What each measure actually is
The mean is the arithmetic average: add up every value and divide by how many there are. It uses every data point, which makes it informative, but also means every data point can pull it. The median is the middle value once the data is sorted — the point where half the values fall below and half above; with an even number of values it is the average of the two middle ones. The mode is the value that appears most often; a data set can have one mode, several, or none.
Consider the data set 2, 3, 3, 4, 8. The mean is 20 divided by 5, which is 4. The median is the middle value, 3. The mode is 3, the only repeated value. All three differ even in a tiny data set, because each answers a different question: the mean asks 'what is the balancing point?', the median asks 'what is the typical middle case?', and the mode asks 'what is the most common outcome?'
- Mean = sum of values ÷ count; it uses every data point.
- Median = the middle value once the data is sorted; half the data lies on each side.
- Mode = the most frequent value; there can be one, several, or none.
How outliers and skew pull them apart
The key practical difference is how each measure reacts to extreme values. The mean is sensitive to outliers because it is computed from every number, so one very large or very small value drags it in that direction. The median is resistant: it depends only on the position of the middle value, not how far the extremes stretch. The mode ignores magnitude entirely and cares only about frequency.
Income is the classic example. In a neighborhood where most households earn between 40,000 and 70,000 but one earns 5 million, the mean can land near 200,000 — a figure that describes almost nobody — while the median, around 55,000, describes the typical household far better. This is the fingerprint of skew: right skew pulls the mean above the median, left skew pulls it below, and in a symmetric distribution the two nearly coincide. Comparing them is itself a quick test for skew.
- The mean is sensitive to outliers; the median is resistant; the mode is unaffected by magnitude.
- Right skew pulls the mean above the median; left skew pulls it below.
- When the mean and median are close, the distribution is roughly symmetric.
When to use the mean
The mean is the right choice when data is roughly symmetric and free of extreme outliers, and when you need a measure that feeds into further calculation. Because it uses all the data and has clean mathematical properties, it is the foundation of variance, standard deviation, standard error, and nearly every inferential test. Compute a confidence interval or run a t-test and you are working with means.
Use it for the average height of a group, the average score on an exam with a fairly normal spread, or the average temperature over a month — cases where no single value dominates. Once a distribution becomes badly skewed or contains genuine outliers, the mean stops being representative and you should reach for the median instead, or report both.
- Best for roughly symmetric data without extreme outliers.
- Uses every value and underpins standard deviation, confidence intervals, and most tests.
- Ideal when you will do further statistical calculation with the center.
When to use the median
The median is the better summary whenever data is skewed or contains outliers, because it reports the genuine middle without being distorted by extremes. This is why incomes, house prices, response times, and hospital lengths of stay are almost always reported as medians: a handful of very large values would make the mean misleading. It answers the question most people actually have in mind when they ask about a 'typical' value.
It is also the natural choice for ordinal data — ranked categories such as survey responses from 'strongly disagree' to 'strongly agree' — where the spacing between categories is not truly equal and averaging them is questionable. With skewed data, report the median alongside the mean; the gap between them tells the reader how skewed the distribution is.
- Best for skewed data or data with outliers, such as income or house prices.
- The right choice for ordinal (ranked) data where category spacing is unequal.
- Reporting median and mean together reveals the direction and size of skew.
When to use the mode
The mode is the only measure of center that works for categorical (nominal) data, where the values are labels rather than numbers. You cannot take the mean or median of eye color, favorite product, or preferred language, but you can identify the most common category — and that is what the mode gives you. When a business asks which item sells most, the answer is a mode.
The mode also marks the peak of a distribution and reveals when data is multimodal — two or more distinct clusters, which often signals that different groups have been mixed together. Its limitation is that for continuous numeric data exact repeats are rare, so it is usually reported alongside the mean or median. The safest habit is to identify the level of measurement of your data first, then let that decide which measure of center is valid. StatRise has a separate calculator for each of the three, and the Data Lab reports mean, median, and mode together for every numeric column of a CSV you paste in.
- The only valid measure of center for categorical (nominal) data.
- Reveals the most common outcome and exposes multimodal data with several clusters.
- Less useful for continuous numeric data, where exact repeats are uncommon.