In social sciences we often have to do with aggregated data which pose both methodological and practical problems. Erroneous inference of individual characteristics and behaviors from this kind of data, described and named by Robinson as 'ecological fallacy', has been known for more than fifty years now. Several statistical approaches were developed to deal with the problem and produce valid and reliable results. This paper provides on overview of the most popular and frequently used methods of ecological inference. First one is ecological regression as developed by Goodman in a direct answer to Robinson's article. The last one is quite new entropy maximization approach. The purpose of paper is to give Polish reader a review of useful techniques in analysis of aggregate data along with their pros and cons. .