When constructing each question think about the words you use. Avoid jargon and technical terms whenever possible. Try not to use words which may have a double meaning or be misinterpreted, as some words have different meanings for different groups of people. Don’t use emotive words. Make sure the question is not ambiguous. And, above all, avoid questions which will cause annoyance, frustration, offence, embarrassment or sadness. You should never make someone feel uncomfortable, for whatever reason, as a result of filling in your questionnaire.

1. Questions should be kept short and simple. This will avoid many of the problems outlined above. Check that a question is not double-barrelled, that is, two questions in one. If it is, ask two questions rather than one. Also, avoid negative questions – the type which have ‘not’ in them as this can be confusing, especially when a respondent is asked to agree or disagree.

2.Make sure that your questions don’t contain some type of prestige bias. This phrase refers to questions which could embarrass or force respondents into giving a false answer. They might do this if they do not want to look ‘bad’ in front of the researcher, or they might do it because it is expected behaviour. Questions about income or educational qualifications might illicit this type of response, so you need to be careful about how you try to obtain this information.

3. Some issues may be very sensitive and you might be better asking an indirect question rather than a direct question. Promising confidentiality and anonymity may help, but many respondents can, understandably, be sceptical about these promises. If you ask an indirect question in which respondents can relate their answer to other people, they may be more willing to answer the question.

Using closed-ended questions

If you are constructing a closed-ended question, try to make sure that all possible answers are covered. This is particularly important for time and frequency questions such as ‘how often do you . . .’ You need to make sure that all the frequencies are covered so that respondents aren’t constrained in their answers and tick a box which isn’t right for them. Also, you want to make sure that you don’t artificially create opinions by asking someone a question about which they don’t know, or don’t care. You need to make sure that you include a ‘don’t know’ category in this case.

Avoiding leading questions

Don’t asking leading questions. The question ‘How often do you wash your car?’ might seem innocuous enough. However, it makes two assumptions. Firstly, it assumes that the respondent has a car and secondly, it assumes the respondent washes his car. It could be considered a prestige bias question.Would a respondent feel bad if they didn’t have a car and therefore would tick ‘four times a week’ anyway?Would they feel bad if they don’t ever wash their car but feel the researcher expects them to? If you need to ask this question, you should ask a filter question first to find out whether the respondent actually owned a car. Then you would need to ask: ‘If you wash your car, how many times a year?’ By wording the question in this way and by being careful about the frequency list, you’re not leading the respondent into answering in a certain way.