6. Results
6.1. Analytical strategy
For this study, we used hierarchical linear modeling (HLM;
Raudenbush & Bryk, 2002) to account for the nested nature of
the data. The two-level HLMs nested children within classrooms,
and predicted their residualized gain (i.e., Time 2 scores with
Time 1 scores as covariates) from the classroom-level measures
of teacher behavior management. Models were built separately
for each child outcome, namely print awareness and vocabulary.
First, an unconditional model without any predictor variables was
tested to compute the intraclass correlation coefficients (ICCs). ICC
refers to the amount of variance in individual child outcomes that
is explained by classroom-level variables. Then, we entered child
variables at level 1 and teacher variables at level 2.