Scoring Repeated Standardized Tests to Estimate Capacity, Not Just Current Ability

Abstract

Changes to educational policies have proliferated testing data to include multiple-administration assessments that repeatedly measure student performance over time. Psychometric models—extended for this type of data—estimate quantities typically associated with assessments that are given once, such as ability at a specific time point. This article considers how multiple-administration assessment offers the opportunity for models to estimate novel quantities that are not available from traditional single-administration assessments but may be of interest to educational researchers and stakeholders. Specifically, dynamic measurement models can directly estimate capacity—the expected future score once the construct of interest has fully developed. Preliminary evidence for this approach shows it may be less susceptible to effects of socioeconomic status and may improve predictions of future performance. An example with real-life operational assessment data is provided. Extensions and limitations for educational assessment are also discussed.