Survey Design

Aims and Objectives
The initial design stage of a research project is crucial; there are two main questions that you need to think about:
    • Why are you conducting a survey?
    • What are you hoping to achieve?
A well designed survey has a well thought through aim. Without a clear purpose the survey will collect information that is useful to no-one, and waste both your time and that of your respondents. Having a clear understanding of what you hope to achieve from the survey will guide you in composing the questions and deciding upon the question types you will use.

Question Types
There are several different types of question that you can use on your survey, each with their own merits and problems. It is important that you use the right type of information to get the type of answers that are meaningful to you.

Single answer – use this when you only want the respondent to select one response from the list of options (e.g. yes or no). Make sure that respondents would not want to provide more than one answer to the question.

Multi-choice – use this when you want the respondent to select more than one response from the list of options (e.g. the reasons that you like an area or particular store).

Grid – use this when you want the respondent to rate several statements on the same scale. This usually works bet when using a version of the Likert scale.

Semi-open (‘other’ responses) – For every question that you have not provided a definitive list of responses for, you should include the opportunity for respondent to provide any ‘other’ responses.

Open – this is where there are no predefined responses and the respondent can give whatever answer they want. This type of question is useful for getting qualitative information from respondent’s, but can be harder to analyse if there are a lot of respondents.

Other question options – particularly when using online survey tools (such as my favourite, SurveyGizmo), there a lot of different question options that can make the survey experience better for respondents. These includes sliders and a variety of pictographic questions.

Additional Tips
Questions need to be clear and easy for everybody to understand and you need to know your audience. If you are surveying senior executives or academics then you can ask complex questions on difficult subject matters, however, if you are surveying the general public you must make sure that all potential respondents can understand the questions, otherwise you may end up biasing the results.

Make use of routing and skip logic to ensure that only relevant questions are asked. Particularly with telephone and online surveys, it is important that respondents do not see, or are not asked questions that are not applicable to them. Asking respondents irrelevant questions will lead to them quitting the survey.

Questions should be free from bias and not lead the respondent to a particular conclusion. This short clip from the classic Yes, Prime Minister demonstrates this to perfection.
Lastly, you should make sure that you avoid double questions. By this I mean asking two questions in one, such as ‘how satisfied are you with the value and quality of the services you receive?’ If the respondent has different views on the value and quality of the services they will have trouble answering the question.

It is important for you to consider whether you want the survey to be anonymous:
  • If the survey is anonymous, respondents may be more inclined to take part and be honest
  • If not anonymous you can link responses to other information you have, such as transactional or demographic data

Compulsory Questions
For telephone and online surveys it is possible to make questions compulsory, i.e. respondents have to provide an answer to the question.  If you make any questions compulsory, you should include a ‘don’t know’ option.  ‘Don’t know’ is a valid answer to many questions and should be provided. ‘Refused’ is also an important option to include, particularly for compulsory questions and demographics.  It is vitally important that you do not force people into selecting an incorrect answer.

Rating Questions
Rating questions on a Likert scale (e.g. strongly agree/ agree/ neither nor/ disagree/ strongly disagree – see above) are widely used in research and can be very useful.  However, how many options should you provide, 3, 5, 7, more? I suggest that you have more than 5 options, otherwise it becomes difficult to tell the difference between the meaning of the responses.  You’ll also notice that I have only suggested odd number – this is because you should always include a ‘neither/ nor’ option in the middle of the scale.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s