GPOWER: A general power analysis program
Open accessThe DOI matches its Crossref record; 55 of the 61 references it cites were found in a registry or an index; none of the 30 sources its claims cite is checked for support yet.
Overview
CoChat Summary
TL;DR
GPOWER is a menu-driven power-analysis program for IBM-compatible and Macintosh computers that computes a priori, post hoc, and compromise power for common behavioral-research tests. The paper argues it improves on existing tables and tools by offering higher-precision calculations, broader options, and graphical display of results.
Key findings
- GPOWER supports seven test types: t tests for means, t tests for correlations, other t tests, F tests in fixed-effects ANOVAs, F tests in multiple regression/correlation analyses, other F tests, and χ² tests.
- It performs three kinds of power analyses: post hoc power for given sample size/effect size/α, a priori sample-size planning for given effect size/α/power, and compromise analyses for given sample size/effect size/β:α ratio.
- The authors report that, compared with Cohen’s tables, GPOWER and Cohen (1988) “agree quite well”; power differences larger than .03 were rare, and for their accuracy-mode validation, GPOWER’s results were correct up to 5 significant digits when inputs were not too extreme.
- In validation against published or reference values, the paper reports perfect 4-digit agreement in 61 of 65 cases for χ² power, perfect 3-digit agreement in 22 of 24 cases for F power, 599 of 600 exact agreement cases for 6-digit noncentral t power, and 140 of 140 agreement cases for 6-digit χ² power; for F power there were 32 differences out of 1,440 comparisons, none affecting the first 5 digits.
Methods
- The paper is primarily a program description and validation study, not an experiment with human participants.
- It compares GPOWER’s accuracy-mode calculations with published values from Patnaik (1949), Cohen (1988), and results from SAS routines for noncentral distributions.
- The main analysis is numerical comparison of power values across test families and parameter settings, including checks of agreement at the digit level and examples showing where speed-mode approximations can diverge from accuracy-mode results.
Limitations
- Stated in the paper: the speed mode may be very misleading for extreme values of the parameters; the authors therefore recommend basing sample-size or critical-value decisions on accuracy mode calculations.
- Stated in the paper: a priori analyses are not available in the “Other t Tests” and “Other F Tests” options because sample size \(N\) is not linked to degrees of freedom in those procedures.
- Evident from the text: the program assumes users are already familiar with statistical power concepts and Cohen’s effect-size conventions; this is a usability limitation because the paper says the present versions “assume” that background.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 7 Not valid · 25 Not checked · 69 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“For example, the speed mode of GPOWER calculates a power of .8340 for one-tailed correlation t tests based on N = 8 pairs of values (thus, df = 6), α = .01, and a very large population correlation (ρ = 0.9).”
plainn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“GPOWER computes (1) power values for given sample sizes, effect sizes and α levels (post hoc power analyses);”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Absolutely no differences in 140 comparisons were observed for 6-digit χ² power values.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Again, however, none of the 32 differences from a total of 1,440 comparisons concerned the first 5 digits.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Substituting n' for n in Equation 1 shows that this adjustment indeed leads to the correct denominator degrees of freedom (v) in all possible cases.”
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
32 more claims
Create a free account to see all 101 claims and to verify the 30 references still open.
- Probably not a substantial claim. 58
- It may state two separate claims as one. 17
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The results established a need for nine participants per group (f = 0.25 and actual power = 80.70%) to detect differences with an assumed Type I error of 0.05 and a Type II error rate of 0.20 (statistical power = 80%).”
From Effect of water-based aerobic training on anthropometric, biochemical, cardiovascular, and explosive strength parameters in young overweight and obese women: a randomized controlled trial (2025) · cites this paper as Erdfelder, Faul & Buchner, 1996
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The procedures provided by G*Power 3 to test effects in between-subjects designs with more than two groups (i.e., one-way ANOVA designs and general main effects and interactions in factorial ANOVA designs of any order) are identical to those in G*Power 2 (Erdfelder et al., 1996).”
From G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences (2007) · cites this paper as Erdfelder et al., 1996
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“We determined the sample size based on a power analysis using G*Power (Erdfelder, Faul, & Buchner, 1996).”
From Emotion understanding and the moral self-concept as motivators of prosocial behavior in middle childhood (2020) · cites this paper as Erdfelder et al., 1996
The cited source says what the claim attributes to it.
In the citing paper
“A power analysis in G*Power35 indicated that based on an alpha level of 0.05, a logistic regression model with 40 participants yields 80% power to detect a moderate effect size (odds ratio [OR] 3.1) and 96% power to detect a large effect size (OR 5).”
From Automated Speech Analysis to Differentiate Frontal and Right Anterior Temporal Lobe Atrophy in Frontotemporal Dementia (2025) · cites this paper as [35]
The cited source says what the claim attributes to it.
In the citing paper
“Based on the results of the study of Shariat Moghani et al. [41], the sample size was calculated 39 people by using the G-power software [42] for stress variable with considering the M1 = 7.07 (mean score of stress subscale of DASS), M2 = 4.59 (assuming 35% reduction due to intervention), SD1 = SD2 = 4.33, one-sided α = 0.05, and Power = 80%. There were 43 people in each group in the final sample”
From Effect of acceptance and commitment therapy on mood, sleep quality and quality of life in menopausal women: a randomized controlled trial (2022) · cites this paper as [42]
3 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
Statistical Power Analysis for the Behavioral Sciences
Muller, K. (1989) · Technometrics · doi:10.1080/00401706.1989.10488618 · Cited by 99,167
Statistical power analysis for the behavioral sciences
(1990) · Computers, Environment and Urban Systems · doi:10.1016/0198-9715(90)90050-4 · Cited by 65,269
Statistical Power Analysis for the Behavioral Sciences (2nd ed.).
Lachenbruch, P. A., & Cohen, J. (1989) · Journal of the American Statistical Association · doi:10.2307/2290095 · Cited by 37,395
Statistical Power Analysis for the Behavioral Sciences
Cohen, J. (2013) · doi:10.4324/9780203771587 · Cited by 25,373
Numerical recipes in C: the art of scientific computing
(1993) · Choice Reviews Online · doi:10.5860/choice.30-5638 · Cited by 17,933
Cohen, J. (1992) · Psychological Bulletin · doi:10.1037//0033-2909.112.1.155 · Cited by 8,771
Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences.
Tufte, E. R., Cohen, J., & Cohen, P. (1979) · Journal of the American Statistical Association · doi:10.2307/2286442 · Cited by 5,180
Introduction to Statistical Analysis.
Latshaw, V. V., Dixon, W. J., & Massey, F. J. (1957) · The American Mathematical Monthly · doi:10.2307/2309989 · Cited by 5,072
Numerical Recipes--The Art of Scientific Computing.
Fritsch, F. N., Press, W. H., Flannery, B. P., Teukolsky, S. A., & Vetterling, W. T. (1988) · Mathematics of Computation · doi:10.2307/2007938 · Cited by 4,845
Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences.
Fong, G. T., Cohen, J., & Cohen, P. (1985) · Journal of the American Statistical Association · doi:10.2307/2287930 · Cited by 4,475
Research works printed but not linked: 10
- Bargmann, R. E., &Ghosh, S. P. (1964).Noncentral statistical distribution programs for a computer language (IBM Research Report RC-1231). Yorktown Heights, NY: IBM Watson Research Center.
- Bredenkamp, J. (1972).Der Signifikanztest in der psychologischen Forschung [The test of significance in behavioral research]. Frankfurt, Germany: Akademische Verlagsgesellschaft.
- Bredenkamp, J., &Erdfelder, E. (1985).Multivariate Varianzanalyse nach dem V-Kriterium [Multivariate analysis of variance using the V-criterion].Psychologische Beiträge,27, 127–154.
- Buchner, A., Faul, F., &Erdfelder, E. (1992).GPOWER: A priori-, post hoc-, and compromise power analyses for the Macintosh [Computer program]. Bonn, Germany: Bonn University.
- Erdfelder, E. (1984). Zur Bedeutung und Kontrolle des β-Fehlers bei der inferenzstatistischen Prüfung log-linearer Modelle [On significance and control of the β error in statistical tests of log-linear models].Zeitschrift für Sozialpsychologie,15, 18–32.
- Faul, F., &Erdfelder, E. (1992).GPOWER: A priori-, post hoc-, and compromise power analyses for MS-DOS [Computer program]. Bonn, Germany: Bonn University.
- Hardison, C. D., Quade, D., &Langston, R. D. (1983). Nine functions for probability distributions. In SAS Institute, Inc. (Ed.),SUGI supplemental library user’s guide, 1983 edition (pp. 229–236). Cary, NC: SAS Institute, Inc.
- Koele, P., &Hoogstraten, J. (1980).Power and sample size calculations in analysis of variance (Révész Berichten No. 12). Amsterdam: University of Amsterdam.
- Onghena, P. (1994).The power of randomization tests for single-case designs. Unpublished doctoral dissertation. Leuven, Belgium: Katholieke Universiteit Leuven.
- Patnaik, P. B. (1949). The non-central χ2- andF-distributions and their applications.Biometrika,36, 202–232.
Cited by
Items 1,001–1,022 of 1,022
Monfaredi, Z., Malakouti, J., Farvareshi, M., & Mirghafourvand, M. (2022) · BMC Psychiatry · doi:10.1186/s12888-022-03768-8 · Cited by 36
“[41], the sample size was calculated 39 people by using the G-power software [42] for stress variable with considering the M1 = 7.”
MethodHow much do focal infarcts distort white matter lesions and global cerebral atrophy measures?
Wang, X., Valdés Hernández, M. C., Doubal, F., Chappell, F. M., & Wardlaw, J. M. (2012) · Cerebrovascular Diseases · doi:10.1159/000343226 · Cited by 31
“G * Power 3 was used for sample size calculations [13] .”
MethodTorres, J. R., Martos, I. C., Sánchez, I. T., Rubio, A. O., Pelegrina, A. D., & Valenza, M. C. (2015) · Archives of Physical Medicine and Rehabilitation · doi:10.1016/j.apmr.2015.06.008 · Cited by 26
Association of amygdala volumes with cortisol secretion in unipolar depressed patients
Schuhmacher, A., Mössner, R., Jessen, F., Scheef, L., Block, W., Belloche, A. C., Lennertz, L., Welper, H., Höfels, S., Pfeiffer, U., et al. (2012) · Psychiatry Research: Neuroimaging · doi:10.1016/j.pscychresns.2011.09.007 · Cited by 25
Vila, M. R., Todorovic, M. S., Tang, C., Fisher, M., Steinberg, A., Field, B., Bottros, M. M., Avidan, M. S., & Haroutounian, S. (2020) · British Journal of Anaesthesia · doi:10.1016/j.bja.2020.02.002 · Cited by 22
Depression and social networks in community dwelling elders: a descriptive study
Wilby, F. (2011) · Journal of Gerontological Social Work · doi:10.1080/01634372.2010.540074 · Cited by 21
“de/aap/projects/gpower ( Erdfelder, Faul, & Buchner, 1996 ).”
Horigian, V. E., Weems, C. F., Robbins, M. S., Feaster, D. J., Ucha, J., Miller, M., & Werstlein, R. (2013) · The American Journal on Addictions · doi:10.1111/j.1521-0391.2013.12031.x · Cited by 21
Moritz, S., Spirandelli, K., Happach, I., Lion, D., & Berna, F. (2018) · Journal of the International Neuropsychological Society · doi:10.1017/s1355617718000097 · Cited by 17
de la Rosa, I. A., Perry, J., Dalton, L. E., & Johnson, V. (2005) · Research on Social Work Practice · doi:10.1177/1049731505277004 · Cited by 16
Fornaini, E., Matera, C., Nerini, A., Policardo, G. R., & Di Gesto, C. (2021) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph18179319 · Cited by 16
“A power analysis using G*Power [41] indicated that a sample of 211 was sufficient to detect medium effects (effect size = 0.25) with 80% power using an ANOVA with six groups (3 × 2) and with alpha set at 0.05.”
MethodInfluentialBe Friendly, Not Friends: How LLM Sycophancy Shapes User Trust
Sun, Y., & Wang, T. (2026) · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791079 · Cited by 16
“The measurements were rated on the 7-point Likert scale (1 = Strongly Disagree, 7 = Strongly Agree) unless otherwise noted.”
Telles, S., Vishwakarma, B., Gupta, R. K., & Balkrishna, A. (2019) · Medical Science Monitor Basic Research · doi:10.12659/msmbr.914956 · Cited by 16
Cerea, S., Sansoni, M., Scarzello, G., Groff, E., & Ghisi, M. (2022) · Supportive Care in Cancer · doi:10.1007/s00520-022-07334-6 · Cited by 15
Living in fear: Low-cost avoidance maintains low-level threat.
Xia, W., Eyolfson, E., Lloyd, K., Vervliet, B., & Dymond, S. (2019) · Journal of Behavior Therapy and Experimental Psychiatry · doi:10.1016/j.jbtep.2018.09.001 · Cited by 14
Quattropani, M. C., Geraci, A., Lenzo, V., Sardella, A., & Schimmenti, A. (2022) · Journal of Child & Adolescent Trauma · doi:10.1007/s40653-022-00450-9 · Cited by 13
Vonk, J. M. J., Morin, B. T., Pillai, J., Rosado Rolon, D., Bogley, R., Baquirin, D. P., Ezzes, Z., Tee, B. L., de Leon, J., Wauters, L., et al. (2025) · Neurology · doi:10.1212/wnl.0000000000213556 · Cited by 10
Lenzo, V., Sardella, A., Musetti, A., Petralia, M. C., Grado, I., & Quattropani, M. C. (2022) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph191911930 · Cited by 10
“Moreover, the Italian versions of the following self-report instruments were administered: The Hospital Anxiety and Depression Scale (HADS) [21,22] measures anxiety and depression at a given point in time.”
MethodBen Cheikh, I., Marzouki, H., Selmi, O., Cherni, B., Bouray, S., Bouhlel, E., Bouassida, A., Knechtle, B., & Chen, Y.-S. (2025) · PeerJ · doi:10.7717/peerj.19020 · Cited by 10
“Previously to the recruitment procedure, a sample size estimation was conducted using statistical software (G*Power software, version 3.1.9.4; University of Kiel, Kiel, Germany) ( Erdfelder, Faul & Buchner, 1996 ).”
InfluentialHoch, E., Scheiter, K., & Stalbovs, K. (2023) · British Journal of Psychology · doi:10.1111/bjop.12620 · Cited by 7
Stigma or Support? How Gender, Race, and Intersectionality Impact Perceptions of Exonerees
Bettens, T., Warren, A. R., & Pica, E. (2026) · Women & Criminal Justice · doi:10.1080/08974454.2024.2386654 · Cited by 4
“An a priori power analysis using G � Power (Erdfelder et al., 1996) indicated a required 199 participants to detect a small-to-moderate effect (0.20) with a power of 0.80 and alpha of 0.05.”
InfluentialThe accuracy of four approximations to noncentralF
Bradley, D. R., Russell, R. L., & Reeve, C. P. (1998) · Behavior Research Methods, Instruments, & Computers · doi:10.3758/bf03200682 · Cited by 2
Pe, S., Bergomi, L., Nicora, G., Simonelli, C. A., Kour, P., Diaz, E., Alendal, G., Hernáiz Ferrer, A. I., Corso, V., Bortolotto, C., et al. (2026) · Machine Learning and Knowledge Extraction · doi:10.3390/make8070216
“1) [52] was used to perform sample size calculations.”
Metadata
- Authors
- Erdfelder, E., Faul, F., & Buchner, A.
- Affiliations
- University of Bonn (DE)
- Christian-Albrechts-Universität zu Kiel (DE)
- Universität Trier (DE)
- Corresponding author
- Erdfelder, Edgar
- Faul, Franz
- Buchner, Axel
- Venue
- Behavior Research Methods, Instruments, & Computers
- Published
- 1996-03
- Volume
- 28
- Issue
- 1
- Pages
- 1-11
- ISSN
- 0743-3808 (ISSN-L)
- 1532-5970
- OpenAlex
- W2069863414
- Cited by
- 4,864
- References
- 61
- Citations by year
- 2026: 170 · 2025: 264 · 2024: 301 · 2023: 297 · 2022: 345 · 2021: 372 · 2020: 278 · 2019: 232 · 2018: 195 · 2017: 133 · 2016: 129 · 2015: 153 · 2014: 172 · 2013: 265 · 2012: 184
- FWCI
- 50.60
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- www.springer.com/tdm (tdm)
- Locations
- Behavior Research Methods, Instruments, & Computers (publishedVersion)
- MADOC (University of Mannheim) (submittedVersion)
- Topics
- Social Sciences › Decision Sciences › Statistics, Probability and Uncertainty › Meta-analysis and systematic reviews
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Clinical Trials
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Keywords
- statistical power analysis, post hoc power analysis, a priori power analysis, sample size determination, effect size, t-test, F-test, chi-square test, statistical software
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.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
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.