For the entire class of 2008 (the current sophomore class), the actual mean Verbal and Math SAT scores were 703 and 704, respectively. I say that the values we derived by using only 10% of the class were consistent with these global values because the mean values for our subsample:
Verbal SAT = 691.5 +/- 7.6, Math SAT 695.7 +/- 6.8
are the same within 1.5 and 1.2 standard deviations, respectively. That is, there is no statistically significant difference between this 10% of the class and the whole class. If I made up 100 different subsamples each containing 100 students (10% of the class), I would find 13 more than 1.5 sigma away from the overall mean, and 23 more than 1.2 sigma off. Thus, the discrepancies of 12.5 and 8.3 points in our values are just the normal, random statistical fluctuations one should expect when looking at only 10% of a 1000-student class.