Head-to-head
Average vs Median
Both the average and the median are trying to answer 'what is a typical value?' They disagree whenever the data is skewed, and the size of the disagreement is itself information about the data.
The average — strictly the arithmetic mean — adds every value and divides by the count. That makes it sensitive to extremes: one very large number pulls it up, one very small number pulls it down. The median sorts the values and takes the middle one, so it does not care how extreme the extremes are, only how many values there are.
Which is correct depends on the question. If you want the total divided evenly — average revenue per customer, average speed over a distance — the mean is the right tool, because it is the number that reconstructs the total. If you want to describe a typical member of a skewed population — incomes, house prices, wait times — the median is more honest, because it cannot be dragged by a handful of outliers.
One practical test: ask whether the total would still make sense if one value were ten times larger. If it would, use the mean. If the answer is that the total would be dominated by that one value, report the median and say so, because a mean quoted without that caveat is technically correct and practically misleading.
Average (mean)
How To Calculate The Average (Mean, Median And Mode)
Asking for the average is really asking which of three different summaries you want, and they disagree whenever the data is lopsided.
Median
How To Calculate The Median
The median is the number that sits in the middle, and it is the statistic that survives an outlier when the average does not.
How They Differ
| Attribute | Average (mean) | Median |
|---|---|---|
| Uses | Every value in the data set | Only the middle value (or the two middle values) |
| Formula | Sum ÷ count | Middle value of the sorted list |
| Sensitive to outliers | Yes — one extreme value moves it | No — extremes only matter if they change the count |
| Can be a value that is not in the data | Yes, often | No, for an odd count |
| Reconstructs the total | Yes — mean × count = sum | No |
| Best for | Symmetric data, rates, per-unit totals | Skewed data: income, house prices, response times |
| In this library | Arithmetic average of a list of numbers | Middle value after sorting |
The Same Numbers, Both Ways
Five values: 12, 15, 18, 22 and one outlier at 60.
- Values
- 12, 15, 18, 22, 60
- Count
- 5
| Line | Average (mean) | Median |
|---|---|---|
| Sum | 127 | 127 |
| Result for all five | 25.4 | 18 |
| Result without the 60 | 16.75 | 16.5 |
| Change caused by the 60 | +8.65 | +1.5 |
Adding one large value moves the average by 8.65 but the median by only 1.5. The average is now 25.4 while the middle value is 18 and four of the five values sit below the average — a fair summary of the total, and a misleading one of the group.
When To Use Each
Average (mean)
- The data is roughly symmetric, so both measures agree anyway.
- You need a number that multiplies back to the total — per-capita figures, average order value.
- The values are rates or ratios over equal units, such as speed over equal time intervals.
Median
- The data is skewed: incomes, house prices, hospital wait times, app session lengths.
- There are outliers you do not want to remove but do not want to let dominate.
- You need a value that is guaranteed to be typical — the median is always an observed value for an odd count.
Common Questions
When are the average and median the same?
When the data is perfectly symmetric. Any skew pulls them apart, and the direction of the gap tells you the direction of the skew: mean above median means a long right tail, as with income.
Which one does 'average income' usually mean?
Statistical agencies report both, and they differ a lot. Mean household income is pulled up by high earners, so median income is the better guide to what a typical household earns. If a figure is quoted without saying which, assume mean and treat it as an upper bound.
Can the median be calculated for an even number of values?
Yes — take the mean of the two middle values. For 10, 20, 30, 40 the median is 25, which is not one of the data points. That is the one case where the median can be a synthetic value.
Does the median have a formula?
Not a closed-form one, because it depends on order rather than magnitude. For grouped data it is estimated by interpolation within the median class, which is why published medians are sometimes approximate.
What should I use for grades or test scores?
Both, plus the distribution. A class average of 72 with a median of 85 means a few very low scores are dragging the mean — information you lose entirely if you report only one of the two.
More Comparisons
- Salary vs Hourly Pay
A $60,000 salary is $28.85 an hour at 2,080 hours, or $30.00 at 2,000. The hours assumption decides the answer, and it is the part most comparisons leave out.
- ROI vs CAGR
ROI measures the whole return; CAGR turns it into a compound annual rate. A 100% return over five years is 20% a year by division but 14.87% a year in reality.
Sources
- Measures of central tendency — mean, median, mode — National Institute of Standards and Technology, e-Handbook of Statistical Methods
- Household income — mean vs median — U.S. Census Bureau
- Consumer expenditure survey — averages and medians — U.S. Bureau of Labor Statistics
Both methods sit in Math.