A Comparison of Observed Score Equating Anchor Set Construction

• In Event: Westin Roundtable Session 2

Abstract

This simulation study examined two anchor set types, mini and midi, used in observed score equating. Linear and non-linear equating method error results were compared while the percentage of anchor items was varied. Equating standard error, bias, and root mean squared error were used as evaluation indices. The study findings indicated that mini and midi anchor sets performed similarly overall, particularly in the middle of the score distribution, and were the most similar when the percentage of anchor items was small. Likewise, as the percentage of anchor items increased so did the equating error and bias for both anchor types. Although this finding contradicts the common belief that more anchor items improve equating, it supports recent equating research.

Authors

• Joshua David MacInnes, University of North Carolina - Greensboro

• Devdass Sunnassee, University of North Carolina - Greensboro