Understanding response drop-off
Your overview page includes a chart with the response counts for each question.
Response counts can decrease from question to question within a survey for several reasons. This example shows some reasons for response drop-off, and how we classify survey responses and non-responses.
- Non-Response: A user doesn’t answer any questions in the survey, like User #1.
- Screen-out Response: A user doesn't choose a target answer for a screener question. In the example, the target answer for Question 1 is Yes, so User #2 gave a Screen-out Response.
- Partial Response: A user stops midway through a survey for a reason other than screening out. User #3 screened into Question 1, but then exited the survey. User #4 screened into Question 1, answered Question 2, but exited before answering Question 3. Both are Partial Responses.
- Complete Response: A user answers all questions in a survey, like User #5.
In the Question response drop-off graph (first graph in this article), there’s a large drop-off from Question 1 to Question 2, from 575 to 210 responses, because Question 1 is a screener question. The drop-off from Question 2 to Question 3 is small due to Partial responses. Question 3 to Question 4 has another large drop-off, from 210 to 100 responses, but not due to a screener.
If you see a large drop-off in your survey responses when there isn’t a screener, look for a reason why users prematurely exit the survey. Too many partial responses can introduce bias: for example, if you ask about equipment for an obscure sport, users who haven’t heard of that equipment may not answer the question at all. Then your results may be biased towards users who have heard of that equipment. In that case, you may want to introduce a screener or reword the questions.