How Survey Software Can Help Prevent Dropout
Dropout is when people start taking the survey, but then don’t finish it. That is always a problem. When people drop out in the middle, that can make the data a little less reliable, since his three might be something non-random about who chooses to quit in the middle. Your survey software should help you determine problem points in your questions.
One feature that distinguishes the most sophisticated survey software from other packages is that they include dropout analysis. Dropout analysis is a report that shows at what points survey you are losing participants. You could potentially learn where people drop out without any special features in your chosen tool by producing reports and seeing how many people answered each question. If you have any skipping or branching in your survey, that can make comparing how many people answer each question problematic. People who start answering questions, but then stop in the middle would look the same as people who skip the question because of skip/branching logic. The most capable programs can produce dropout analysis reports that look at each individual and whether they should have skipped any questions based on logic, and not count them as having dropped out because of not answering a question they didn’t see.
A complete dropout analysis report should include information such as the label/titles of the questions, the type of question – multiple-choice, numeric answer, grid or matrix or text answer – what percentage of your respondents answered the question, the change from the previous question and what percentage of them dropped out of the survey at that point. You might find out that certain questions are particularly problematic from this perspective.
You should also be able to include information such as whether an answer was required to each question. Requiring answers can increase the dropout rate, since some people might not wish to answer certain questions. It is still valid to require answers in some situations, but it can be useful to know when doing so for particular questions causes dropout. In some cases there might be a simple fix for the issue. For example, if you presented a question with a list of choices without a “none of the above” or “don’t know” option and a significant number of people refuse to answer it, adding one or both of those choices might reduce the dropout rate at that question.
Another very useful feature that the most capable programshave is the ability to show demographics on the dropout report. That kind of data can help you determine whether certain kinds of people are more likely to drop out at particular questions than other kinds of people. You can use this knowledge to address any issues in the question and answer choice wording to which those people might be particularly sensitive or consider whether you really wish to present a question at all.