Meta-Analysis: A Systematic Approach to Scientific Inquiry
Beard, Margaret T., Bailey, Catherine B., Marsh, Verdell, Journal of Theory Construction and Testing
Abstract: Meta-analysis is a perspective of many techniques of measurement and statistical theories used in synthesizing findings from research studies. Various definitions of meta-analysis are presented. Uses and elements of meta-analysis are presented. Uses and elements of meta-analysis are described. This is a systematic approach to examine research, formulate hypotheses, conduct literature reviews, establish inclusion/exclusion criteria for articles, statistically synthesize and combine effect sizes from studies, search for moderating variables to explain effects, and report results. Computers have allowed meta-analysts to expand the technique and to develop programs specific for analyses. As with any method, efforts to reduce bias should be employed. The history and current status of the meta-analytic approach are described.
Key Words: meta-analysis, effect size, and homogeneity
Meta-analysis is a perspective of many techniques of measurement and statistical theories used in synthesiz ing findings from research studies. The term has come to encompass all methods and techniques of quantitative research synthesis. Meta-analysis is: (a) the statistical analysis of a collection of analyses that is used to integrate and summarize empirical findings that have resulted from existing studies (Glass, 1976); (b) a method designed to bring order to diverse and complex literatures by summarizing and synthesizing the findings of empirical studies (Maruyama, 1991); (c) a statistical tool for summarizing empirical results across studies to reach a quantitative generalization (Arthur, Bennett, & Huffcutt, 1994); (d) a body of statistical methods derived from the literature that is used for the purpose of drawing an omnibus conclusion based on summary information obtained from a collection of independent investigations (Proskin & Volpe, 1994); and (e) a method for systematically examining a body of research, carefully formulating hypotheses, conducting an exhaustive literature review, establishing inclusion/exclusion criteria for articles, recording and statistically synthesizing and combining data and effect sizes from the studies, searching for moderator/ mediator variables to explain effects, and reporting results ( Rosenthal & Dimatteo, 2001). As such, meta-analysis provides the investigator with a disciplined process of summarizing research findings with techniques that measure strengths of relationships that are more sophisticated than those used in qualitative reviews (Lipsey & Wilson, 2001).
One of the earliest efforts of combining research results was introduced by Karl Pearson (1904), who reviewed studies on the effects of treatments for enteric fever in South Africa and India. In 1932, R.A. Fisher developed a method based on the product of (-values for obtaining an overall (-value. Glass (1976) coined the term, meta-analysis as an approach for summarizing quantitative research findings from individual studies. Shortly thereafter, the empirical case for summarizing studies was made with indications that narrative review procedures led to imprecise or inaccurate characterizations of cumulative research results (Cooper & Rosenthal, 1980). Glass, McGraw and Smith (1981) proposed a view of the summarization of research studies as a new application of analysis of variance and multiple regression procedures with study outcomes in the form of effect sizes, treated as the criterion variable and the features of studies, as the predictor variables. Cooper (1982) introduced the term "integrative review" to describe the application of a research process to analyze a collection of studies. As meta-analytical procedures became more popular and philosophy of science and conceptual issues were identified, methods for dealing with sampling error and biases were developed (Hunter & Schmidt, 1990; Rosenthal, 1991). By the year 2000, meta-analysis was an accepted method of scientific inquiry. …