9 Tips to Writing Good Survey Questions

Today, surveys still carry much importance in gathering metrics and understanding people's opinions. In business, customers provide valuable feedback on products and services. In education, children's measures of intelligence are gathered through "survey tests". In Economics, unemployment rates are calculated by the Bureau of Labor Statistics via phone, mail, and online surveys. The point is, asking questions is important!

It is essential to any market researcher to write questions that are understandable and can provide accurate and consistent results. What is a good question? A good survey question produces answers that are reliable and valid measures of something we want to describe. Just the style of writing a question can alter its meaning to the participant. Here are a few ways to write questions that can make a survey more effective.

  1. Keep the language very simple. A survey question is not a test of your vocabulary skills. The use of language that is incomprehensible to the participant will only distort the findings. So avoid framing a question like “What is the frequency of your culinary visits to the nearest non gourmet eatery in a week”. Keep it simple with “How many times per week do you go to the nearest fast food joint”.
  2. Ensure Answers are Consistent. Answers should be structured so that there is only one way to answer. For example, the open-ended question "What type of shampoo do you use?" can be answered with a brand, a shampoo type, or none-I don't use shampoo. Also ensure that all possible choices are included, if you suspect there could be other choices, include an "Other" category. Mineful's online survey software allows you to include the "Other" category and adds a space for the respondent to elaborate.
  3. Avoid using negatively worded questions. Double negatives are a no-no. Instead of asking “Should the government not provide free medical support to pregnant women?” rephrase the questions as “Should the government provide free medical support to pregnant women?”.
  4. Use close ended questions as much as possible. Any marketing analyst knows that analyzing close ended questions provides more opportunites for advanced analysis. Yes, open-ended questions allow for text analysis, but survey reseach methodology goes far and beyond when analyzing close ended question. As an example, avoid using a question like “What kind of food do you usually order in a restaurant?”. Instead change it to “Which of the following do you usually order when you visit a restaurant?".
  5. Avoid leading questions. Adding a personal opinion to a question when it is not necessary can bias the answers of the participants. For example, the question “Has the new, modern and state of the art audio system installed in the auditorium improved the voice quality” is sure to lead the participant to answer “yes”. This can be avoided by just asking “Has the voice quality in the auditorium improved”.
  6. Do not assume participants know. There are some terms that not everyone would be familiar with, or have sufficient knowledge about. For example while asking” Are you in favor of Proposition 13” it has been assumed that everyone knows what it means. Who is your sampling frame (the total population you are interested in describing)? Do they understand what these technical terms mean? Is the language adequate for them?
  7. Minimize choices. Survey questions like ranking items by importance should not have more than 5-6 options. As the number of items increases the reliability of the answer falls. If a choice is likely to obtain less than 5 responses, it is recommended to merge it with another category or inside an "Other" option.
  8. Pre-test your questionnaire. Once the survey is complete, before sending it to the target audience, send it as a test to a small group of people you know to check if the responses obtained are consistent. They will also be able to tell you if any question was confusing or should be altered to make it easier to answer.
  9. Ensure responses are valid. There are several ways to determine if the responses you obtained are reliable and trustworthy.
    • Evaluate the strength of anticipated relationships - if you expected 50% and got 80% that should ring a big alarm.
    • Compare to similar questions - Surveys sometimes include similar questions at the begining and at the end to ensure that the respondent is consistent with their answers. Similarly, one can compare responses of the same person at different points in time.
    • Construct test questions that lead respondents to only one answer - For example, "This question is very long...Answer the third choice for validation purposes". This helps validate that the respondent is not speed responding the survey in order to obtain their incentive.
    • Reinterview in person or by phone - Only for highly speculative responses that carry significant weight. For example, at BLS, they would reinterview companies like Boeing and GE before releasing the unemployment rates since their employee numbers carry significant weight for the national unemployment rate.


Post a Comment

Back to TOP