Millisecond Forums

Generally, a log entry is produced when there is some interaction with the server. That is:
- When a script is launched.
- When data is uploaded.
- When certain errors occur (e.g. a file cannot be retrieved).
- When the finish action occurs (e.g. redirection to the default finish page or a different web site).

Terminating a script via CTRL+Q at an early point -- i.e., when no data has been collected yet and there is thus nothing to upload -- could in theory account for missing log entries (option A).

Terminating the web player in other ways (killing the task) would also prevent log entries from being created; There would be no way to transmit any data or events to the server anymore once the program is suddenly terminated. However, you should not see any finish events etc. in the logs in this case either (option B).

Option C is theoretically possible, but should hopefully be relatively unlikely. Without further information -- e.g. direct feedback from an affected participant -- I don't see a way to make a firm conclusion one way or the other.

Option D is possible if, for example, the participant's internet connection broke down temporarily, i.e., no way to transmit the data and no way to create a log entry or transmit an error. In this case, Inquisit would have prompted the participant to store the data file locally and contact the researchers (that is Inquisit's fallback option if data transmission to the server fails for whatever reason).

I may be able to say more if given a few specific examples of participant logs where certain parts are apparently missing.

Howdy Dave,

If you can access my logs that would be easiest. Nevertheless here is an example where a section was not logged and there is no data file (top) and one that is complete (bottom). Missing from the top the the HHPercent_Likert_raw log and there is no data file. So they were able to complete the consent form, missed the HHPercent_Likert, but then completed the rest of the questionnaire. 



Thoughts?

Jeff

The upper log would suggest to me that HHPercent_Likert  was either not run at all for this participant (perhaps a mistake in case you use multiple <batch> elements / conditions?), that the script was terminated at an early point (when there was no data collected / recorded yet), or that the data could not be uploaded (e.g. temporary issue with the internet connection).

It's also possible that something else happened entirely which is not evident from the logs -- IIRC, there was a bug in 5.0.1.0 that could in theory prevent data from being recorded properly under certain circumstances when executing scripts via <batch>, it may be that this is what happened in this case (unfortunately there is no way to be certain based on the logs alone). Looking at the time elapsed between the data upload for the consent script and the data upload for HH_AMP may provide an indication, though. I don't know the typical completion times for each of the scripts involved, but perhaps you can tell whether approx. 9 minutes (18:09 to 18:18) would typically be enough time to complete both HHPercent_Likert and HH_AMP, or just one (HH_AMP).

In case the study is still active, I would definitely recommend updating it to use the most recent Inquisit Web version via its settings (in case you haven't done so already).

Hope this helps.

Hey Dave,

The answer is helpful.

I checked some of the options I described.
-Using Ctrl+Q during instructions results in no log/data file being created.
-Using Ctrl+Alt+Del to run task manager to kill the process during any point prior to finishing the section results in no log/data file being created, but also kills the rest of the survey.
(I cannot test creating errors, and I highly doubt an internet connection would have been the issue.)

Good idea to check the time between consent to AMP: 9 minutes may or may not be long enough to complete both sections. Depends on the speed of the user/competency with computers. I checked some subjects that had completed the sections and time to complete both ranged from 5 min to 20min. This individual seemed to have completed the remainder of the survey in a short amount of time. Still, I can't say whether or not this was a failed data recording. This was an issue for <2% of the responses so far. 

I have updated to 5.0.4.0.

I'll just keep fingers crossed the update resolves any issues!

Jeff

Update:

I see that this issue is continuing to happen with the updated 5.0.4.0 Web version. It still seems to be a very rare circumstance, but multiple sections are not uploading for a few participants. There is clearly enough time between data uploads that the in between time should have had additional sections uploaded.

The questionnaire is being run predominantly on workplace computers in health care settings where there could be complications with firewalls, antivirus or general administrator settings. 

I wouldn't know where to start or how to troubleshoot this. 

Ciao,

Jeff