A New Approach to Avoiding Problems of Scale in Interpreting Trends in Mental Measurement Data

Extracting policy-relevant information from large national surveys of educational achievement is ordinarily a nontrivial task. It is made more treacherous when the data are expressed on scales that are not uniquely determined. The paper begins with a critical analysis of a recent attempt to interpret the findings on reading achievement obtained by the National Assessment of Educational Progress (NAEP). It then describes a new approach to the quantification and interpretation of change and demonstrates its appropriateness for repeated cross-sectional designs such as NAEP. Limitations imposed by the survey design and the nature of the measurements are highlighted.