Effects of competition outcome on testosterone concentrations in humans: An updated meta-analysis.
Level 1 - systematic review of randomized trials
Meta-analysis of human experimental and observational studies on competition outcome.
PubMed 27720891 · doi:10.1016/j.yhbeh.2016.10.002
What was done
An updated meta-analysis of human studies published over 35 years examining the effect of competition outcome (winning vs. losing) on acute changes in testosterone concentrations. The analysis synthesized 60 effect sizes from >2,500 participants, evaluating sex differences and moderators including study context (laboratory vs. field/sports venues) and the timing of pre-competition baseline hormone sampling.
What was found
Across all studies, winners had a larger increase in testosterone concentrations relative to losers (Cohen's D = 0.20), though heterogeneity was high. Context substantially moderated the effect: - Field settings (e.g., sport venues): D = 0.43 - Laboratory settings: D = 0.08 overall; D = 0.15 in men, and D = -0.04 in women - Baseline sampling timing: Sampling pre-competition testosterone >10 minutes before competition showed a stronger effect (trim-and-fill adjusted D = 0.38) than sampling within 10 minutes of competition start (D = 0.09).
Why it matters
This provides updated quantitative evidence that the human winner-loser testosterone effect is largely context-dependent, appearing reliably in real-world sports competitions but remaining negligible in artificial laboratory tasks, particularly in women. It also establishes that sampling baseline testosterone too close to competition onset obscures outcome-related hormonal changes.
Limits
The primary pooled effect showed substantial unexplained heterogeneity. The effect did not replicate in laboratory settings for women (D = -0.04) and was very small for men in labs (D = 0.15). The abstract does not specify the exact types of sports or tasks included, control for circadian variation, or detail publication bias metrics beyond mentioning trim-and-fill on one sub-analysis.