Matthew K. Tabor takes to his blog to write an impassioned response to a piece I wrote, backed up by over 250,000 student records on the class of 1999, arguing the SAT and ACT mattered little in college admissions. His evidence? He The New York Daily News found a student who gasp! graduated high school unprepared for college.

Like Tabor, I was skeptical of the ability of a college to estimate a likely student’s ability on the basis of their high school GPA. It makes a certain sense that a national exam would be a better judge of a student’s potential, because everyone takes the same test in the same conditions, whereas high school GPAs are dependent on individual teachers who give out grades, susceptible to grade inflation, and reflect to a certain degree the student’s course choices. To top it all off, as Tabor points out, a high school diploma is no guarantee of real learning, as measured by whether a student will be successful in college.

But to take it from there to where Tabor jumps requires a leap of logic that deliberately ignores the best evidence at hand.

First the leap of logic. Tabor seems to be arguing that because high schools operate independently and somewhat subjectively, they can’t be trusted. Instead, we should have some type of external check on their success. Maybe we could create a test of some sort that all high school students intent on going to college would take. The test could tell us who the promising students are, i.e. who we should let into college, and we could ignore those phony high school measures.

Except we already have this system, and the best evidence says it doesn’t work. What Crossing the Finish Line says–and remember this is really unimpeachable research with data on more than 250,000 students entering college in 1999 (about 10 percent of all beginning college and university students that year) conducted by noted scholars in the field–is that high school GPA is 3-5 times better at predicting a student’s eventual graduation from college than SAT and ACT scores. When the researchers included a dummy variable for the high school the student attended, not even a proxy for school quality, but a simple dummy variable for the school itself, GPA mattered even more. SAT scores became negative, meaning they were actually a detriment to an admissions officer’s decision.

To prove his point, Tabor mentions a hypothetical student with a 3.0 high school GPA who possibly didn’t learn that much in four years of secondary schooling. Lucky for us, the authors of Crossing the Finish Line did some tests on this exact student. If you hold her GPA constant, here are her expected graduation rates according to varying SAT scores:

Below 800: 39%

800-890: 49%

900-990: 52%

1000-1090: 54%

1100 and above: 51%

In other words, college graduation rates don’t change much no matter what a student scored on the SAT. High school GPA, on the other hand, matters a lot more. Here are the six-year graduation rates if we hold her SAT score constant at 1100 and change their GPA:

Below 2.67: 35%

2.67-2.99: 39%

3.00-3.32: 51%

3.33-3.66: 57%

3.67 and above: 72%

Lest you think these cut points were chosen at random, the authors published charts for higher and lower achievers and found similar results.

Now, Tabor has a point about how this study defines “success.” Simply earning a college degree is a good-but-not-perfect measure of learning. A college degree must stand for something concrete, and it must enable a student to earn a well-paying job in a respectable field. But while Tabor completely discards college graduation as “an abysmally low bar,” he doesn’t offer any better measures for measuring success (although we have a few ideas). Higher education researchers have long defined “success” as getting students into college or getting them through their first year. Crossing the Finish Line should help move the conversation by focusing on completion, and Tabor too easily discards it.