Abstract
Many educational studies are carried out in intact settings, such as classrooms or groups in which individual data were collected before and after a treatment. Researchers advocate either the use of individual scores as the unit of analysis or class means. Both approaches suffer from conceptual and methodological limitations. In this article, the use of hierarchical ANCOVA for analyzing quasiexperimental data including baseline measures is designed and promoted. It is illustrated with a realworld data set collected from a curriculum study. Results showed that the hierarchical ANCOVA is a conceptually and methodologically sound approach, and is better than ANCOVA based on individual scores or ANCOVA based on class means. The potential of using hierarchical ANCOVA designs for curriculum studies is discussed in terms of statistical power and congruence with study plans.
DOI
10.22237/jmasm/1146457080
Included in
Applied Statistics Commons, Social and Behavioral Sciences Commons, Statistical Theory Commons
