How do you come up with good survey questions? A good survey ensures that you get reliable answers to your research questions. On this page, we provide you with a checklist and tips on how to write good survey questions. If you want to learn more about creating a good method and the reasoning behind it, such as the number of required respondents, then check out our method page.
It is important to start your survey with an introduction and any necessary instructions, so that the respondent knows what to expect. Make sure to include the exact text of the introduction in your thesis (for example, as an appendix). In the introduction, you can indicate the following:
Indicate the duration of time to complete the survey (try to keep it short if possible! Research shows that many respondents drop off when faced with a long list of questions);
Provide a good reason for why a respondent should participate in this research (relevance/importance of the research);
Who you are and what the subject is;
Describe whether the results will be processed anonymously. For the reliability of your survey, it is important to protect the privacy and confidentiality of respondents as much as possible. So only ask relevant socio-demographic questions and try to store your data set anonymously;
Thank them in advance for participating.
If necessary, an instruction explaining how to fill out the questions. The more detailed the instructions, the more consistent the measurement will be.
Would you like some help in writing the survey questions?
What is important when writing survey questions?
Below, we provide a checklist with the 10 tips you should consider:
Formulate your survey questions based on theory. This is a very important point, first finish your literature study! It is tempting to come up with your survey questions first, but you can only create a reliable and valid survey when you base it on your theoretical framework. Below is an example of an operationalization model, read more about it on our page reliability and validity. In addition to designing your own survey, it is advisable to choose an existing validated questionnaire from theory where possible. The advantage of this is that the questionnaire has already been tested for reliability and validity, but be careful: you can’t just adjust the questionnaire. The choice between a custom questionnaire or an existing questionnaire depends on your subject. Formulate survey questions with the target group and your research questions in mind. Keep the survey questions simply formulated and avoid double negatives. So avoid, for example, the statement: “I am not always prepared for new situations”. It is better to leave out the word “not” to avoid confusion. Ask mostly closed questions and not too many open questions. Closed questions are easier to analyze (in SPSS or other analysis programs). Many respondents will type very little text. It is sometimes interesting to ask an open question when you want to know unknown underlying reasons or motives. Often the answer category ‘not applicable’ is forgotten when certain questions are asked. Try to include this when possible, this will prevent respondents from just answering something. Try to ask general questions first and then specific questions. For example, first ask about the general impression of the store and then specifically about, for example, the shopping experiences such as the expertise of the employees. When you want to know the opinion of a respondent on certain statements, using the Likert scale is interesting. A Likert scale has been proven to be a very reliable, commonly used method in many studies and is easy to fill in. Usually a 5-point scale is used, from completely agree (1) to completely disagree (5). However, in various studies it has been indicated that a 7-point Likert scale leads to more accuracy (especially in online surveys). If you have asked several survey questions with Likert scale answer categories, you can often perform interesting analyses (in science, the Likert scale is generally accepted as being at the rational level).
