Paper Summary
Share...

Direct link:

Testing the Measurement Invariance of Bifactor Models: An Application on Chinese Middle Schools

Sun, April 19, 2:15 to 3:45pm, Hyatt, Floor: East Tower - Purple Level, Riverside East

Abstract

Measurement invariances are pre-requisite of meaningful comparisons of the target constructs. Achievement tests usually comprise of multiple test format including items that could be administrated independent or several items based on the same task and need to be administrated as a testlet. Testlet formats potentially violate the local independence assumption and affect the parameter estimates and scoring. Bifactor models can exam the testlet effects and primary construct effect simultaneously. The present paper studied the measurement invariance using bifactor models to scrutinize to which extend the measurement is invariant across several middle schools using the same district-wide test in China.

Author