Gaps, Clusters, and Unusual Features

A histogram is never just a shape to name. When you see a gap, a stretch of the number line with no data in it, that's a clue that you're looking at more than one population squeezed into one dataset. Real processes tend to produce smooth, connected spreads of values. A sudden empty stretch usually means something changed between the low group and the high group: a different type of subject, a different method of measurement, a different time period. Clusters are the flip side of the same clue; a cluster is just a pile of values sitting together, separated from other piles by those gaps.

Describing a distribution fully means covering four things every time: shape (skewed, symmetric, unimodal, bimodal), center (mean or median), spread (range, IQR, or standard deviation), and any unusual features (gaps, clusters, or outliers). Skip the last part and your description is incomplete, even if the numbers are right. Saying 'this distribution is skewed right with a median of 25' tells only half the story if there's also a gap splitting the data into two clumps.

This habit of checking for unusual features connects directly to the next question in your course: how shape and outliers affect your choice between mean/standard deviation versus median/IQR. A gap or cluster tells you the data might not be one coherent group, which matters just as much for choosing a summary statistic as it does for choosing a graph. Unusual features aren't noise to smooth over; they're often the most informative part of the dataset, pointing you toward a better question than the one you started with.