No data written if experiment crashes -> workarounds? (case 27293)

[khudson]

We are running MediaLab v2014 on Windows 7 computers, and occasionally our experiment file crashes (for a variety of reasons that aren't the focus of this particular post). Unless the experiment successfully runs through to completion, no responses are saved in the data file, meaning the participant has to repeat the entire experiment.

This was acknowledged as a issue for DirectRT back in 2008 (http://www.empirisoft.com/support/showthread.php/1421-Saving-data-in-case-of-crash), but that rationale doesn't explain why MediaLab wouldn't be able to write in real time, or at least save periodically throughout the experiment (e.g., at the end of each questionnaire).

Is there a way to write data throughout the course of the experiment that we're missing? Any insight would be much appreciated, as alternatives like breaking out our questionnaires into smaller experiments to force data writing more frequently don't fit well with the study design.

Thanks!

[jason_reed]

We're sorry to hear that you have experienced issues involving loss of data during your use of MediaLab. MediaLab does require that any .que files are successfully run to completion to record the responses to items within those .que files. If your current .que files are not currently successfully running to completion, I will be happy to help where I can to get those files working.


Please send me the .que files and other files necessary to run them. You can either upload them here or send them to me directly at support@empirisoft.com (please reference case 27293). Please also tell me what happens on your computer when the .que files don't work (i.e., what happens on the computer screen, which specific item names in the .que file do not work).

[khudson]

Jason -

Thanks for your response. Our individual questionnaires do run to completion; we occasionally have issues when the questionnaires are compiled into a single experiment, which then at times crashes, and no data are saved for the questionnaires that were completed. Should data save after each completed questionnaire run within an experiment, or is MediaLab programmed to save all questionnaire data within an experiment only after the entire experiment completes successfully? And if it's the latter, what is the workaround to save by questionnaire instead of by experiment without having to individually run each questionnaire outside of an experiment file?

Karen

We're sorry to hear that you have experienced issues involving loss of data during your use of MediaLab. MediaLab does require that any .que files are successfully run to completion to record the responses to items within those .que files. If your current .que files are not currently successfully running to completion, I will be happy to help where I can to get those files working.


Please send me the .que files and other files necessary to run them. You can either upload them here or send them to me directly at support@empirisoft.com (please reference case 27293). Please also tell me what happens on your computer when the .que files don't work (i.e., what happens on the computer screen, which specific item names in the .que file do not work).

[jason_reed]

Please send me the .que files and other files necessary to run them. You can either upload them here or send them to me directly at support@empirisoft.com (please reference case 27293). I'd like to run them on my copy of MediaLab to see how they work.

[khudson]

Jason -

I will need to check with the PI to see if these are something we can send along. In the meantime, can you answer this question:

At which point during the course of running an experiment file does MediaLab write the individual questionnaire data to the data file? Does it write in real time, at the end of each completed questionnaire, at the end of the entire experiment, or some other time/way?

Please send me the .que files and other files necessary to run them. You can either upload them here or send them to me directly at support@empirisoft.com (please reference case 27293). I'd like to run them on my copy of MediaLab to see how they work.

[jason_reed]

When running a .exp file, all parts of that .exp file must be successfully completed before any data is recorded.

[khudson]

Thanks Jason, that explains the problem. Is this something that will be addressed in future versions of MediaLab? It seems quite risky to have a lengthy experiment file if nothing writes until the entire experiment concludes. It would be great if MediaLab could write the data after each questionnaire, if not in real time.

When running a .exp file, all parts of that .exp file must be successfully completed before any data is recorded.

[jason_reed]

I will be sure to pass this suggestion to our programmers.

[ROLLAND]

Thank you for sharing issue publicly. It helped alot.

[FIDEL]

Decent, need to accomplish something like this with mine.