@golden, @Nebula, this is how I would set up the dichotomy tests: I would allow each pole of the dichotomy to be independent and then graph the results. Ideally, you would get a regression line with a high R value showing an inverse relationship, and the points would be clustered on either side in a bimodal distribution. However, either of the properties might not work out.


A theoretical reason you'd get bimodal distribution is if two processes where antagonistic, and it was mentally easier to use more of one or the other. For example, the antagonistic relationship between the introverted and extroverted version of a cognitive process is the reason why all ignore our ignoring function in favor of the base.