ABSTRACT
Critics of significance testing claim that this statistical framework promotes discrepancies by using arbitrary thresholds (α) to impose reject/accept dichotomies on continuous data, which is not reflective of the biological reality of quantitative phenotypes. Here we explore this idea and evaluate an alternative approach, demonstrating the potential for meta-analysis and related estimation methods to resolve discordance generated by the use of traditional significance tests. We selected a set of behavioral studies proposing differing models of the physiological basis of Drosophila olfactory memory and used systematic review and meta-analysis approaches to define the true role of lobular specialization within the brain. The mainstream view is that each of the three lobes of the Drosophila mushroom body play specialized roles in short-term aversive olfactory memory [1-5], but a number of studies have made divergent conclusions based on their discordant experimental findings [6-8]. Multivariate meta-regression models revealed that short-term memory lobular specialization is not in fact supported by the data, and identified the cellular extent of a transgenic driver as the major predictor of its effect on short-term memory. Our findings demonstrate that meta-analysis, meta-regression, hierarchical models and estimation methods in general can be successfully harnessed to identify knowledge gaps, synthesize divergent results, accommodate heterogeneous experimental design and quantify genetic mechanisms.