Search

Experimentation and Generalization in the Social Sciences The paper discusses the issue of the generalization of experimental results in the social sciences. Specifically, it emphasizes the importance for generalization of: (1) the interaction between the experimental treatment and the characteristics of the observed group, a factor of external validity of the experiment, known as selection-X interaction; (2) a reference model for formulating a theory of the selection-X interaction. In this model, in addition to the experimental treatment and to the outcome it produces, great attention is paid to context and mechanism variables. The paper highlights the differences between context and mechanism variables, providing a demarcation criterion with regard to function but not to substance. Three different types of mechanism are presented, all conforming to the model of causal betweenness, but distinguished according to the types of the relationship between the experimental treatment and the mechanism. Special consideration is given to the dispositional mechanisms because the many logical issues and impediments involved in the use of dispositional concepts for the purposes of explanation and prediction. Finally, a data-gathering procedure referred to as a retrospective post-test one is presented as a way to objectively control relations between variables in the causal betweenness model.