Mean, Median, and Which One to Trust

The mean is what you get when you take every value's contribution seriously: add them all up, divide by how many there are. Picture the data points as weights on a seesaw. The mean is the spot where the seesaw balances. That means one huge value sitting far out on one side can drag the whole balance point toward it, even if every other value stays put. The median ignores the actual sizes of the numbers and just counts position. Sort the data, find the middle value (or average the two middle values if there's an even count), and that's it. A number that's barely above the middle counts exactly the same as a number that's a million miles above the middle, as long as it's still on the high side.

That difference is why the two can tell very different stories. In a symmetric distribution with no outliers, mean and median land close together. In a skewed distribution, or one with outliers, the mean gets dragged toward the long tail while the median holds its ground. A right-skewed dataset (a few very large values) pushes the mean above the median; a left-skewed one pulls the mean below it. The size of that gap is itself a clue about shape and skew, which connects directly to how you describe a distribution's shape later in this unit.

Because the mean is sensitive and the median is resistant, choosing which one to report is a judgment call, not a formula. For skewed data or data with outliers, the median usually gives a more honest sense of a 'typical' value, and you'd pair it with the interquartile range rather than standard deviation.