Three answers to one question
Suppose you ask five people how many minutes they spent on a task and record 24, 26, 28, 28, and 94. What was a typical time? The answer depends on what you mean by “typical.” The mean is 40 minutes: add all five times and divide by five. The median is 28 minutes: put the times in order and take the middle value. The mode is also 28 minutes: it appears more often than any other value.
None of these calculations is wrong. They answer different questions. The mean accounts for every recorded value, so the 94-minute task pulls it upward. The median identifies the middle observation and changes less in response to that one unusually long task. The mode tells you the most frequently recorded time. Choosing among them is part of interpreting the data, not just doing arithmetic.
Use the mean when every value should count
The mean is often useful for numerical data when you want a figure that reflects the total. If a team spends 200 minutes across five tasks, the mean of 40 minutes tells you how that total would be distributed if each task took an equal share. It can help with questions about overall workload or resource use.
The mean can also describe a central value well when the observations are fairly balanced on either side of it. For 8, 9, 10, 11, and 12, the mean and median are both 10. But when a few values are much larger or smaller than the rest, the mean may describe the total better than it describes any individual experience. In the task-time example, four of the five people finished in 28 minutes or less, even though the mean is 40.
Use the median to describe the middle observation
The median divides ordered data in half: at least half the observations are at or below it, and at least half are at or above it. With an odd number of values, select the middle one. With an even number, average the two middle values. For example, the median of 12, 15, 19, and 40 is halfway between 15 and 19, or 17.
The median is particularly helpful when a distribution is skewed, as with the task times above. It keeps one extreme observation from shifting the reported center as much as it shifts the mean. That does not make the extreme value unimportant: a 94-minute task may deserve investigation. It simply means the median can give readers a clearer picture of the middle of the group. When reporting a median, consider also showing the range or some other indication of spread so that variation remains visible.
Use the mode to identify what occurs most often
The mode is the value or category with the highest frequency. It is especially useful when numbers cannot meaningfully be averaged. If a survey asks people to choose a preferred meeting format—online, in person, or hybrid—the most common response is the mode. A mean or median of those labels would not make sense.
A dataset can have more than one mode when values tie for the highest frequency. It can also have no useful mode: if every recorded task time is different, naming each one as equally frequent tells you little. For numerical data, the mode may change depending on how values are rounded or grouped. Times recorded to the nearest minute, for example, may repeat more often than times recorded to the nearest second. Say how the data were recorded if that detail affects the result.
Choose the measure that fits the question
Start with the kind of data you have. For categories, look to the mode. For numerical values, ask whether you care most about the combined total, the middle observation, or the most frequent value. Then inspect the distribution before settling on one number. Are most values close together? Is there a long tail of high or low values? Are there distinct clusters that a single average would hide?
When the measures disagree, reporting more than one can be more informative than declaring a winner. For the five task times, you might write: “The median time was 28 minutes; the mean was 40 minutes because one task took 94 minutes.” That gives readers both a sense of the usual experience and a reason the total is higher. Whatever you choose, name the measure rather than calling it simply “the average,” and include the number of observations and relevant context when you can.
