Is the replication crisis worst in psychology or medicine?

Written by AI. Help improve this answer by adding to the sources section. When the sources section is updated this article will regenerate.

The question of whether the replication crisis is worse in psychology or in medicine hinges on how “worse” is defined—percent of studies that fail, prominence of the failures, or downstream social cost. The four sources listed focus more heavily on psychology, but they also give a glimpse into biomedical research (specifically Alzheimer’s).

Psychology

In 2015 the Open Science Collaboration attempted to replicate 100 high-profile psychology papers and could reproduce the original statistical result only 36 % of the time [1].
A 2023 commentary argues that roughly 75 % of day-to-day psychology findings are likely false, citing continual failed replications and persistent questionable research practices [2].
A detailed reassessment of stereotype-threat research finds the mean effect dramatically smaller than early reports and documents multiple non-replications, highlighting that even celebrated social-psychological effects are shaky [4].

Medicine (biomedical research)

A 2025 New York Times opinion piece describes large-scale data fabrication in a landmark line of Alzheimer’s studies; subsequent attempts to reproduce the work have failed and drug programs built on the disputed findings were halted [3].
The article stresses that fraud and irreproducibility in basic biomedical science can derail billion-dollar drug programs and patient hopes, but it does not quantify an overall failure rate for medicine [3].

Comparative assessment

Quantified evidence: Psychology offers a systematic 100-paper audit (36 % success) plus meta-analyses of specific paradigms. The biomedical source supplies anecdotes of spectacular failure but no field-wide replication audit.
Severity by proportion of failed studies: On the available numbers, psychology looks worse (≥ 64 % non-replication in a random sample [1] and possibly higher according to the 75 % claim [2]).
Severity by real-world impact: Biomedical failures like the Alzheimer’s case can divert years of drug development and billions of dollars [3], so the societal cost per failed result may be higher even if overall percentages are unknown.

Where the authors disagree

Sources [1] and [2] both characterise psychology’s problem as widespread and quantifiable.
Source [3] does not claim medicine’s crisis is smaller—it highlights egregious instances and suggests systemic vulnerability—but offers no numerical baseline.
Because the biomedical piece focuses on fraud whereas the psychology pieces emphasise routine non-replication, they frame “crisis” differently.

Public-discourse timeline 2015 Publication of “Estimating the Reproducibility of Psychological Science,” launching mainstream discussion of psychology’s replication crisis [1]. 2018–2022 Large numbers of stereotype-threat and other social-psychology studies fail to replicate; Inzlicht’s overview crystallises doubts [4]. 2023 The Unsafescience newsletter claims three-quarters of everyday psychology claims are false, keeping the topic in public view [2]. 2025 The New York Times reports on alleged fraud in foundational Alzheimer’s papers, shifting part of the replication conversation toward medicine [3].

Conclusion With the present evidence, psychology shows the higher documented non-replication rate, while medicine demonstrates fewer audited numbers but potentially higher stakes per failure. Ongoing, field-wide replication projects akin to psychology’s 2015 audit would be required before one can definitively say the crisis is “worse” in one domain over the other.

Sources

Question

Is the replication crisis worst in psychology or medicine?