Similarities Between Cross-Sectional and Longitudinal Growth in Large, Rasch-Scaled Reading Achievements

• In Event: Rasch Models: Considerations and Use

Abstract

Cross-sectional studies are generally considered inferior to longitudinal studies. This paper shows that the growth rates from overlapping cross-sectional and longitudinal cohorts on high-profile national reading tests are virtually identical. Both sets of tests consist largely of multiple-choice items with a potential for guessing. The Rasch model, used to equate the tests in both designs, did not account for guessing. A correction for guessing in the Rasch model in cross-sectional designs has been shown to result in an increased rate of growth. Because of the equivalence of longitudinal and cross-sectional growth, the paper applies the correction obtained from the cross-sectional design to the longitudinal design, resulting in an increased, more accurate rate of growth in reading in the longitudinal design.

Authors

• David Andrich, University of Western Australia

• Ida Marais, The University of Western Australia