Mean, median, and mode are the first things you learn in statistics, and they look so simple that it is easy to assume any of them will do. They will not. Each measure of center answers a slightly different question, and each can be quietly misleading for the wrong data. This guide shows how to compute all three and, more importantly, when to distrust each one.
The Three Measures, Computed
All three are "measures of center" — single numbers meant to summarize where a dataset sits — but they locate the center differently.
The mean is the arithmetic average: add every value and divide by the count. For the dataset 2, 4, 4, 6, 9, the mean is (2 + 4 + 4 + 6 + 9) ÷ 5 = 25 ÷ 5 = 5.
The median is the middle value when the data is sorted. For 2, 4, 4, 6, 9 — already sorted, five values — the middle one is the third: 4. With an even count, the median is the average of the two middle values. For 2, 4, 4, 6, the median is (4 + 4) ÷ 2 = 4.
The mode is the value that appears most often. In 2, 4, 4, 6, 9, the value 4 occurs twice and every other value once, so the mode is 4. A dataset can have no mode, or several.
Here all three are close. The lesson of this article is the cases where they pull apart.
When the Mean Misleads: Outliers and Skew
The mean uses every value, which sounds like a strength but is also its weakness: a single extreme value drags it.
Take five employees earning $40k, $42k, $45k, $48k, and $50k. The mean is $45k — a fair summary. Now the owner joins with a $500k salary. The new mean is ($40k + $42k + $45k + $48k + $50k + $500k) ÷ 6 = $725k ÷ 6 ≈ $120,833. Not one of the six people earns near $120k. The mean has been pulled far from the group by one outlier.
This is why median household income, not mean, is the standard economic statistic — incomes are right-skewed, with a long tail of high earners stretching the mean upward. The general rule: when data is skewed, the mean is pulled toward the longer tail. In a right-skewed distribution the mean sits above the median; in a left-skewed one it sits below.
When the Median Misleads: Ignoring Magnitude
The median's strength is the mean's weakness reversed: it is resistant to outliers, because it only cares about the middle position, not the size of extreme values. The salary example proves it — the median of those six salaries is the average of the third and fourth values, ($45k + $48k) ÷ 2 = $46,500, barely moved by the $500k owner.
But that resistance is also a blind spot. The median throws away information about magnitude. Compare two regions, each with a median home price of $300,000. In one, the priciest home is $400,000; in the other, it is $4 million. The medians are identical, yet the regions are nothing alike. The median tells you the middle and nothing about the spread or the extremes on either side.
There is also a practical limit: you cannot do arithmetic with medians the way you can with means. The mean of a combined group can be found from the group means and sizes; the median of a combined group cannot be recovered from the separate medians.
When the Mode Misleads: Not Really a Center
The mode is the only measure that works for categorical data — you cannot average "blue, red, blue, green," but you can name the most common category. For that kind of data, the mode is the only option.
For numerical data, though, the mode is often a poor summary of center. Consider 1, 2, 2, 98, 99, 100. The mode is 2, sitting at the very bottom of the data while most values cluster near 100. The most frequent value is not always near the middle. The mode can also be unstable: change one data point and the mode can jump, while the mean and median barely move.
Choosing the Right One
A short decision guide:
- Categorical data (colors, brands, yes/no): the mode is the only choice.
- Numerical data that is roughly symmetric with no severe outliers: the mean is the most informative, since it uses every value.
- Numerical data that is skewed or has outliers (incomes, home prices, response times): the median gives the most honest sense of a typical value.
A useful diagnostic: compute the mean and the median together. If they are close, the data is roughly symmetric and the mean is safe. If they differ noticeably, the data is skewed — report the median, and treat any single "average" quoted without context with suspicion.
Getting Help
A measure of center is only half the summary; spread tells the other half, and these ideas feed directly into understanding confidence intervals. For the rest of the descriptive-statistics toolkit, browse the Statistics & Data study guides.
Conclusion
Mean, median, and mode each locate the center of a dataset, but none is universally right. The mean uses all the data and is pulled by outliers and skew. The median resists outliers but ignores magnitude. The mode is essential for categorical data and unreliable as a center for numerical data. Match the measure to the data — symmetric numbers favor the mean, skewed numbers favor the median, categories require the mode — and always check the mean against the median to see what the data is hiding.
