@KVB I am only successful in this contest because of one simple principle my friend -
In Rasch terms, "True" valiance is the "adjusted" variance (observed variance adjusted for measurement error). Error Variance is a mean-square error (derived from the model) inflated by misfit to the model encountered in the data.
Without this knowledge I would be joining Diggity aka the new Dad in last place. God Bless.