Mean versus median wage

Every occupation table on this site shows two summary figures side by side: a median wage and a mean wage. They answer different questions, and for most occupations they are not the same number. The median is the headline figure here, and this page explains why.

Two ways to summarize a distribution

The median is the wage in the middle: half of the workers in the occupation earn less and half earn more. The mean is the arithmetic average: total wages paid divided by the number of workers. If every worker in an occupation earned the same, the two would be identical. They diverge whenever the distribution is lopsided.

Wages are almost always lopsided in the same direction. There is a floor near the minimum wage and no ceiling, so a small number of very high earners pull the mean upward while leaving the median where it is. For that reason the mean is usually higher than the median, and the gap between them is a rough measure of how long the upper tail is.

How BLS computes them

The Bureau of Labor Statistics does not collect an individual wage for each worker. In the Occupational Employment and Wage Statistics (OEWS) survey, employers report how many workers in each occupation fall into each of twelve wage intervals. BLS then estimates the median by finding the interval that contains the middle worker and interpolating within it, and estimates the mean by multiplying each interval’s employment by a mean wage for that interval drawn from a separate BLS survey, the National Compensation Survey. The Handbook of Methods describes both calculations.

Two consequences follow. First, the published median is an estimate, not an exact midpoint; it carries sampling error like every other OEWS figure. Second, because the top interval is open-ended, BLS cannot compute an exact mean for workers in it and instead assigns a fixed value, which is why very high wages are reported as “at or above” a top code (see top-coded wages).

Reading the two numbers together

A median tells you what a typical worker in the occupation earns. A mean tells you what the occupation pays per worker on average, which is the number an employer’s payroll budget or a national wage bill would use. Neither is wrong; they are different statistics.

When the mean is far above the median, the occupation’s pay is spread widely at the top. Chief executives, physicians and securities sales agents show large gaps. When the two are close, pay is compressed, as it is for cashiers or fast-food workers, whose distribution sits near the legal floor.

This site follows the family rule that the median is shown first and the mean is never shown alone. A page that reported only the mean would make a highly unequal occupation look better paid than most of its workers experience, and a page that reported only the median would hide the top of the distribution. The percentile table shows both ends at once.

When comparing occupations or places

Compare medians with medians. Comparing one occupation’s mean with another’s median mixes two statistics and will overstate the first. The same applies across states: the state tables on the states pages and the national figures on the occupation pages both carry a median and a mean, so like can be compared with like.

Also note what both figures leave out. OEWS wages are for wage and salary workers only, exclude the self-employed, and cover base pay and some incentive pay but not overtime premiums, benefits or employer retirement contributions. A mean that excludes bonuses understates the compensation of occupations where bonuses are large; a median does too, but less visibly.

The short version: read the median to learn what a typical worker earns, read the mean to see how much the top pulls the average, and read the two together to judge how unequal the occupation is.