NCJ Number
158708
Journal
Evaluation and The Health Professions Volume: 18 Issue: 3 Dated: (September 1995) Pages: 304-314
Date Published
1995
Length
11 pages
Annotation
This article highlights prevalent misunderstandings of how meta-analysis is conducted and interpreted, as well as its relation to primary research and narrative review.
Abstract
Meta-analysis, the quantitative synthesis of research, has become established as an accepted form of scholarly and practical inquiry since it was articulated by Glass (1976). In areas where the number of research studies make it difficult for researchers to distill consensual findings, meta-analytic tools can assist in the detection of patterns across studies. Meta-analytic approaches are indicated when measures of central tendency, variance, and the probabilistic relations among salient variables in replications need to be defined in the service of theoretical or practical development. Some misunderstandings of meta-analysis are that there is one best meta-analytic method; that meta- analysis is better than narrative review; that meta-analysis is objective; that meta-analysis culminates the research in a given domain; and that the most important product of a meta-analysis is the average effect size. Each of these misunderstandings is countered in this discussion. The author argues that the "meta" in meta-analysis does not suggest the method is beyond reproach or necessarily superior to primary research or narrative review; rather, its scope of investigation is broader than primary research, its aims more comprehensive than primary research, and its procedures more public than narrative review. For all its quantification, statistical sophistication, and systematic organization, meta-analysis is still dependent on human decisionmaking and should be viewed as a partner to, rather than a master of, narrative review and primary research. 12 references