Reply #54: Interesting article

[mgr]

I need to look closer at it again, since I cannot tell if the take off point is in the Model I and Model II ANOVAs, or these are elements of an alternative. Unfortunately, as a biologist, I think we have been taking into account treatment effects for some time--I am not quite sure if the same terminology crosses over exactly to psychology--it was these ANOVAs that I thought could potentially address exit polling at the precinct level when recombining.

My own temperment suggests I would favor combining the two approaches into one, and agree treatment effects are significant even if the population is too small to show significance. One should never presume randomness or drift just because the sample size is too small.

One of my favorite books is Dialectical Biology (Levin and Lewontin)that also addresses an alternative to the null hypothesis model. Give it a gander.

Mike