Abstract

This article explores the difficulties raised for social scientific investigation by the absence of experiment, critically reviewing realist responses to the problem such as those offered by Bhaskar, Collier and Sayer. It suggests that realist arguments for a shift from prediction to explanation, the use of abstraction, and reliance upon interpretive forms of investigation fail to demonstrate that these approaches compensate for the lack of experimental control. Instead, it is argued that the search for regularities, when suitably conceived, provides the best alternative to experiment for the social sciences