their confidence intervals are completely separate. This strongly suggests that the true means are genuinely different. (We'll formalize this idea in Chapter 23 with hypothesis testing.)