Create a survey

Compiling a questionnaire

Questionnaire is your contact with respondents and therefore it is very important how you ask questions. Follow next guidelines:

  1. Creating you own questions means more freedom when forming and shaping your questionnaire. But there are doubts about objectivity, reliability (in what rate we get the same results on same units and in the same conditions) and validity (if we measure what we intended to measure). In order to improve these three components multiple measurement scales and questionnaires were developed. As we can see it is a good idea to check if scales that measure information we need already exist.
  2. Don’t force respondents to answer – don’t use hard reminders – but check if respondents provided information that was asked for. If for example, you want to find out the year of birth, limit input filed in a way that only four numeral characters can be inserted. A warning should just remind the respondent that provided information was wrong and ask for a correction, but it shouldn’t disable continuation of filling out a survey. Prohibiting continuation can force respondents to provide made up data or make them stop filling out the survey.
  3. Questions should be simple and understandable. Jargon, slang and professional words should be avoided.
  4. Question order is important. Next guidelines should be followed:
    • Question order should be logical. For example, questions about events should match order of the events or questions about valuing a certain experience should follow questions about describing that experience;
    • Questions about the same topic should be together. Constant switching between topics decreases probability of well-thought-out answers. If the questions are organized according to a topic, respondents can focus on just one topic at the time;
    • Topics (and questions) should follow from those most important to respondents to those least important to respondents. Research shows that topics more important to respondents have higher response rates;
    • Sensitive (hard) questions should be at the end. You don’t want respondents to stop filling out the survey;
    • Demographic questions are usually at the end of the questionnaire;
    • Questions with similar parts should be together – for example, questions with the same scale should be put together. In this way, we reduce the cognitive burden of the respondents.
  5. First question should be appropriate for anyone (the only exception is when we want to determine “adequacy” of the respondents):
    • It should be simple for respondents to read, understand and answer in a couple of seconds. This is not the place for long questions with many possible answers or for open questions;
    • It should be interesting to increase the probability of respondents to participle in a survey.
  6. Measurement scales should be consistent throughout the survey. If the scale from 1 to 5 was made, that scale should be used in the survey. Naming of highest and lowest values should also be consistent (1 should always be the lowest agreement and 5 always the highest and the other way around).
  7. Be careful when creating the first and last page of the questionnaire. The first page should motivate respondents, therefore you shouldn't write instructions on it. The last page should be simple; it usually involves a thank you note.
  8. Determine which type of question is best for each question.
  9. If the survey consists of many questions it can also have multiple pages. Too much scrolling is not desirable. It is also not desirable to put only one question on each page, because that increases the time of filling out the survey and consequently increases the likelihood of respondents to leave the survey.
  10. Use conditions and branching if necessary. Never include text like "If you answered 'yes' on the question Q1, answer question Q2, if you answered 'no', go directly to question Q3".
  11. Check if all possible answers are enabled with closed questions. If you are not sure, use the option “other”.

<< Back to Few recommendations