Z:gnu-www-ja-frank--156ef1-Earlier we noted that one poss/en

Earlier we noted that one possible explanation for the observed differences between economics students and others is that economics students are more likely than others to be male. To control for the possible influences of sex, age, and experimental condition, we performed the ordinary least squares regression reported in Table 2. Because each subject played the game twice, the individual responses are not statistically independent. To get around this problem, we limited our sample to the 207 subjects who either cooperated with, or defected on, each of their two partners. The 60 subjects who cooperated with one partner and defected on the other were deleted from the sample. The dependent variable is the subject's choice of strategy, coded as 0 for &ldquo;cooperate&rdquo; and 1 for &ldquo;defect.&rdquo; The independent variables are &ldquo;econ&rdquo; which takes the value 1 for economics majors, 0 for all others; &ldquo;unlimited,&rdquo; which is 1 for subjects in the unlimited version of the experiment, 0 for all others; &ldquo;intermediate,&rdquo; which is 1 for subjects in the intermediate version, 0 for all others; &ldquo;limited,&rdquo; which is the reference category; &ldquo;sex,&rdquo; coded as 1 for males, 0 for females; and &ldquo;class,&rdquo; coded as 1 for freshmen, 2 for sophomores, 3 for juniors, and 4 for seniors.