The WLS has 24 complete interviews and is taking a short respite from the field while data is delivered and the search is made for any preliminary indications of problems that have slipped through the first two pretests (or, have been introduced by any of the changes that we've made since the second pretest). Looking at the codebook reminded me of what I think wins the award for the most philosophically intriguing question on the telephone survey. Imagine that you are in the middle of a phone survey and the interviewer asks you:
"During the past four weeks, have people who do NOT know you understood you completely when you speak?"
My answer to this when I was a mock respondent was "Um, you tell me." I'm not sure what we would do if somebody did say "no" to this, given that the interviewer taking their "no" for an answer would be presuming that the interviewer was indeed able to understand what the respondent was saying. It's like a self-answering question, or at least answering "no" is a self-negating denial. Or maybe there should be an option for the interviewer to indicate that the respondent had given an unintelligible response, which would be the real way of answering the question in the negative. But, then, you would still have to wonder how the interviewer and respondent had made it this far in the phone interview with the respondent being unable to be understood by the interviewer.
The question is from the Health Utilities Index, a general health measure which otherwise has a variety of virtues (and is designed to be also administered in contexts other than the telephone), and so it's not something we are free to modify. So far all the actual WLS respondents have answered "yes" to the question. Looking back, though, I see that there was a case in the first pretest who did answer "no." I wonder if it was someone caught up in the epistemic angst of feeling like one can never be completely understood. Or maybe they just feel like they are chronically, hopelessly misunderstood, although that would be perhaps more likely if this was a survey of thirtysomething brooders rather than people in their sixties. I wish I could go back to the pretest 1 cases and look for clues to explain that one respondent's "no" response, but I don't even have the time to be making these blog entries!