How far apart two groups are

Compare two numerical distributions with similar variability shown on a shared scale (e.g. dot plots) in two distinct ways: (1) judge their actual visual overlap directly from the displays — how much of each group's data lies among the other group's values; and (2) separately, find the difference between their centres and express it as a multiple of a clearly named measure of variability of either group, using matching measures (difference in means ÷ MAD, or difference in medians ÷ IQR), interpreting a larger multiple as centres that are further apart relative to the typical spread. Recognise that the same difference in centres means more when the variability is small, and that for similarly shaped, roughly symmetric, single-peaked distributions a larger multiple usually goes with less overlap — but that the ratio alone does not determine overlap: skew, clusters, gaps or outliers can make the visible overlap greater or smaller than the ratio suggests, so overlap is always checked on the display; and this simple ratio is only appropriate when the two variabilities are similar