Quoting Stephen R. Covey:
Begin with the end in mind.
It seems a simple phrase, but it takes more thought that it first appears. The questionnaire is the key part of your survey. Asking the wrong questions can be worse than receiving wrong answers. There is no turning back once the questionnaire is out.
Consider the following question: “Do you have a computer?”
Does this mean you want to know if the participant owns one themselves, or has access to one. Perhaps someone uses one at work, but does not have one privately, is the answer to this question yes or no?
What do you mean with computer? Is this a laptop or a desktop? Or does this even include the latest innovations of smart phones, BlackBerry, PDA, iPhone, netbooks etc.
As you can see, such a simple question throws up a whole range of potential pittfalls.
It is therefore important when designing a questionnaire to have these kinds of pittfalls in mind. Every question should be unambiguous and clear.
Types of questions
There are different types of questions, mainly grouped into two sections.
- Open questions – this gives participants the option to write their response on your question. When designing your questionnaire you will have to take into account the space you give to the participant. If you want to limit this, you will have to make this clear in your question, by saying that for the answer a maximum of xx words can be used
- Closed questions – these can be grouped into a further four sections. The participants have to choose which answer to pick. When designing your questionnaire you will have to take into account the potential sensitivity of the question. To avoid getting ‘wrong’ or misleading answers, you can opt for either give the possiblitity of leaving the answer blank, or for a “rather not say” option. This last option guarantees that the person has read the question and answered it, and not skipped through the survey.
Nominal questions: the answer to these questions have no relationship and no mathematical order. Examples of nominal questions are gender of the participant, Yes / No answers, relationship status
Ordinal questions: the answer to these questions do have a relationship, but no mathematical order. An example of ordinal questions are ranking answers from best to worst. The most common used answer option is the Likert scale.
A Likert item is simply a statement which the respondent is asked to evaluate according to any kind of subjective or objective criteria; generally the level of agreement or disagreement is measured.
Integer questions: the distance between the answers is the same but there is no true zero. This option is mainly used for temperature and therefore not really common within most questionnaires
Ratio questions: the answers are made up of numbers, where you can multiply and divide the answers (i.e. there is mathematical order). Common questions in surveys are age, length, salary and weight.
Final thoughts on questionnaire design
Make certain that when you plan your time for questionnaire design, you include time for testing the questionnaire. Use your friends, family, colleagues, neighbours and anyone else you consider appropriate as test participants. This can give you a good idea if your questionnaire is simple, unambiguous and measures what you want to measure
Include an introduction to the questionnaire and thank the participant for their time. Courtesy still works.
Quality design in a questionnaire will result in more useful data which leads to greater accurracy in analysis.