Coping with Inductive Risk When Theories are Underdetermined: Decision Making with Partial Identification

Controversy about the significance of underdetermination of theories persists in the philosophy and conduct of science. The issue has practical import when scientific research is used to inform decision making, because scientific uncertainty yields inductive risk. Seeking to enhance communication between philosophers and researchers who analyze public policy, this paper describes econometric analysis of partial identification. Study of partial identification finds underdetermination and inductive risk to be highly consequential for credible prediction of important societal outcomes and, hence, for credible public decision making. It provides mathematical tools to characterize a broad class of scientific uncertainties that arise when available data and credible assumptions are combined to predict population outcomes. Combining study of partial identification with criteria for reasonable decision making under ambiguity yields coherent practical approaches to make policy choices without accepting one among multiple empirically underdetermined theories. The paper argues that study of partial identification warrants attention in philosophical discourse on underdetermination and inductive risk.

💡 Research Summary

The paper bridges a long‑standing philosophical debate about the underdetermination of scientific theories with the practical problem of inductive risk in policy‑making. It argues that the econometric literature on partial identification provides a rigorous way to represent scientific uncertainty when data and credible assumptions together can only bound, rather than pinpoint, population parameters. By treating these bounds as the logical consequence of underdetermination, the author shows that inductive risk—traditionally defined as the probability of being wrong when accepting or rejecting a hypothesis—can be quantified directly from the width of the identified set.

The author critiques the “lure of incredible certainty” that pervades much of applied economics and other empirical sciences, where researchers feel pressure to present point estimates and definitive policy recommendations despite weak identifying assumptions. Traditional reliance on Occam’s Razor or other simplicity criteria is shown to be insufficient for decision‑making because such criteria do not incorporate the welfare consequences of errors.

Two substantive case studies illustrate the approach. In medical treatment choice, conventional hypothesis testing fixes Type I and Type II error rates (often 5 % and 10‑20 %) without regard to the relative harms of false positives versus false negatives. By applying the minimax‑regret decision rule—derived from the identified set and a loss function—the paper demonstrates how to choose treatments that minimize the worst‑case welfare loss across all admissible models. In climate policy, the author shows how to incorporate the entire range of temperature‑rise projections generated by multiple climate models into policy design, thereby avoiding over‑confident forecasts and embracing robust, ambiguity‑aware strategies.

A central contribution is the systematic mapping of terminology across disciplines. “Underdetermination” in philosophy corresponds to “partial identification” in econometrics and to “ambiguity” in decision theory; “inductive risk” aligns with Type I/II error probabilities and with the concept of maximum regret in statistical decision theory. This translation facilitates communication between philosophers, econometricians, and policy analysts.

Section 5 integrates partial identification with decision‑theoretic criteria, arguing that when the identified set is set‑valued, decision makers should not be forced to pick a single hypothesis. Instead, they can adopt criteria such as minimax‑regret, which respects both the epistemic uncertainty and the societal stakes of the decision. The paper concludes that partial identification is not merely a technical curiosity but a practical tool for responsibly handling scientific uncertainty in public decision‑making, and it calls for greater attention to this framework in philosophical discussions of underdetermination and inductive risk.