March 25, 2010

How do you decide how to word your survey questions?

Q: Do you ever take suggestions from the general public on the proper wording of questions? One question I have seen multiple times in your polls is along the lines of “How important is religion in your life?” (The exact wording may vary slightly from time to time.) Even though I am an atheist, and I never participate in any kind of religious ceremony or partake of the activities of religious organizations, the way the question is asked means I have to answer “Very important” if I am being honest. Religion is probably the MOST important negative in my life — I am constantly having to do battle over some issue where religion is the instigator and the main force for ill. Might I suggest that you re-work your question to separate out those who find religion “important” in a positive way from those who find it “important” in a negative way?

We are always open to input about our questions and have occasionally made changes in response to ideas and suggestions that come from outside the Pew Research Center. It’s certainly possible that there are other people like you who say that religion is important in their life but mean it in a negative way. But we think that’s likely to be a very small portion of all those who say religion is important. In the 2007 Religious Landscape Survey conducted by the Pew Forum on Religion & Public Life, 56% of all respondents answered that religion was very important in their lives. Yet among self-described atheists in the study, just 3% did so. This suggests to us that most people interpret this question the way we intend it to be interpreted.

Even if we have concerns about certain survey questions, we sometimes continue to use them because their trend over time is too valuable to lose. A good example can be found in the series of agree/disagree questions in our ongoing survey of political values. Survey methodologists have found that the agree/disagree format is susceptible to bias because certain types of people may be more likely than others to agree with a statement presented in this format than to the same statement presented in a more balanced format. But because we first began asking these questions in 1987, their value in tracking changes in opinion over time is very high; accordingly, we still periodically use these questions (though in 1994, we began a new series with many of the same concepts presented in a “balanced alternative” format).

The question about importance of religion has an even longer history than our values series (a version dates back to the mid-1960s), so we would be reluctant to change it unless we thought it was seriously misleading.

Scott Keeter, Director of Survey Research, Pew Research Center