Why Most Published Research Findings Are False.¶
Ioannidis, J. P. A. (2005). Why Most Published Research Findings Are False. PLoS Medicine, 2(8).
Cited by¶
11 citations across 11 artifacts.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Epistemic Humility
- Scientific methodology: Designing experiments to falsify rather than confirm (the Popperian 1959 approach), accepting refutation and updating models, reporting confidence intervals and uncertainty quantification, publishing null results, recognizing publication bias and the replication crisis Ioannidis (2005) named in his analysis of why most published research findings are false.
This sourceArgues that low power, small effect sizes, flexible analysis, and bias make most positive findings non-replicable. SUPPORTS marker 216 (replication crisis / why most published research findings are false).
Supported in partVerified against the work's full text
Ioannidis argues most published findings are false and discusses selective reporting, but does not use the term replication crisis or cover Popperian falsification or confidence-interval reporting.
“There is increasing concern that most current published research findings are false.”
- Scientific methodology: Designing experiments to falsify rather than confirm (the Popperian 1959 approach), accepting refutation and updating models, reporting confidence intervals and uncertainty quantification, publishing null results, recognizing publication bias and the replication crisis Ioannidis (2005) named in his analysis of why most published research findings are false.
- Garbage In, Garbage Out
- In scientific research, derived results inherit primary-data quality, and the replication crisis is in part a GIGO story.
This sourceArgues that small studies, weak designs, bias, and selective reporting limit the reliability of primary findings, a structural driver of the replication crisis.
Supported in partVerified against the work's full text
Ioannidis shows bias, low power and selective analysis make many published findings false, linking this to nonreplication; he does not frame it as derived results inheriting primary-data quality.
“The extent that observed findings deviate from what is expected by chance alone would be simply a pure measure of the prevailing bias.”
- In scientific research, derived results inherit primary-data quality, and the replication crisis is in part a GIGO story.
- Hypothesis Testing (Null vs. Alternative)
- The frameworks agree in many applied settings (Bayesian analysis with diffuse priors often produces similar inferences to frequentist) but differ conceptually in what "probability" attaches to—long-run error rates in frequentism; degrees of belief in Bayesianism
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
- The frameworks agree in many applied settings (Bayesian analysis with diffuse priors often produces similar inferences to frequentist) but differ conceptually in what "probability" attaches to—long-run error rates in frequentism; degrees of belief in Bayesianism
- Parallel Independent Inspection
- In science, peer review, replication studies, multiple-author meta-analyses, and post-publication review distribute the inspection of a single artifact across diverse inspectors.
This sourcePeer review and replication as distributed inspection, and how correlated bias (a shared confound) degrades it.
Supported in partVerified against the work's full text
Ioannidis supports multiple teams and low-bias meta-analyses weighing the totality of evidence, but never discusses peer review or post-publication review as distributed inspection of an artifact.
“Second, most research questions are addressed by many teams, and it is misleading to emphasize the statistically significant findings of any single team. What matters is the totality of the evidence.”
- In science, peer review, replication studies, multiple-author meta-analyses, and post-publication review distribute the inspection of a single artifact across diverse inspectors.
- Reproducibility & Replicability
- The three-stage verification tells a clear story: (α) The original pattern reproduces in held-out data (not a spurious artifact), but (β) selection bias explains much of the observed effect (customers adopting the feature combination were pre-selected on lower-default-risk characteristics), and (γ) randomized promotion produces much smaller causal effects than observational analysis suggested
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
Supported in partVerified against the work's full text
Ioannidis supports the idea that many published findings fail to replicate and are false, argued by modelling, but does not document replication failure across psychology, medicine and economics.
“the high rate of nonreplication (lack of confirmation) of research discoveries is a consequence of the convenient, yet ill-founded strategy of claiming conclusive research findings solely on the basis of a single study”
- The three-stage verification tells a clear story: (α) The original pattern reproduces in held-out data (not a spurious artifact), but (β) selection bias explains much of the observed effect (customers adopting the feature combination were pre-selected on lower-default-risk characteristics), and (γ) randomized promotion produces much smaller causal effects than observational analysis suggested
- Selection Bias
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
- Speculative Bubble
- Science: Speculative research fashions, where a popular hypothesis or method attracts disproportionate funding, citations, and effort because it is already attracting them, then deflates when results fail to keep pace with expectation—a reflexive inflow of scholarly attention that overshoots the evidential base, a pattern Ioannidis (2005) connects to inflated false-positive rates in fields where investigative attention rushes ahead of replication.
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
Supported in partVerified against the work's full text
Ioannidis's Corollary 6 links 'hot' fields with many competing teams to lower odds that findings are true, but does not describe self-reinforcing attention outrunning replication.
“The hotter a scientific field (with more scientific teams involved), the less likely the research findings are to be true.”
- Science: Speculative research fashions, where a popular hypothesis or method attracts disproportionate funding, citations, and effort because it is already attracting them, then deflates when results fail to keep pace with expectation—a reflexive inflow of scholarly attention that overshoots the evidential base, a pattern Ioannidis (2005) connects to inflated false-positive rates in fields where investigative attention rushes ahead of replication.
- Statistical Power
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
- Statistical Significance (p-Value)
- The persistent misinterpretation of p-values as posterior probabilities is not a failure of education alone—it reflects that the Bayesian quantity is often the decision-relevant one while the frequentist p-value is the more easily computed
This sourceFoundational analysis of how publication bias, low statistical power, and flexible analytic choices produce a literature in which most positive findings fail to replicate—motivating epistemic humility about scientific claims.
- The persistent misinterpretation of p-values as posterior probabilities is not a failure of education alone—it reflects that the Bayesian quantity is often the decision-relevant one while the frequentist p-value is the more easily computed
- Technical Debt
- Scientific research — claims accepted without the replication that would have grounded them, and methods built on standards that later work must reconcile.
Supported in partVerified against the work's full text
Ioannidis shows findings get claimed as conclusive on single studies and often fail confirmation, but says nothing about methods built on standards that later work must reconcile.
“the high rate of nonreplication (lack of confirmation) of research discoveries is a consequence of the convenient, yet ill-founded strategy of claiming conclusive research findings solely on the basis of a single study assessed by formal statistical significance”
- Scientific research — claims accepted without the replication that would have grounded them, and methods built on standards that later work must reconcile.
Domain-specific¶
- File Drawer Problem
- Meta-analysis methodology — file-drawer correction as a routine diagnostic step: funnel-plot asymmetry, trim-and-fill, p-curve, Egger regression, and selection models estimating the hidden mass and the corrected effect. Clinical-trial registries — registration before data collection (clinicaltrials.gov, EU CTR) instituted precisely to make the conducted population visible regardless of outcome, eliminating the filter prospectively. The replication crisis — the file drawer as one proximate mechanism behind over-stated, non-replicating effects in psychology, biomedicine, and economics (the inflation of the published over the conducted literature)
Supported in partVerified against the work's full text
“Bias can entail manipulation in the analysis or reporting of findings. Selective or distorted reporting is a typical form of such bias.”
- Meta-analysis methodology — file-drawer correction as a routine diagnostic step: funnel-plot asymmetry, trim-and-fill, p-curve, Egger regression, and selection models estimating the hidden mass and the corrected effect. Clinical-trial registries — registration before data collection (clinicaltrials.gov, EU CTR) instituted precisely to make the conducted population visible regardless of outcome, eliminating the filter prospectively. The replication crisis — the file drawer as one proximate mechanism behind over-stated, non-replicating effects in psychology, biomedicine, and economics (the inflation of the published over the conducted literature)
Verification¶
Does it exist? Confirmed. This work's DOI resolves to a registered record, which fixes its identity. That is all it fixes.
Does it back the claim? Read against the text for 7 of 11 citations: 7 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:8842c316b057 · see in the full table