摘要：Conventionally, multilevel analysis of covariance (ML-ANCOVA) has been therecommended approach for analyzing treatment effects in quasi-experimental multilevel designswith treatment application at the cluster-level. In this paper, we introduce the generalizedML-ANCOVA with linear effect functions that identifies average and conditional treatment effectsin the presence of treatment-covariate interactions. We show how the generalized ML-ANCOVAmodel can be estimated with multigroup multilevel structural equation models that offerconsiderable advances compared to traditional ML-ANCOVA. The proposed model takes intoaccount measurement error in the covariates, sampling error in contextual covariates,treatment-covariate interactions, and stochastic predictors. We illustrate the implementation ofML-ANCOVA with an example from educational effectiveness research where we estimateaverage and conditional effects of early transition to secondary schooling on readingcomprehension.
关键词：multilevel analysis of covariance; Average effects; Multilevel structural equation modeling; conditional effects; quasi-experimental designs
