"In all five datasets, we investigated if rates of autism diagnosis significantly differed by gender by first conducting χ2 tests (Model 1, unadjusted), and the

"In all five datasets, we investigated if rates of autism diagnosis significantly differed by gender by first conducting χ2 tests (Model 1, unadjusted), and then by conducting logistic regressions adjusted for age and educational attainment as covariates (Model 2, adjusted)." It appears the kernel density estimates is being used as a display of the data, not the actual method of statistical analysis, or perhaps I missed something.