Millisecond Forums

Inquisit online - Missing datafiles

https://www.millisecond.com/forums/Topic10032.aspx

By Lizzy - 3/21/2013

Hi,


I'm collaborating with a research management company to collect data using an online study. I have two experiments registered online, and want to run 75 participants through each. I included a wrap-up link at the end of the online experiments that refers participants back to a webpage of the research management company so thy can check how many participants have finished the study.


The problem now is that their numbers and my numbers don't match. According to their data (which I can also download), 150 participants have already reached the wrap-up link. However, when I look at my datafiles, I only have datafiles for about 100 participants. 


We tried to figure out what this discrepancy could be due to, but we have no idea. For existing half-way the experiment doesn't bring people to the wrap-up links. I made sure that in the last section of my experiment, participants have to click a button on the screen to continue, so even if they do not enter any other responses, this button click is recorded.


I was hoping you could offer some suggestions as to what might cause the discrepancy in numbers.



All best


Lizzy

By Dave - 3/21/2013

The problem now is that their numbers and my numbers don't match. According to their data (which I can also download), 150 participants have already reached the wrap-up link. However, when I look at my datafiles, I only have datafiles for about 100 participants.


I would first ask said company how exactly they determine that number. For example, their page hits may be inflated by search engine bots / web crawlers.

By kent62 - 11/17/2014

Hi I am having this issue too, I am getting my participants to complete training online.  I am using a batch function, they have to fill out an initial form, and then complete a training task.  I am getting more initial tasks than training data, though they are saying they have completed the whole exercise. Did you ever find out what was happening?  

Kind Regards
By Dave - 11/17/2014

The discrepancy may be due to this
https://www.millisecond.com/forums/FindPost14889.aspx (symptom)
https://www.millisecond.com/forums/FindPost14905.aspx (reason)

The solution would be to have participants ensure that TLS 1.0 is enabled on their systems (cf. http://blogs.msdn.com/b/kaushal/archive/2011/10/02/support-for-ssl-tls-protocols-on-windows.aspx ).

The above, of course, only applies if your "initial task" (for which you report to receive more data) takes place outside of Inquisit (e.g. on some survey platform). It's not quite clear from your description. If *both* "inital" and "training" tasks are run via Inquisit, something else is amiss. In particular, this would suggest an issue with / mistake in the "training" script (the fact that data is available for the "inital" task inidcates that (a) scripts run fine and (b) data uploading works okay in principle).
By kent62 - 11/17/2014

Hi Dave,

the initial one is also in Inquisit, the two are linked using a batch file.  Surely if this of a script error I would see this in all of my participants all of the time, rather than just a few, randomly? And only in one script, instead of 3 separate ones (all of which I took from the task library and only modified the instructions minority)?  I am missing about 3-6 data files in each of my 3 conditions, but none of the initial tasks (a survey on which they confirm their mobiles are off).

All the Best
By kent62 - 11/17/2014

A student has also just messaged me that her computer froze on the final page, could that affect the upload?  
By Dave - 11/17/2014

If the program for some reason (seemingly) freezes and someone force-quits it as a consequence, yes, that could affect the upload. The fact that this only happens with 1 of your 3 scripts does suggest some sort of systematic issue to me (not necessarily a technical one). Which script exactly is that? For example, it might be that your script's "final page" expects some input from participants that they are not aware of (e.g. pressing a certain key or keys to move past) and they're under the impression that the program froze. Then they force-quit.
By kent62 - 11/18/2014

I have actually just found a couple in one of my other scripts.  It is happening in both my dual and single n back tasks (both adaptive from Jaeggi, 2010).  While some report freezing others have reported that they had no issues and successfully reached the inquisit web thank you page.  They have noticed nothing unusual.  I have sent out emails to all of them asking for computer and browser information to see if there are any trends.  So far I have identified 9 participants whose data is unaccounted for (so far 3 have confirmed that they noticed nothing strange).