Williams VSL. Exploring statistical adjustment of results from the National Assessment of Educational Progress (Technical Report #102). 1999. Research Triangle Park, NC. National Institute of Statistical Sciences.

Motivated by a concern for improper comparisons of states based on published test score aggregates, a series of statistical adjustments was applied to data from the 1990 and 1992 Trial State Assessments (TSAs) of the National Assessment of Educational Progress (NAEP) Using the method of direct standardization, the analyses adjusted for demographic factors, home background variables, and school-related variables. The differences between unadjusted and adjusted NAEP averages were small, the mean overall change for states being a one-point increase on the NAEP mathematics score scale. The largest single difference was an increase of ten NAEP scale points for California for adjustment based on three student-reported variables related to student’s home background (number of parents living in the household, responses to the NAEP background question “How often is another language spoken in your home?,” and a home environment composite variable describing the number of reading materials available in the home). The states of the Southeastern US showed the greatest average increases across all adjustments, followed by the Western states, the states in the Northeast, and finally the Central US states. The adjustment analyses resulted in a generally homogenizing effect, that is, differences among states were reduced. Although the results of this exploration are preliminary and the continued investigation of plausible adjustment variables is advised, it is recommended that reports containing unadjusted state NAEP means be augmented with adjustment results based on select variables that are generally considered beyond the control of the school and school systems.

Share on: