Generalization: Conceptions in the Social Sciences

A major goal in the behavioral and social sciences is to offer conclusions that are valid for observed particulars extensive to abstract entities. Four different approaches to generalization are considered in this article. The first describes random sampling with its advantages and limitations. The second considers purposive selection in contrast to randomization. The third covers the possibility of generalizing causal statements and its connection to meta-analysis. And the last refers to extrapolation from a sample to nonsampled entities. This article concludes by arguing that random sampling, though limited, is the best formal method for generalization.