Counting Diplomas and 9th-Graders

Tuesday, May 23, 2006

Jay P. Greene and Marcus A. Winters defend higher dropout rate estimates:

We do not share [Larry] Mishel and [Joydeep] Roy's faith in the accuracy of graduation rate estimates from surveys. Even very well-done surveys like [the National Education Longitudinal Study] have difficulty identifying and tracking dropouts. Young people who do not graduate high school are on the margins of society and do not make themselves easily available to researchers.

Instead, we estimate graduation rates for the Class of 2003 by dividing the number of regular diplomas issued by public high schools by an estimate of the number of ninth-graders who entered four years earlier, adjusting for population changes in the interim. This is not a precise technique, but it is hard for it to be very far off. The number of students enrolled and diplomas awarded are relatively easy for schools to count and are likely to be reliable. Our technique adjusts for that ninth-grade "bulge," and we have checked the accuracy of that adjustment against census population counts to confirm that it is on target. Using our method, we estimate that the national public high school graduation rate was 70 percent for the Class of 2003.

It is relatively easy to figure out whether our estimate of 70 percent or Mishel and Roy's estimate of 82 percent is more likely to be accurate -- just divide the number of regular diplomas issued by the Census population for the graduating class. According to the 2000 Census, there were 4,106,867 15-year-olds, of whom we calculate about 208,832 were enrolled in private school, which leaves roughly 3,898,035 students who could have been in the public graduating Class of 2003. If 82 percent of those students received regular high school diplomas, there should have been about 3,196,389 diplomas awarded. But according to public schools, there were only 2,719,947 regular diplomas awarded. If Mishel and Roy's estimate is right, there ought to be about 476,442 more diploma recipients than the number actually reported.

Dividing the number of diplomas actually issued by the number of students in the graduating Class of 2003 yields an estimated graduation rate of 70 percent -- exactly what our method produces.

Mishel and Roy have to explain how public schools underreported the number of diplomas by almost a half-million. They cannot explain away the results by pointing to an incorrect population change since there is no population change in the ratio of diplomas to 15-year-olds. They cannot explain away these results by pointing to immigrants who never enroll in school. All 15-year-olds must enroll in school by law. If they can explain how we are missing a half-million diplomas, their estimate would be reasonable. But they have not been able to do so and thus must be wrong.

© 2006 The Washington Post Company