Good Survey Questions Lead to Good Data

When writing a survey, a question is a question, right? Wrong.  Good survey questions can make the difference between solid results and weak data.  So how do you know what makes a good survey question?  Let’s look at some common mistakes to avoid.

Question Mark: Good Survey Questions

Be Specific

When writing a survey, it’s easy to forget certain aspects of your questions such as time frame.  “Have you ever donated to a political campaign?” is a completely different question than “Have you donated to a political campaign within the last five years?”  Yet the only difference is time frame.  Try to avoid wording questions like: “Have you donated to a political campaign?” which doesn’t specify a time frame at all.  Good survey questions never leave respondents wondering.  Tell them exactly what you mean and leave no room for interpretation.

Cover All Bases

Your respondents should never be left thinking that there’s no answer that describes them.  You should always include at least one catch-all answer that describes anyone who doesn’t choose any of the other answers.  “All of the above,” “none of the above,” and “Other, please specify” are all examples of these catch-all answers.

Randomize

When taking a survey, respondents sometimes experience something called “first order bias.”  This means they naturally focus more on the options listed toward the top than those listed at the bottom.  By changing up the order of your answers, you can eliminate most of the skewing that can come along with first order bias.  However, continue to list answers such as “all of the above” or “none of the above” at the bottom of the list.

Keep it Simple

If your questions are so complicated that respondents have to read them multiple times just to understand what you mean, you need to simplify.  Good survey questions are always simple and to the point.  Stay away from industry jargon and wordy phrases.  Make sure to read your questions as if you are the survey respondent seeing them for the first time.  If necessary, ask someone else to read your questions to see if they make sense and aren’t too complicated or wordy.

Give Clear Instructions

The last thing you want is for your survey respondents to wonder if they’re responding correctly.  Doubts may lead to respondents not completing the survey or doing so incorrectly.  If you want respondents to select all that apply, say so.  If you want respondents to choose the answer that best applies, say so. Don’t leave any room for interpretation, because people will interpret differently, leaving your results skewed and incomplete.

Photo Credit: Question mark made of puzzle pieces from Flickr

ABOUT THE AUTHOR: Lev Mazin
Lev is the CEO and Co-founder of AYTM.com - Ask Your Target Market, the leading innovator in DIY online market research. Having a graphic design background, Lev is an expert in UI/UX with over 15 years of experience. Prior to AYTM, Lev worked with such clients as Oracle, Tiffany & Co, Maserati, Harry Winston, Whole Foods Market, and Jamba Juice helping them build their brands and multimedia/web applications.
  • Anonymous

    This is a great resource for anyone putting together a survey! Some of the points like randomizing the order of the answers make so much sense, but are often overlooked. I also like the catch-all. It’s all about making the feedback process easier and simpler for the target market.

  • Damien-Maddox

    Would split-testing surveys go under the randomize section here? I’ve seen it done for websites, but I imagine you could randomly display one of two very similar surveys to see which has a larger response, which gives better results, etc….

  • Ash

    Dats very useful Information..Thank You!