“It’s so easy,” she said, “just put it into Survey Monkey.”
The truth is, it is not that easy, for a multitude of reasons. Yes, it is easy to make a survey using a free online tool, but there are other factors, such as the experience for the person taking the survey and the experience the data scientist encounters with the tool. What about the survey design itself? It’s a topic we don’t often discuss and, all too often, people create surveys that should only be used to provide anecdotal (rather than validated) insights.
I often use the expression “you can spend 20 years researching…”, but it is true, especially here, where you can even get a PhD in psychometrics. That being said, I’ll focus this article on the upfront part of designing a survey: designing the purpose and developing the questions. If you are interested in longitudinal design, meta-analysis, or Bayesian probabilities, we have a few colleagues on staff who are happy to have a chat over coffee.
It may seem basic, but asking “what are the goals/purpose/use of this survey?” leads to a lot of other questions that inform survey design. Some of our clients know very little about their target customers and use surveys as exploratory tools in the customer research process. We also have clients who use surveys as a way to validate their research processes. For example, we are currently interviewing a large group of customers to understand how they may perceive and react to a change in the industry. Since we can’t interview everyone, we are starting with a small relevant sample. Their feedback so far hasn’t been homogenous, which will allow us to better design a survey for a larger sample size, drilling down in the right areas to provide more valid results. Read more »