Meta-Analysis

Meta-analysis is a method for systematically combining data from multiple studies to make statistical conclusions about a specific research question. As it involves an increased number and a greater diversity of subjects, the conclusions reached through meta-analysis are statistically stronger than the results of a single study. There are many strengths associated with conducting meta-analyses, including increased statistical power, increased accuracy, and reduced uncertainty regarding the conclusions from various individual studies. In addition, the process of conducting a meta-analysis can reveal additional context for understanding the results of individual studies as well as suggest potential questions for future research. For these reasons, a meta-analysis of well-conducted, rigorous studies is considered one of the highest levels of evidence in the literature. Although mathematicians and scientists have explored various methods for summarizing the results of different studies since the 17th century, in 1976 Gene Glass developed the formalized technique known today as meta-analysis. Since that time, researchers in many fields such as education, psychology, communication, and the biomedical sciences have employed the use of meta-analysis to combine data and make inferences from multiple studies.

This entry provides an overview of the steps involved with conducting a meta-analysis: (a) formulating the research problem, (b) searching the literature, (c) analyzing the data, and (d) disseminating the results as well as the strengths and [Page 1163]weaknesses of this method. Formulating the research problem involves carefully defining the key constructs for the research question being studied. A detailed literature search requires outlining a set of inclusion and exclusion criteria that are used to retrieve and screen primary studies for analysis. The statistical analysis consists of extracting and coding the data, calculating the effect sizes for each primary study, examining the heterogeneity, and computing summary statistics. The methodology, results, and a detailed discussion of the meta-analysis are then compiled in a structured report for dissemination.