Tests of alignment among assessment, standards, and instruction using generalized linear model regression

Abstract

An essential component in school accountability efforts is for assessments to be well-aligned with the standards or curriculum they are intended to measure. However, relatively little prior research has explored methods to determine statistical significance of alignment or misalignment. This study explores analyses of alignment as a special case of the generalized linear model (GLM). A general approach for such analyses is suggested, and examples are given for analyses with traditional alignment and GLM regression using data from two previously published studies. Results from the GLM are compared with ordinary least squares (OLS) regression. Findings show that the GLM method allows more informative analysis of differences between source documents than alignment indices alone, including determination of whether marginal discrepancies are statistically significant or not.