Does The SAT Predict College Success?

One of the hottest debates roiling American campuses today is whether the SAT and other standardized tests should continue to play a dominant role as a college admissions criterion. The main point of contention in this debate is whether the SAT or equivalent scores accurately gauge college preparedness, and whether they are valid predictors of college success, most particularly in comparison with high school grades. Behind this ostensible concern is the expressed fear that over-reliance on collegiate admissions tests will reduce “access” to college on the part of low-scoring applicants, many of them from poor or minority families and, thus, risk making American colleges and universities less demographically diverse.
First, let me address “access” and diversity: According to the most recent (2007) data, 45 percent of all colleges or universities, and 66 percent of public ones, have no admissions criteria at all. In the public sector – which accounts for three-quarters of all higher education slots – among the 34 percent of schools with some kind of admissions screen, 69 percent accept more than half of their applicants. Even among the remaining somewhat selective institutions, the majority either do not require admissions test scores or they accept most low-scoring applicants, with the result that the average verbal SAT for all college applicants is 532, and that for the math SAT is 537 (both out of a potential score of 800).
Second, regarding the sincerity of the most vociferous admissions test opponents: Virtually all of the schools calling for abandonment or down-grading of SATs and comparable admissions test have always been highly selective – and intend to remain so. There should be absolutely no confusion on this score. These places have no intention of becoming academically more diverse, meaning they are not planning to admit academically inferior poor or minority students. As predominantly rich institutions, they have an army of admissions officers able to pore over every applicant’s high school transcript and other evidence of academic ability to keep recruiting the best and brightest students, even absent admissions tests. Actually, even with their “test-optional” policies, they will have access to most applicants’ SAT scores anyway, because academically strong applicants will continue to take the tests to keep all their collegiate options open. If one were inclined to take a conspiratorial view of these institutions’ motives, one might suspect that they were mounting this concerted campaign to assure that America’s public colleges and universities remain unselective, derailing the rising admissions aspirations of those ambitious public institutions that threaten to cut into their current monopoly of gifted high school graduates.

Allied with selective private colleges in the war on SATs are elite public institutions like UC Berkeley that, in implementing their high selectivity standards, used to limit admissions to applicants with high SAT scores, but reserved a limited number of places for lower scoring African Americans and Latinos. Now, barred by state law and numerous court rulings from practicing such a blatant double standard, they have soured on standardized tests altogether.
Having lain to rest the groundless fears of denied access and diminished diversity, let’s dive into the empirical heart of the controversy: whether SAT or similar tests predict collegiate academic success with reasonable accuracy. In making such a determination, we need first to have a way of measuring it. Given enormous variability in the rigor and grading standards of college courses, the best possible metric of academic success is degree completion. From the perspective of both students and society, the reason to go to college is to earn a degree, a goal whose importance is reinforced by national data that indicates the enormous economic premium associated with possessing a baccalaureate. So the empirical question behind the SAT debate can be phrased thus: how strongly are baccalaureate graduation rates correlated with students’ admissions test scores, especially when compared to similar correlations based on high school grade point averages.
Fortunately, we can find answers to this question by looking at the experience of the largest university system in the United States – the State University of New York (SUNY) with over 200,000 students enrolled in its sixteen baccalaureate colleges and research universities; a set of distinctive campuses with highly varied student profiles and admissions requirements. As provost of SUNY for many years, I was deeply involved in reviewing campus admissions criteria, and I oversaw the university’s institutional research operation which maintains a comprehensive student unit record system.. Given the sheer scale of SUNY, and the diversity among its campuses with respect to admissions criteria and student academic outcomes, we have a controlled experiment of sorts.
During the period that I served as Provost, all campuses were encouraged to clarify their admissions criteria and, once having decided on a specified level of selectivity, to be consistent in implementing it. Many campuses opted to remain in their historic selectivity tier (which can be characterized as “intermediate” relative to national peers), but a significant number chose to raise their admissions standards. The SUNY admissions selectivity template considers both an applicant’s high school grade point average (GPA) and his/her SAT or equivalent standardized test scores, but campuses looking to raise their admissions standards focused more on SAT scores than grades. High school GPAs at SUNY have been remarkably consistent both across campuses and over time. Thus, by comparing graduation rates at SUNY campuses that raised the SAT admissions bar with those that didn’t – in the context of more or less stable high school grades – we can get a pretty clear idea of whether higher SAT scores lead to higher graduation rates.
The short answer is: they do. Looking at changes in admissions profiles and 6 year graduation rates of the entering classes of 1997 and 2001 at SUNY’s 16 baccalaureate institutions, a period during which some campuses became more selective and others did not, this is what we find. Among this group – encompassing a broad band of selectivity from nearly open admissions to highly selective – nine campuses chose to increase their selectivity after 1997. This group included two nationally ranked research universities (Buffalo, Stony Brook) and seven regional colleges (Brockport, Cortland, New Paltz, Old Westbury, Oneonta, Potsdam and Purchase). As noted, the move to raise selectivity standards had a much greater impact on entering students’ SAT scores than on their GPAs. For the rising selectivity campuses, SAT score increases between 1997 and 2001 ranged from 4.5 percent (Cortland) to 13.3 percent (Old Westbury), while high school GPAs increased only between 2.4 and 3.7 percent, a gain almost identical to that at campuses that chose not to raise their SAT standards.
The percentage increases in six year graduation rates at the rising selectivity campuses – just over a four year period – were dramatic, ranging between 10 percent (at Stony Brook whose graduation rate went from 53.8 to 59.2) to 95 percent (at Old Westbury which went from 18.4 to 35.9). Most revealingly, the seven SUNY campuses that stuck with their prior selectivity profiles, meaning their entering students’ SAT scores between 1997 and 2001 were stable or rose only modestly, actually saw their graduation rates decline. Even Binghamton, always the most selective of SUNY’s research universities, maintained a flat SAT profile and saw its graduation rates decline by 2.8 percent. The most compelling evidence that higher SAT scores predict higher graduation rates can be gleaned by looking at the experiences of campuses with nearly identical student profiles in 1997.
I will highlight the graduation rate experiences of three pairs of comparable SUNY campuses that, between 1997 and 2001 took divergent paths with respect to SAT admissions: two research universities with about 17,000 students (Stony Brook/Albany), two large urban colleges (Brockport/Oswego) with enrollment of about 8,000, and two 5000 student small town liberal arts colleges (Oneonta/Plattsburgh). In each case, in 1997 these pairs had similar admissions profiles with respect to high school GPAs (high 80s for the universities; mid-80s for the colleges). The only distinguishing difference among the campuses in each pair was that by 2001, one school admitted freshmen with significantly higher SAT scores, and the other one didn’t. At all of these schools high school GPAs of entering freshman rose modestly between 1997 and 2001: about 2 percent.
In each case, the campus that raised its SAT bar saw a substantial gain in graduation rates – in only four years! Stony Brook and Albany: Between 1997 and 2001, Stony Brook increased its average entering freshman SAT score by 7.9 percent to 1164, and its graduation rate rose by 10 percent to 59.2; in this interval Albany’s freshman SAT increased by only 1.3 percent and its graduation rate actually fell 2.7 percent to 64.0. Brockport and Oswego: Brockport increased its average freshman SAT score by 5.7 percent to 1080, and its graduation rate increased by 18.7 percent to 58.5; at the same time Oswego’s freshman SAT rose only 3.0 percent and its graduation rate fell 1.9 percent to 52.6. Oneonta and Plattsburgh: Oneonta raised its freshman SAT score by 6.2 percent to 1069, and saw its graduation rate rise by 25.3 percent to 58.9; Plattsburgh’s freshman SAT increased by only 1.3 percent and its graduation rate fell sharply by 6.3 percent to 55.1.
Other examples: SUNY Old Westbury, always the university system’s academically weakest campus, was able to improve its graduation rate by over 95 percent in four years after it instituted higher SAT requirements, increasing its entering students’ SAT scores by 13.3 percent between 1997 and 2001. Purchase College, a highly specialized place centered on the fine and performing arts, increased SAT scores by 10.3 percent, and saw graduation rates rise by 22.4 percent.
The common denominator in all of these SUNY examples is that among campuses with highly differentiated missions, across a very wide band of admissions selectivity, all enrolling freshmen whose high school grade point averages improved by the same modest amount (about 2 to 3 percent), only those campuses whose SAT scores improved substantially over this very brief four year interval saw gains in the most valid measure of academic success: graduation rates.
Is this the empirical last word on the SAT controversy? Probably not, but it should go a long way in countering those that challenge the value of standardized tests in predicting academic success. Given the sheer diversity and size of New York’s State University system and the systematic implementation of clear cut shifts in admissions policy by many of its campuses we have as good a laboratory for examining this hypothesis as any in the higher education landscape. As fresh data becomes available, this pattern may change, but I doubt it.