The economist Heather Sarsons doesn’t usually think too much about medicine or healthcare.
SARSONS: When I started my Ph.D., I was pretty interested in looking at why we still see gender and race inequality in labor markets.
Heather’s a professor at the University of British Columbia, and most of the time, her work focuses on discrimination.
SARSONS: A lot of the models in economics focus on two types of discrimination: taste based, which is the idea that people just may not like other people from a certain group, and then statistical discrimination.
Statistical discrimination can also be harmful but in theory it doesn’t arise from prejudice, or racial or gender bias. Instead, it relies on obvious traits — like race or gender — to make generalizations about a person or a group.
SARSONS: If you take the example of say, women, studying math, if historically women are less likely to have invested in math and science skills, then employers knowing that might hold women to a higher standard. And so a woman really has to send a strong signal that I’m very good at math. I’ve really studied this.
But it can be exhausting to have to send a strong signal.
SARSONS: It at some point becomes sufficiently costly to have to study so hard, take so many difficult courses just to signal that you’re as good as men. And basically the employer’s beliefs are reinforced. Women don’t invest in those skills and the employer is correct in thinking that women are less likely to have invested in these skills.
The scenario Heather describes is a self-fulfilling one that’s often based on perceptions. It could be perceptions about anything: how hard-working someone is, how smart or capable they are, how affable they are. But the challenging thing about perceptions is that they’re hard to prove with numbers. If you’re creative though, it’s not an impossible problem to solve.
SARSONS: I was interested in this question about how we attribute success and failure. If someone performs really well at their job, do we think that, that person’s really great at this job? Or do we think that they kind of got lucky or got help? If they performed poorly one day, do we think that they just had a bad day, but they’re still good? Or do we think this person isn’t very well suited to this job?
Heather was looking around for a setting to explore these questions and stumbled upon a study on referral patterns between physicians. Which got her thinking:
SARSONS: If I refer a patient to someone, the patient doesn’t do well, do I then refer to someone else or do I keep referring to that surgeon?
SARSONS: Does my decision depend on the gender of the surgeon who performed the surgery?
From the Freakonomics Radio Network, this is Freakonomics, M.D. I’m Bapu Jena. Today on the show: we’re going to talk with Heather Sarsons about the paper she eventually wrote to try to answer that question.
SARSONS: Other men are kind of unaffected by the actions of one man, whereas other women are affected by what happens to one woman.
But first, we’ll talk with a surgeon about how she’s trying to fix bias and discrimination in her field — for everyone’s benefit.
SALLES: When we don’t support women physicians that means we’re not supporting the patients who are being cared for by those women physicians.
* * *
SALLES: My name is Arghavan Salles. I am, a surgeon by training, but, actually at the moment, I’m not doing any clinical work. I do research and program development in the areas of diversity, equity, and inclusion.
JENA: And where do you work?
SALLES: Ah, I work at, Stanford University School of Medicine.
JENA: I’ve heard of that. Don’t be ashamed.
Dr. Arghavan Salles describes her decision to become a doctor as “rational.”
SALLES: I was doing biomedical engineering as an undergraduate. And I, for lots of reasons, decided I did not want a career in engineering. And so I thought, what other career pathways are available to me without having to take additional courses? And medicine was right there.
In medical school, she sometimes felt dissatisfied by the rote memorization that’s usually required. That all changed, though, in her third year.
SALLES: The third year was like a totally different environment. You’re in the hospital learning something new every day. But why did I choose surgery? I thought it was fun. I was lucky that I had trained with or had exposure to several really awesome women surgeons. And they were inspiring to me. They were the type of person I thought I wanted to be.
Surgery is an area where female physicians are particularly underrepresented: on average, just 12 percent of surgeons are female. Arghavan wasn’t necessarily aware of this statistic when she was training to become a surgeon but soon enough, she could feel it.
SALLES: When I was an engineering student and then in medical school, I was so blissfully unaware of gender bias. And then I started my residency and it was the strangest experience because all of the things that I had done that had made me, reasonably successful to that point in life which is show up early, stay late, do all the extra work, being super diligent. All of that, I kept doing and for the first time it was met not with positive energy, it was met with a lot of negativity and I really didn’t know how to understand that.
To try to understand it, during her surgical training Arghavan ended up also studying social psychology.
SALLES: I started learning about stereotype threat, which impacts people’s performance and started to understand the experiences that I had had for two years in the hospital as a surgical resident. And, that shifted everything that happened for me afterward, even to now.
Stereotype threat refers to a situation where an individual feels like they could be at risk of confirming a negative stereotype about a group they identify with — like women. As she started her career as a busy surgeon, Arghavan started to notice she was being treated differently. For instance, when surgeons operate, they need certain tools that are specific to the case and that the surgeon feels comfortable using. Those needs vary from case to case and from surgeon to surgeon. So, surgeons have something called preference cards that tell operating room managers which tools to have ready before an operation starts.
SALLES: It’s like a recipe for how we’re gonna bake this cake, which is doing this specific case. And I worked for four months without preference cards.
It wasn’t clear from our discussion why Arghavan didn’t have preference cards, whether because of discrimination, oversight, or some other reason. In any event, not having preference cards could affect not just her, as the surgeon, but also the staff meant to support her and, of course, her patients. She suspected her experience wasn’t unique, and wanted to better grasp what was happening — and also why it was happening.
SALLES: Most of us show up to an interaction with all the knowledge and experience of everything that’s come before. When people say things like women need to just be more confident, why are women not confident you think? It’s because of all the messages we’re constantly receiving about how we are too aggressive or not aggressive enough, or too outspoken, or too quiet, or too direct, not direct enough. So changing ourselves won’t solve it. It’s changing how we all perceive women’s behaviors that’s going to solve the problem.
JENA: Can you talk to me a little bit about your work in this area on bias and microaggression and other discriminatory practices towards women in the field?
SALLES: So we’ve done some work looking at, written evaluations, how people write about trainees. Looking for differences based on the gender of the trainees. And, we found that people are much more likely to speak about men trainees using superlatives, like how their performance was outstanding, for example and they’re more likely to make clearly positive statements about their future with saying things like, “Sure to be a leader in academic medicine,” but they don’t use that same language for women. People are more likely to hedge with women’s evaluations and say things like, “she works really hard, I think she might do well.” We found those types of differences in the evaluations. We also looked at bias among healthcare workers. We looked at data from the Implicit Association Test on gender and career, which basically uses reaction times to say, do people associate men or women more with career and men or women, more with family? What we found was that both men and women are more likely to associate men with careers and women with family. And then when we had a sample of surgeons take a similar I.A.T. that was looking at specialty and gender, both men and women surgeons were more likely to associate men with surgery and women with family medicine. And so that explains kind of why it is that people are so surprised even today to see a woman surgeon or to see a woman physician, because that’s just not what we expect.
JENA: Another thing that I think it’s hard for people to appreciate and it took me a while to appreciate, is that there’s an incentive problem that is created here. When there’s inequity or disparate treatment of people, they react differently and they will respond differently, not only in that instance, but in the future as well. And what are we missing out on as a result of those changes that people make to how they act, how they behave, what goals they strive for.
SALLES: The other study that I think is really relevant to the conversation we’re having now is a study we did where we interviewed about 45 surgeons across the country to understand the different challenges that people face in building their practice. And the paper was specifically about the relationships between women physicians and women nurses. And admittedly, we didn’t interview nurses, so we don’t have data on the perspectives of the nurses or their intentions. But what came through very clearly in almost every single interview, was that the surgeons believe that nurses expect different things from women physicians than they do of the men. That they expect women physicians to do what we call performative niceness, which is just being super extra sweet and saccharine in all of our interactions. That they expect us to be more available, to show up early, to stay late, that they expect us to take on additional work that is not physician work, I’m not saying that nurses actually expect these things. I’m saying that this is what the surgeons believe the nurses expect. And so having these beliefs altered the behavior of the women physicians to the point where they described baking cookies and bringing in brownies and asking people about their children’s soccer games and doing all of this to build social capital explicitly so that they could gain cooperation for the patient care activities and support that they needed. The reason we wrote this paper was to say, this is additional labor that women are having to do, that the men do not have to do.
JENA: So take for example, your study on the implicit association test. How do we distinguish a belief versus the actual statistical issue, which is that there are many more men in neurosurgery, than there are women? And that proportion is much different in family medicine. So how much of the I.A.T. reflects just the reality or almost like a statistical understanding of people (AS^Mm-hmm.) versus telling us something about expectations? So for example, in the surgeon nurses study, it’s hard to distinguish between are surgeon’s perceptions of nurses coming from what women physicians are doing because we actually know they’re doing a lot of those things. Or is it coming from an actual differential expectation?
SALLES: It’s a great question and I can say, especially for our interview study that you just referenced, we know clearly what’s happening there because people told us, I mean, we had women who described not having engaged in those behaviors and having faced severe consequences, negative consequences. Now is that what’s happening on the nursing side? I don’t know and actually we’re working on a study to understand that. But what seems very clear from both men and women surgeons we interviewed is that they believe there is a different expectation for women’s behavior. And they believe this because they’ve seen what happens when women don’t do those things. When I was like a second year resident in the I.C.U., I watched the other residents who were more senior to me, I watched how they interacted with nurses. And so I would do like the men did and walk into a room and say, “Hey, I’d like to do this, I’d like to do that.” And I’d walk out only to find that they were then upset with how I had spoken with them. Or I would put in orders and follow up later to see what the results were and notice that the tests weren’t ever sent. And that experience didn’t really appear to be there for the men.
JENA: And we care obviously about the wellbeing of our healthcare workers and equity and the incentives that they have to invest in their careers and their occupation, but there’s a potential impact on patients as well.
SALLES: Oh, absolutely. When we don’t support women physicians that means we’re not supporting the patients who are being cared for by those women physicians.
So, what about the patients?
SARSONS: Maybe women don’t wanna take on risky patients because they know that they’ll be punished if something goes wrong.
After the break: what can referral patterns between physicians tell us about discrimination in medicine?
SARSONS: There’s a gap between, what happens to men and women.
I’m Bapu Jena, and this is Freakonomics, M.D.
* * *
SARSONS: I’ve done some work looking at who gets credit when men and women work together on a paper.
That, again, is the economist Heather Sarsons.
SARSONS: In economics, the way that co-authorship works is you write the paper and you list the authors alphabetical, by last name. If one person contributed more to the paper, that’s not gonna be immediately obvious. And you do see that men tend to get credit for papers if they’re co-authored with other women.
As Heather told us earlier, a lot of her work focuses on discrimination, which can come in a variety of flavors. It occurred to her that one flavor could be how people attribute success and failure.
SARSONS: If someone performs really well at their job, do we think that, that person’s really great at this job? Or do we think that they kind of got lucky or got help? If they performed poorly one day, do we think that they just had a bad day, but they’re still good? Or do we think this person isn’t very well suited to this job?
Heather wasn’t sure how to answer these questions, until she came across a paper from 2020, by the economist Dan Zeltzer, that looked at referral patterns between physicians and surgeons.
SARSONS: He shows that female physicians are more likely to refer to female surgeons and men to men, and it was a really nice dataset to try to test this question of whether there’s bias in how people attribute successes and failures.
In that data, Heather could see which physicians were referring to which surgeons and also, the outcome of each surgery — which was a crucial element.
SARSONS: I tried to trace out how physicians’ referral patterns changed after they referred a patient to someone and that patient either did well in the surgery or did poorly in the surgery.
But Heather wasn’t only interested in how a good or bad outcome changed a physician’s decision to continue referring to a particular surgeon.
SARSONS: Specifically, I look at whether that decision depends on the surgeon’s gender.I find that, if physicians refer someone to a surgeon, and the patient doesn’t do well, the physician is always, a bit less likely to refer to that surgeon moving forward. But there’s a gap between, what happens to men and women. So female surgeons are much less likely to receive referrals from that referring physician in the future, whereas male surgeons are just slightly less likely.
JENA: Can you put a quantity on it? How much less likely are female surgeons to get subsequent referrals?
SARSONS: The number of referrals drops by about 20 percentage following a bad event. And for men it’s, much smaller, closer to 3 percent or so.
SARSONS: And then after one of these good events, there you don’t really see much of a difference in how the male and female surgeons are treated. Both the male and female surgeons are equally likely to continue getting referrals after a patient is just kind of fine after the surgery. One other important thing to note is that this effect is really concentrated among, what I call new referral relationships. (BJ^Mm-hmm.) So if I’m a physician and I’ve been referring to someone for a very long time, and there’s a bad outcome, then I don’t really change my referrals that much for a man and a woman, I’m kind of equally likely to continue to refer to them even after a bad outcome.
JENA: Why do you think you observe this asymmetric relationship in response from the referring doctor? So if there’s a good outcome, there’s no difference between male and female surgeons in subsequent referrals. Whereas if there’s a bad outcome, female surgeons are less likely to be referred to versus male surgeons.
SARSONS: My interpretation of this is that if someone does have a lot of information about a surgeon, then these biases aren’t really present. I’m treating men and women very similarly, and it’s more if I’m just starting out referring to someone, then that’s where this bias might kick in. If we still think that women are less likely to be represented, that this is a more male career then we might be misattributing these errors or these bad outcomes to someone’s ability, than to luck if the surgeon’s a woman.
JENA: When I first saw this paper the way I thought about it in my mind was that, when a male surgeon has an unexpected outcome we can roughly think about it as the cost of doing business. Surgeries are difficult, they’re complicated, bad things can happen. Whereas when a female surgeon has a bad outcome, it’s somehow reflective of the quality of her as a surgeon or of her surgical skill, which is a very different inference to make based on an otherwise similar outcome.
SARSONS: My sense is that the bad outcomes are much more salient than the good outcomes.
JENA: So you see that the female surgeon who has a bad outcome, she herself experiences a decline in referrals. How big is that in terms of, let’s say their billings or revenue? Is it like a substantive chunk?
SARSONS: If you look across all of the referrals that they’re receiving, a woman gets about 0.25 fewer referrals, in the quarter after a bad event occurs relative to the quarter before. But keep in mind that in this data set, I’m only seeing them getting about, one to two referrals per quarter. So it’s a pretty big effect, but then when you average across all the people that they’re getting referrals from, it’s not too large, in that respect. But I think that’s where this unique aspect of this labor market comes in, where if that was the only person that surgeon was getting referrals from, that would really be a big hit to their career.
JENA: Is the decline in referrals to that female surgeon only coming from the physician who made that initial referral, whose patient had a bad outcome? Or is this socialization or network effect where that female surgeon actually experiences, fewer referrals from other doctors connected in some way to that initial referring physician?
SARSONS: I do see some decline in the referrals that the surgeon receives from other physicians who are in the same practice as the referring physician. (BJ^Huh.)
JENA: That is interesting. Is there a broader effect on female surgeons in general? For example, female surgeons in the same specialty, do they, see declines in referrals? Is the referring doctor gonna say, “Look, you know, I’m inferring something about the quality of female surgeons from this one data point and I’m actually gonna let that influence my referral decisions to other female surgeons who are in the same specialty.”
SARSONS: The most depressing part of the paper in my eyes was you do see that the physician who made this referral to a female surgeon when the bad event happens, also becomes less likely to refer to other female surgeons who are in that same specialty. (BJ^Wow.) So if I refer someone to a neurosurgeon, the neurosurgeon’s female and the patient doesn’t do well, then I become less likely to refer to other female neurosurgeons moving forward and that’s not something that we see when we look at referrals to other men. Other men are unaffected by the actions of one man, whereas other women in that same specialty, they are affected by what happens to one woman.
JENA: Obviously, you can’t speculate as to what might be going through doctors’ minds, but do you have any general thoughts about why women surgeons are being punished in a way that men aren’t?
SARSONS: There’s an idea in psychology that, if there’s one group of people who are really underrepresented in, say, a given occupation, then you tend to group them together and update about the entire group based on the actions of one person in that group. That can also be a rational thing to do. If I just have very little experience with people from a certain group, then I’m learning about that group by interacting with one person. The thing is that when I look at these good events, you don’t see that other women get more referrals after a patient does really well. It really seems to be just when these bad events occur, which is in line with some of these concepts in psychology about seeing someone from the majority group as an individual, but seeing the minority group as a group and inferring a lot about the minority group, if you have a bad interaction with someone from that group.
JENA: There’s a lot of ways that this could affect female surgeons. We know that there’s a volume outcome relationship with surgeons, and the more they operate, the better they become. And that might also be true for certain types of patients. So patients who are somewhat risky or specialized, getting experience in those patients is important for surgical, development. Do, you see that the composition of patients that get referred to a surgeon changes after they have a bad outcome?
SARSONS: Female surgeons see this drop in referrals, but if they do continue to get referrals, they tend to get different types of procedures moving forward. So patients who are less risky in procedures that have a lower death rate, for example. If this is a profession where it’s really important to get a lot of practice doing these things and women get fewer chances, then that could hurt their career going forward.
JENA: And how long do these effects last?
SARSONS: The drop in referrals is quite persistent. So a year and a half after, a patient death, you still see, women receiving fewer referrals. The results on patient risk and the types of referrals that you’re receiving are a bit noisier, but they do, persist for about a year or so after the event as well.
JENA: Are there situations where the biases that you’re talking about, might be beneficial? These heuristics, they emerge for a reason. From an evolutionary perspective, they occur because they make our lives easier in some ways.
SARSONS: I guess if you have one bad outcome and then you say, okay, well I’m not gonna do that anymore, I guess it could be helpful. I think the issue is when you’re doing that differently based on something like gender or race. And it’s not saying that you should necessarily be referring to the women more after a bad outcome, but maybe you should actually be referring to the men less after a bad outcome
JENA: Is there a way that this then ultimately affects patients?
SARSONS: If we think that physicians are wrongly moving away from women who just had bad luck or were in a tricky situation, and referring those patients to other surgeons, then patients could be hurt because they’re not going to that surgeon who’s actually very good and just had, one complicated procedure .
Beyond receiving fewer referrals and less complex cases, there’s another, more profound implication for female surgeons.
SARSONS: If people are giving women fewer chances to succeed, it could put them on a different track altogether.
And that’s something Dr. Arghavan Salles knows all too well. Not long ago, after years of feeling unsupported in both her clinical and research work, she made a big decision.
SALLES: I started looking for other jobs. My mental health was suffering quite a lot during this time. I was very concerned that if I took another academic surgery job, that it could be potentially even worse. I was unemployed for a while and then was able to co-create the job that I have now, which is pretty much a perfect fit for me as perfect as a job can be.
When we think about discrimination in medicine, Heather’s work shows us that the effects may not stop at those who work in the field. And Arghavan’s examples aren’t just about slights to women. If a surgeon’s instruments aren’t ready because they didn’t have preference cards, that could mean more time under anesthesia for the patient. And if a surgeon hasn’t been able to acquire skills because one bad outcome led to a persistent drop in referrals, that could impact patients too. It could also affect the number of women who want to get into the field at all. It probably already has.
The extent of these issues is hard to know but important to understand. We have the data to figure out whether and how workplace discrimination could ultimately find its way to patients. Which means it’s something all of us should care about, no matter which side of the stethoscope we’re on.
That’s it for today’s show. I’d like to thank my guests, Dr. Arghavan Salles and Heather Sarsons. And thanks to you, of course, for listening!
This reminds me of a familiar riddle you probably already know: A man and his son are driving on the highway. They get into a car crash and both the man and his son are severely injured. The son is taken to the emergency department. When he gets there, the surgeon walks into the room and says “I can’t operate on this child. He’s my son.” This riddle stumps a lot of people, all kinds of people, because of the assumptions we make. And as we’ve seen today, those assumptions can have consequences.
Email me your thoughts about this episode! I’m at email@example.com. That’s B-A-P-U at freakonomics.com. Or, leave us a review wherever you get your podcasts. Coming up next week on the show:
DISIS: I’ve been studying tumor immunology for 30 years, and I can tell you that 30 years ago, people really didn’t think the immune system had any role to play in cancer.
So, what do people think now?
DISIS: Literally every day things are being published where I’m like, “Wow, I really need to know this. This is life altering!”
That’s next week on Freakonomics, M.D.
* * *
Freakonomics, M.D. is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and People I (Mostly) Admire. All our shows are produced by Stitcher and Renbud Radio. You can find us on Twitter at @drbapupod. This episode was produced by Julie Kanfer and mixed by Eleanor Osborne. Lyric Bowditch is our production associate. Our executive team is Neal Carruth, Gabriel Roth, and Stephen Dubner. Original music composed by Luis Guerra. If you like this show, or any other show in the Freakonomics Radio Network, please recommend it to your family and friends. That’s the best way to support the podcasts you love. As always, thanks for listening.
SARSONS: I think I’ve gotten into a bad research cycle of pointing out problems and then not coming up with solutions.
JENA: There’s nothing wrong with that.
- “Gender Differences in Recognition for Group Work,” by Heather Sarsons, Klarita Gërxhani, Ernesto Reuben, and Arthur Schram (Journal of Political Economy, 2021).
- “Unpacking the Status-Leveling Burden for Women in Male-Dominated Occupations,” by M. Teresa Cardador, Patrick L. Hill, and Arghavan Salles (Administrative Science Quarterly, 2021).
- “Physician Specialty Data Report,” by the Association of American Medical Colleges (2021).
- “Gender Homophily in Referral Networks: Consequences for the Medicare Physician Earnings Gap,” by Dan Zeltzer (American Economic Journal: Applied Economics, 2020).
- “Assessing Gender Bias in Qualitative Evaluations of Surgical Residents,” by Katherine M. Gerull, Maren Loe, Kristen Seiler, Jared McAllister, and Arghavan Sallesb (The American Journal of Surgery, 2019).
- “Estimating Implicit and Explicit Gender Bias Among Health Care Professionals and Surgeons,” by Arghavan Salles, Michael Awad, Laurel Goldin, Kelsey Krus, Jin Vivian Lee, Maria T. Schwabe, and Calvin K Lai (JAMA Network Open, 2019).
- “Interpreting Signals in the Labor Market: Evidence from Medical Referrals,” by Heather Sarsons (Working Paper, 2017).
- “The Most ‘Unique, Excellent, and Promising’ Episode,” by Freakonomics, M.D. (2021).
- “What Are Gender Barriers Made Of?” by Freakonomics Radio (2016).
- “The True Story of the Gender Pay Gap,” by Freakonomics Radio (2016).