IAT randomization

[sanspeur]

Hello Empirisoft Support,

I have been using DirectRT and MediaLab for several experiments, and I
have had a very positive experience. Thank you for putting together
these excellent pieces of software.

I wanted to ask you a quick question about IAT randomization in
DirectRT. I noticed that in your sample IATs ( IAT-Practice.csv and
IAT-Race.csv), you elected to use randomization without replacement. I
was just wondering if this was done consciously, and whether this type
of randomization is preferred over true randomization. I was also
wondering if you think it might be worthwhile to create separate
StimList files for every block in which the StimList is called (e.g.
adject1-1.txt, adject1-2.txt, etc.) to reset the randomization for
each block. In short, I would like to know your opinion on which of
these three randomization methods is most appropriate for the IAT (1.
randomization without replacement, 2. randomization without
replacement with multiple identical StimLists, or 3. true
randomization).

Thank you for your time.

[jarvis24]

Sanspeur--thank you for the kind words about our software--of course, feel free to post a quick review here: http://www.empirisoft.com/support/showthread.php?t=46 :)

Regarding the IAT design, we tried to replicate the classic designs of yesteryear as closely as possible. Being out of social psych for a while now (and into software/hardware) I wouldn't want to advise you on this. I would seek opinions from a source such as the SPSP list serve. Let me know if you'd like details on how to do that. In general though, randomization with replacement (or in our lingo "true") is rarely as useful in a practical way as it is in a theoretical way. Randomization without replacement simply gives you the ability to ensure equal n's--but that can trump the niceties of "true" randomization. But again, I'd consult "true" IAT researchers or the social psychology world in general for design questions like this. I can say though that Tony Greenwald gave our sample IAT a test run many years back and gave it thumbs up. Hope that helps!