[This is clearly the wrong topic for this thread but I don't see how to create a new topic.]
In studies that use the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998), typically only the overall "IAT effect" is reported. This is essentially the mean difference in response latencies between compatible (e.g., flower + pleasant) and noncompatible (e.g., insect + pleasant) combined tasks.
In a study I conducted for a research methods class last semester I used the IAT to measure acculturation in a sample of Chinese immigrants. The IAT has been shown to have advantages over self-report acculturation measures, but there's a problem. Acculturation is thought to be bi-dimensional, and in the way IAT data are normally analyzed it would be impossible to tell whether participants in my study have positive associations with BOTH China and the U.S.
But the data is there, and all that may be needed is a novel analysis of the latencies. Here's my idea, using this study as an example (use your imagination to extrapolate the principle to your own work). If you treat China + pleasant and U.S. + pleasant as subscales, then you can calculate their mean latencies independently. The response times are not inherently connected-- it is possible to separate them. For more statistical power you could then subtract China + unpleasant responses from China + pleasant (providing a total of 40 trials in a typical IAT), and do the same for the U.S. + pleasant and U.S. + unpleasant (also 40 trials). Then these two results could be compared to baseline latencies, i.e. the remaining trials in which participants are assigning stimuli to one category and are under low cognitive load.
When I tried this analysis with my sample I got a nice range of responses: some participants appeared to have positive associations with both China and the U.S., some were more strongly associated with one or the other (which the overall IAT effect showed), and some had response times that indicated negative associations with both. These patterns neatly map onto Berry's acculturation quadrants: integration (+culture of origin +host culture), assimilation (-culture of origin +host culture), separation (+culture of origin -host culture), and marginalization (-culture of origin -host culture).
Unfortunately I can't say whether my analysis has any external validity, and I'm curious to know if anyone else has experimented with subdividing IAT data like this. In my case the sample size was very small, and to keep participants from dropping out I included only two self-report measures-- neither of which would validate the bi-dimensional acculturation I claim to have measured. Because this part of my analysis didn't really go anywhere it didn't make it into my paper, but I am happy to post it if anyone has actually read this far and is interested.
My aim in posting here is to find out whether other researchers have attempted to analyze IAT data in the way I've described, and what the results were. If anyone has any information about this please reply to this post or PM me. Thanks!
- Badges
- Science topic
- Similar topics
- Psychology
- Social Psychology