Mode

The mode is the most frequently occurring value in a dataset — the only measure of central tendency that's meaningfully defined for nominal categorical data, where values are labels with no inherent order or numeric meaning, since neither a mean nor a median can be computed on a variable like favorite color or payment method. This makes the mode the default "typical value" answer whenever the data in question is categorical rather than numerical: reporting that credit card is the most common payment method is a mode, and it's the only one of the three classic central-tendency measures that question can even be meaningfully answered with.

A dataset can have one mode (unimodal), two modes (bimodal), or more than two (multimodal), and the presence of multiple modes is itself a meaningful signal about the underlying data that a single-number summary like the mean or median would completely hide. A bimodal distribution of customer order sizes — one cluster of small, frequent orders and a separate cluster of large, infrequent bulk orders — suggests two distinct underlying customer behaviors mixed together in the same dataset, a pattern invisible if only a mean or median is reported, since both of those would simply land somewhere between the two clusters, in a region that might contain very few actual observations.

On numerical data, the mode is less commonly used than the mean or median specifically because it's the measure most sensitive to how finely the data is measured or binned: a continuous variable like height or weight, measured to several decimal places, often has no two values exactly identical, making the "mode" technically undefined or arbitrary at that level of precision, while the same data binned into broader ranges (height rounded to the nearest inch) can produce a clear, meaningful mode — which is why the mode on continuous data is typically computed after binning (and visualized via a histogram, whose tallest bar visually represents the modal range) rather than on the raw, unbinned values directly.

The mode's relationship to the mean and median also serves as a quick diagnostic of distribution shape: in a perfectly symmetric unimodal distribution, all three measures coincide at the same value, while in a skewed distribution they separate predictably, with the mode sitting at the distribution's peak, the mean pulled furthest toward the direction of the skew, and the median landing between the two — a relationship often summarized with the rule of thumb that in a moderately skewed distribution, mean, median, and mode fall in that predictable order, though it's an approximation rather than a strict mathematical law and doesn't hold for every distribution shape.

Read the mode as the central-tendency measure built for a different job than the mean or median: it's the only one of the three that applies to categorical data at all, its presence in multiple distinct peaks is a signal — easy to miss if only a single summary number is reported — that a dataset may actually contain more than one underlying group mixed together, and on continuous data its usefulness depends heavily on how the data is binned, which is why it's most often read directly off a histogram rather than computed from raw, unbinned values.