just another statistic
Okay... I am now convinced that I am reaching the correct conclusions for the wrong reasons. Just a little more tinkering to convince the software that I really do want it to run a multiple regression testing for interaction between my two factors, and eventually I'll get there.