Are Causal Analysis and System Analysis Compatible Approaches?

Abstract

In social science, one objection to causal analysis is that the assumption of the closure of the system makes the analysis too narrow in scope, that is, it considers only ‘closed’ and ‘hermetic’ systems thus neglecting many other external influences. On the contrary, system analysis deals with complex structures where every element is interrelated with everything else in the system. The question arises as to whether the two approaches can be compatible and whether causal analysis can be integrated into the broader framework of system analysis. This article attempts a negative answer on the grounds of fundamental differences in their assumptions and suggests using system analysis as a post‐hoc comparative tool.

Acknowledgements

I wish to thank Phyllis McKay Illary, Jon Williamson, Guillaume Wunsch, and the participants of the seminars at the Centre for Reasoning (University of Kent, February 2008) and of the IUC Conference (Dubrovnik, April 2008) for stimulating and helpful comments on an earlier draft of this paper. Financial support from the FRS‐FNRS (Belgium) is also gratefully acknowledged.

Notes

[1] As mentioned above, causal models can also be used in an inductive way, e.g. data mining. This is, for instance, the approach of Spirtes, Glymour and Scheines (1993), and, to some extent, of Pearl (2000). Inductivist approaches claim that causal relations can be inferred from data without the burden of extra‐statistical and causal assumptions made in their H‐D counterparts. However, it goes far beyond the goal of this paper to discuss the success of inductive causal models. Consequently, the scope will be limited to causal models that employ an H‐D methodology.

[2] Feedback loops that are ordered in time are admissible: X_t causing Y_t′ which in turn causes X_t″ but at a later time.