I'd like to ask you about item fit statistics. I understand that to arrive at infit and outfit values if done by hand, one would first need to calculate the difference between the observed score (which is a 0 or a 1 if it's a dichotomous item) and the modeled probability.
Now, my question concerns this, in order to arrive at the modeled probability, we'd need to know the true values of item difficulty and person ability. If I'm not mistaken, when calculating the fit statistics, Facets uses the parameter estimates. Is there a correction implemented?
It's not really a big deal, I'm just interested.