A few unusually high values can raise the mean even when most people see no change. Worked examples show when to look at the median and the spread.
A company can report rising average pay even if most employees have had no raise. If a small number of people receive much larger salaries, they can push up the mean for the whole group. The arithmetic may be correct while the headline gives the wrong impression of an ordinary employee's experience. The same problem appears in research summaries. “Participants improved on average” does not tell you whether most improved a little, a few improved a lot, or some improved while others did worse. To understand the result, you need more than the centre of the group.
The examples below start with five reading totals so every calculation is visible. They show how the mean differs from the median, why neither is automatically the better measure, and how the spread changes the story. These are invented numbers for explanation; the point is to know what to ask when a real report compresses many people's results into one figure.
The page counts are 4, 5, 5, 6, and 30. Add them and you get 50. Divide by five readers and the mean is ten pages. The mean is what people often call the average. Ten is a correct mean here. Yet four of the five people read fewer than ten pages. Now line the counts up from smallest to largest. The middle value is five. That is the median. In this group, it gives a different sense of a typical reader. Neither number is a lie. They answer related but different questions. Trouble starts when a report gives one number and lets you assume the other.
Suppose the fast reader reads 80 pages instead of 30. The counts are now 4, 5, 5, 6, and 80. The total is 100, so the mean is 20. The median stays at five. The first four people's results have not changed. Yet the mean has doubled. If a headline says average reading doubled, it is true for this set. If it makes you picture everyone reading twice as much, it gives the wrong impression. When a few large values could dominate, ask for the median or a display of individual results alongside the mean. The additional information shows whether the reported average resembles most people’s experience.
Do not decide that the mean is bad and the median is always good. If you need to know total demand, the mean can be very useful. Suppose you are supplying paper. The person who uses 80 pages still uses them. You cannot plan total stock from the middle person's use alone. The median helps describe the middle. The mean connects directly to the total when you know the group size. Choose the measure that fits the question, and state what it is. OpenStax explains mean and median with more examples. Our five-reader set is designed to make their difference easy to see.
Consider two groups with a mean score of five. Group A scores 4, 5, and 6. Group B scores 0, 5, and 10. The center is the same. The spread is not. In Group A, everyone is near five. In Group B, the results are far apart. A reader who only sees the average would miss that difference. A lesson or tool may help some people and hurt others. One number cannot tell that whole story. Look for the range or another measure of spread. A range shows the lowest and highest values, though it can be pulled by unusual cases. Other measures summarize spread in other ways.
Suppose a group average rises from five to six. You may imagine each person gaining one point, but many patterns could produce that change. Three people might each gain one. One might gain three while the others stay the same. One might gain five while two each lose one. The average change is one in all three cases. If the study follows the same people, their paired results can help show who changed. If it compares different groups, do not talk as if each person was tracked through the change. The wording should match the design. Average scores were higher is not always the same claim as people improved.
A firm may report average pay for all staff. That may include a few people with very high pay. It might instead report the median worker's pay. Those can paint different pictures. A study might report only people who finished the program. People who struggled may have left early. Then the final average may make the course look too good. Ask who is included and who is left out. The calculation can be correct while the choice of group makes the claim misleading.
Do not assume bad intent each time a report uses a mean. Ask for the missing view. Often a second number is enough to make the first one useful.
When an average is used to describe a typical person, check whether a few extreme values could be pulling it away from most people's experience. Ask for the median or a simple display of the results. When the question concerns a total, the mean may be exactly the useful measure. Also check who was included in the calculation. An average among people who finished a course may tell a different story from an average among everyone who started. The calculation and the group both matter. A report becomes clearer when it states which average it uses and shows enough of the distribution to explain what that number leaves out.
