TCS Daily

Ego, Testosterone, and the Academy: Why the Controversy Over Larry Summers is Important

By Arnold Kling - February 21, 2005 12:00 AM

"if one is talking about physicists at a top twenty-five research university, one is not talking about people who are two standard deviations above the mean. And perhaps it's not even talking about somebody who is three standard deviations above the mean. But it's talking about people who are three and a half, four standard deviations above the mean in the one in 5,000, one in 10,000 class."
-- Lawrence Summers

Harvard President Lawrence Summers gave an informal talk at the National Bureau of Economic Research that some of his opponents denounced as sexist. Now that a transcript of the talk has been posted, it is clear that it as near a perfect example of judicious, thoughtful speculation as any imperfect human being might produce.

I have three daughters. Two are in college (not Harvard), and one is in high school. Their educational experience is already polluted by the anti-intellectualism represented by the hostility to Summers' talk. To fight this pollution, the faculty at Harvard and other high-caliber universities owe Lawrence Summers nothing less than full vindication and endorsement.

True or False

At the University of Maryland, my oldest daughter, Rachel, took a class in which one test included a question in which she was asked to respond to the statement "Gender is socially determined." This was given, not as an essay question, but as a machine-graded true-false choice. Having read the textbook for the class, Rachel knew that the machine would treat "true" as the correct answer. She herself believes that the answer is something other than "true." Perhaps, if given an opportunity, she could have written a thoughtful, balanced essay on the topic. Evidently, however, her professor does not have a sufficiently open mind to be willing to face such an essay.

The question facing Lawrence Summers as he gave his talk was, "True or false: the explanation for the high ratio of males to females in physics, math, and engineering at universities like Harvard is cumulative sex discrimination." Evidently, the textbook answer is "true." Instead, Summers gave a thoughtful, balanced essay answer that was something other than "true." For that, many modern academics, including some smug critics at MIT and other prestigious institutions, believe he deserves a bad grade. Shame on the critics. Praise to his defenders.

The Becker Test

At one point in his talk, Summers trotted out a thesis first articulated by Nobel Laureate Gary Becker, who argued that discrimination harms the discriminator. Becker's point is that it is inefficient to allow non-economic factors to affect a market decision. Therefore, the discriminator will achieve sub-optimal results, either in terms of consumption or profits.

Suppose that in math and science, some departments discriminate against women. Then according to the Becker-Summers argument, those departments, because they are willing to choose second-rate men over first-rate women, will be at a disadvantage relative to departments that are either nondiscriminatory or which discriminate in favor of women. Summers argues that we do not see evidence of these "profit opportunities." As Summers admits, this is a limited argument, because it does not speak to possible discrimination earlier in life. However, it does give pause to anyone who wants to give the automatic "True" response of discrimination.

Gender Differences

Having argued that the Becker Test indicates fails to indicate discrimination, Summers continues:

"So my best guess, to provoke you, of what's behind all of this is that the largest phenomenon, by far, is the general clash between people's legitimate family desires and employers' current desire for high power and high intensity, that in the special case of science and engineering, there are issues of intrinsic aptitude, and particularly of the variability of aptitude, and that those considerations are reinforced by what are in fact lesser factors involving socialization and continuing discrimination. I would like nothing better than to be proved wrong"

Summers argues that to be a professor at a top university is to be at the very top of one's profession, just as a corporate CEO is at the top of a firm. He says that to reach the top of a profession, one must dedicate an inordinate amount of time. He says that this need for professional dedication conflicts with family responsibilities for both men and women, but this tends to take more of a toll on women.

Summers' other point concerns statistical distributions. On a variety of attributes, statistical measures show that men have higher variance than women. Thus, if you look at the very top or at the very bottom of the distribution, you will find a larger share of men, while if you look in the middle, you will find a slightly larger share of women. He conjectures that this difference at the extremes exists for some attribute that is important in math and some branches of science. If to be at the top of one of those fields you need a genetic trait that is found only once in every 5000 or 10,000 people, and if rare genetic traits are more often found in men, then when you look at the top of those fields you will see more men.

Male-Dominance Behavior

At this point, I am going to engage in some anecdotally-based observations on gender differences. To his credit, Summers mostly avoided such stereotyping.

I believe that it is characteristic of males more than females to focus on the "dominance hierarchy." Males are very concerned about "whose is bigger," and this shows up particularly when a group of males gets together for the first time. They compare, they boast, and they try to assert superiority. I noticed this when my first-year graduate school class at MIT met on the lawn the day of student orientation. It made me uncomfortable then, and such behavior has made me uncomfortable ever since.

Another example was in 1998, when my Internet company, Homefair, merged with a similar company, The School Report. At the first meeting of the executives of our two companies, our CEO and the other company's CEO turned the occasion into a male-dominance contest between themselves. In the process, the rest of us, used to being treated as equal partners, instead were treated as lackeys. It took some effort to lower the testosterone levels enough to restore organizational balance and harmony.

Male-Dominance Behavior can be summarized in the following table. It shows how someone in male-dominance mode bases his treatment of another person on his perception of that person's place in the relevant pecking order.

Relative Ranking



Higher than Self

Potential Relationship


Equal to Self


Show Off/Put Down

Lower than Self



My experience is that in male-dominated organizational groups, one observes this sort of behavior. Guys tend to flatter the boss in the belief that this will help them build a useful relationship. They try to make their rivals look bad and try to make themselves look good. They disregard people who they think are not in a position to promote them or to compete with them.

My sense is that women find male-dominance behavior annoying. They particularly dislike being treated as "irrelevants" during meetings. I can understand their point of view. I avoid the American Economic Association meetings, in part because I am sickened by the flattery and the Show Off/Put Down. Above all, being treated as an "irrelevant" rather than as a fellow human being by people you once thought of as friends and colleagues is a highly discouraging experience.

So to Lawrence Summers' list of possible reasons that women are under-represented in some fields, let me add annoying male-dominance behavior. To the extent that one must put up with or join in such behavior to succeed in largely-male fields, I could see where otherwise qualified women might not have the taste for it.

Self-regarding Attribution Bias

Another of my anecdotally-observed gender differences concerns self-regarding attribution bias. Self-regarding attribution bias means that you internalize success and externalize failure. Suppose that you have bought and sold many stocks over the years, with mixed results. If you tend to view the profitable investments as reflecting skill, while viewing the losses as reflecting bad luck, then you have this bias.

In my observation, more men tend to have this bias, and more women tend to have the reverse bias. That is, men tend to credit themselves with success and blame others for failure. Women are more likely to credit others for their success and to blame themselves for failure. I believe that the process for selecting CEO's for large organizations tends to select for self-regarding attribution bias. I believe this accounts for the susceptibility of corporations to scandals - the Darwinian process in corporate America too often rewards men who think they are immune to any adversity. I think it is far less likely that women - or, more generally, someone who is capable of internalizing failure and recognizing external factors in success - would lead a company to scandal.

Venture capitalists, because they believe in swinging for homeruns, seem particularly inclined to select in favor of self-regarding attribution bias. This tendency led to the excesses of the dotcom era, as documented in The New, New Thing, for example. Venture capital is another field where women are notoriously under-represented, particularly as CEO's of venture-funded firms, even though women have a high rate of entrepreneurial success otherwise.

The Biggest Danger

Perhaps male-dominance behavior and self-regarding attribution biases are issues that our society needs to address. Perhaps we would be better off if we saw less of those forces at work selecting CEO's - or college Presidents.

But the biggest danger on campus today is not discrimination against women. It is discrimination against thinkers like Lawrence Summers and my daughter, who should not be forced to affirm propositions that they believe are complex and far from completely true.


TCS Daily Archives