Are Experiments Possible?

Abstract

The paper critically examines the inferential logic underlying Mill's methods of agreement and difference as it is applied in present-day behavioral science where the methods in question are embodied in the socalled `treatment-control random design' (Mayo, 1996). The reputed goal of the design is to be able to arrive at judgments of sufficient causation (e.g. `X is a sufficient cause of [condition for] Y'). This is accomplished by `neutralizing' the influence of the extraneous variables inherent in a particular (experimental) set of conditions by way of creating another (control) set of conditions which is identical to the experimental prototype in every way except for one element (the independent variable). Any difference between the effects of the two sets (the experimental effect) is, then, said to reflect or be a product of the independent variable in question. As I argue, the above-mentioned inference is rooted in the generally tacit assumption of effect independence, that is, the view that the `extraneous' and the `independent variable' components comprising the experimental condition end up exerting completely separate or isolated effects on the dependent variable under consideration. It is argued that the said assumption, given both the inherent conceivability and the empirical ubiquity of various interaction effects, has no logical foundation and, thus, cannot be affirmed with any degree of certainty. The non-acceptance of the assumption calls into question the logical viability of clinical process research.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.