ABSTRACT
The goal in any science is the production of cumulative knowledge. Ultimately, this means the development and testing of theories that explain the phenomena
that are the focus of the scientific area in question. One example is theories that identify the school factors associated with student achievement. Unless we can precisely calibrate relationships among variables (for example, leadership style or school climate and student achievement), we do not have the raw materials out of which to construct theories. There is nothing consistent for a theory to explain. For example, if the relationship between instructional leadership and student achievement varies capriciously across different studies from a strong positive to a strong negative correlation and everything between, we cannot begin to construct a theory of how leadership might affect achievement. This implies that there is a need for researchers in any field to conduct reviews of research and in particular meta-analyses (which are a more rigorous means of review) to examine and integrate the findings across studies and reveal the simpler patterns of relationships that underlie research literatures, thus providing a more robust basis for theory development. Moreover, meta-analysis can correct for the distorting effects of sampling error, measurement error and other artefacts that can produce the illusion of conflicting findings and obscure ‘real’ underlying patterns (Hunter and Schmidt 2004) by means of the calculation of average effect sizes for relationships.
