Survey Construction Tips

Each semester when some of my research students decide to use surveys as their primary data collection instrument, I go through some of the same theory again.

But each time I do it, I get more clarity in explaining exactly why we do what we do.

All surveys need an informed consent section at the beginning, with a "next" button that by clicking, our participants agree to our research conditions.

We should always get volunteers to test our questionnaire, and ask them to time how long it took them, so we can put a realistic time estimate in the informed consent section; so our research participants know what they're letting themselves in for.

I find it very interesting how theory easy it is for us to present our survey questions in a way that will "turn off" our research participants. By that, I mean that we tell our research participants in our consent section that our survey data will be completely anonymous; we get them to agree to participate by clicking the next button; then the first questions we ask are very personal questions about demographics. Our participants immediately feel that we are invading their privacy.

While what we are asking them does not necessarily make them immediately identifiable, it undermines the spirit of our psychological contract with our participants.

Instead, we need to warm up our participants, by asking them questions that are more related to our actual research question. We start with questions which are fairly simple, then as our participants get used to how we ask questions, we can ask our more important and complex questions.

Then lastly, we can have a section on what are the absolutely important demographics that we absolutely have to know in order to cross-tab without other answers for our research to be successful.

We aim to have only 10 to 15 questions overall. Part of our question will include the instructions on how each question needs to be answered. We ask all our questions - where possible - and exactly the same way. If we are using Likert scales, we use the same Likert scale throughout our questionnaire.

We cluster similar questions alongside each other. We use closed questions before asking open ended questions, to give our participants time to consider their answers before they give them.

I would recommend we normally close with a large and welcoming text box, asking participants if they want to provide any additional information, or make any comments on the survey. This text field often contains an amazing amount of "gold" when you extract and analyse your data.

After this we ask any demographic data questions that we need. By this stage we have built a relationship, of sorts, with our participants; they will be more willing to trust us.

Then lastly we have a thank you page, which also contains our email address, should the participants wish to get in touch with us.

This should be common sense. But it is surprising how often we can all forget these simple principles.


Sam

• References: Remenyi, D. (2012). Field Methods for Academic Research: Interviews, Focus Groups and Questionnaires. UK: Academic Conferences Limited.