This doesn’t happen at random, it’s just consistent with your script: you only start and log the “timeout” Timer element if no key was pressed during the exposure period, otherwise you just never even deal with that Timer element, so it won’t appear in your results files. Here’s a step-by-step reading of your script (the “any time”s correspond to whether the participant pressed a key that would stop the timers prematurely):
0ms print "fixationpoint" ; print "primero" start playing "primero" ; start 700ms timer and wait [ ... ] 700ms start playing "segundo" ; start 1150ms timer and wait [ ... ] 1850ms remove "fixationpoint" ; print "num" listen to keypresses on left/right arrow (will be logged) associate keypresses with [ stop "exposure" ; stop "timeout" ] start 300ms "exposure" timer and wait [ ... ] any time remove "num" btw 1850 **if NO key was pressed (2150ms)**: & 2150ms [ start 5000ms "timeout" timer (will be logged) and wait ] any time stop listening to keypresses btw 1850 **if correct key was pressed*: & 7150ms [ play "coin7" and wait ] any time end of current trial btw 1850 move to next trial & 7150 + audio ms
Consistently, you’ll find that for item 60, 3956ms elapsed between the reported keypress and the end of your Header, which necessarily means that no key was pressed during the exposure timer (which takes place between 1850 and 2150ms). For item 68, however, the keypress happened 2044ms after the end of your Header, which means approximately 100ms before the end of the “exposure” timer (or, 200ms after it started). So the failure command on the Key was just not validated (ie. a key had been pressed indeed at that point) and the “timeout” timer was just never started nor told to be logged. So there’s not really any data to recover here, because the “data” doesn’t correspond to anything that actually happened.
Don’t you get the desired reaction time by subtracting the Header-end time event from the keypress time event though, the way I just did?
You could also add log onto your “exposure” Timer element, which will add a line reporting when it started (and ended), so you can subtract the start time event from the keypress. It would probably more accurately reflect the reaction time you’re after, as I understand your design