median
/ MEE-dee-an /
Line up everyone in a room from shortest to tallest and then point at the person standing in the very middle: their height is the median. It is not the average height, it is simply the value that splits the group into two equal halves, with as many readings above it as below.
The median of a set of measurements is the middle one once they are sorted in order. If there is an odd number of readings, it is the single central value; if there is an even number, it is the average of the two in the middle. Unlike the mean, the median ignores how far the extreme values stray — it cares only about their position in the lineup.
It matters because that very indifference to extremes makes the median robust: one wildly wrong reading shifts the median hardly at all, even though it can drag the mean far off. The trade-off is that the median throws away some information that the mean uses, so with clean, well-behaved data the mean is usually the more efficient and preferred summary.
Five results are 4.1, 4.2, 4.3, 4.4 and 9.0. The mean, dragged up by the 9.0, is 5.2, but the median, the middle value, stays at a much more representative 4.3.
An extreme value pulls the mean but barely touches the median.
When data are symmetric and free of outliers the mean and median nearly coincide; a large gap between them is itself a hint that the data are skewed or contain an outlier worth investigating.