Taking composition and similarity effects into account: theoretical and methodological suggestions for analyses of nested school data in school improvement research

School improvement research is faced with a school teaching staff, which is not a simple homogeneous entity. The compositional attributes of the teaching staff – such as diversity – can have a crucial influence on school processes. Whether the teaching staff is highly fractured, consists of sharply dissociated subgroups, or has shared beliefs, affects the adoption of school improvement programs differently. However, school improvement research has not yet taken into account what different compositions of the teaching staff mean from a methodological viewpoint. It is true that the use of multilevel analysis is standard in school improvement research and that it considers nested school data. However, this method alone only takes averaged measures of teaching staffs into consideration but not their different compositions. In this contribution, we argue that school improvement research has to consider, theoretically and methodologically, how compositional attributes of the teaching staff can be conceptualized. We first discuss some advancements in the conceptualization of group composition from research on small groups and organizations. We then incorporate suggestions for different diversity typologies from small group research to describe the compositional attributes of the teaching staff. Additionally, we address how the composition of the teaching staff influences each teacher differently, depending on the specific position a teacher has within the teaching staff. We further suggest incorporating the Group Actor-Partner Interdependence Model (GAPIM; Kenny, DA, Garcia RL, Small Group Res 43:468–496, 2012) as a methodological approach for assessing these compositional influences. In addition to classic multilevel analysis, the GAPIM also considers the effects of the other teachers on staff and the similarity and dissimilarity of a teacher to the other members of the teaching staff. Finally, we illustrate the possibilities of the theoretical and methodological endorsem