In this section, I briefly outline the history of meta-analysis. My goal is not to be exhaustive in detailing this history (for more extensive treatments, see Chalmers, Hedges, & Cooper, 2002, Hedges, 1992, and Olkin, 1990; for a history intended for laypersons, see Hunt, 1997). Instead, I only hope to provide a basic overview to give you a sense of where the techniques described in this book have originated.
There exist several early individual attempts to combine statistically results from multiple studies. Olkin (1990) cites Karl Pearson’s work in 1904 to synthesize associations between inoculation and typhoid fever, and several similar approaches were described from the 1930s. Methods of combining probabilities advanced in the 1940s and 1950s (including the method that became well known as Stouffer’s method; see Rosenthal, 1991). But these approaches saw little application in the social sciences until the 1970s (with some exceptions such as work by Rosenthal in the 1960s; see Rosenthal, 1991).
It was only in the late 1970s that meta-analysis found its permanent place in the social sciences. Although several groups of researchers developed techniques during this time (e.g., Rosenthal & Rubin, 1978; Schmidt & Hunter, 1977), it was the work of Gene Glass and colleagues that introduced the term “meta-analysis” (Glass, 1976) and prompted attention to the approach, especially in the field of psychology. Specifically, Smith and Glass (1977) published a meta-analysis of the effectiveness of psychotherapy from 375 studies, showing that psychotherapy was effective and that there is little difference in effectiveness across different types of therapies. Although the former finding, introduced by Glass, would probably have been received with little disagreement, the latter finding by Smith and Glass was controversial and prompted considerable criticism (e.g., Eysenck, 1978). The controversial nature of Smith and Glass’s conclusion seems to have had both positive and negative consequences for meta-analysis. On the positive side, their convincing approach to the difficult question of the relative effectiveness of psychotherapies likely persuaded many of the value of meta-analysis. On the negative side, the criticisms of this particular study (which I believe were greater than would have been leveled against meta-analysis of a less controversial topic) have often been generalized to the entire practice of meta-analyses. I describe these criticisms in greater detail in Chapter 2.
Despite the controversial nature of this particular introduction of metaanalysis to psychology, the coming years witnessed a rapid increase in this approach. In the early 1980s, several books describing the techniques of metaanalysis were published (Glass, McGraw, & Smith, 1981; Hunter, Schmidt, & Jackson, 1982; Rosenthal, 1984). Shortly thereafter, Hedges and Olkin (1985) published a book on meta-analysis that was deeply rooted in traditional statistics. This rooting was important both in bringing formality and perceived statistical merit to the approach, as well as serving as a starting point for subsequent advances to meta-analytic techniques.
The decades since the introduction of meta-analysis to the social sciences have seen increasing use of this technique. Given its widespread use in social science research during the past three decades, it appears that metaanalysis is here to stay. For this reason alone, scholars need to be familiar with this approach in order to understand the scientific literature. However, understanding meta-analysis is valuable not only because it is widely used; more importantly, meta-analysis is widely used because it represents a powerful approach to synthesizing the existing empirical literature and contributing to the progression of science. My goal in the remainder of this book is to demonstrate this value to you, as well as to describe how one conducts a meta-analytic review.
Source: Card Noel A. (2015), Applied Meta-Analysis for Social Science Research, The Guilford Press; Annotated edition.