A headline says the average payment is 200. Most of the people in the story received 100. Before deciding that one of those statements must be wrong, ask which average was calculated.
“Average” often means the arithmetic mean, but it can also refer to a median or another summary. The word is not enough to explain how the figure was produced or whether it describes a typical person’s experience.
Try five fictional payments
Imagine payments of 100, 100, 100, 100 and 600 units. The total is 1,000, so the mean is 200. Put the values in order and the middle one—the median—is 100. Both calculations are correct. The large fifth payment pulls the mean upward.
This example is invented to show the arithmetic; it is not a dataset about wages, benefits or any real organisation. That distinction matters whenever an illustration appears beside a news claim.
Find the measure in the original release
The Office for National Statistics discusses mean and median in its guide to earnings estimates, including how high values can influence the mean in a skewed distribution. The median is useful in that setting, but it does not answer every possible question.
If you want to understand the total amount being distributed, the mean and population size can be relevant. If you want to describe the middle observation, the median may be more informative. Neither number shows the whole spread.
Ask who is counted
A carefully calculated figure can still be misunderstood if its population is unclear. Does it include only full-time employees, all workers, households or people who responded to a survey? Are the amounts before or after a particular deduction? What period do they cover?
Write the unit beside the number. “Per household per month” is different from “per person per year”. When comparing two figures, check that these definitions match before interpreting the gap.
Look beyond one central value
Where available, inspect a distribution, range or percentiles. These can show whether observations cluster tightly or vary widely. Also read uncertainty notes and any explanation of changes in the sample or method.
For a practical reading habit, underline three items in the original release: the named measure, the population and the time period. If the headline drops one of them, put it back into your own summary before sharing the story.
A useful question is not “Which average is the honest one?” It is “Which question does this calculation answer?” That small change makes a numerical claim easier to assess without treating every difference as deception.