Comparing standardized effect sizes (Cohen's d with 95% CIs) between Carney, Cuddy & Yap (2010, N=42) and the high-powered preregistered replication by Ranehill et al. (2015, N=200).
Simulate 1,000 randomized trial draws under a true population null effect (true d = 0.0) to observe how small sample sizes (N=42) inflate false positive rates (Type I Error) compared to N=200.
| Outcome Dimension | Carney 2010 (N=42) | Ranehill 2015 (N=200) | Replication Verdict |
|---|---|---|---|
| Testosterone Level | d = +0.86, p = 0.05 | d = +0.06, p = 0.42 | Failed to Replicate |
| Cortisol Level | d = -0.89, p = 0.02 | d = -0.02, p = 0.88 | Failed to Replicate |
| Financial Risk-Taking | d = +0.64, p = 0.03 | d = +0.13, p = 0.34 | Failed to Replicate |
| Subjective Feeling of Power | d = +0.68, p = 0.01 | d = +0.41, p = 0.001 | Replicated (Self-Report) |
* Note: While subjective self-reported feelings of power showed moderate consistency, physiological biomarker changes (testosterone rise & cortisol reduction) and actual behavioral risk choices showed no significant effect in the high-powered preregistered study.