On the challenges of drawing conclusions from p -values just below 0.05
Open accessThe DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index.
Abstract
In recent years, researchers have attempted to provide an indication of the prevalence of inflated Type 1 error rates by analyzing the distribution of p-values in the published literature. De Winter & Dodou (2015) analyzed the distribution (and its change over time) of a large number of p-values automatically extracted from abstracts in the scientific literature. They concluded there is a 'surge of p-values between 0.041-0.049 in recent decades' which 'suggests (but does not prove) questionable research practices have increased over the past 25 years.' I show the changes in the ratio of fractions of p-values between 0.041-0.049 over the years are better explained by assuming the average power has decreased over time. Furthermore, I propose that their observation that p-values just below 0.05 increase more strongly than p-values above 0.05 can be explained by an increase in publication bias (or the file drawer effect) over the years (cf. Fanelli, 2012; Pautasso, 2010, which has led to a relative decrease of 'marginally significant' p-values in abstracts in the literature (instead of an increase in p-values just below 0.05). I explain why researchers analyzing large numbers of p-values need to relate their assumptions to a model of p-value distributions that takes into account the average power of the performed studies, the ratio of true positives to false positives in the literature, the effects of publication bias, and the Type 1 error rate (and possible mechanisms through which it has inflated). Finally, I discuss why publication bias and underpowered studies might be a bigger problem for science than inflated Type 1 error rates, and explain the challenges when attempting to draw conclusions about inflated Type 1 error rates from a large heterogeneous set of p-values.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Power failure: why small sample size undermines the reliability of neuroscience
Button, K. S., Ioannidis, J. P. A., Mokrysz, C., Nosek, B. A., Flint, J., Robinson, E. S. J., & Munafò, M. R. (2013) · Nature Reviews Neuroscience · doi:10.1038/nrn3475 · Cited by 7,641 · cited 1×
“At the same time, we can be certain power varies substantially across studies and research disciplines (e.g., Button et al., 2013 ; Sedlmeier & Gigerenzer, 1989 ), and therefore the p -value distribution can be more accurately modeled by summing multiple p -value distributions across di ff erent…”
BackgroundSimmons, J. P., Nelson, L. D., & Simonsohn, U. (2011) · Psychological Science · doi:10.1177/0956797611417632 · Cited by 6,567
Negative results are disappearing from most disciplines and countries
Fanelli, D. (2012) · Scientometrics · doi:10.1007/s11192-011-0494-7 · Cited by 1,248 · cited 6×
“Indeed, both Pautasso (2010) as Fanelli (2012) have provided support for the idea that negative results have been disappearing from the literature, which raises the possibility that the relative differences in p-values between 0.”
BackgroundResultInfluentialDo studies of statistical power have an effect on the power of studies?
Sedlmeier, P., & Gigerenzer, G. (1989) · Psychological Bulletin · doi:10.1037/0033-2909.105.2.309 · Cited by 748 · cited 2×
“At the same time, we can be certain power varies substantially across studies and research disciplines (e.g., Button et al., 2013 ; Sedlmeier & Gigerenzer, 1989 ), and therefore the p -value distribution can be more accurately modeled by summing multiple p -value distributions across di ff erent…”
Backgroundp -Curve and Effect Size: Correcting for Publication Bias Using Only Significant Results
Simonsohn, U., Nelson, L. D., & Simmons, J. P. (2014) · Perspectives on Psychological Science · doi:10.1177/1745691614553988 · Cited by 717
Registered Reports: A Method to Increase the Credibility of Published Results
Nosek, B. A., & Lakens, D. (2014) · Social Psychology · doi:10.1027/1864-9335/a000192 · Cited by 716 · cited 1×
“…recent years, but I believe it is worthwhile to point out that designing well-powered studies with high informational value (e.g., Lakens & Evers, 2014 ) and preventing publication bias (e.g., Nosek & Lakens, 2014 ) are at least, and I personally believe even more, important to improve our science.”
BackgroundSelf-citationThe alleged crisis and the illusion of exact replication
Stroebe, W., & Strack, F. (2014) · Perspectives on Psychological Science · doi:10.1177/1745691613514450 · Cited by 597
Cumming, G. (2008) · Perspectives on Psychological Science · doi:10.1111/j.1745-6924.2008.00079.x · Cited by 451
Performing high‐powered studies efficiently with sequential analyses
Lakens, D. (2014) · European Journal of Social Psychology · doi:10.1002/ejsp.2023 · Cited by 414
Self-citationKühberger, A., Fritz, A., & Scherndl, T. (2014) · PLoS ONE · doi:10.1371/journal.pone.0105825 · Cited by 399
A peculiar prevalence of p values just below .05
Masicampo, E. J., & Lalande, D. R. (2012) · Quarterly Journal of Experimental Psychology · doi:10.1080/17470218.2012.711335 · Cited by 266 · cited 6×
“…p = 0.05 observed by Leggett et al. (2013) does not replicate in other datasets of p -value distributions for the same journal in later years ( Masicampo & Lalande, 2012 ), in psychology in general ( Hartgerink et al., unpublished data ; K¨uhberger, Fritz & Scherndl, 2014 ), or in scientific…”
BackgroundResultInfluentialLakens, D., & Evers, E. R. K. (2014) · Perspectives on Psychological Science · doi:10.1177/1745691614528520 · Cited by 266 · cited 2×
“…very salient in recent years, but I believe it is worthwhile to point out that designing well-powered studies with high informational value (e.g., Lakens & Evers, 2014 ) and preventing publication bias (e.g., Nosek & Lakens, 2014 ) are at least, and I personally believe even more, important to…”
BackgroundSelf-citationMeta-analysis using effect size distributions of only statistically significant studies.
van Assen, M. A. L. M., van Aert, R. C. M., & Wicherts, J. M. (2015) · Psychological Methods · doi:10.1037/met0000025 · Cited by 265
Worsening file-drawer problem in the abstracts of natural, medical and social science databases
Pautasso, M. (2010) · Scientometrics · doi:10.1007/s11192-010-0233-5 · Cited by 136 · cited 2×
“Indeed, both Pautasso (2010) as Fanelli (2012) have provided support for the idea that negative results have been disappearing from the literature, which raises the possibility that the relative di ff erences in p -values between 0.041–0.049 and 0.051–0.059 observed by De Winter & Dodou (2015) are…”
BackgroundResultInfluentialde Winter, J. C., & Dodou, D. (2015) · PeerJ · doi:10.7717/peerj.733 · Cited by 78
The Life of p : “Just Significant” Results are on the Rise
Leggett, N. C., Thomas, N. A., Loetscher, T., & Nicholls, M. E. R. (2013) · Quarterly Journal of Experimental Psychology · doi:10.1080/17470218.2013.863371 · Cited by 64
Cited by
Anderson, S. F., Kelley, K., & Maxwell, S. E. (2017) · Psychological Science · doi:10.1177/0956797617723724 · Cited by 406
p-Curve and p-Hacking in Observational Research
Bruns, S. B., & Ioannidis, J. P. A. (2016) · PLOS ONE · doi:10.1371/journal.pone.0149144 · Cited by 210
“However, other empirical surveys have documented an increased prevalence of p -values of 0.041 – 0.049 in the scientific literature over time [4,5], and the spurious excess of statistically significant findings in various types of both observational and experimental research [6,7] that have been…”
BackgroundInfluentialBishop, D. V. M., & Thompson, P. A. (2016) · PeerJ · doi:10.7717/peerj.1715 · Cited by 90
“…p-values in the .04-.05 range is small (see Head et al, 2015, Table 3) More general problems with drawing inferences from binomial tests on p-curves Lakens (2015) noted that to model the distribution of p-values we need to know the number of studies where the null hypothesis or alternative…”
Background
Metadata
- Authors
- Lakens, D.
- Venue
- PeerJ
- Publisher
- PeerJ
- Published
- 2015-07-30
- Volume
- 3
- Pages
- e1142
- ISSN
- 2167-8359 (ISSN-L)
- OpenAlex
- W934038905
- Cited by
- 44
- References
- 16
- FWCI
- 3.29
- Type
- journal article
- Language
- en
- Open access
- gold
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.
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
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.