Issues in Conducting Linkages between Distinct Tests

Abstract

Educational measurement practitioners are often asked to link scores on tests that are built to different content specifications. The goal in linking distinct tests is often similar to that for equating scores across different forms of the same test: to provide a set of comparable scores across the two measures. Traditional equating methods can be applied but results cannot be interpreted in the manner of equated scores. This article proposes a linkage process that consists of four stages to follow in linking distinct tests: choosing an appropriate linkage type and methodology, linking scores and computing summary measures, evaluating the quality of the linkage and determining what to report, and making recommendations for the interpretation and use of the linkage results. The complete linkage process is illustrated by discussing practices and decisions made at each stage for a linkage conducted between ACT Composite and SAT I V+M scores using equipercentile methods. Prediction is explored as an alternate form of linking in situations when use of the equipercentile results is not appropriate.