Shape: Symmetry, Skew, and Modes

Look at a histogram in one pass and you can pull three things from it: shape, center, and variability. Shape comes first because it tells you how to read the other two. A symmetric distribution has roughly mirror-image halves, and its mean and median land close together. A skewed distribution has one side stretched into a long tail. That tail, not the tall bump, gives the skew its name. The bump sits on the side with more typical, clustered values; the thin tail reaches toward the extreme, rarer ones.

Home prices make this concrete. Most houses cluster in an ordinary price range, forming a tall bump on the left side of the histogram. A handful of mansions stretch the right side into a long, thin tail. That tail pulls the mean up while the median, which only cares about position, barely moves. So the mean exceeds the median, and the distribution is skewed right, even though the tall bump is on the left. The direction of the pull, not the location of the bump, decides the name.

Shape also reacts to how data got collected. Pour two different groups into one histogram and you can create a shape neither group had alone. If two classes have different typical scores, combining them often produces two separate bumps, a bimodal distribution, one per group. Recognizing that pattern tells you the single histogram is secretly hiding two populations. Shape also decides which summary numbers to trust: for symmetric data, mean and standard deviation describe the distribution well, but for skewed data, the mean gets dragged toward the tail, so median and interquartile range describe center and spread more honestly.