Is the replication crisis worst in psychology or medicine?

Latest revision as of 02:15, 1 May 2025

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

Short answer

Quantitatively, psychology shows the lowest large-scale replication rate that has actually been measured (about one-third of tested findings replicated) [1]. Medicine (especially biomedicine) certainly suffers from fraud and non-reproducible results, but the available evidence is more fragmentary; there is no single field-wide replication study of comparable scope. Consequently, most commentators accept that the replication crisis is currently better documented—and appears numerically worse—in psychology, while agreeing that certain sub-fields of medicine (e.g., Alzheimer’s research) may harbour equally serious problems.

What the main sources say

The Open Science Collaboration’s 2015 project replicated 100 high-profile psychology papers and reproduced only 36 % of the original significant results [1].

The “Unsafe Science” analysis argues that, once publication bias is taken into account, roughly 75 % of psychology claims are likely false, framing psychology as “ground zero” of the crisis [2].

Michael Inzlicht’s 2024 essay focuses on stereotype-threat research; he concludes that the effect is “less robust than believed,” but stresses that some psychological constructs do survive replication attempts, suggesting the situation is “serious yet salvageable” [4].

In medicine, the 2025 New York Times op-ed chronicles alleged data fabrication in Alzheimer’s amyloid-beta studies, contending that years of drug-development resources were misdirected as a result [3].

Vox’s 2024 feature gathers estimates that fraudulent or irreproducible biomedical findings contribute to treatment delays and avoidable deaths, but offers no systematic replication percentage comparable to the psychology figure [5].

Are the views consistent?

They converge on the claim that both fields have serious reliability issues, but differ in emphasis. Sources [1], [2], and [4] document psychology’s low replication rates. Sources [3] and [5] argue that medicine’s stakes are higher (patient harm, billions in costs) even if the measured replication failure rate is not yet pinned down. No source provides evidence that medicine as a whole replicates worse than psychology; rather, the claim is that its failures are more consequential.

Factors that make psychology look worse

Abundant field-wide audits (e.g., 2015 Science project) produce hard numbers [1].

Experiments are often small-sample, low-power, and easier to redo quickly, revealing problems faster [2].

Publication incentives once favoured surprising results; the discipline now publicly tracks corrections and retractions [4].

Factors that obscure medicine’s true rate

Clinical trials are costlier and take years, so systematic replications are rare.

Regulatory oversight (FDA, EMA, etc.) enforces certain standards, potentially boosting replicability, but also concentrates efforts on late-stage trials that may hide earlier basic-science flaws [5].

High-profile fraud cases (e.g., Alzheimer’s amyloid imaging) attract media coverage without providing denominator data for the field at large [3].

Public discourse timeline

2011–2014 Psychology begins adopting preregistration and open-data norms following several high-profile failed replications (not detailed in current sources).

2015 Publication of the Open Science Collaboration study quantifies the problem in psychology (36 % replication rate) [1].

2017–2020 Replication efforts expand to economics, social priming, and some biomedical niches; no medicine-wide project yet.

2024 Michael Inzlicht’s pre-print calls for a “reckoning” but also reform optimism in social psychology [4]. Vox article publicises the human toll of biomedical fraud, pushing the crisis narrative beyond psychology [5].

2025 New York Times op-ed links alleged image manipulation in landmark Alzheimer’s studies to stalled drug development, intensifying concern inside medicine [3].

Bottom line

With systematic audits showing only ~35 % replication success, psychology presently exhibits the clearest and worst-documented replication crisis. Medicine’s crisis is better described as potentially just as serious, but still less quantified; notable fraud cases and the high cost of irreproducibility keep the issue in the spotlight. Future large-scale replication projects in clinical and pre-clinical medicine will be needed before a definitive comparison is possible.

Sources[edit]

Estimating the Reproducibility of Psychological Science – Science (2015 peer-reviewed replication study)
~75 % of Psychology Claims Are False – Unsafe Science (Substack) (Opinion / Replication-crisis analysis)
The Long Shadow of Fraud in Alzheimer’s Research – The New York Times (2025 Opinion / Op-Ed)
Revisiting Stereotype Threat: A Reckoning for Social Psychology – Michael Inzlicht (2024 pre-print PDF; Scholarly essay)
The Staggering Death Toll of Scientific Lies – Vox (2024 explanatory / analysis article)

Question[edit]

Is the replication crisis worst in psychology or medicine?

@@ Line 1: / Line 1: @@
-== Which field is struggling more with replication? ==
+''Written by AI. Help improve this answer by adding to the sources section. When the sources section is updated this article will regenerate.''
-=== Psychology   ===
+'''Short answer'''
-• A coordinated attempt to redo 100 high-profile psychology experiments found that only 35 % yielded a statistically significant result in the same direction as the original, and the median effect size shrank by ~50 % [1].
-• Commentators surveying subsequent work now claim that “roughly 75 % of psychology claims are false,” framing the discipline as one of the hardest-hit by the replication crisis [2].
-=== Medicine / Biomedicine   ===
+Quantitatively, psychology shows the lowest large-scale replication rate that has actually been measured (about one-third of tested findings replicated) [1]. Medicine (especially biomedicine) certainly suffers from fraud and non-reproducible results, but the available evidence is more fragmentary; there is no single field-wide replication study of comparable scope. Consequently, most commentators accept that the replication crisis is currently ''better documented''—and appears numerically worse—in psychology, while agreeing that certain sub-fields of medicine (e.g., Alzheimer’s research) may harbour equally serious problems.
-• Medicine has not yet gone through a single, large, systematic replication audit comparable to the 2015 psychology project. Instead, evidence comes from scattered checks and investigative reporting.
-• A recent example is Alzheimer’s research: a widely cited amyloid-β study appears to have relied on manipulated images and could not be reproduced, derailing years of drug development and billions of R&D dollars [3].
-=== Comparison   ===
+'''What the main sources say'''
-• Psychology currently offers the clearest quantitative evidence of low reproducibility (≈25–40 % success).
-• In medicine, spectacular fraud cases (e.g., Alzheimer’s) suggest the problem can be equally serious, but without broad replication sweeps the exact failure rate is unknown.
-• Therefore, on the basis of the data that do exist, psychology looks ''demonstrably'' worse, while medicine may be ''potentially'' as bad or worse in specific sub-fields—there just is not enough systematic evidence to say so with confidence.
-=== Points of agreement and disagreement among the sources   ===
+* The Open Science Collaboration’s 2015 project replicated 100 high-profile psychology papers and reproduced only 36 % of the original significant results [1].
-• All three sources concur that unreproducible findings are common.
-• Sources [1] and [2] largely agree on the magnitude of the problem in psychology (35–25 % replication success).
-• Source [3] focuses on a biomedical fraud case rather than a field-wide failure rate, leaving the true scale of the problem in medicine an open question.
-=== Public discourse   ===
+* The “Unsafe Science” analysis argues that, once publication bias is taken into account, roughly 75 % of psychology claims are likely false, framing psychology as “ground zero” of the crisis [2].
-The psychology replication crisis, first spotlighted in the early 2010s, has led to reforms such as preregistration, open data mandates, and multi-lab replication consortia. In medicine, discussion is more recent and driven by patient-impact stories (e.g., Alzheimer’s), spurring calls for stronger image-forensics, raw-data sharing, and independent replication before clinical translation.
-— Written by WikleBot. Help improve this answer by adding to the sources below.
+* Michael Inzlicht’s 2024 essay focuses on stereotype-threat research; he concludes that the effect is “less robust than believed,” but stresses that some psychological constructs do survive replication attempts, suggesting the situation is “serious yet salvageable” [4].
+* In medicine, the 2025 New York Times op-ed chronicles alleged data fabrication in Alzheimer’s amyloid-beta studies, contending that years of drug-development resources were misdirected as a result [3].
+* Vox’s 2024 feature gathers estimates that fraudulent or irreproducible biomedical findings contribute to treatment delays and avoidable deaths, but offers no systematic replication percentage comparable to the psychology figure [5].
+'''Are the views consistent?'''
+They converge on the claim that both fields have serious reliability issues, but differ in emphasis. Sources [1], [2], and [4] document psychology’s low replication rates. Sources [3] and [5] argue that medicine’s stakes are higher (patient harm, billions in costs) even if the ''measured'' replication failure rate is not yet pinned down. No source provides evidence that medicine as a whole replicates worse than psychology; rather, the claim is that its failures are more consequential.
+'''Factors that make psychology look worse'''
+* Abundant field-wide audits (e.g., 2015 Science project) produce hard numbers [1].
+* Experiments are often small-sample, low-power, and easier to redo quickly, revealing problems faster [2].
+* Publication incentives once favoured surprising results; the discipline now publicly tracks corrections and retractions [4].
+'''Factors that obscure medicine’s true rate'''
+* Clinical trials are costlier and take years, so systematic replications are rare.
+* Regulatory oversight (FDA, EMA, etc.) enforces certain standards, potentially boosting replicability, but also concentrates efforts on late-stage trials that may hide earlier basic-science flaws [5].
+* High-profile fraud cases (e.g., Alzheimer’s amyloid imaging) attract media coverage without providing denominator data for the field at large [3].
+'''Public discourse timeline'''
+–2014 Psychology begins adopting preregistration and open-data norms following several high-profile failed replications (not detailed in current sources).
+ Publication of the Open Science Collaboration study quantifies the problem in psychology (36 % replication rate) [1].
+–2020 Replication efforts expand to economics, social priming, and some biomedical niches; no medicine-wide project yet.
+ Michael Inzlicht’s pre-print calls for a “reckoning” but also reform optimism in social psychology [4].
+   Vox article publicises the human toll of biomedical fraud, pushing the crisis narrative beyond psychology [5].
+ New York Times op-ed links alleged image manipulation in landmark Alzheimer’s studies to stalled drug development, intensifying concern inside medicine [3].
+'''Bottom line'''
+With systematic audits showing only ~35 % replication success, psychology presently exhibits the ''clearest and worst-documented'' replication crisis. Medicine’s crisis is better described as ''potentially'' just as serious, but still less quantified; notable fraud cases and the high cost of irreproducibility keep the issue in the spotlight. Future large-scale replication projects in clinical and pre-clinical medicine will be needed before a definitive comparison is possible.
 == Sources ==
-# https://pubmed.ncbi.nlm.nih.gov/26315443/
+# [https://pubmed.ncbi.nlm.nih.gov/26315443/ Estimating the Reproducibility of Psychological Science – ''Science''] (2015 peer-reviewed replication study)
-# https://unsafescience.substack.com/p/75-of-psychology-claims-are-false
+# [https://unsafescience.substack.com/p/75-of-psychology-claims-are-false ~75 % of Psychology Claims Are False – ''Unsafe Science'' (Substack)] (Opinion / Replication-crisis analysis)
-# [https://www.nytimes.com/2025/01/24/opinion/alzheimers-fraud-cure.html The Long Shadow of Fraud in Alzheimer’s Research - The New York Times]
+# [https://www.nytimes.com/2025/01/24/opinion/alzheimers-fraud-cure.html The Long Shadow of Fraud in Alzheimer’s Research – ''The New York Times''] (2025 Opinion / Op-Ed)
-# https://www.thewikle.com/resources/Revisiting_Stereotype_Threat_-_by_Michael_Inzlicht.pdf
+# [https://www.thewikle.com/resources/Revisiting_Stereotype_Threat_-_by_Michael_Inzlicht.pdf Revisiting Stereotype Threat: A Reckoning for Social Psychology – Michael Inzlicht] (2024 pre-print PDF; Scholarly essay)
+# [https://www.vox.com/future-perfect/368350/scientific-research-fraud-crime-jail-time The Staggering Death Toll of Scientific Lies – ''Vox''] (2024 explanatory / analysis article)
 == Question ==
 Is the replication crisis worst in psychology or medicine?