Despite the fact that women comprise half the population and hold over half the degrees in higher education, the number of women in high-ranking positions and in the sciences is disproportionately small. The current consensus is that discrimination, preferences for work hours, and fields of study account for this occupational difference between men and women. In Do Women Shy Away From Competition? Do Men Compete Too Much?, Muriel Niederle and Lise Vesterlund find that women generally like competition less than men. They explain that this difference contributes to the paucity of women in top jobs.
The authors came to this conclusion through an experiment they created. In this experiment, 80 volunteers completed a task that paid a non-competitive piece rate and then a competitive winner take-all tournament. In each system, the groups consisted of two men and two women. Of course, this group division was only important in the winner take-all system because there they were actively competing against each other. In the non-competitive game, participants received $0.50 for every set of five two-digit numbers they correctly added in five minutes. In the winner take-all system, the three losers made nothing but the winner who summed the most sets of five two digit numbers received $2.00 per correct answer. After playing in the non-competitive and competitive schemes, the authors asked the volunteers to choose whether they wanted to play the competitive or non-competitive game for the third task.
The first finding was that the average man and woman were equal in their abilities. For the non-competitive game, women solved on average 10.15 problems while men answered 10.68. There was also no significant gender difference in performance in the tournament. Women solved on average 11.8 problems correctly, and men 12.1. Of the 20 tournaments, 11 were won by women and 9 by men. These gender differences are not statistically significant. This improvement between the non-competitive and competitive games may be caused by learning (because the non-competitive game came before the competitive game) or by the increased performance incentives of the tournament. The authors noted however that the increase in performance from the piece rate to the tournament does not differ by gender.
Despite their similarities in aptitude, for the final game, 73% of the men preferred the competitive tournament game compared to only 35% of the women. Both men and women who won the first tournament game were more likely to enter the tournament game for the third round, which is rational behavior. However, men were much more likely to enter at every ranking from the first competitive game. Men in first place, second place, third place, and fourth place were more likely to play the competitive game than women in those respective positions. The authors then controlled for probability of success in the competitive system by comparing men and women with the same scores in the initial competitive trial. Among the winners from each initial competitive game, women were 38% less likely to re-pick the competitive game. The authors concluded that among the highest-performing individuals men were more likely to enter the competitive game and, consequently, profited more. However, among the lower performing participants, men chose the competitive game too often and, as a result, made less.
The current explanation is that discrimination and differences in the preferable number of work hours are major causes of the small number of women at the top. However, these two factors were absent and still a difference in the highest earners of the two genders was observed. The scoring was objective; there was no discrimination. The time for each task was limited so the idea that women work less because they have a stronger domestic tie to the house and children cannot account for the difference here either. With this experiment, the authors demonstrate that there are other factors contributing to the differences of the number of men and women in top positions.
The authors believe that although men are not actually better at this task, they are more confident about their ability to win in the competitive system. Men and women were both overconfident in their skills, but men were much more so. The authors asked the participants after the competitive game if the believed that they had won. 75% of men and over 43% of women believed that they had won their tournament. (If men and women had the correct level of confidence, only 25% of each gender would have believed that they had won or played the competitive tournament game for the third trial.)
The authors then studied if risk, feedback aversion, or additional factors affected the gender difference in compensation choice. They demonstrate that even when controlling for these factors, there was a large gender gap in tournament entry. They concluded that there are large gender differences that affect the decision of choosing a competitive environment, and this fact must be taken into account when studying why women are underrepresented in many top jobs.
I think an interesting twist to the games that would better mimic the competition of achievement would be to have active scoring during the game. In this system, the number of questions each player had answered correctly at that point in the five minute game would be illuminated on the scoreboard for all the competitors to see. I would do this in the non-competitive and competitive games because in the competitive game it is possible that someone would get behind and simply give up. It would be interesting to see if women think that they should do worse and whether this would become a self-fulfilling prophecy. I wonder how their scores in the previous game, which were essentially anonymous, would match up to their scores in the new game, where the scores are clearly presented throughout the game. Also, this new would further delineate the differences of risk and competition. The authors try to control for risk in later rounds, but I think this new game would give a more complete answer. I question the validity of their controlling for risk. I see someone who enjoys seeing the score of the other competitors and trying to beat them as competitive, while an individual who does not choose the competitive tournament game for the third task as being risk averse instead of not competitive.
I believe that the experiment was taking the right approach but was too limited in scale to produce anything truly meaningful. For one, I believe that having only 40 men and 40 women compete is not enough. One-hundred of each would provide a much more powerful statistical measure. To me the difference in the number of questions answered between women (10.15) and men (10.68) is substantial. I believe that as more data is produced, these two numbers will converge.
The difference between a world class 100 meter sprinter and a good high school sprinter is not much more. In that case, the world class runner receives a million dollar sponsorship from Nike and the high schooler makes nothing—the difference is huge. You can continue this analogy to their being very few people who are at the top and who can be CEOs and everyone else.
The authors discuss how both genders performed better in the competitive game. I believe that they should have differentiated between this coming from a learning effect or from competitiveness. To test the difference, they would just have to have the contestants play a third non-competitive to see if scores continue to improve from a learning effect.
Also, I am uncomfortable with the implication that people who play the riskier game are more likely to take risks in life to become the high-powered people. I would play the riskier game because I would rather have a 25% chance of making twenty dollars than a 100% chance of making five dollars. Five dollars does not buy anything but twenty dollars would be nice. My logic is something along the lines of why people play the lottery despite the lower expected return. It would be interesting to create a tournament game where the expected return was lower than in the non-competitive game. In this case, people who selected the non-competitive game would be rational. Maybe women are more rational and, therefore, better managers and decision-makers.
Furthermore, when choosing professions, the risk of losing from competition may not be drastic. If a woman aims to be a CEO and fails, she will likely have more power and a bigger paycheck than if she chose not to compete. The drawback to competing is minimal. In this game, it was large.
Otherwise, I thought the paper was a pleasure to read but lengthy. It is a good paper and one that should generate discussion.