As surveys get more complicated, with skip patterns and more questions in general, the risk of survey participants making mistakes increases. Planning ahead can help to mitigate some of that risk, mistakes will always happen. What is most important is that you have a clear understanding of how to deal with these mistakes.
I have listed the two most common approaches PortMA takes with skipped questions in a survey. Determining which of these methods works best for you will go a long way to making data cleaning a much easier process.
Dealing with skipped survey questions
Skipped questions are a common error when patrons take a survey.
The easiest solution to this is to make all of the questions in your survey require a response. Any online survey tool has this capability. This isn’t without its drawbacks. Some survey participants will choose to abandon the survey rather than answer a question they would prefer to skip.
There are also two other traditional approaches for preventing skips in your data.
1. Exclude routinely skipped questions from reporting
The first is simply to ignore it. This works well if your survey is simple and does not rely upon interdependence between your questions.
I recently had a set of over 200 paper surveys sent to me where the second to last question was simply skipped. The question was largely irrelevant. It asked if the consumer had sampled a competitor’s product.
Because it only influenced one aspect of the reporting, we were able to move forward and simply exclude those results when we analyzed opinions on competitive brands.
2. Omit routinely skipped questions from the overall results
This should be done only if the question is pivotal to the results, or if your survey had a logic progression that was disrupted by the skipped question.
By doing this, you will help to keep your data clean and better ensure the accuracy of your analysis.
We recently had a survey which started by asking patrons if they were a current customer. Survey participants who said “yes” were directed to one set of feedback questions. Those who said “no” were directed to another. Those that skipped the question however, were directed to both sets.
Anyone who skipped the first question had to be deleted, as they were supplying at least one response that was out of place. It was better to delete the data than risk having false results in the analysis.
Photo Source: https://www.flickr.com/photos/deanmccoyphotos/