Research synthesis and meta-analysis

Abstract

As the volume of primary research across all fields of social science continues to grow at rapid rates, research synthesis has become more important today than at any other time in history. With the development of metaanalysis, a set of procedures for summarizing the quantitative results from multiple studies, the rigor, systematicity, and transparency of research syntheses was greatly improved. However, a number of developments, including the creation of the Cochrane Collaboration and Campbell Collaboration, have heightened the profile of meta-analysis in recent years. Furthermore, recent advancements in analytic strategies, including the use of a random effects model of error, the development of meta-regression, and improved methods for dealing with missing data and data censoring, have enhanced the popularity, efficiency, and trustworthiness of meta-analyses. We begin this chapter with a brief history of meta-analysis and research synthesis. We then describe the different stages of a rigorous research synthesis. Next, we outline a set of generally useful meta-analytic techniques and follow this with a discussion of some of the difficult decisions that research synthesists face in carrying out a meta-analysis. We conclude by addressing some broader issues concerning criteria for evaluating the quality of knowledge syntheses in general and meta-analyses in particular. A general theme of the chapter is that social scientists who are conducting research syntheses need to think about what distinguishes a good synthesis from a bad synthesis. This kind of effort is crucial for assessing the value of existing research syntheses and for promoting high-quality research synthesis in the future.