A Meta-Analysis of Growth Trends from Vertically Scaled Assessments

A vertical scale, in principle, provides a common metric across tests with differing difficulties (e.g., spanning multiple grades) so that statements of absolute growth can be made. This paper compares 16 states' 2007-2008 effect size growth trends on vertically scaled reading and math assessments across grades 3 to 8. Two patterns common in past research on vertical scales, score deceleration (grade-to-grade growth that decreases over time) and scale shrinkage (variability in scale scores that decreases from lower to higher grades), are investigated. Pervasive, but modest, patterns of score deceleration are found for both math and reading. Limited evidence of scale shrinkage was found for reading, and virtually no evidence was found for math. In addition, linear regression was used to show that little of the considerable variability in the growth effect sizes across states could be explained by readily identifiable characteristics of the vertical scales. However, many scale characteristics were not well documented in available technical reports. The most important of these characteristics, along with their implications for interpretations of growth, are discussed. The results serve both as a normative baseline against which other scaling efforts can be compared