No matter the method you choose, you must be thoughtful about the questions you ask. The following are some suggestions to ensure that the questions you design meet the standards of best practice.
One question at a time
Each question you design for your survey should be about one thing and one thing only. If you find yourself wanting to ask two questions, design them as separate questions instead of squeezing them into the same sentence. Asking two questions in one can lead to confusion and force respondents into answering a certain way. For example, the question ‘Which candidate did you vote for in the last election?’ actually contains two questions: 'Did you vote in the last election?' and 'If so, which candidate did you vote for?'
Clear questions that don’t “lead” the respondent to a certain answer
When devising questions, you must always keep your assumptions in check and ensure that you remain as neutral in your phrasing as possible. It is very easy for your own attitudes and opinions to sneak into your language. For example, asking the question “How much will prices go up next year?” shows an assumption that prices are increasing, and could lead a respondent to answer differently than had he or she been asked “Do you think prices will go up or down next year?”
Allowing for lack of opinion/knowledge
Not all questions will be applicable to all of your participants. It is important to allow people the option to say they don’t know or the question is not applicable (N/A). For ethically sensitive questions, you may want to include an option of ‘I’d rather not say.’
Careful wording of negatives
Questions should be as simple and clearly worded as possible. Asking someone to choose the answer that is NOT true can lead to misunderstandings and inaccurate data. Likewise, double negatives can inadvertently ask the opposite of what you meant. For example, instead of asking “How often do you feel you aren’t unsupported?” you should ask “How often do you feel supported?” instead.
Asking the same question two ways
There is value in asking questions on the same topic in both positive and negative formats. Doing so allows you to check that people are understanding what you are asking, and are consistent in their answers. In the above example, asking how often a person feels supported and also asking how often a person feels unsupported could act as a validation tool. If they say they rarely feel supported and rarely feel unsupported, this would be a discrepancy that you would want to look into more carefully.
Pilotingis when you test out your questions on at least one or two people first – preferably from the population which you will be sampling - and get their feedback. Note any times you had to reword or explain a concept. Ask them what they would have liked to be asked, or if you are missing anything.
How many questions?
Think of beneficiaries. In general, adults can maintain uninterrupted focus for around 20 minutes at a time. For children, a good guideline is that their attention span is roughly their age in minutes. Therefore, if you are asking a twelve-year-old to fill out a survey, it should take no more than 12 minutes. It is also important to remember that the amount of time it takes you to fill out the questionnaire or survey is not necessarily an indicator of how long it will take your respondent to do so. This is something else that should be tested in your pilot by members of the population.
For more information on asking questions, a good resource is Chapter 11 of Bryman, A (2012) Social Research Methods. Oxford: Oxford University Press.
The Research Methods Guidebook (produced by the Economic and Social Research Council, Researcher Development Initiative, and Institute of Education, University of London) provides excellent guidance on asking questions.