Millisecond Forums

I have been using the IAT for my research. I have been attempting to test the IAT to make sure that it is working correctly. In doing so I have attempted to take it with different biases by taking it more slowly for one group and then the other. When I score the results of my test trials, I am getting inconsistent D score directionality. In some instances, I get a negative number and in other cases, I get a positive number when I should be getting a D score in the same direction (both positive or negative).

Has anyone run into this problem before? All of the code matches the code from the millisecond website. 

I was wondering if anyone knows a better way to test the IAT or has any ideas why I am not getting consistent results.

Thanks.