Medical research suggests that testosterone has positive effects on mood (t
hereby reducing the chances of depression), and social science research fin
ds testosterone to be related to antisocial behavior; risk behavior, unempl
oyment and low paying jobs, and being unmarried-factors known to be positiv
ely related to depression. Analysis of a sample of 4,393 men finds a parabo
lic model best fits the data. The relationship between testosterone and dep
ression is inverse for men with below average testosterone and direct for t
hose with above average testosterone. The relationship disappears for those
with above average testosterone when controls for antisocial and risk beha
viors and the absence of protective factors such as marriage and steady emp
loyment are in the equation. The relationship is unchanged for those with b
elow average testosterone. The results help explain the difference between
medical and social research findings. Mechanisms accounting for the finding
s are explored.