The original race IAT by Greenward implemented 12 (6 european americans and 6 african americans) faces and 16 words (8 positive and 8 negative), which makes the trial structure very tricky.
In OpenIAT version, 10 faces and 10 words are used.
The implementation of race IAT in other flatform (i.e. Minno.js) also matched the numbers of words and faces.
It’s very confusing how the original design counterbalanced the stimuli and categories.
Anybody implemented it to OpenIAT have some idea how they dealt with it?
After writing down ALL 200 trials in original race IAT,
It seems NOT counter-balanced. some faces shown more than 3 times in certain block, the other is shown only once. Still very confused in random sampling method. I guess the original one only counterbalanced the condition, and sample the stimuli randomly.
Wow, really? Not balancing stimuli seems like a very bad idea. in my own IATs I didn’t just balance the stimuli itself, but also the words per category, on properties like average word length.
Yes, if you are interested, you can write down all the words and face images by yourself in Harvard IAT page.
Even their newest version, The numbers of words and faces give no idea to counterbalance stimuli level (in trial level, of course, it’s counterbalanced.)
Njah, I’ll just trust you on your word