Generalization Bias in Science

Abstract

Many scientists routinely generalize from study samples to larger populations. It is commonly assumed that this cognitive process of scientific induction is a voluntary inference in which researchers assess the generalizability of their data and then draw conclusions accordingly. We challenge this view and argue for a novel account. The account describes scientific induction as involving by default a generalization bias that operates automatically and frequently leads researchers to unintentionally generalize their findings without sufficient evidence. The result is unwarranted, overgeneralized conclusions. We support this account of scientific induction by integrating a range of disparate findings from across the cognitive sciences that have until now not been connected to research on the nature of scientific induction. The view that scientific induction involves by default a generalization bias calls for a revision of the current thinking about scientific induction and highlights an overlooked cause of the replication crisis in the sciences. Commonly proposed interventions to tackle scientific overgeneralizations that may feed into this crisis need to be supplemented with cognitive debiasing strategies against generalization bias to most effectively improve science.

Cultural Bias in Explainable AI Research: A Systematic Analysis

“…Hence, the unrestricted conclusions in most of the reviewed XAI papers were hasty generalizations, i.e., claims whose scope was broader than warranted by the evidence and justification provided by the researchers. Since psychological research suggests that explanatory needs likely differ between WEIRD and non-WEIRD populations, as discussed in Section 2, the pervasive insufficiently supported extrapolations that we found from WEIRD samples to other populations may indicate a cultural "generalization bias" (Peters et al, 2022) toward WEIRD populations in many currently available XAI user studies.…”

Dead rats, dopamine, performance metrics, and peacock tails: Proxy failure is an inherent risk in goal-oriented systems

“…Indeed, Watve admonishes that we did not "adequately cover proxy failure in academia" where "proxy failure has reached unprecedented and unparalleled levels." This is perhaps the right place to disclose that proxy failure in academia was in fact one of the main areas of research that lead up to the current target article Peters et al, 2022). So our main justifications for neglecting the topic in the present article are that (i) we have written about it previously and (ii) we felt sure it would be raised by others.…”

“…So perhaps the similarities that were so frequently noted were superficial. Perhaps we were succumbing to an overgeneralization bias (Peters, 2022). Or perhaps, trying to explain everything would end up explaining nothing.…”

Teleonomy, legibility, and diversity: Do we need more “proxynomics”?

“…Several other comments presented examples of closed proxy failure in academia. Indeed, Watve admonishes that we did not “adequately cover proxy failure in academia” where “proxy failure has reached unprecedented and unparalleled levels.” This is perhaps the right place to disclose that proxy failure in academia was in fact one of the main areas of research that lead up to the current target article (Braganza, 2020, 2022; Peters et al, 2022). So our main justifications for neglecting the topic in the present article are that (i) we have written about it previously and (ii) we felt sure it would be raised by others.…”