265 1 In the first part of our paper (Vorobeichik, Kozlov, 2012) we discussed methodology of studying impact regions and outlined the typical errors made by researchers while harvesting the outcomes of passive experiments. The other tasks of equal importance are correct data analysis and accurate presentation of the results in publications. Keeping in mind the urgent need of quantitative research synthesis in the field of impact ecology, in this paper we discuss the basic requirements to the presentation of the information in scientific publications.
META ANALYSIS AS AN EFFECTIVE TOOL FOR INTEGRATING THE RESULTSOF INDEPENDENT STUDIES Search for regularities by generalization of particular results is the most challenging goal of any scientific research, which is however difficult to achieve. While the statistical analysis is widely used in ecology since the first quarter of the XX century, no formalized methods for the quantitative analysis of the accumulated infor mation existed until the late 1970's. As the result, narra tive reviews suffered from inevitable, sometimes uncon scious, subjectivity: their conclusions were non repeat able due to the absence of strict criteria for both selection of publications for the review and evaluation of the quality of the published data. 1 The article was translated by the authors. This problem has been solved with the develop ment of meta analysis. This relatively modern statisti cal method allows to quantitatively combine the results of several independent studies addressing the same problem (Shitikov et al., 2008;Borenstein et al., 2009). The core of the meta analysis is the transfor mation of the diverse data extracted from publications (selected by clearly defined criteria) into a common measure of effect size. These effect sizes are then anal ysed to find common trends and identify factors explaining differences between the outcomes of indi vidual studies (Gurevitch, Hedges, 2001). Impor tantly, meta analysis explicitly accounts for the repre sentativeness of each case study, giving larger weight to the studies based on a larger number of replicates.Not every problem deserves application of meta analysis. Meta analysis is most effective when the amount of accumulated information is large, and the preliminary analysis has demonstrated that, in an aver age, effect sizes are minor, results of case studies are contradictory, and/or there is the reason to suspect research or publication bias. Impacts of industrial pol lution on biota obviously belong to this kind of problem.Objectivity of the meta analysis is achieved through the combination of precisely formulated cri teria for selection of primary studies with the use of the statistical approach for integration of their outcomes. Therefore, in theory, conclusions of meta analysis are reproducible: any scientist, by using the described cri Abstract-In the present paper we briefly outline the principles of meta analysis, which gradually substitutes narrative reviews in modern ecology and becomes a standard for generalization...