Mean, Median and Mode: Measures of Central Tendency

Mean, median and mode are three ways of answering one question: what is a typical value in this data? They often give different answers, and the gap between them tells you something about the data.

The three measures in one example

Seven people wait for a bus. Their waits, in minutes, are 3, 4, 4, 5, 6, 7 and 30. One person arrived just after the bus left.

MeasureHow to get itAnswer
MeanAdd them up and divide by how many: 59 ÷ 78.43 minutes
MedianSort them and take the middle one (the 4th of 7)5 minutes
ModeThe value that appears most often4 minutes

Open this example in the mean, median and mode calculator

Six of the seven people waited 7 minutes or less, yet the mean says 8.43. That single 30-minute wait pulls the mean up. The median ignores how far out the 30 is. It only cares that the 30 is in the top half. So here the median describes a typical wait much better.

When the median is the better choice

Use the median when your data have a long tail or a few extreme values. Common examples:

  • House prices. A handful of mansions push the mean far above what most homes sell for.
  • Income. Government statistics on household income usually report the median for this reason.
  • Time to finish a task, recover from an illness or respond to a ticket. These have a floor at zero and no ceiling.

A quick check: if the mean is well above the median, the data are probably skewed to the right. If it is well below, they are skewed to the left.

When the mean is the better choice

The mean uses every value, so it changes when any single value changes. That makes it more precise when the data are roughly symmetric with no wild outliers, such as heights, test scores or measurement errors. Most of the tests on this site (t-tests, ANOVA, regression) compare means, so you will meet it far more often than the median.

The mean also has a property the median lacks: if you know the mean and the count, you know the total. Seven waits with a mean of 8.43 add up to 59 minutes. You can’t recover a total from a median.

What the mode is good for

The mode is the only one of the three that works for categories. You can’t average “red, blue, blue, green”, but you can say blue is the most common. For numbers it is less useful. Many data sets have no repeated value at all, and some have two or more modes. A data set with two clear peaks (bimodal) is often a sign that two different groups have been mixed together, such as the heights of men and women in one list.

Grouped data and weighted data

If your data come as a frequency table, such as 5 people aged 20 to 29 and 12 aged 30 to 39, you can still estimate the mean and median. Use the midpoint of each class as a stand-in for the values in it. The grouped data calculator does this and shows the working.

If some values count more than others, such as a final exam worth 50% and two quizzes worth 25% each, you need a weighted mean.

Common slips

  • Forgetting to sort before finding the median. The median of 7, 3, 30, 4 is not 3 or 30. Sorted, it is 3, 4, 7, 30, and the median is (4 + 7) ÷ 2 = 5.5.
  • Taking the middle position, not the middle value, with an even count. With an even number of values, average the two middle ones.
  • Reporting only the mean for skewed data. If the mean and median are far apart, report both, and say why.

Check yourself

Five salaries are $32k, $35k, $38k, $41k and $250k. Which measure would you report as the typical salary, and what is it?

The median, $38k. The mean is $79.2k, which is higher than four of the five salaries.

What is the median of 12, 15, 9, 20, 11, 14?

Sorted: 9, 11, 12, 14, 15, 20. The two middle values are 12 and 14, so the median is 13.

A shoe shop wants to know which size to stock most of. Mean, median or mode?

The mode. The shop needs the size people buy most often. A mean shoe size of 8.37 is not a size anyone wears.