Here’s the myth: physician salary data is straightforward. Pull a number from AAMC, pull a number from MGMA, line them up, and boom—you know which specialty pays more.
Wrong.
That’s not analysis. That’s spreadsheet cosplay.
AAMC and MGMA are both useful. I use both. But they are not interchangeable, and they do not answer the same question. One is often telling you what academic physicians are paid in institutional settings shaped by rank, promotion, and protected time. The other is usually telling you what the broader compensation market looks like, often with much more attention to productivity, practice type, and percentiles. If you compare them casually, you’ll get bad conclusions fast.
And the word “salary” itself is already doing too much work. You need to separate base salary from total compensation. Then total compensation from bonus. Then bonus from wRVU-driven productivity pay. Then all of that from benefits, retirement match, sign-on money, loan repayment, and the giant hidden variable everyone loves to ignore: overhead. Especially if you’re comparing employed physicians with owners or partners.
Educational use only. This is not financial, legal, or tax advice, and compensation figures vary widely by market, contract structure, workload, and career stage. If you’re using compensation data for a contract, negotiation, or career decision, sanity-check it with qualified advisors.
The other lazy mistake? Reading specialty pay headlines without asking what kind of work bought that number. Forty hours a week? Sixty-five? Q4 overnight call? Pure clinic? Heavy procedures? Half-day admin blocks? Big city academic center? Rural private practice with zero local competition? I’ve watched residents stare at a “lowest-paid specialty” list as if it’s gospel, when the list was really just a mashup of unlike populations. That’s not truth. That’s bad methodology dressed up as insight.
Start With the Right Question: Are You Comparing Compensation or True Earnings?
The first question isn’t “Which specialty pays more?” The first question is: what exactly are you measuring?
Salary usually means base pay. Fine. But base pay can be a small piece of the picture. Total compensation may include incentive bonuses, quality bonuses, call pay, medical director stipends, retention money, and retirement contributions. A wRVU-based model may reward volume and procedure intensity in ways a flat academic salary won’t. Then there’s benefits—health insurance, CME, malpractice, disability coverage, retirement match—which absolutely matter even if they don’t show up in a social media graphic.
And “true earnings” can diverge even further. An employed outpatient pediatrician with predictable hours and little call might report lower compensation than a procedural specialist, but the work structure may be cleaner, the volatility lower, and the nonclinical burden lighter. Meanwhile, a private practice physician may show stronger gross compensation but absorb more business risk, staffing headaches, and overhead drag.
That’s why headline rankings are usually junk food. They tell you almost nothing about hours worked, nights on call, inbox burden, admin time, or whether the physician is spending two days a week teaching residents and writing grants instead of maximizing billable encounters. If you don’t control for workload and setting, you’re not comparing earnings. You’re comparing noise.
What AAMC and MGMA Actually Measure—and Why That Matters
AAMC is, first and foremost, an academic medicine lens. Its compensation data is built around faculty salaries, academic departments, rank, and institutional context. Assistant professor versus associate professor versus full professor matters. So does whether someone is in a department of pediatrics, psychiatry, medicine, or surgery. Academic jobs often include teaching, research, committee work, and protected time that directly shape compensation.
MGMA is different. It’s much more of a market compensation lens. It typically focuses on compensation distributions—median, mean, percentile ranges—and often includes productivity measures like wRVUs, plus breakdowns by practice type or ownership structure. That makes it more useful when you want to know what the broader employment market is paying, especially outside pure academic faculty roles.
Here’s where people get sloppy. They’ll grab a single AAMC mean for an academic specialty and compare it to an MGMA median from a broader market sample. That’s bad statistics and worse career advising. Mean and median are not the same. Academic faculty and mixed practice populations are not the same. Department-based salaries and ownership-influenced market compensation are not the same.
The same specialty can look dramatically different depending on source. Psychiatry in an academic center with protected teaching time may look modest on an AAMC table. Psychiatry in a high-demand employed market or private setting may look very different in MGMA. Same specialty label. Different population. Different incentives. Different conclusion.
So the rule is simple: don’t compare a single AAMC number to a single MGMA number unless you’ve accounted for methodology. If you skip that step, the data isn’t helping you. It’s tricking you.
How to Compare Specialties Without Being Fooled by Apples-to-Oranges Numbers
If you want a fair specialty comparison, use a boring method. Boring is good. Boring is how you avoid nonsense.
Start with the same year. Compensation markets move. A number from one reporting year and another from a different year can reflect changing demand, inflation, staffing shortages, payer pressure, or post-COVID employment shifts rather than a true specialty difference.
Next, match the role type and employment setting. Academic faculty should be compared with academic faculty. Hospital-employed physicians with hospital-employed physicians. Private practice or partner-track jobs with similar arrangements. Don’t stack an assistant professor in general pediatrics against an owner-operator private practice dermatologist and pretend you’ve discovered a law of nature.
Then match the metric. Compare median to median. Mean to mean. Percentile to percentile. This sounds obvious, yet people mess it up constantly. The median tells you the middle of the distribution. The mean can be pulled upward by high earners. Percentiles tell you spread. They answer different questions.
After that, adjust for workload whenever possible. Compensation per hour is often more revealing than annual compensation. So is compensation per clinical FTE. In productivity-heavy fields, compensation per wRVU matters because it shows whether a job pays well for the work generated or simply pays more because the physician is producing a massive volume of billable work. Those are not the same thing.
Then come the confounders. And yes, they matter a lot. Subspecialty mix can distort broad categories. “Internal medicine” may hide outpatient primary care, hospital medicine, and niche practice models. Call intensity changes the lived value of compensation. Procedure volume drives pay in ways a specialty label alone won’t capture. Region matters. Rural markets may pay more because recruitment is hard. Early-career versus late-career physicians may sit in entirely different parts of the distribution.
If you’re building a ranking table for lower-paid specialties, define the population tightly. Same source. Same year. Same employment setting. Same summary statistic. Same FTE assumption. If you can add hours, call burden, or wRVUs, even better. That ranking still won’t be perfect, but at least it won’t be fake.
Why Low-Paid Specialty Rankings Are Often Misleading
Let’s kill another bad idea: lower salary does not automatically mean a specialty is “worse paid.”
That’s an unserious conclusion.
Some specialties appear lower on compensation lists because they generate fewer wRVUs, rely less on procedures, or are overrepresented in academic settings. That doesn’t mean the jobs are bad. It means the payment model rewards different work differently. Pediatrics is the classic example. Family medicine too. Psychiatry often gets misread the same way, although market demand can make compensation much stronger than people expect depending on the setting. Some internal medicine subspecialties look modest on one list and solid on another because practice structure changes everything.
Academic physicians often accept lower compensation in exchange for things that don’t fit cleanly into a salary ranking: teaching, research, prestige, a narrower clinical scope, protected time, or less relentless production pressure. That tradeoff is real. So is the opposite tradeoff. I’ve seen physicians leave academia, add volume, take more call, and increase pay substantially—while also becoming more operationally squeezed. More money, yes. Better job? Not automatically.
And median pay hides huge variation. A family physician in a coastal academic system may look very different from a rural employed family physician with recruitment leverage. A psychiatrist in a university department may earn far less than one in a community market desperate for coverage. Same specialty. Different world.
That’s why “lowest-paid specialty” lists are often methodological artifacts with nice formatting.
A Practical Framework for Reading Reports and Building Your Own Comparison
Use a simple workflow.
First, define the question. Are you trying to compare academic career paths? Negotiate your first employed contract? Decide whether a specialty is financially viable for your goals? Those are different questions, so they need different datasets.
Second, choose the dataset that matches the question. AAMC is stronger for academic salary benchmarking. MGMA is usually more helpful for market benchmarking and contract review. Use the right tool. Don’t use a hammer as a thermometer.
Third, isolate the specialty group and write down the exact metric you’re using: median total compensation, mean salary, 75th percentile compensation, compensation per wRVU, whatever it is. Then document the limitations. If the sample is academic faculty only, say that. If ownership income isn’t captured, say that. If hours aren’t available, say that too.
I tell residents and fellows to build a spreadsheet with columns for source, year, specialty, subspecialty if relevant, setting, academic rank if relevant, median, mean, percentile data, hours, call, wRVUs, clinical FTE, and notes. Boring spreadsheet. Extremely useful.
For negotiations, never cite a number unless you know exactly what population it represents. I’ve watched people walk into a contract conversation waving a compensation figure that referred to a totally different setting, then wonder why they got nowhere. If you’re using AAMC, say it’s academic faculty benchmarking. If you’re using MGMA, know whether you’re pointing to median market compensation, a percentile target, or a productivity benchmark.
Your final checklist should be ruthless: same source, same year, same metric, same setting, same workload assumptions, and clear notes on what’s missing. If you can’t check those boxes, don’t trust the ranking.
Bottom Line: Use the Data, Don’t Worship the Data
AAMC helps you understand academic compensation. MGMA helps you understand market compensation. That’s the core distinction, and if you ignore it, everything downstream gets shaky.
Fair specialty comparisons require matching the same metric, the same year, and the same practice setting. Better comparisons also account for hours, call, productivity, and clinical FTE. That’s how you move from headline salary gossip to something actually useful.
The contrarian truth is this: the label “lowest-paid specialty” is often less a fact than a methodological mess. It can reflect academic overrepresentation, lower procedural volume, different lifestyle tradeoffs, or simple bad comparisons. So use the data. Absolutely. Just don’t bow down to it.
Because a bad comparison table can make a perfectly good specialty look weak. And that mistake is everywhere.
