Monday, August 13, 2007

Question Order Effects

The order you ask questions can influence how people answer survey questions. That's why you often see the general all-purpose "how would you evaluate the president's job" first, not after a long series of questions that ask about all the stuff that is wrong with the country.

Order matters.

One of my earliest academic efforts was a question-order paper. Hell, it even won first place in the theory and methodology division at AEJMC (must have been a weak year because I didn't even mention agenda setting, and I don't think T&M allows a paper that doesn't mention agenda setting. It's a law).

So how about political knowledge? Well, according to one study, asking difficult political knowledge questions early in a survey will dampen how much respondents say they pay attention to the media and are interested in politics. That's kinda interesting. Maybe they get these tough questions and become more self aware of their actual knowledge or interest or media use, or perhaps it artificially dampens their true sense of how much they care and keep up with public affairs. Give people a "buffer" item that allows them to think it's okay to not know so much and the effect is less.

The upshot is such questions, like sensitive items people hate to answer, should go later in the survey. We often preface questions today to make it okay for people to say they don't vote (always inflated) and now you will see respondents told that some people don't keep up with the news, then they are asked "how about you?" This makes good sense. We want accurate measures, not socially desirable ones.

No comments: