Abstract

This chapter presents a heterodox critique of regression analysis in economics in the sense that it reviews whether critical realist concerns should lead to the abandonment of regression analysis. A review of the different ways in which econometrics has tried to cope with the generally agreed difficulties of identification is presented. The critical realist position is then examined, and its argument that the general econometric approach is indicative of an ontologically illegitimate application of a closed system to a structured open system. The chapter closes by arguing that set within the context that causes of events might relatively endure, but are subject to change, regression analysis can be linked to other forms of evidence in an empirical strategy.