Mean, median and mode are measures of central tendency: each one tries to summarize a data set with a single typical value. They sound like the same question, but they are not. The mean answers “how much would each one get if everything were split equally”; the median answers “which value separates the lower half from the upper half”; the mode answers “which value shows up most often”. Picking the wrong one does not produce an arithmetic error, it produces a wrong conclusion, and that is where many news stories, reports and test questions trip readers up.
Arithmetic mean
Add all the values and divide by the number of values. It is the most widely used measure because it uses every data point, and for that very reason a single very high or very low value moves it.
With the scores 6, 7, 7, 8 and 9: the sum is 37, there are 5 scores, and the mean is 37 divided by 5, which is 7.4. Notice that no student scored 7.4. The mean does not have to be one of the values in the set.
Weighted mean
When the values matter differently, multiply each one by its weight and divide the total by the sum of the weights, not by the number of values.
A test scored 6 with weight 2 and another scored 8 with weight 3: the weighted sum is 2 × 6 + 3 × 8 = 12 + 24 = 36, the weights add up to 5, and the mean is 36 ÷ 5 = 7.2. With three tests scored 5, 7 and 9 and weights 1, 2 and 3, you get (5 + 14 + 27) ÷ 6 = 46 ÷ 6 ≈ 7.67: the heaviest test pulls the result toward its own score. This is how a course grade with a heavier final exam works.
Median
Sort the data from smallest to largest. If the count n is odd, the median is the value exactly in the middle, at position (n + 1) ÷ 2. If n is even, there is no single middle value: the median is the mean of the two middle ones.
- For 9, 4, 7, 10, 8, sorting gives 4, 7, 8, 9, 10. There are 5 values, the 3rd is the middle one, and the median is 8.
- For 10, 4, 8, 7, sorting gives 4, 7, 8, 10. There are 4 values, the middle ones are 7 and 8, and the median is (7 + 8) ÷ 2 = 7.5.
Mode
It is the value that repeats most. In 3, 5, 5, 6, 9 the mode is 5. In 2, 2, 4, 4, 7 there are two modes, 2 and 4, and the set is bimodal. If every value appears the same number of times, we say there is no mode.
The mode is the only one of the three that works with categorical data. In a survey about how students get to school, if “bus” is the most common answer, it is the mode. There is no mean or median of ways to get to school. In a store, the best-selling shoe size is also a mode: among the pairs sold in sizes 7, 8, 9, 9, 9, 10, 11, 11, the mode is 9, the size worth keeping in stock.
Worked example: salaries at a small company
A hypothetical company has 7 employees whose monthly pay, in thousands of dollars, is 2, 2, 3, 3, 3, 4 and 20. The last one belongs to the owner. Which number best describes the typical salary?
- Sum: 2 + 2 + 3 + 3 + 3 + 4 + 20 = 37. Divide by 7: 37 ÷ 7 ≈ 5.29 thousand.
- The data is already sorted. There are 7 values, so the middle one is the 4th: the median is 3 thousand.
- The 3 appears three times, more than any other value: the mode is 3 thousand.
- Six of the seven people earn 4 thousand or less. The mean of 5.29 thousand is pulled up by the 20 thousand salary and does not describe the typical employee; the median and the mode do.
Check. For the mean, the deviations from it must add up to zero: the first six values sit below 5.29 and the 20 sits 14.71 above, and the exact sum of the deviations is 0. Another way to check: remove the 20 thousand salary. The mean of the remaining six drops to 17 ÷ 6 ≈ 2.83 thousand, just over half, while the median stays at 3 thousand. One value changed the mean and left the median alone, which is exactly the difference between the two.
When to use each one
| Situation | Best measure |
|---|---|
| Symmetric data, no extreme values (heights in a class) | mean |
| Extreme values or skewed distribution (income, home prices) | median |
| Categorical data or the most common value (best-selling size) | mode |
| Values of different importance (weighted grades) | weighted mean |
A rule of thumb: when the mean and the median are close, the data is roughly symmetric and either one works. When they are far apart, as in the salary example, the gap itself is information: extreme values are pulling the mean, and it is worth reporting both.
Common mistakes
- Finding the median without sorting. In 9, 4, 7, 10, 8 the middle of the original list is 7, but the median is 8.
- Forgetting the even case. With an even count, the median is the mean of the two middle values, not the first of them.
- Dividing a weighted mean by the number of scores. In the example with weights 2 and 3, dividing 36 by 2 gives 18, a “mean” higher than any score. Divide by the sum of the weights.
- Using the mean for income or prices with a few very high values. The arithmetic is right, but it describes a person almost nobody is.
- Assuming the mode is always unique. A set can have two modes or none.
Frequently asked questions
Why are incomes usually reported with the median?
Because a few very high incomes pull the mean up. The median is the value that splits people into two halves, and those extremes do not affect it.
Can a data set have more than one mode, or none?
Yes. With two values tied as most frequent the set is bimodal. If every value appears equally often, we usually say there is no mode.
Does the median have to be one of the values in the set?
No. With an even number of values, the median is the mean of the two middle ones and may not appear in the data, like the 7.5 of 4, 7, 8 and 10.
Where does the weighted mean show up in school grades?
Whenever assignments count differently: multiply each grade by its weight, add everything and divide by the sum of the weights. A final exam worth more than a quiz is the classic case.