AN EQUATING METHOD CONSIDERING THE VARIOUS CONDITIONS AMONG ANCHOR TESTS

Abstract

Weighted Error Method (WEM) for equating several tests with different levels of difficulty was proposed. As a way of research, different weights to error components were given in proportion to the number of common items between two tests and the number of subjects administered. Using a reparameterizing technique, test difficulties would be estimated by usual weighted least squares method. Difficulty levels of 20 actual tests were equated by WEM. Engelhard and Osberg's method and data (1981, 1983) were used to evaluate WEM numerically, and as Engelhard and Osberg's results reasonable estimations were found.