Meta-analysis: a tool for constructing theories or evaluating interventions or simply proving everyday assumptions?

Abstract

Read online

After a brief consideration of the development of meta-analyses as a joint discussion of results from a research area across development stages 0, 1, 2, it is concluded that the present form 2.0 is unsuitable to serve as a basis for theory building. Further development of this tool into a meta-analysis 3.0 is necessary for this purpose which requires the validity of the independent variables in the primary studies, the reduction of the error variance of the dependent variables, a stability of the effects over the primary studies and a quantitative comparison between observed and predicted effects in the primary studies. In the current meta-analyses 2.0, a concrete single-case approach creates the impression that mainly everyday ideas are investigated, which one would like to generalize to a population of other conditions. Furthermore, the results of the existing meta-analyses are either homogeneous and very small or heterogeneous. Meta-analysis 2.0 searches for the instability of the measurements under a specific topic with methods of induction. The procedure of a meta-analysis 3.0 is described in general and carried out hypothetically and with an empirical example. It searches for the stability of quantitative reconstructions of data over different topics with the method of abduction. The conclusion can be summarized as that meta-analysis 3.0 is indispensable as a tool for theorizing, and theorizing presupposes meta-analysis 3.0. The link between this interdependence is abduction in contrast to induction as a research strategy.

Keywords

• meta-analysis
• theory testing
• reliability
• validity
• error theory
• abduction