ABSTRACT
Explanatory patterns regarding situational differences in reading comprehension performance may be best captured by multidimensional reader profiles. Data from 56 third- and fifth-grade students were collected to investigate the applicability, scope, and convergent validity of a reader profiling scheme based on Alexander's (2005) reader profile framework and then compared with results from a hierarchical cluster analysis and a Bayesian cluster analysis. The reader profiling methodology used identified examples of all six of Alexander's reader profiles at each grade level, along with an additional hypothesized profile, the interest-reliant reader. The reader profiles related as expected to reading outcomes on a researcher-designed comprehension measure and a standardized comprehension assessment, with a few exceptions, and explained variance in those outcome measures better than the use of cluster analysis, except for the third-grade standardized scores. Finally, interesting differences emerged in the proportions of elementary students assigned to each profile across the grade levels.
Note
Acknowledgments
Thanks to Erin Monahan, Nadine Williams, and David Selber for their help in data collection and coding. Thanks also to Patricia Alexander for her comments on earlier drafts of this manuscript.
Notes
1. Coefficient H is a latent-variable approach to calculating reliability. It uses factor loadings rather than raw scores to determine the consistency of items within a measure.