There's a person on Twitter who uses his account almost exclusively to search for #wagegap and respond that it doesn't exist. There's what he calls earnings gap, he maintains, in which women earn less than men because they work less difficult jobs for fewer hours—but this is not the same as his definition of the wage gap, in which women receive unequal compensation for performing the same jobs.

@Lexi_Hoth Women need to more hours and aim for harder higher paying jobs, it's that simple.

— Simon Jones (@Wateronatrain) May 18, 2015

@KatieBoWill Earnings* gap

— Simon Jones (@Wateronatrain) April 22, 2015

It's not an unpopular opinion, as a cursory review of the comments on any story about the gender pay gap will reveal. Even the literature backs him up—in healthcare, studies have suggested that gender differences in physicians' salaries can be accounted for by women's tendency to enter the lower-paying primary care field and work fewer hours.

In some ways, Simon's not wrong, although he's got his semantics mixed up. The term "wage gap" refers to the difference between the median male and female earnings, expressed as a percentage of male earnings, according to the Organisation [sic] for Economic Co-operation and Development. The term doesn't imply any causation one way or another. In fact, it's a remarkably neutral term—a quantitative way to assess how much money people of both genders make.

In some sectors of healthcare, that neutral, quantitative gap is growing. In 2008 in New York, male physicians at the outset of their career earned $16,819 more on average than newly trained female docs. In 1999, the difference was just $3,600. More starkly: Between 1987 and 1990, there was a 7% wage gap between male and female physician assistants ($41,304 to $38,185). Between 2006 and 2010, that gap had skyrocketed to almost 30%—$86,236 to $61,010.

There's a host of reasons this might be happening, but before we get into those, it's important to note that this is in direct contrast with what's happening outside of the healthcare industry. Pew Research Center data based on hourly earnings of full- and part-time workers found a 16-cent pay gap this year. In 1980, it was 36 cents. For young women in 2015, there is only a 7-cent gap. It’s not at parity, but the general trend has been down.

Why is that gap widening, in the face of the national trend? Are the wage-gap critics right—are women themselves to blame for the disparity in compensation? Explanations like preference for lower-paying specialties are appealing, because they suggest that the status quo is not an injustice that obligates the industry to do something about it—but do they reveal the whole story?

A series of theories, supported and debunked

Dr. Emma Simmons had been out of medical school for about five years when she went to work for a brand new primary care practice. Now the associate dean of student affairs in the University of California, Riverside School of Medicine, Simmons knows that there was a certain disparity in experience between herself and her two male colleagues at the practice. But when she found out the disparity in their salaries, Simmons says, "I saw flames in my eyes."

"Obviously you have to correct for the amount of training, education and experience, but even that being said: I was making maybe half of what my two male partners were making," Simmons said. "That was eye-opening to me."

Simmons went directly to her superior and said, "Do you realize this is a problem?"

She got a significant raise almost immediately. According to Simmons, the practice leaders grasped instantly the cold, dark possibility of a PR nightmare.

That kind of account is often dismissed as anecdotal or an outlier in favor of other, less simple explanations. But if basic discrimination is not the culprit of the wage gap—and to be clear, there is little evidence that it is—what does the literature say about the most common alternative theories?

Productivity

Productivity in healthcare, like value, is a tough term to define. Most studies (which have returned mixed results) measure it in terms of patients seen over a given time period, suggesting that the more patients a physician sees, the more "productive" he or she is. This is a crude measurement at best because it fails to account for the things that, increasingly, physicians are being compensated for: improved quality and outcomes. And since the industry overall is grappling with how to measure quality, from an analytical standpoint, there is no way to test the hypothesis that male physicians are more "productive" than female physicians.

Simon Jones might suggest that, more importantly, women don't work as many hours, and therefore earn less—one of the most common "mansplainers" for the wage gap—and to a certain degree, he'd be right.

Hours

Most of the literature points unequivocally to the fact that women work fewer hours. A comprehensive 2011 study published in Health Affairs did find that a great percentage of newly-trained female physicians (38.1% compared to 24.5% of men) planned to spend fewer than 40 hours a week on patient care. And a smaller percentage of new female docs (23.4% compared to 37.3% of men) plan to spend more than 50 hours a week on patient care. In a perfectly fair world, that shouldn't impact compensation within those two brackets—physicians who were willing to work longer hours would be compensated accordingly, regardless of gender.

Yet when researchers restricted their analysis to only the new docs who planned to spend more than 40 hours per week on patient care, the results weren't substantively altered. Men still won out. Remember, this is physicians at the very beginning of their careers, a moment in which, theoretically, experience levels across genders should be pretty much equal.

Specialty choice

So, Simon might suggest, women are obviously choosing lower-paying specialties like primary care. This would seem logical: According to a 1990-2003 study, the absolute number of female primary care physicians is rising (although the proportion of female docs who choose to enter the primary care field is decreasing). 13.9% of women are in pediatrics, v. 5% of men. Conversely, there is a smaller proportion of women in the higher-paying specialities (1.3% v. 4.1% in cardiology, 1.8% v. 3.4% in diagnostic radiology).

This argument raises two questions. First, if that were true, the playing field should be level within a given specialty. According to the same Health Affairs study, that's not the case at all. In fact, across 20 specialities, in only two did women earn more than men at the outset of their careers—gastroenterology and general surgery—and in both cases the difference was not statistically significant.

Again, it's important to note the reason that researchers used starting salaries here. It's the most "pure" measurement because other diluting, difficult-to-measure factors haven't yet entered the equation. The power of observable characteristics like gender and specialty choice to explain the wage gap with any statistical certainty diminishes over time.

The second big problem with the "women are choosing lower-paying specialties" argument is that it doesn't look at why women are choosing lower-paying specialties. Does it express a preference, or are women choosing the specialties that are open to them? This is why it’s important to look at both the adjusted and unadjusted figures—wages both within specialties and irrespective of specialty.

Assistant professor of healthcare policy and medicine at Harvard Medical School Dr. Anupam Jena uses the example of black-white income disparity to explain how the mechanism might work in healthcare.

"The classic example is saying, should we control for college attainment," Jena says. "If you don't control for college attainment, the difference in black and white income is X. But when you do control for college attainment, then the difference goes down."

That drop might suggest that researchers should control for college attainment because college graduates earn more money and white people are more likely to graduate from college than black people.

"Does it past the sniff test?" Jena asks. "Maybe not because one of the reasons black people earn less that white people is because they are less likely to be able to go to college. You don't want to wipe that mechanism under the rug."

In healthcare, the mechanism is still undetermined. Do women face unspoken (or spoken) discrimination in certain specialties that drive them towards more traditional "female" departments like pediatrics? And even if that’s not the case—even if gender preference is the driver—that still doesn't explain why the median female income is still less than men's in the same specialty.

What about women at the top?

The best man and the best woman are going to earn the same amount, right?

That may be true on a case-by-case basis, but that doesn't address why there are so many more men at the top of the pay scale than women. In the Health Affairs study, only 5.9% of reported salaries of either gender fit into the top compensation bucket (which changed over time to reflect rising salaries overall). Broken down by gender, the study revealed that 8.3% of men fit into the top compensation bucket, compared to only 2.3% of women. In fact, "the fact that more men were top-coded than women implies that the imputation approach probably understated the difference between men's and women's salaries," the researchers wrote.

Personal choice

But when women leave the workforce to get married and have babies, they return with less experience than their male counterparts, so—the argument goes—they get paid less.

Researchers are split on the degree of the effect of motherhood on this disparity, but it's definitely some.

According to Jena, some unpublished research shows that if you look only at women physicians who don’t have children, the gap decreases dramatically. "[Female income] isn't quite the same as men, but it's pretty close," Jena says.

"Female physicians who do have kids, that's where you start seeing the income drop," Jena said. "The hypothesis is that female physicians not only have to shoulder the workload of being a physician but also the workload of being a mother and a wife and whatever household responsibilities come with raising a family."

The Health Affairs study characterizes the impact of family choices on the overall pay gap as much smaller. Remember that study looked at physicians at the very outset of the careers—a time period in which few would have been choosing to have a child. And, as that study notes: "Importantly, much of the previous research has found that family status typically has a comparatively small effect on female physicians' incomes… when other factors such as specialty are controlled for."

Then there's the comparison with the legal profession. Physicians and lawyers are a remarkably similar bunch. Both acquire some serious human capital in the form of education and tend to be a self-selected group of driven, ambitious individuals. Like in healthcare, the legal profession has its own persistent pay gap.

Studies of the legal industry estimate that characteristics related to family status account for about half of the pay gap in that profession, which leaves half of the pay gap unaccounted for. So even if you assume that family choices have a similar impact in the healthcare industry, there's still a big chunk of the disparity that's left unexplained... and growing.

Social factors

Inevitably, the conversation then turns to difficult-to-prove reasons for the pay gap, most of which have to do with how women are "taught" to behave in society. The notion that women don't ask for the higher salary, popularized by Cheryl Sandberg's "Lean In", is probably at least partly to blame.

"The wage gap may be widening because women continue not to be as aggressive in negotiations for their initial salaries which sets the stage for all subsequent raises and negotiations," Simmons said—perhaps thinking of own experience in the new practice. "We tend to want to please and be conciliatory and to not take too much from 'the institution' or others."

This is such a well-known theory that it is probably becoming a self-fulfilling prophesy, through a kind of perverse cognitive dissonance effect. A theory pioneered by Claude Steele in 1999 suggests that when people are faced with a threat to their sense of self—what they believe to be true about themselves—they are motivated to correct the misconception by either reasoning away an inconsistency or modifying their behavior. Studies show that women who are asked to give their gender before taking a math or science test perform worse than women who are not asked to identify their gender. The indication is that they conform their behavior to the social perception that women are "bad" at math and science.

This also might explain why women aren't being offered higher salaries to begin with—an observable phenomenon demonstrated by unconscious bias studies. In a randomized, double-blind study published in PNAS in 2012, science faculty from a research university looked at the identical job application of a student who was randomly assigned either a male or a female name for a lab manager position. Not only did the faculty rate the male applicant as significantly more competent and hireable that the female applicant, they also awarded the student a higher starting salary. Men and women faculty members were equally likely to exhibit bias against the female student—discrimination has no less impact if it's not overt. (Arguably, it just becomes harder to identify and mediate.)

Notably, all of these studies deal with perception, not ability—they don't indicate that women aren't capable of being tough negotiators. A 2010 study found that gender differences in assertive negotiating are mediated by women's fear of backlash and attenuated when negotiating on behalf of others—as the Health Affairs study noted, "recent studies have demonstrated gender differences driven by social contexts and constraints, as opposed to a deficient negotiating capacity in women."

In part, the hesitancy may stem from the dearth of female mentors after whom women can model their behavior.

"We don't have the number and variety of female mentors still that males do—especially in leadership roles," Simmons says. "This problem is even more tenuous for women who come from black and Latino heritages. Until we have more of a critical mass of women with stable power, it will continue to be an uphill climb."

Data support that a "critical mass" is necessary to reduce gender disparities in the workplace—but it's tough to apply a numerical value to the tipping point.

"I think it's important to recognize that being on equal footing doesn't necessarily mean 50-50," said Ana Mari Cauce, dean of arts and sciences at the University of Washington. "There are always going to be gender differences, people are going to be attracted to different things and there will be some gender component to it. I also have no doubt that we will get to the point that there will be a good critical mass of women across probably every scientific discipline. But I am not sure that there is some magical 50-50 that once we reach it we will know everything is fine."

But why is it getting worse?

Do all of the theories for the wage gap point to a lack of agency on the part of women? Almost all but one.

Every data set that has looked at income trends based on gender has some limitation. Dr. Jena's study used exclusively U.S. Census data, which couldn't account for specialty, for example. But one thing that no major study has been able to account for is certain non-remunerative characteristics of employment—a flexible schedule or generous vacation time, for example.

"Instead of being penalized because of their gender, female physicians may be seeking out employment arrangements that compensate them in other—nonfinancial—ways, and more employers may be beginning to offer such arrangements," the Health Affairs study notes. 

This is maybe the most reasonable explanation for why things are getting worse, not better. (The other, less kind alternative is that healthcare is an insular industry that hasn't kept pace with sophisticated business sensibilities of the consumer world. A fair charge? Maybe. It might explain why the pay gap for executives is shrinking: MBA programs teach the expression "glass ceiling," while med schools don't always.)

The number of women in healthcare is rising. As of 2011, almost half of all U.S. medical students were women and women were projected to make up approximately one-third of the total physician population by the start of the decade. As more and more women enter the healthcare workplace, it's possible that they have reshaped the job marketplace.

"Physician practices may now be offering great flexibility and family-friendly attributes that are more appealing to female practitioners but that come at the price of commensurately lower pay," write the authors of Health Affairs study. "Such an explanation not only is consistent with the pattern observed in the data, but it also suggests that the continued integration of women into the physician workforce is reshaping the practice and business of medicine in ways that need to be measured by variables that are more subtle and comprehensive than salary."

If this is the culprit behind the growing wage gap, it raises its own host of unanswered questions. In light of sobering physician burnout statistics, the industry is giving more attention to quality of life issues. Do women value quality-of-life than their male counterparts? Does that vary across specialty? The changes in the wage gap have been inconsistent across job type:

Those data are from Jena's census study, and he notes that the change in the physician salary gap—from 20% to 25%—is probably not all that significant. The larger shift he points to is the widening gap in physician assistant pay.

"During this period of time [from 1987 to 2010], the work of physician assistants changed," Jena says. "Now it's a relatively lucrative job in healthcare. Like being a nurse practitioner, it’s a relatively prized job. As the demand for that particular occupation is going up—reflected by the doubling of income, which you don't see anywhere else—that difference is probably starting to grow because whatever is driving that disparity in general, that increased demand is stretching out the male-female differences and making the more pronounced."