Variance overlapping isn't a reason to discount this analysis. Also note that the variance narrows substantially for older age groups. So effectively what the variance says is that some younger people are vastly more responsible drivers than other young people but older people tend to be more responsible in general.