Subscores and Subgroups

• In Event: Reliability and Subscores

Abstract

There is an increasing level of interest in subscores for their potential diagnostic
value. Haberman (2008) suggested a method to determine if subscores have added value over the
total score. Performance of subgroups, for example, those based on gender or ethnicity, on subtests is often of interest. We
examined whether the knowledge of the subgroup membership of the examinees leads to a better
estimation of the true subscore in applications of the method of Haberman (2008). Analyses of
data from several operational testing programs show that the knowledge of the subgroup membership
of the examinees does not lead to a better estimation of the true subscore and that the value
added of the subscores does not vary over the subgroups.

Authors

• Sandip Sinharay, CTB/McGraw-Hill

• Shelby Haberman, Educational Testing Service