Power Posing: Brief Nonverbal Displays Affect Neuroendocrine Levels and Risk Tolerance
Psychological Science, 2010 · paper · Psychology
Tests whether holding expansive "high-power" versus contracted "low-power" poses changes hormones and behavior. The authors report that high-power posers showed higher testosterone, lower cortisol, greater feelings of power and more risk tolerance.
Checked: PDF from bu.edu, fetched 5 Oct 2026. Page numbers are those of that PDF.
2 findings
Reference or label Reference "Galinsky, A.D., Gruenfeld, D.H., & Magee, J.C. (2003). From power to action. Journal of Personality and Social Psychology, 87, 327–339." gives a volume inconsistent with its year.
- p. 7 References
to action. Journal of Personality and Social Psychology, 87, 327–339.
- p. 7 References, Briñol & Petty
Psychology, 84, 1123–1139.
- p. 7 Reference list
Journal of Personality and Social Psychology, 87, 327–339.
- p. 7 References
Galinsky, A.D., Gruenfeld, D.H., & Magee, J.C. (2003). From power to action. Journal of Personality and Social Psychology, 87, 327–339.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that JPSP volume 87 is the 2004 volume and that the 2003 article is in vol. 85, pp. 453–466. Within the file, the p.7 image shows this entry with vol. 87, while Briñol & Petty (2003), in the same journal and year, is vol. 84. The Galinsky et al. entry is cited as 2003 on p.2.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference "Dabbs, J.M. (1991). Salivary testosterone measurements... Physiology & Behavior, 42, 815–817." appears to give the wrong volume (49).
- p. 7 Reference list
Physiology & Behavior, 42, 815–817.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that Dabbs (1991) appeared in Physiology & Behavior vol. 49, pp. 815–817. The p.7 image shows vol. 42. The files cannot settle this; the same journal's 1979 entry (Coe et al.) is vol. 23, which only roughly suggests vol. 42 would be earlier than 1991.
Second check: another model agreed
Report a problem with this finding
8 minor (typos and the like; most checked by the judge only)
Wording Hormone tests "F(1, 39) = 4.29" and "F(1, 38) = 7.45" have error df equal to N minus 2 after the Note 2 exclusions (41−2, 40−2), with no df spent on the stated covariates (sex, Time 1 level, the other hormone); also, the abstract's "for both male and female participants" is not supported by any sex-specific analysis reported.
- p. 5 Results, testosterone and cortisol
F(1, 39) = 4.29, p < .05; r = .34
- p. 5 Methods, covariates
we used participant’s sex as a covariate in all analy- ses
- p. 2 Abstract
behavioral changes for both male and female participants
- p. 5 Results
F(1, 38) = 7.45, p < .02; r = .43
Only the authors can say what is right here.
Judge's note: p.5: "we used participant's sex as a covariate in all analyses... controlling for Time 1... Analyses with cortisol controlled for testosterone, and vice versa." Note 2 (p.7) excludes 1 testosterone outlier and 2 cortisol outliers. A model with these covariates would have about 36 and 35 error df, not 39 and 38. The abstract (p.2) makes a claim about both sexes, but the results report only pooled analyses with sex as a covariate. The r-value parts of this cluster (notes 3–4) are handled under C008.
Second check: another model agreed
Report a problem with this finding
Wording Pretest effect size "t(94) = 21.03, p < .001; r = .99" does not correspond to the printed t and df under the t-to-r conversion used elsewhere in the paper (≈ .91).
- p. 3 Power poses, pretest
t(94) = 21.03, p < .001; r = .99
Only the authors can say what is right here.
Judge's note: Checked p.3 image. r² from t: 21.03²/(21.03²+94) = .825, so r ≈ .908, not .99 (r² = .98). The paper's other effect sizes do follow sqrt(F/(F+df)): .55, .60, and Φ = .30 all match. A d-based r from the printed means and SDs gives about .84, which is not .99 either. The paper does not define the r metric for this pretest, so the authors should say how .99 was computed.
Second check: another model agreed
Report a problem with this finding
Wording "we had 19 pretest participants rate the comfort..." but tests reported as "t(16) = 0.24 ... t(16) = 0.77 ... t(16) = –0.82"; the df fit neither a paired test on 19 people (df 18) nor an independent test (df 17), and no missing data are mentioned.
- p. 3 Power poses, comfort pretest
poses on comfort, t(16) = 0.24, p > .80
- p. 3 Method, Power poses, second pretest
comfort, t(16) = 0.24, p > .80; difficulty, t(16) = 0.77
- p. 3 Power poses, comfort pretest
we had 19 pretest participants rate the comfort
- p. 3 Power poses, comfort pretest
comfort, t(16) = 0.24
Only the authors can say what is right here.
Judge's note: p.3 image confirms N = 19 and t(16) in all three tests, and that all participants made all four poses (within-subject design). I found no statement anywhere in the text about exclusions or missing pretest data. Only the authors can say what the analysed N was.
Second check: another model agreed
Report a problem with this finding
Wording Error df of 41 in "F(1, 41) = 17.40", "F(1, 41) = 22.55" and "F(1, 41) = 9.53" do not fit N = 42 in two groups (df 40); for testosterone, Note 2 also excludes one outlier, which would give df 39.
- p. 5 Saliva analysis, sex differences
Time 1, F(1, 41) = 17.40, p < .001, r = .55
- p. 5 Results, feelings of power
F(1, 41) = 9.53
- p. 5 Saliva analysis, sex differences
F(1, 41) = 17.40
Only the authors can say what is right here.
Judge's note: Checked p.5 (results and saliva analysis) and Note 2 on p.7. The method gives N = 42 (26 female, 16 male), so a two-group comparison has df 40, and fewer for testosterone after the exclusion in Note 2. I found nothing that explains df = 41. The r values match either df.
Second check: another model agreed
Report a problem with this finding
Wording Hormone effect sizes "F(1, 39) = 4.29 ... r = .34" and "F(1, 38) = 7.45 ... r = .43" do not correspond to the printed F and df under sqrt(F/(F+df)) (≈ .31 and ≈ .40), the conversion that reproduces the paper's sex-difference r values.
- p. 5 Results
F(1, 38) = 7.45, p < .02; r = .43
- p. 5 Results
F(1, 39) = 4.29, p < .05; r = .34
Only the authors can say what is right here.
Judge's note: p.5 image checked. r² from F: 4.29/43.29 = .099 (r ≈ .315) and 7.45/45.45 = .164 (r ≈ .405). Both printed r values would instead imply about 33 error df, so they may come from a covariate model whose df differ from those printed (see C005). The paper does not say how r was computed; the authors should state which model and df each r comes from.
Second check: another model agreed
Report a problem with this finding
Wording The additional study is described as "(N = 49)" but reports "F(1, 48) = 4.38, p < .05, r = .30". A two-group comparison of 49 participants has error df = 47, and r from these values is about .29.
- p. 6 Discussion, additional study
F(1, 48) = 4.38, p < .05, r = .30
- p. 6 Discussion, additional study
findings from an additional study (N = 49)
Only the authors can say what is right here.
Judge's note: Checked page 6. This is the same N − 1 pattern as in the main study, and it is not explained.
Second check: another model agreed
Report a problem with this finding
Typo In-text citation "(Strack, Martin, Stepper, 1988)" lacks the ampersand; it should read "Strack, Martin, & Stepper, 1988".
- p. 3 Introduction, embodied cognition paragraph
(Strack, Martin, Stepper, 1988)
Judge's note: The p.3 image shows the comma-only form. The reference list on p.7 reads "Strack, F., Martin, L.L., & Stepper, S. (1988)", and other three-author citations in the text use &.
Second check: none (judge only)
Report a problem with this finding
Typo "86.36% took the gambling risk (only 13.63% were risk averse)": 3/22 = 13.636% rounds to 13.64%; as printed the two shares sum to 99.99%.
- p. 5 Results
(only 13.63% were risk averse)
- p. 5 Results, gambling task
86.36% took the gambling risk (only 13.63% were risk averse)
Judge's note: The p.5 image confirms both values. 86.36% matches 19/22, so its complement 3/22 should be 13.64%. The printed value looks truncated rather than rounded.
Second check: none (judge only)