Since 1997, U.S. News has published a measure called "graduation rate performance." It's a way to acknowledge that some colleges and universities are doing a better job than expected in graduating their students. It counts for 5 percent of the total model for the National Universities and Liberal Arts Colleges rankings.
For the 2008 edition of America's Best Colleges, we added the percentage of Pell grant recipients as part of this computation. One of the criticisms about our rankings had been that our system didn't encourage schools to take low-income, at-risk students and that the only way to do well in the rankings is to buy students with high SAT/ACT scores through merit scholarships. Many colleges and universities have the mission of educating high percentages of students who are from low-income families, who tend to have lower graduation rates than non-Pell grant students. By including Pell grant data in the graduation rate performance calculation, schools (with high Pell grant percentages) will now get credit in our methodology for educating their low-income students because they will now do better in the graduation rate performance indicator than they did before we made this change.
How does the graduation work? We predicted a school's 2006 six-year graduation rate for the class that entered in 2000 based on data about those incoming students, as well as some of the institution's characteristics. This predicted value was then compared with the actual graduation rate of that class. We then calculated a ratio of the actual graduation rate over the predicted graduation rate, and that value was used in our rankings. So, if a school has a higher actual graduate rate than our predicted rate it's "overperforming," and this college is enhancing achievement. If the school's actual graduate rate is less than the predicted rate, it's "underperforming," and the school should be doing more for its students.
We come up with our predicted graduation rates using a regression model based on the SAT and ACT scores and high school class standing of those students who entered six years ago, per-student educational expenditure data from two of the first four years that this class was enrolled, percentage of Pell grant recipients, and whether a school is public or private.