Importance of Equating High-Stakes Educational Measurements

Abstract

Many examination agencies, policy makers, media houses, and the public at large make high-stakes decisions based on test scores. Unfortunately, in some cases educational tests are not statistically equated to account for test differences over time, which leads to inappropriate interpretations of students' performance. In this study we illustrate the consequences of not equating scores on parallel forms of a test. The study used data from a high-stakes primary school exit exam in Malawi. A spiraling process was used to create two randomly equivalent groups of examinees who took two different forms of a test administered in different years to allow for a randomly-equivalent groups equating analysis. The study revealed that: (1) test difficulties were dissimilar across test forms signifying that equating was necessary; (2) changes in pass rates across years did not necessarily signify changes in performance of students; and (3) classification of students into grade categories across forms were different before equating, but similar after equating. The results illustrate that equating is instrumental in promoting fairness and in facilitating more accurate classification decisions and score reporting to stakeholders.