Multilevel model development. The researchers used relationships found in the single-level SEM models (i.e., classroom, school, and district) to inform the development of a multilevel model. Analyses reported here represent a multilevel model in a single diagram that contains paths between predictor variables at each level of analyses and the outcome variable. In the multilevel model, each path contains three path coefficients, which one may interpret as regression coefficients. To create datasets for each level of analysis, researchers aggregated variables from lower levels, assigned the mean score to the higher level, and assigned variables from higher levels to lower levels. As an example, for school-level analysis, researchers aggregated classroom-level variables (e.g., teacher beliefs about technology) to the school level by calculating the mean scale score for teachers within a given school and then assigning that mean to the school level. Conversely, at the classroom level, researchers assigned values for school-level variables (e.g., principal’s use of technology) to each teacher, which remained constant across all teachers within a school. Table 5 presents a summary of variables used in analyses.