Clients often ask a simple question about a screener: which answers screen a respondent out? The Screenout Reasons block on the Insights page answers it from your test runs. It lists the routing questions that ended test runs with the outcome Screenout, each with the answer responsible.
Why this is hard without SurveyTester
A screenout rarely happens on the question that caused it. Q3 can decide the outcome while the respondent still sees Q4 to Q6 before landing on the end page. Looking at the last answers of a screened-out interview therefore points at the wrong question, and going through the survey script for every path takes time. Counting by hand also breaks down once a project has hundreds of test runs in several languages.
How it works
SurveyTester already knows which questions change the survey flow: the routing questions, also called filter questions. The block combines them with the outcome of each test run:
- It reads every finished test run with the outcome Screenout, from bots and from manual testers. Repeated test runs count as well.
- It finds the page the test run ended on.
- Going backward from the end, it picks the nearest routing question that decides on that end page, and takes the answer the test run really gave. A routing question on the very last answered page counts as well when the survey ended right there.
Questions that are no routing questions are never named. If Q3 caused a screenout at Q7, the answer to Q6 stays out, even though it came later.
Reading the block
Open Project Details › Insights and find Screenout Reasons. If you do not see it, switch it on in the block selector at the top of the page.
Each row is one distinct cause:
| Column | Meaning |
|---|---|
| Routing question | The question responsible, with its page below it |
| Answer that leads to Screenout | The answer given, with its label and value |
| Ended on page | The page or pages the test runs ended on |
The block is a clear statement per answer, not a list of test runs, so there are no counts per row. A repeated test run confirms a row without inflating it. The line above the table says how many test runs were read and for how many no routing question is known.
Use the language and market selectors to look at a single audience, for example when a screener differs between countries.
When no routing question is known
Some screenouts cannot be explained yet, for example when a typed age screens a respondent out, or when routing questions were never detected on the project. These test runs appear in the separate section No routing question known, with the answers on their last answered page.
Treat these as candidates to check, not as causes. To turn them into real rows, run Detect Routing Questions on the Pages view, or mark the question as a routing question yourself. A numeric question you mark yourself is then shown with the number the test run gave.
Good to know
- Routing questions are detected from up to the latest 500 test runs of the whole project. The language and market selectors narrow the test runs read by the block, not the routing questions.
- The outcome Screenout is used as it is stored. It can also come from the expected outcome of a test case, so configure your end pages for reliable results.
- A screenout that depends on two questions together is shown with the nearest routing question only.
- If the survey changed during the project, older test runs may still follow the old routing. Open text, decimal and date answers are not shown.