ‘puts words into their<br />

mouths’, where the question influences the answer,<br />

perhaps illegitimately. For example (Morrison<br />

1993: 66–7) the question ‘When did you stop<br />

complaining to the headteacher’ assumes that<br />

the interviewee had been a frequent complainer,<br />

and the question ‘How satisfied are you with<br />

the new Mathematics scheme’ assumes a degree<br />

of satisfaction with the scheme. The leading<br />

questions here might be rendered less leading by<br />

rephrasing, for example: ‘How frequently do you<br />

have conversations with the headteacher’ and<br />

‘What is your opinion of the new Mathematics<br />

scheme’ respectively.<br />

In discussing the issue of leading questions, we<br />

are not necessarily suggesting that there is not a<br />

place for them. Indeed Kvale (1996: 158) makes<br />

a powerful case for leading questions, arguing<br />

that they may be necessary in order to obtain<br />

information that the interviewer suspects the<br />

interviewee might be withholding. Here it might<br />

be important to put the ‘burden of denial’ onto<br />

the interviewee (e.g. ‘When did you stop beating<br />

your wife’). Leading questions, frequently used<br />

in police interviews, may be used for reliability<br />

checks with what the interviewee has already said,<br />

or may be deliberately used to elicit particular<br />

non-verbal behaviours that give an indication of<br />

the sensitivity of the interviewee’s remarks.<br />

Hence reducing bias becomes more than<br />

simply: careful formulation of questions so that<br />

the meaning is crystal clear; thorough training<br />

procedures so that an interviewer is more aware<br />

of the possible problems; probability sampling of<br />

respondents; and sometimes matching interviewer<br />

characteristics with those of the sample being<br />

interviewed. Oppenheim (1992: 148) argues, for<br />

example, that interviewers seeking attitudinal<br />

responses have to ensure that people with known<br />

characteristics are included in the sample – the<br />

criterion group. We need to recognize that the<br />

interview is a shared, negotiated and dynamic<br />

social moment.<br />

The notion of power is significant in the<br />

interview situation, for the interview is not simply<br />

a data collection situation but a social and<br />

frequently a political situation. Literally the word<br />

‘inter-view’ is a view between people, mutually, not<br />

the interviewer extracting data, one-way, from the<br />

interviewee. Power can reside with interviewer<br />

and interviewee alike (Thapar-Björkert and Henry<br />

2004), though Scheurich (1995: 246) argues that,<br />

Chapter 6

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!