In first explicating the notion of quantitative literature review for the social sciences, Glass (1976) argued that knowledge is not built from any individual study, but from the integration of findings from many studies. Individual studies do not so much yield knowledge as evidence with which knowledge can be built. Knowledge is socially constructed. To overemphasize a single study's findings or integrate research only impressionistically leaves researchers knowing less than the evidence

more »

... , insufficiently exploiting the wealth of data scattered in separate studies. Quantitative research integration, or meta-analysis, has a history in both the physical and social sciences that precedes Glass' formulation (Bangert-Drowns 1986; Hedges 1987) . Most generally, meta-analysis is a perspective rather than a method, a recognition that research findings can be interpreted probabilistically in the context of collections of studies. The meta-analytic perspective is consistent with, and perhaps newly empowers, communal and cumulative activities of science in refining method and transforming data into knowledge (Schmidt 1992). A number of writers initially responded with skepticism or even overt hostility to this apparently new method of inquiry (e.g., Eysenck 1978). It is hard now to find critics opposed to meta-analysis in principle (Wachter 1988). However, two kinds of concerns are still expressed about meta-analysis. The first suggests that quantitative review communicates an appearance of precision and comprehension which is in fact unreal and thus misleading. The second concern is that meta-analysis is not doing what it claimed it could do: settle important theoretical and practical questions in the midst of contradictory research findings. These concerns arise from the fact that there is plenty of room for subjectivity and imprecision in meta-analysis (Guzzo et al. 1987; L'Hommedieu et al. 1988; Wanous et al. 1989) . In spite of advances in meta-analytic method that are meant to increase the precision of literature review, meta-analysis is still, in many ways, a very human enterprise. Though in principle meta-analysis offers simple means for