Evidence, External Validity, and Explanatory Relevance

Nancy Cartwright

in Philosophy of Science Matters

Published in print March 2011 | ISBN: 9780199738625
Published online May 2011 | e-ISBN: 9780199894642 | DOI:
Evidence, External Validity, and Explanatory Relevance

Show Summary Details


Evidence-based policy commends randomized controlled trials (RCTs) as gold-standard evidence for predictions about whether policies will “work” (so-called “effectiveness predictions”), and largely because RCTs establish causal conclusions without need for theory. But what makes RCTs evidence for effectiveness at all? A usual label for this problem, “external validity,” conceals a host of problems and a productive answer. This chapter employs Achinstein's formula, “evidential relevance = explanatory relevance,” to argue that the evidential relevance of an RCT to an effectiveness prediction is conditional. If the effect in question is governed by the same causal laws in the RCT and in the target, these laws could explain the effect in both, thus securing the explanatory relevance of the RCT result to the effectiveness prediction. But the explanation, and in consequence the evidential relevance, is conditional on (a) sharing these causal laws and (b) the target possessing the requisite auxiliary factors that call the shared laws into play. Finding the requisite auxiliaries involves what this chapter calls “horizontal search”; finding shared laws, if they exist at all, involves “vertical search”: locating or creating the right kind and level of abstraction. Both generally require a great deal of theory, which RCT advocates had hoped to avoid.

Keywords: RCT; randomized controlled trial; external validity; evidential relevance; explanatory relevance; horizontal search; vertical search; abstraction; theory; Peter Achinstein

Chapter.  6135 words. 

Subjects: Philosophy of Science

Full text: subscription required

How to subscribe Recommend to my Librarian

Buy this work at Oxford University Press »

Users without a subscription are not able to see the full content. Please, subscribe or login to access all content.