Imagine I run a psychophysical experiment. I have generated a sequence of trial conditions which are presented to each participant in same order. But not all of participants are equally good at passing the requirements during the trial. So, some trials are skipped during the experiment and that skipped condition is inserted into the procedure later.
But what if this skipped trial is the last condition in the sequence? Participant will notice that the same condition is being presented to him repeatedly (provided he fails it more than once) - which is undesirable. How do I deal with this?
Maybe it is better to simply present conditions in random order to all participants and ignore the failures altogether, because when analysed, all participants' data will be pooled together, anyway, for statistical analysis.
$\begingroup$
$\endgroup$
4
-
$\begingroup$ Without a more detailed background of the effects you are trying to measure, I believe it will be hard to provide generalized advise. You describe an experimental method, but don't describe any of the dependent or independent variables. Also, what made you choose this particular methodology? Any similar studies you are basing yourself on? Could you elaborate? $\endgroup$– Steven Jeuris ♦Oct 4, 2021 at 7:24
-
2$\begingroup$ @StevenJeuris, IV is an image presented to a subject; DV is his accuracy of identifying it under a specific condition; some subjects fail to meet this specific condition during some trials - it is voluntary; when he fails to meet this condition, the trial is aborted and nothing can be done to recover except repeating the trial all over; the methodology is generally called 'tachistoscopy' and is well-elaborated elsewhere; i do not know of any similar studies because the fail rate in my experiment is very high - only 13% of trials are valid due to technical reasons $\endgroup$– ivan866Oct 4, 2021 at 14:28
-
$\begingroup$ Thank you. Could you please edit that information into your question? $\endgroup$– Steven Jeuris ♦Oct 4, 2021 at 16:24
-
1$\begingroup$ In my experience, failed trials are skipped. This is often done to reduce variability between sessions - eg, a poorly performing participant would end up spending longer in the session, getting more practice, more fatigue, etc, causing a confound. If you are facing a particularly high fail rate, then consider making the experiment easier, limiting the number of repetitions, or including dummy trials between repetitions. $\endgroup$– Arnon Weinberg ♦Oct 4, 2021 at 16:32
Add a comment
|