The Problem of Parallelism in Multi Level Modelling

The way in which a test item operates depends in part on the characteristics of the test taker. Every analytic framework has some way of allowing those perameters to take on different values. The historical roots are fascinating, but it just scales up from there. In multi level modelling, if we have a level one predict or a level two predictor, there's a cross level interaction. And i think about it both to myself and when i try to teach this is, do you believe any given perameter, a factor loadate, a regression, coefficion, a co variance, a residual variance? Pick your favorite perameter.