Measuring Changes in Educational Attainment over Time: Problems and Possibilities

To attempt to measure trends over time seems a perfectly natural thing to do. It is a common activity, for example, when studying mortality rates, heights of children, or economic indicators such as the retail price index. In education the popular notion of "standards" is often linked closely to questions about changes in attainment or achievement over time. The Assessment of Performance Unit (APU) in Britain and the National Assessment of Educational Performance (NAEP) in the United States have both given a high priority to making inferences about trends in attainment over time (Gipps & Goldstein, 1983; Wirtz & Lapointe, 1982). Despite much interest in the issue, there has been curiously little attempt to define precisely what is meant by "trends over time," or to discuss the attendant measurement problems. First, in order to set the scene, the ways in which the issue is perceived in areas other than education are briefly reviewed. Next a discussion of the logical basis of the most commonly used methods underpinning analyses of trends over time is provided. Finally, alternative formulations of the issue are presented, together with suggestions for their implementation.