TheJobsMarket
What Jobs Pay

Finding Out What a Job Actually Pays

Short answer

Three sources dominate salary information — employer surveys, self-reported figures and scraped job postings — and they disagree because they measure different things, not because two of them are lying. The one worth believing is the one that names its method, its population and its reference period.

Before any figure on this site or anywhere else is useful, you need to know what kind of number it is. Almost every disagreement between salary sources dissolves once you can tell an employer survey from a self-report from a job posting, and almost none of them tell you which they are.

Three different things called the same word

An employer survey asks businesses what they pay the people on their payroll. A self-reported dataset asks workers what they earn. A posting scrape reads what employers advertise. These produce different numbers about genuinely different populations, and the gap between them is usually wider than the gap between any two sources of the same kind.

The questions that settle it

Who was counted, and who was left out. What period does the figure describe. Is it a median or a mean, and is that stated. Does the source name its own source. A publisher that will not answer those four questions is not a source, it is a repeater — and a great many salary figures circulate for years after the underlying number has moved.

Reading a distribution properly

A percentile table is not a ladder you climb with seniority. The 90th percentile is not what you get after twenty years; it is what the top tenth of everyone doing that work earns right now, and it includes people in different industries, different employers and different cities. Treating it as a career path is the most common way the data gets misread.

When there is no figure at all

Some occupations are too small, too new or too concentrated for a survey to publish without disclosing individual employers. That is a real gap, not a failure to look, and the honest response is to say so rather than to substitute an estimate from somewhere weaker.

Articles in this section

Common questions

Which salary website is the most accurate?

None of them is accurate in the abstract, because they answer different questions. For a typical figure across everyone doing the work, the employer survey is the most defensible. For a role the survey covers badly, or for a specific large employer, self-reported data can be better despite being noisier.

Should I trust a figure with no date on it?

No. Wage data moves annually and a figure with no reference period cannot be checked or corrected. The absence of a date is usually a sign the publisher copied it from somewhere else.

What is the difference between median and mean salary?

The median is the middle of the distribution and the mean is the average. Where a few very high earners pull the average up, the mean can sit well above what a typical person earns, which is why the median is the more useful of the two.

Why does my job title not appear in the data?

Wage surveys are organized by occupation code, not by job title. Titles are invented by employers; codes describe what the work is. Matching your role to the right code is usually a question of what you do all day rather than what your offer letter calls you.

Is scraped job-posting data useless?

No, but it answers a narrower question: what employers are currently advertising, which is a leading signal about demand. It is a poor guide to what people are actually paid, because postings skew toward the top of a range and toward roles that are hard to fill.

How do I check a salary claim myself?

Find the occupation code, go to the published survey table for it, and compare. If the claim is about a specific place, use the metropolitan table rather than the national one. If the source will not tell you which table it used, there is usually a reason.

Where to go next

Pay by Occupation

Once you can read the data, the next question is what your own occupation actually pays.

Read it →
CS

Charles Slocs

Data and research

Charles Slocs builds the data side of this site — pulling the federal wage and employment series, matching job titles to occupation codes, and working out what the numbers do and do not support. He writes the pages that are mostly a question about evidence: what a survey measured, how wide the spread really is, and which published figure is out of date.

All articles by Charles Slocs →