« a little more on interviews. emphasis on the "little" | Main | encouragement from the mighty b: »
August 31, 2006
closing out the category, alex
Foddy, William. Constructing Questions for Interviews and Questionnaires: Theory and Practice in Social Research. Cambridge: Cambridge UP. 1993.
reality-check-type stuff to remember about interviewing:
"the relationship between what respondents say they do and what they actually do is not always very strong" (usually more of a problem related to (a) things ppl are embarrassed about or otherwise afraid to reveal or think the researcher would judge them poorly if he/she knew or (b) recall of events or (c) the correlation between attitudes & actions, i.e. the chinese tourists example) (3)
many (incredibly obvious but also some less so) things (like wording, context, situation, question-length & complexity) influence respondants' answers to questions, s.a. their ordering; respondants will answer a question differently based on the influence of the questions that preceded it (7)
when writing questions: be sure you're crystal clear on your goal b4 you write any specific questions, keep the general-goal question handy when writing specific questions, & keep asking yourself 'why do i want to know this'--demand substantive relevance answers; curiosity isn't enough (32)
language is a barrier (40)--respondants may/will hear things differently from how researchers hear/intend them. you need to know what the question means to your respondant b4 you'll know what to do w/the response
"funnel sequence" (kahn and cannell 1957) of general-to-specific questioning works b/c "general questions are the most easily justified in terms of their relevance to the avowed purpose of the study" so "they serve as natural leads for more specific questions" (62)
it's been historically common for researchers to try to tell respondants as little as possible about the study's aims, in order to keep from inappropriately influencing responses, but the fundamental flaw in that approach is human nature: "when respondants are given only the vaguest of information about the purposes of the studies in which they are asked to participate, they are put in the position of having to make guesses about the researchers' requirements" (71)--& they're going to be just as influenced by those guesses about purpose as by knowing the real thing, only as a researcher you can't account for those influences b/c if they come from guesses inside respondants' heads, you don't know what they are.
"respondants do their best to answer all questions that are put to them--even those they have trouble interpreting. it seems that when respondents have difficulty interpreting a question they use all available clues to help them....clues afforded by the way...the question is worded; clues generated by the accompanying sets of response options; clues contained both in previous related questions and in the answers the respondents have already given...and clues afforded by the overall definition of the situation" (or what they imagine it to be). "the problems that arise when different respondents are influenced by different clues...can be lessened...by clearly defining the topic in specific, concrete terms and clearly spelling out the reason for asking the question so that there is no reason for respondents to fall back onto contextual clues in order to clarify the question for themselves" (75).
among the "topics that are seen to be 'socially undesirable' and are therefore under-reported": "illegal and contra-normative behavior" including "committing a crime" (118) <--be aware that this could get in the way when students' perceptions of questions lean in the direction of their perceptions of plagiarism-hysteria
the tricksy thing about open questions: although they have lots of benefits in terms of being open to respondents' presentations of their attitudes & reference-frames, their own readings of questions & situations, & their responses based on their values & concerns rather than as limited by pre-determined response options, when respondents fail to give complete or adequate answers, "probing" for those answers easily becomes overly directive, which might skew answers and, potentially, "turns [open questions] into closed questions" (135)
plus, coding is always murky (138).
piloting questions: is a good idea. "demario (1983:101) has proposed that interviewers work in pairs when piloting questions, so that one of them is free to concentrate fully on conducting the interview while the other is free to listen to and observe both the interviewer's and the respondent's behavior" and "converse and presser (1986) suggest that interviewers who carry out pilot surveys should be encouraged to make copious notes in the margins of the questionnaires while they are doing the interviews and that they should take part in group discussions once they have completed them, to help identify problem questions" (185)
one method of question-testing: "asking respondents to rephrase questions in their own words" to see what they're hearing (186)
Posted by ttobryan at August 31, 2006 10:18 AM