ABSTRACT
The use of essentialist language and an exaggeration of results showing "difference" suggests a tendency to assume, rather than wait to see if one finds, gender difference. By treating each difference in point estimates of the means as though it were a statistically meaningful, definitive "success", this "test" in effect assumes that the standard errors corresponding to the estimated means are negligible—in essence, zero. One could ignore the differences in performances among the various individual players, ignore the differences in the sizes of various teams, and ignore the variety of locations from which the teams were drawn. But it is very difficult to think of an interesting real-world question about gender and risk-aversion to which such a team-and-tournament model could correspond. Each team plays only one game. Suppose that the decision about which sex "wins" the tournament is based on the number of games won, no matter whether games are blowouts or squeakers.
