Annual meeting of the American Educational Research Association location:San Diego date:13-17 April 2009
This paper investigates the “explanatory power” of school process variables based on teacher data constructed through multilevel factor analysis, compared to school process variables developed using conventional factor analysis plus aggregation. To date, multilevel factor analysis has only scarcely been used in educational effectiveness research. The school process variables were associated with student achievement and non-cognitive outcomes. Our results indicated that some of the school process variables created through multilevel factor analysis were associated to a different extent to achievement than the variables obtained using conventional factor analysis plus aggregation. Therefore, these results underline the importance of using the correct constructional technique to develop school process variables.