Dr. Lise Vesterlund (University of Pittsburgh) delivered a seminar to us on Friday afternoon. Interesting and provocative stuff–everyone should read the paper (Niederle, Segal, and Vesterlund, forthcoming) with the appropriate hesitation and humility to prevent wide and sweeping inferences. Links are at the bottom of this post.

The paper covers a set of experiments where men and women were asked to add up 5 2-digit numbers in their heads, repeatedly for 5 minutes. In the first task, they were paid $0.50 for every correct answer. In the second task, they competed against their group (2 men and 2 women in the Pittsburgh experiment, 3 men and 3 women in the Boston version) in a winner(s)-take-all version. In Pittsburgh, the one with the highest score received $2.00 for every correct response, everyone else recieving nothing. In Boston, the top two performers each received $1.50 per response, everyone else recieving nothing. In the third task, they were given the opportunity to choose to either compete in the tournament again or get paid a piece-rate again, and then, in the fourth task, they were given the opportunity to submit a previous piece rate performance for competition or piece-rate payment. In the Boston experiment, they added two more tasks involving a form of Affirmative Action, in which the winners were the highest woman performer and the highest performer from the remaining five in a tournament, then with the opt-in tournament. Finally, having seen none of the results the whole time (making all decisions based on priors and their own performance), subjects were asked to self-assess relative to their group, and in Boston, relative to the group members of the same gender.

In both studies, the women systematically opted in less often than the men, controlling for performance. So, in Pittsburgh, men and women performed equally. In Boston, the men performed slightly better (due to a few outliers at the top of the distribution), but in both cases, controlling for performance, when given the opportunity to enter the winner-take-all tournament, men were far more likely to enter the competition (73% of all men entered in Pittsburgh, 35% of all women).

The upshot I took from the presentation was that there could be three major driving reasons why the women in the experiment were less likely to enter: beliefs about performance, desire to compete, and risk preference. In reverse order,

Risk preference

It’s unclear how much of an impact, if any, this has on the gender gap. When comparing the decision to submit a previous performance to a tournament–basically, to take a gamble, rather than a sure bet–those who thought they performed best were equally likely to submit their performance, male or female. It doesn’t appear that risk alone, the desire to take a high-stakes gamble on a past performance over a sure thing, can account for much of the difference.

Desire to compete

When looking at the desire to enter into a high-stakes gamble on a future performance, however, men were much more likely to enter than women, even controlling for their own beliefs about their chances. About 50% of the women who thought they ranked highest entered the tournament, while about 80% of the men who thought they ranked highest entered. That’s a 30 percentage-point difference. Which is big. So there’s something about the prospect of the pressure of competition that men seem to like and women don’t. Cultural? Genetic? Who knows. I think it’s probably cultural, but a lot of people seem to think there’s a genetic basis–competition for mating or food, etc. The distinction, though, between future performance or past performance is important, I think, because it has a lot to do with the state of mind while performing.

This has some implications for how we can encourage high-performing women (and discourage low-performing men, I suppose). If it’s cultural, then praising competition among women could get them more interested in entering high-stakes professions where their abilities can do a lot of good. If it’s genetic, then changing the structure and mechanism of workplace advancement could be another way to encourage capable women to work their way into positions of power.


Finally, and I think this is probably the most striking thing that I heard, both men and women are wrong about their own performances. Dr. Vesterlund said that they were both overconfident, but that’s not really correct. Instead, they both suffer from the Lake Wobegon effect, which I think has been misdefined as, again, systematic overconfidence. Instead, I think it’s that everyone tends to shift their self-assesment toward “above-average,” and so everyone below the 70th %ile (or so) is overconfident, while everyone above that line is underconfident, and people on that dot tend to have pretty good estimates (by sheer chance, really) of their abilities. There’s a paper on competence and the ability to self-assess, but for the life of me, I can’t remember the citation, so I’ll try to come up with it and repost.

So everyone’s beliefs are off. Low-performing men overestimate their ability by more than low-performing women do. High-performing women underestimate their ability by more than high-performing men do. Basically, the above-average line that men and women move their estimates toward (at least in the computation of the sum of five two-digit numbers) is different. An unbiased estimate of performance for an average performer would be (on a rank of 1-4, 1 being best) 2.5, right? Half above, half below. In Pittsburgh, the men and women performed equally well, and so both averaged about 2.5. The mean male estimate of own rank was 1.4 while the mean female rank of estimate was 1.825. Fully 3/4 of the men (30 of the 40) thought they had done the best in their group. Only 17 of the 40 women thought they had done so.

To me, that seems huge. Even setting future competition aside, women were as likely to submit a past performance for competition as men, controlling for self-estimated performance. But if self-estimated performance is so highly skewed, we still have women being far less likely to compete than men. Additionally, Vesterlund and Niederle find them far less likely to compete than they should be willing to do to maximize total social gain.

So to recap, even when they outperform their male counterparts, women don’t mind risk that much, they don’t enjoy timed competition, and they don’t think they’re as good as they should. Men enjoy competition, even when they know they’ll probably lose, and they think that they’ll lose more rarely than they actually will.

The conversation after the presentation was as interesting as the presentation itself. The business world is real-time winner-take-all–no wonder so few CEO’s are women. So are politics and many other professions. More than that, how have we managed to inculcate such lowered self-assessments into the young women who were the subjects of this study? Are these results applicable across the population?

If we let people know how they do, will men be disappointed and stop striving? Maybe women will be so pleasantly surprised to find out that they’re better at almost everything then they think they are (or better than people they thought they were worse than, at least), and will turn that sense of new ability into heightened ambitions and desire to compete.

Girl or boy, Pickle will not suffer from a lowered self-assessment. That’s not like a pledge, it’s just an admission. The Delaney family has never been one that liked to encourage humility for humility’s sake, certainly not in measures of performance. Perhaps in measures of value as a human being or comparative worth or self-righteousness, but if you’re good at math, you’re good at math. If you’re smart and good-looking and can spell and like to read heavy stuff and derive functional forms for everyday occurrences and you’re funny and quick of wit, a formidable disputant and everyone likes you, then hey, you’re one of us, no surprise there, right?

Still, Dr. Vesterlund had a very good point. What can we do as a society, really? We have blocks in our path. Even her most liberal friends, who want to raise their children gender-neutral, don’t give their sons dolls and skirts. They just give their daughters baseballs and math books and pants. In America, growing up gender-neutral means being brought up as boys have historically been brought up. What does that say about our value of a traditionally female upbringing? It’s got to be possible to get the parts of that that would benefit boys and girls without the parts that would damage them both, in my opinion. It wouldn’t have killed me to learn how to cook a decade earlier than I did. I did my laundry from age 9 onward. I still can’t sew a button very well, but I hope any son I have can. To be fair, I had a doll (I stole it from my cousin Kaity and named it “Alice Cooper,” and from age two to three continually had to explain to people that Alice Cooper was a boy’s name, not a girl’s name, which makes me both precocious and incredibly weird), so maybe I’m not the best example, but then again, maybe I am.

Information’s probably good. I always like standardized tests, because they give you a chance to calibrate your self-assessment, and I guess because I usually did well. But hopefully my kids will do well. And if they don’t, then hopefully they’ll be able to acknowledge that standardized tests are just measures along one axis, and not even one that should count for much, in the long run. Still, I think, better to know thyself, painful or no. Hopefully more people feel the same way.

Dr. Vesterlund’s previous paper, which formed the first half of the presentation (and covered the Pittsburgh study), can be found at the link below, with the presentation in the second link.

Do Women Shy Away from Competition?


You can find Dr. Vesterlund’s home page at Pitt here: Dr. Vesterlund’s home page