CoChat

Issues with data and analyses: Errors, underlying themes, and potential solutions

Open access
Brown, Andrew W., Kaiser, Kathryn A., Allison, David B. · 2018 · Proceedings of the National Academy of Sciences

The DOI matches its Crossref record; 95 of the 95 references it cites were found in a registry or an index; 22 of the 71 sources its claims cite were checked for support.

Abstract

Some aspects of science, taken at the broadest level, are universal in empirical research. These include collecting, analyzing, and reporting data. In each of these aspects, errors can and do occur. In this work, we first discuss the importance of focusing on statistical and data errors to continually improve the practice of science. We then describe underlying themes of the types of errors and postulate contributing factors. To do so, we describe a case series of relatively severe data and statistical errors coupled with surveys of some types of errors to better characterize the magnitude, frequency, and trends. Having examined these errors, we then discuss the consequences of specific errors or classes of errors. Finally, given the extracted themes, we discuss methodological, cultural, and system-level approaches to reducing the frequency of commonly observed errors. These approaches will plausibly contribute to the self-critical, self-correcting, ever-evolving practice of science, and ultimately to furthering knowledge.

Overview

CoChat Summary

TL;DR

This paper reviews data and statistical errors in empirical research, categorizes their themes and contributing factors, and proposes methodological, cultural, and system-level remedies. It argues that many invalidating errors are common, consequential, and often correctable, but that current correction processes are inconsistent and slow.

Key findings

  • The authors define “invalidating errors” as factual mistakes or substantial deviations from accepted procedures that, if corrected, might alter a paper’s conclusions, and they group errors into themes including bad data, data management, statistical analysis, logic, and communication.
  • Cited prevalence examples include: only 28% of 149 sampled medical journal studies were considered “acceptable”; 76% of conclusions or abstracts in 196 rheumatoid arthritis drug trials contained “doubtful or invalid statements”; 55% of psychology articles had errors, with 18% having gross errors; and “statcheck” found one in eight psychology articles had errors that could affect conclusions.
  • Specific statistical problems discussed include misanalysis of cluster-randomized trials, incorrect effect-size calculations in meta-analyses (errors found in 17 of 27 meta-analyses), and the “differences in nominal significance” error; consequences include retractions and an estimated $28 billion spent on irreproducible preclinical research in the United States in 2015, about half of all preclinical research.
  • Proposed responses include commenting on studies rather than authors, respecting due process, developing uniform and timely error-correction procedures, and using education, gatekeeper functions, incentives, resources, and cultural shifts to normalize error correction.

Methods

  • Design: a review/commentary that presents a case series of relatively severe data and statistical errors coupled with surveys of some error types to characterize magnitude, frequency, and trends.
  • Data or sample: a nonsystematic evaluation of the literature and selected published examples, drawing on cited surveys and meta-research across medicine, psychology, nutrition, and obesity research.
  • Main analysis: narrative thematic synthesis of error types, contributing factors, consequences, and potential solutions, rather than a systematic review or pooled quantitative analysis.

Limitations

  • The paper states that its provisional themes represent the authors’ opinions from their own vantage point and that there is no mathematical proof these are the best themes; the evaluation of the literature is nonsystematic.
  • The paper states that categorizing errors requires a degree of subjectivity and that others have struggled with it; it also explicitly does not specifically address fabrication and falsification because they are defined by intentionality and egregiousness.
  • Evident from the text: the prevalence figures are drawn from heterogeneous cited studies with different methods and fields, so they are not presented as pooled or systematically estimated rates.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1073/pnas.1708279115

Cites

95 of 95 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

16 Valid · 6 Not valid · 36 Not checked · 70 Own results and interpretations

34 more claims · 7 with their cited sources checked

Create a free account to see all 128 claims and to verify the 49 references still open.

See all claims — free account
  • Probably not a substantial claim. 71
  • It may state two separate claims as one. 41

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

2 retracted

Show all 95 references

Research works printed but not linked: 4

  1. ; American Council on Science and Health From Mice to Men: The Benefits and Limitations of Animal Testing in Predicting Human Cancer Risk (American Council on Science and Health, 3rd Ed, New York, 1991).
  2. F Benford, The law of anomalous numbers. Proc Am Philos Soc 78, 551–572 (1938).
  3. DS Chawla “We should err on the side of protecting people’s reputation:” Management journal changes policy to avoid fraud—interview with Patrick Wright. Retraction Watch . Available at retractionwatch.com/2016/07/13/management-journal-changes-its-policy-to-avoid-fraud-qa-with-editor-patrick-wright/. Accessed June 22 2016. (2016).
  4. M Scudellari “Strange. Very strange:” Retracted nutrition study reappears in new journal. Retraction Watch . Available at retractionwatch.com/2017/03/28/strange-strange-retracted-nutrition-study-reappears-new-journal/. Accessed September 22 2017. (2017).

Cited by

See all 159

Metadata

Authors
Brown, A. W., Kaiser, K. A., & Allison, D. B.
Affiliations
University of Alabama at Birmingham (US)
Venue
Proceedings of the National Academy of Sciences
Published
2018-03-12
Volume
115
Issue
11
Pages
2563-2570
ISSN
  • 0027-8424 (ISSN-L)
  • 1091-6490
OpenAlex
W2790042388
Cited by
157
References
95
Citations by year
2026: 7 · 2025: 17 · 2024: 15 · 2023: 24 · 2022: 34 · 2021: 26 · 2020: 18 · 2019: 11 · 2018: 4
FWCI
23.80
Type
journal article
Language
en
Open access
Locations
  • Proceedings of the National Academy of Sciences (publishedVersion)
  • PubMed Central (publishedVersion)
  • PubMed (publishedVersion)
  • IUScholarWorks Open (Indiana University) (submittedVersion)
Funders
  • HHS | NIH | National Institute of Diabetes and Digestive and Kidney Diseases (R25DK099080, P30DK056336)
  • HHS | NIH | National Institute of General Medical Sciences (R25GM116167)
  • HHS | NIH | National Heart, Lung, and Blood Institute (R25HL124208)
  • HHS | NIH | National Institute on Aging (P30AG050886)
Topics
  • Physical Sciences › Computer Science › Information Systems › Research Data Management Practices
  • Social Sciences › Decision Sciences › Statistics, Probability and Uncertainty › Meta-analysis and systematic reviews
  • Social Sciences › Education › Academic integrity and plagiarism
Keywords
data errors, statistical errors, research errors, error reduction, data collection, data analysis, scientific rigor, research culture
MeSH terms
Scientific Experimental Error, Data Collection, Humans, Quality Control, Research Design, Science, Statistics as Topic, Reproducibility of Results

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-04, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-04 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Issues with data and analyses: Errors, underlying themes, and potential solutions. https://app.cochat.ai/papers/10.1073/pnas.1708279115
The paper
Brown, A. W., Kaiser, K. A., & Allison, D. B. (2018). Issues with data and analyses: Errors, underlying themes, and potential solutions. Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.1708279115

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.