from the department of non-astonishing results

[chhotii]

If you set alpha=0.05, then 5% of your experiments will show a statistically significant effect of gender.

Comments

[quietann]

It does get a lot more complicated if you have >2 categories for gender, though. Very small n in the "other" category can present analysis difficulties...