How Plausible is Using Averaged NAEP Values to Examine Student Achievement?

Abstract

Software limitations and difficulty of use have led to some using averaged values instead of plausible values. This study compared the results of plausible values using AM Statistical Software with averaged values using SPSS for descriptive statistics multiple regressions using the 2009 4th and 8th grade math and reading National Assessment of Educational Progress (NAEP) for the total sample in addition to a subsample of Native American students. Gender and poverty and absenteeism and enjoyment of the subject material were used in separate equations to predict achievement. The subgroup means were almost identical, but differences were evident in the standard deviations. The regressions yielded similar results for both approaches. The results suggested that averaged values might present a viable, although not ideal, approach to plausible values.