Millisecond Forums

Which Inquisit 5 version did you run this under?

I used Inquisit 5.0.6 when creating procedure and
Inquisit 5.0.7 to run it in a lab.

Thanks. You said "I configure batch to run procedure in this order  DD_N; DD_S  DD_W" and "I have never found this error in data from DD_N procedure (run as first) but only in DD_D  run as second) and DD_W ( run as third)."

However, (1) the <batch> script you included runs the scripts in the order DD_S; DD_N: DD_W, not  DD_N; DD_S  DD_W:

<batch>
/ file="DD_S.iqx"
/ file="DD_N.iqx"
/ file="DD_W.iqx"
</batch>

Is the <batch> a different one than you actually used? Or did you never find the error in the 2nd script run (DD_N), not the 1st? Finally, you mention a script called DD_D in the above, which I can't find. I assume this is a typo and you actually meant DD_S?

My apologise for this mess,  to claryfy:

1) This is correct a scripts order in a batch fille:
<batch>
/ file="DD_S.iqx"
/ file="DD_N.iqx"
/ file="DD_W.iqx"
</batch>

2) I have never found error in DD_N and you are right it is second script. DD_D is a typoo

3) Just in case,  please find attached my raw data from this scripts.   Errors are in a column response. 
In column response I should find only valid responses so A, B ora sequence of numbers from WM task, but they are other responses like erasescreen.

Thank you for intrest in this issue.

Best

Thanks for the helpful clarifications and data. Looking at it and trying to reproduce. It would certainly seem that some sort of weird bug is involved, although I'm not yet sure what triggers it. I don't necessarily think it's the <batch> element / order of execution of the scripts -- if the 1st script were unaffected (i.e. nonsensical erasescreen responses only showing up in the 2nd and 3rd script), that would have seemed more plausible. However, while it doesn't rule some extra-weird interaction out completely, only the 2nd script being unaffected would seem to go against the <batch> theory. I'm wondering whether the different working memory load-levels play a role instead. DD_N is the script with zero load and thus doesn't run certain <trial> & <surveypage> elements that the other two scripts execute in between the delay discounting trials --  that's the most significant difference between DD_N (unaffected) and DD_S & DD_W (both affected) I can make out.

Looking through the data you kindly shared, another thing that sticks out is that in a lot of the affected trials (i.e. those with an "erasecreen" response logged), the latencies are too small to be "real" (sometimes as low as 1 or 3 ms, many more significantly below 100 ms). Perhaps the <surveypage> with <textbox> to delay discounting <trial> transition somehow (randomly) messes with the delay discounting trial's input handling. Why that would affect certain trials (e.g. <trial p5>) more often than others, though, I have no idea (yet). It's quite strange.

Can you tell me a bit more about the devices / computers involved in the data collection? I.e. which operating system (type / version), were any of these using touchscreens for input? I've counted 11 subjects (subject numbers 5, 7, 10, 11, 14, 17, 18, 20, 21, 23, 25) with "erasescreen" responses in your data. Some with very few -- e.g. subjects 11 & 17 -- and some with a relatively large number -- e.g. subjects 20, 21 or 25. Are there any other commonalities you can think of, e.g. were those latter subjects' data perhaps all collected on the same device in the lab?

Thanks!

Thank you for all this information,

"the <batch> theory" comes to my mind partially because it seems to me that when I am testing this scripts separately (starting manually one after another - not linked by batch) using monkey, then I never found this errors, but this might be only a coincidence because I do only a few tests.  You are right that my reasoning with scripts order  in batch have a little or no sence.

When it comes to operation systems I can't say a lot, because, this pilot study was done by my colleagues in a lab in other city and I have no access to this lab in this moment.  What I know is that this  script was run on difrent computers, with Windows 7. This scripts were used only mouse and keyboard as input device.  

Data from 20, 21 and 25 participants were collected from different devices, in raw data I have information that this participants started script about the same time so, they probably did it on different computers.


Best