I’ve never understood the argument about variance. The difference between two samples’ means can be statistically significant regardless of their respective variances.