Halfway through the “Bell Curve,” which is an analysis of differences in intelligence between races, I realized what had been bothering me about Charles Murray’s thesis. It wasn’t the accuracy of his analysis, which concerned me, too. It was that he analyzed. The truth, I used to believe, was always beautiful, whether it was what happened in the multiverse at T equals zero, or the historical counterfactual if Neville Chamberlain hadn’t signed the peace accord with Adolph Hitler. After reading Murray’s book, I realized that the truth can be irrelevant, ugly, and utterly useless. Even if the average intelligence of races was truly different, so what? Surely, civilized people must judge each other as individuals, regardless of the veracity of the statistical baggage of their ethnicities.
Murray was castigated, deservedly, for swallowing the bell curve uncritically. But his detractors missed one point. Murray wasn’t just wrong because he was factually wrong or for inquiring. In fact, it was worse, because Murray, it turned out, was wronger than wrong.
Stephen J Gould, in The Mismeasure of Man (man is gender-neutral here) – the most biting critique of the study of human intelligence –recognized the biggest weakness with Murray’s analysis. Gould always had an antipathy to statisticians – his life once depended on his dismissing bland statistics. Gould observed that when using an attribute, such as IQ, to distinguish between two groups, if the within group variation of that attribute is equal to or greater than the difference in the means between the groups, that attribute becomes a useless discriminant. Neither does IQ predict race nor, reassuringly, does race adequately predict IQ.
You should read The Mismeasure of Man. I don’t wish to ruin Gould’s scholarly prose but let me restate his insight – Murray’s analysis was wrong because it was two lazy generalizations, i.e. aggregates of two bell-shaped curves, curves which overlap so copiously that they may as well be one curve. A lazy generalization – what is that, you ask?
What is the central tendency of a distribution but a lazy generalization? The aggregate, the mean, is wrong about everyone but the few closest to the mean, yet is so revered because we mistake the aggregate for the truth. The tyranny of the aggregate is the most extraordinary tyranny of our times. The aggregate is built by people who vary, yet it imposes itself on the individuals, the very variation which creates it. It literally bites the hands which feed it.
Race and gender are thorny issues and are better kept away from Gauss. Recently, it was gender’s turn – a study found that female physicians have better outcomes than male physicians. This is hardly surprising. In Britain, patients prefer female GPs of South East Asian origin. And if patients prefer female doctors, it follows logically that female doctors will have better outcomes.
But no good deed that is quantified goes unpunished. The study caused disquiet, in part, because instead of celebrating the virtues of female doctors, the media picked up on the vices of male doctors. Because if female doctors are doing better, it must mean male doctors are doing worse, or worse, it must mean that male doctors are a public health menace. This is an incredible problem for our hyper rational times – nothing can be seen for its own sake.
Now, if you’re seeking flaws and obvious confounders in the study, don’t waste your time – there are none. The effect size is real and plausible. Nor will it do good suspecting the researchers of maleficence. One of the co-authors, Dr. Anupam Jena, gallantly, and with infinite patience and boundless grace, defended the paper on social media. Though it is arguable, as David Shaywitz noted, that the researchers would have published the results had the opposite been shown – i.e. if male physicians had better outcomes than female physicians, and I hope that good manners, or at least political sense, will stop the researchers from analyzing the outcomes of doctors based on their race, I have no grounds to doubt the good intentions of the researchers, just as I never suspected Murray of anything but good intentions.
But I’m struggling to see the utility of their findings, in the same way I struggled with the Bell Curve. At a time when patient care is being recognized as a team effort, and a team is made up of both male and female doctors, the study’s Y-chromosomal reductionism seems distinctly out of place.
It may be argued that the one silver lining of the study is that female physicians will be compensated fairly. That should be the case, as Dr. Vinay Prasad notes, regardless of what the study found. Fairness is a moral, not scientific, imperative. Have we become so insensate that we need science to tell us to do the right thing?
Nor will I quibble that because the study was not randomized the results aren’t valid – not because of the impossibility of randomizing patients to different genders, but because the findings are plausible. A randomized controlled trial is not the only source of truth.
The study led to good-natured banter on social media. It gave men a chance to self-deprecate. There’s something endearing about men putting themselves down in front of women (incidentally, we’re supposed to be funnier, but that hasn’t been rigorously tested) and no doubt quoting this study will make men appear charming at cocktail parties and on first dates, and what better charm than evidence-based charm?
But the study led to deductive leaps which, though seemingly logical, is troubling when you explore them closer, such as this leap: if a drug boasted the effect size of being seen by female physicians, FDA would approve it in a heartbeat. This strikes me as such an odd conclusion that I’m hesitant in refuting it, hesitant because when something seems so obviously incorrect to me, I feel I’m being like Johnny’s defensive grandmother who said “everyone is marching out of step with Johnny.”
The comparison of genders to a drug such as a statin which, though not uniform in its action is uniform in composition, collapses genders into a homogenous artificial singularity. It implies that all men are precisely the same and that all women are precisely the same, and you don’t have to believe that gender is not a social construct to realize the speciousness of that implication.
Since when did physician quality, undoubtedly difficult to precisely measure, bifurcate between genders as uniformly as anatomy and chromosomes do? Let me be clear. The aggregate of physician outcomes is built by doctors who vary in quality. Not to put too fine a point on this, there are good and better, bad and worse, best and worst, male and female physicians. This variation in competence is true even if females, on average, make better doctors than males.
Let me give you an example to make my point clearer. Men are, on average, taller than women. That doesn’t mean that there aren’t short men and tall women, or that the shortest man is taller than the tallest woman. If you randomly and blindly picked a man and a woman and tried to guess their gender based on who was taller, you’d be wrong so often that it’d be an embarrassing exercise.
My own career is testament to such variation. I was put off internal medicine by a female physician and drawn to surgery, which I briefly dabbled with and then left, by a male surgeon. I eventually chose radiology because I was impressed with the diagnostic reasoning of a female radiologist. I also admired how Mrs. Thatcher dealt with male snarkiness, because it is snarkiness in men that I find the urge to slay. The gender influences in my life have been a coin toss. To impose upon me a female mentor would be just as illogical as imposing a male mentor.
You might tell me to “man up and face the facts.” But in this post-fact world it is easy to miss the fact that facts are nothing but blurred point spread functions. Taut singularities, where none exist, aren’t facts, but lazy generalizations.
The most egregious misuse of the study is the extrapolation to “32, 000 fewer people would die prematurely every year if seen by female doctors”, even by the intellectually cool Atlantic. This is a death every twenty minutes, which makes male doctors deadlier than Jihadi John. Cumulatively, over time, we have more blood on our hands than Pol Pot. Keynes warned that starting with a mistake, a remorseless logician can end up in Bedlam. But you can get to Bedlam by logical reasoning without making an initial error.
Let’s assume that the estimate of “32, 000 preventable deaths” is correct. Holding preventable deaths as normative has implications for society. Logical consistency is a terrible beast. It implies that every day the FDA delays approval of a life-saving drug they have blood on their hands. It implies that when people stop taking statins for primary prevention because the media is skeptical of the benefits of statins, and some die prematurely, as statistically speaking a few will, the media are complicit in murder. When we stop thinking beyond quantification, our species risks losing the plot.
I ask, what should be done differently to male doctors? Should male doctors be forcibly feminized? The Ottomans castrated the men who guarded their harems – a prescription, I suspect, even radical feminists might find unpalatable. If patients prefer seeing Roshni over Raj, there’s very little Raj can do without relinquishing his sexuality.
No doubt, some male doctors have something to learn from some female doctors and clearly the attributes which make a good doctor are more prevalent in female doctors. But they’re also present in male doctors. If we’re still unclear what these attributes are, the medical profession is in a pitiful state of such enormity that one struggles to comprehend. These attributes, if they must be stated, are genuine empathy, kindness, making eye contact, attention to detail, analytical skills, decision making, among others. Find a physician, regardless of gender, with such skills and learn from him or her.
We’re slowly shedding racial and gender prejudices. The cure to lazy generalizations of one polarity isn’t lazy generalizations of the opposite polarity. The truth doesn’t always set you free. We need more perspective, not more research. The physician gender-outcome study is unlikely to help either male or female physicians. It will least help patients, who will be taxed with yet another uncertainty – the deadly risk of being treated by a male physician.
Saurabh Jha is an opinionated radiologist who can be reached on Twitter @RogueRad