Learning patterns/Testing survey questions
What problem does this solve?
When doing a survey, it can be hard to know if the survey works well. It is always best to test surveys before doing a larger collection of data. You can also run a short survey to test the entire process before doing a larger survey
What is the solution?
There are many ways to test surveys. Here are a few common ways:
- Write the survey with a colleague. Since surveys need to be taken and understood by others, it is helpful to have the perspective of at least one other person when starting to write questions
- Invite a few of your target users to read through your survey. You can ask users to read through your survey out loud; this will help to highlight major problems of the survey as they read through the survey and tell you their thinking. The biggest advantage is to find jargon or questions that are hard to understand.
- Test the survey in the survey software. Make sure to test the survey once its in the software, especially if the survey has many skip patterns.
- If you are doing translations, it is imperative to test your survey at all stages to make sure you have no problems from copy/pasting or from translating jargon.
- Test an entire survey from start to finish, before doing a major survey. You want to make sure that you have the right process in place for sampling and that you are getting the information you expected from respondents. After doing a small scale survey (say to 1-200 people), you can then do a much larger survey with more confidence and with less risk that you will have a major data blunder.