Wednesday, October 26, 2005

Survey Question Guidelines

I promised yesterday that I'd tackle a couple of questions posted in a comment. The first question was, "How do you determine what questions to include in a survey?"

Well, that's of course a complicated question that depends on the product, the market, and situation. However, as I've noted previously, the questions you want answered often aren't the questions you should include in a survey. The reason is response bias.

In addition to some of the traditional advice for avoiding response bias, keep in mind that:
  1. The most interesting findings often come from correlating responses to more than one question.
  2. Including one or more open-ended (fill-in-the-blank or essay) questions allows for unanticipated answers and findings.
  3. You will cause respondents to answer randomly or systematically if you include too many questions in your survey.
  4. No matter how much you want them answered, it's pointless asking questions to which your respondents don't know the answers.
  5. In general, as a market researcher, you main goal is to understand customer psychographics, not just demographics.
As you can see, surveys are about much more than just publishing a bunch of questions using SurveyMonkey.com or another tool.

No comments: