Sometimes, it’s interesting to ask a question about data in a vacuum, comparing estimators to particular population quantities, meaning parameter values in a presumed distribution. Mostly that’s a good warm-up in statistical textbooks. As a toolkit in practice, utility is limited. Far more often, data are collected from two or more populations which are then compared to one another to see if there’s a difference. For example, in clinical trials we don’t compare the effectiveness of a drug, given to patients (i.e., people) at large. Rather, we compare the treatment group (patients taking the drug) to a control group (taking a placebo), usually assigned at random 1 .