Benchmarks
What clinical research actually pays.
Three kinds of evidence, never blended: what members report anonymously, what employers filed with the US government, and what they print on job ads. Every number carries its sample size and source; reported cuts publish only at n=5.
0 member reports · 22 government filings · 3 advertised ranges · updated 2026-07-18
Reading the figures
Reported — middle half of member reports; the line is the median.
Filed — exact salaries from public government filings, one mark per person.
Advertised — the range printed on a job ad. Wide by design; context, not a midpoint.
Clinical monitoring (CRA)
Fig. 1 · US · USD · annual base
Is this what you actually make? Add your anonymous datapoint — one minute, no name, no email. Public records: source 1source 2.
Clinical operations / PM
Fig. 2 · US · USD · annual base
Is this what you actually make? Add your anonymous datapoint — one minute, no name, no email. Public records: source 1.
Biostatistics
Fig. 3 · US · USD · annual base
Is this what you actually make? Add your anonymous datapoint — one minute, no name, no email. Public records: source 1source 2source 3.
Statistical programming
Fig. 4 · US · USD · annual base
Is this what you actually make? Add your anonymous datapoint — one minute, no name, no email. Public records: source 1.
Data management
Fig. 5 · US · USD · annual base
Is this what you actually make? Add your anonymous datapoint — one minute, no name, no email. Public records: source 1.
How to read these numbers
- Reported — submitted directly by a person in the role, anonymously. The only source that reflects what people actually accept.
- Filed— exact base salaries employers disclosed to the US Department of Labor in visa paperwork. Public record, per person, verified row-by-row against the government dataset. Filing years vary; older rows understate today's market.
- Advertised — the range printed on a job posting. Compliance bands often span 2–3x and multiple geographies; when a bar runs off the figure's edge, that is the point.
- Currencies are never mixed inside one statistic. Reported cuts smaller than n=5stay unpublished so a lone submission can never be someone's actual salary on display.