There are a number of ways that survey questions may be problematic. You should not ask questions that your respondents are unlikely to know the answers to, or imply a demand for an unreasonable degree of precision.
- If you provide a set range of answers for participants, make sure that they are clear about whether you are requesting that they select only one answer from the set, or that they should indicate all that apply. Be careful that the language you use to express the question is not leading, or loaded.
- Avoid value-laden language which might imply what you expect to be the ‘correct’ answer. Only ask one question at a time. If you find that the text of a closed question (one for which you have provided a series of possible answers for the respondent) contains words like ‘and’ or ‘or’ you may be asking your survey participants to give one answer to two or more different questions. Avoid complex constructions, such as double-negatives, in your wording.
The order in which questions are asked, or the nature of the set of questions in which a given question is embedded may have an influence on the response given.
Ambiguous, or poorly designed questions which are obvious to the respondent will make it difficult for them to respond, and thus give them a bad feeling about the whole questionnaire, or indeed the entire research project with which they are engaging.
And you cannot hope to obtain valid information from poorly designed questions.