Contrary to the impression many students in high school AP Statistics courses may have, statistics is a very rich. intellectually stimulating field. Yet, in a sense the core issues boil down to just two concepts, bias and variance
(Note: bias here is not what you may be thinking!).
Appreciation of these ideas is important for any course, but especially for statistics/data science courses serving as "general education." Statistics is important to one's status as an informed consumer/citizen, and the"graduate" of such a course should be equipped to recognize these two central ideas whenever he/she views data.
Students of statistics, at all levels, and their instructors.
Here I am NOt referring to the first thing taught in any statistical methods class: To estimate a population variance σ2, we use
s2 = (1/(n-1)) Σin (Xi - Xbar)2
with the rather unnatural denominator n-1 rather than n. This makes s2 unbiased, i.e. its average over all possible samples is exactly σ2, E(s2) = σ2.
Long ago, early in my career, I served as a consultant for a major medical chain. There were four hospitals of interest, and the question was, Do they all give the same quality of care to heart attack patients? Let's denote the quality of care by Q.
A major issue was the age A of the patients. One of the hospitals was located in a district that consisted disproportionately of the elderly. Thus it would be unfair to that particular hospital to gauge quality of care by the criterion of mean Q, say estimating E(Q) by the sample mean Qbar at each of the four hospitals. This kind of comparison would be unfair, i.e. "biased" in the casual sense of the word.
The fair criterion--i.e. "unbiased" in the casual sense--would be to estimate say E(Q | A = 80), mean Q for 80-year-olds, and compare these four values.
This casual terminology can be reconciled with a formal statistical formulation, though we will not pursue this here. We simply define bias as "not using relevant covariates."