Exploring meta-analysis

Abstract

Meta-analysis is a research method widely used in medicine and social sciences, where the population of interest is studies that have already been made in the area to be researched. The main aim of meta-analysis is to obtain a combined effect size which helps in hypothesis testing. This is done through different methods to obtain a measure of effect for each study included in the meta-analysis, and a model to combine them. The choice of model depends on the included studies as well as the extent to which conclusions are intended to be generalised. Another concept in meta-analysis is the estimation of the between-study variance, where a number of estimation methods have been developed and improved upon along the years. A whole process must be done in order to conduct a good meta-analysis, starting with a thorough procedure to choose the studies that will be included in the meta-analysis. This process is described in detail in this dissertation. Details on how to report a meta-analysis are given, together with the important statistical aspects and tests that are required. Apart from this, detail is given on the estimation methods used to estimate the parameters of a mixed model. Six estimation methods for the between-study variance are also explained, based on a paper by Veroniki et al. (2016). A published meta-analysis was used as a running example throughout this dissertation, to show various steps of a meta-analysis, as well as to apply and compare the different estimation methods for the between-study variance. Interestingly, the estimation methods that differed from that used in the published meta-analysis led to a different conclusion.