Equating in many-facet situations is more complex than equating two facets, e.g., multiple-choice tests.

 

The general rules are:

 

(a) All facets except one must be equated. That one is usually persons, except in studies of rater behavior.

 

(b) There need to be at least 5 common elements for each facet. This is a rule of thumb. In principle, one element will work. But it gives no quality-control.

 

(c) First do independent analyses of each data set.

 

(d) Then cross-plot the measures for common elements in each facet.

 

(e) Disregard as "common" those elements whose measures are clearly different across datasets.

 

(f) If the remaining common elements for a facet are close to a line parallel to the identity line, than anchor those elements, or combine the data sets.

 

(g) If the remaining common elements for a facets are still somewhat of a cloud, then use group anchoring of that facet to equate it.

 

(h) Continually monitor criterion-level-based success-rates and other indicators to check that the equating is making sense.