Source
Douglas (2000), LIT-139 (Philosophy of Science 67(4), 559–579), read in full in NOTE-145.
What was actually shown
The premise from the cases. Inductive risk arises at the choice of methodology, the characterisation of data and the interpretation of results, as well as at acceptance (THEORY-639).
The argument (§3, pp. 563–565). At each internal stage “one can make a wrong (i.e., epistemically incorrect) choice, with consequences following from that choice”. When those consequences fall outside science and can be foreseen, deciding which error to risk more needs a valuation of them, and that is a non-epistemic value judgement. Whatever threshold or classification rule is adopted answers it implicitly. So “non-epistemic values are required for good reasoning” (p. 565), and for such science “value-free science is inadequate science” (p. 559).
Two roles (§3, §5, §7). Longino’s uncontroversial roles of values (choosing problems, using results, ethical limits on methods) are direct and external. Inductive risk gives an indirect, internal role: values weigh the consequences of possible error, together with how likely error is. Values alone never decide the choice (p. 565). A direct role inside science stays illegitimate. Blinding is endorsed to prevent “the seeing of something because one wants to see it” (p. 572), and “I want X to be true, therefore X is true” remains a bad argument (p. 578).
Against leaving values to the users (§2, p. 563). Jeffrey and McMullin would have scientists report and users apply utilities. Douglas replies that scientists’ claims are authoritative in policy, and that many risks are taken before acceptance, where decision-makers cannot see them.
What could have come out otherwise. This is a normative argument. It could fail at its unargued premise, or at the reply to the division of labour.
What this does not say
- Why scientists bear that responsibility. The premise is explicitly “beyond the scope of this paper” (p. 563), and “required” inherits it.
- That reporting uncertainty cannot replace acceptance. The argument does not engage that rejoinder (NOTE-145, Limitations). The reply to Jeffrey and McMullin is one paragraph.
- How to tell the indirect role from the direct one in practice. A rule for borderline slides chosen for its known regulatory effect may slide from one to the other. The paper does not address this; Douglas’s later book does.
- Who should supply the valuations. The paper argues only that scientists cannot avoid them.
- That most research is affected. That research without foreseeable outside consequences is “rapidly becoming a minority” is asserted (§7).
- What this means for objectivity. That is deferred to “future work” (p. 578).
Connections
- Reiss (LIT-118; THEORY-537) builds the same indirect role into a theory of evidence: higher stakes lower the bar for rejecting an alternative. He does not cite Douglas.
- Longino’s critical contextual empiricism (THEORY-151) also denies that values stay outside science. Douglas’s argument is normative where Longino’s is descriptive (fn. 8).