>need to rule out even slight correlations like smoking and (low) income.

Especially with hair dying, since looking at my environment at least early usage (before graying) is very much correlated with low income/education.