Respondent status types included in experiment export file

Live Surveys
1

Hi,

Hopefully a simple question that others might also find useful.

Under the Results > Data tab, I understand I can export 3 types of results files for: covariates, experiments and timestamps. It was my understanding that only individuals that have completed the survey (those with a “status = 7” in the covariates spreadsheet) would appear in the experiments spreadsheets.

However, I have found this not to be the case. For example, I can see that some respondents with status = 10 (system initiated timeout) are also included in the experiments spreadsheets which I assume is a way of ensuring that any information is not lost?

Please could you confirm which respondent status’ additional to status = 7 are included in the experiments spreadsheets that I can download from the Results tab?

Thanks in advance.

2

If you open any of your surveys and open the Dictionary (in the menu on the left in the Edit section) you find all possible status values under the STATUS data label (close to the top).

The downloads include all respondent sessions that are closed, i.e. where the final status has been determined. That excludes a few status values, 15 (Active) being the most important. Generally, you should be prepared to handle all status values in the downloaded files e.g. by filtering for those you are interested in.

3

Hi,

Thank you - this answers my question and explains which respondents are present in the downloads.

Best,

Rob

4

What is the meaning of “system initiated timeout” and “user initiated timeout”? And how to deal with those responses?

5

The difference between user vs system lies in how the server detects the timeout.

  1. The typical use case for a user-initiated timeout is that the respondent starts a session, gets a phone call, comes back after 2h and tries to continue the survey. The server tries to serve the request and notices that the session has timed out.

  2. The typical use case for a system-initiated timeout is that the respondent starts a session, loses interest and closes the browser. The server would never notice that because there is no further request. However, it collects timed-out sessions on its own at some point and then closes the session as system-initiated timeout.

How to deal with those responses is up to you. You could try to use the partial results, but I suppose many researchers discard them.