Computing RTs according to log file vs. csv file

Online experiments
1

Hi all,

I am new to using psychopy online and was looking at some data today and stumbled upon a discrepancy in RTs computed via the log files vs. the csv files. My reason for testing this was that I would like to eventually compute RTs relative to a stimulus presentation that happens in a different routine than the keyboard component and was hoping I could just use the times from the log file.

The RTs are off by a few milliseconds in all cases (by different amounts) but I’ve pasted one example below:

The csv file gives the RT for one trial as .187545. For the same trial, the log file has these values:

4218.942995000048 EXP Dist6_post_9: autoDraw = true
4219.1259950000085 DATA Keydown: z

The difference is .183.

Am I misunderstanding the times in the log file? I interpreted the first line as the onset time of the stimuli and the second as the time of the key press. I double checked that I had the sync timing with stimulus onset option checked in the builder. In the cases I have checked, the log file difference is always less than the reported RT in the csv file.

I think what I am seeing might be similar to this post that was unanswered: Inconsistent response times

Thank you!
Anthony

2

Your post is similar to mine here: Response Times in Log Files, Key Press vs. Release, and EXP log messages

If I resolve my question on that post, I think you will also find what you are looking for and I’ll follow up.

I actually did find a great match between log file and csv, but assuming that the EXP is logged on the next frame flip approximately, and that assumption is very much under debate. See my post.

I have some questions for you that will help the investigation:

(1) Where do you call Dist6_post_9: autoDraw = true?
(2) Are all of your RT’s computed from the log files just small underestimates of the .csv RT?

Also note that post that was unanswered was not online psychopy, so the coding will be different probably.