Rough set methodology in meta-analysis - a comparative and exploratory analysis

We study the applicability of the pattern recognition methodology "rough set data analysis" (RSDA) in the field of meta analysis. We give a summary of the mathematical and statistical background and then proceed to an application of the theory to a meta analysis of empirical studies dealing with the deterrent effect introduced by Becker and Ehrlich. Results are compared with a previously devised meta regression analysis. We find that the RSDA can be used to discover information overlooked by other methods, to preprocess the data for further studying and to strengthen results previously found by other methods.