The prefix meta comes from the Greek meaning "occurring later or in succession to," "after," "more comprehensive," or "transcending."1 When discussing meta-analyses or meta-studies, the researcher revisits previously completed studies in order to summarize them and achieve a more comprehensive understanding of variables or phenomena. Thus, this approach results in an entirely new study that transcends the original research. The purpose of this article is to describe meta-studies and meta-analyses, provide a detailed look at meta-analyses, and discuss the differences between them.

META-STUDY

Meta-studies of qualitative research are not for faint-hearted especially when one considers that there are a wealth of qualitative methods, many researchers, and a variety of interpretations that need to be made by the synthesizing researcher. The phenomenon of interest must be determined before the meta-study can begin; the inclusion criteria must be established; and the literature within the specified time period must be reviewed critically. When conducting a meta-study, it is most helpful to have teams of reviewers available to critique each others' rationales and to question how a particular study met the inclusion criteria.

There are 2 different types of qualitative meta-studies: one in which the same phenomenon is studied and one in which the researcher synthesizes the results and processes of previous studies.2 The researcher who studies results and processes of identified phenomenon has the luxury of reflecting on the steps taken by the primary researcher to determine where we are presently and where we are going in the future.3

META-ANALYSIS

A meta-analysis is the combination of the results from many quantitative studies, dealing with the same topic, which are put together in a format that provides a systematic review of the studies. In 1904, Karl Pearson conducted the first meta-analysis (not called that at the time) to overcome small sample size among studies. In the 1970s, Glass introduced the term "meta-analysis" to the educational community to compare findings among studies using a common metric such as effect size.4

As in a meta-study, the topic of meta-analyses needs to be determined, and the research steps followed. It is imperative that the researcher make, and keep, a detailed and clearly identified written protocol of the research questions for the project in order to guide the research. Inclusion criteria need to be identified and procedures applied for locating and examining empirical studies for similar constructs and relationships that may be obscured in other approaches. Finally, results are subject to statistical procedures. Locating all studies is a priority when conducting meta-analyses, but this can be very frustrating. Because research with nonsignificant or uninteresting results is rarely published, it may be difficult to find all the needed studies. Unless these studies are found and considered, potential bias will result.

There are a variety of approaches a researcher may use for determining inclusion criteria in a meta-analysis. Vote counting is a method in which the researcher looks at the findings as positive, negative, or nonsignificant. However, conclusions based on vote counting can be very misleading because treatment effect and sample size are not taken into consideration.4

In Glassian or classical meta-analysis, the inclusion criteria are very liberal, including nonsignificant studies because the meta-analysis itself can determine study quality. Study effect meta-analysis is a method in which inclusion rules are more selective, and studies with methodological flaws are excluded. Psychometric meta-analysis requires considerable information from previously published studies in order to correct for effect sizes; however, this information is not always available in published reports. The last step in all these approaches involves calculating the statistical significance of the studies.

CONCLUSION AND DISCUSSION

Meta-studies that explore relationships in quantitative research and meta-analyses that explore qualitative research actually have more in common than one would think. In both types, the researcher must determine a study question that is specific and narrow, collect specific and comprehensive data in the form of primary research, base the study selection on specific criteria, and require an appreciable investment in time, planning, and implementation. The only difference is that the qualitative studies are not subjected to any statistical analyses.

References