MIT and the Reinstatement of the SAT

In 2014, an article in The New York Times Magazine described me as one of the SAT’s harshest and most relentless critics, beginning with my discovery in 2005 that essay readers were trained to value length more than any other factor. My later research demonstrated that untrue details posing as facts and quotations, regardless of their relevance to the essay’s argument, significantly raised an essay’s score. That Times article also describes a 2012 conversation I had with the then-incoming president of the College Board, David Coleman, that convinced him to scrap the mandatory SAT essay. Consequently, that I applaud the decision of many colleges and universities to scrap the SAT is by no means shocking. However, that I strongly support the Massachusetts Institute of Technology’s decision to reinstate the SAT requires an explanation based on my 30-year history with that institution.

At MIT, I served as an associate dean, a member of the Committee on Undergraduate Admissions and Financial Aid, a first-year adviser, and director of Writing Across the Curriculum. I also taught first-year writing, scientific and engineering writing, and a class entitled Classical Rhetoric and Modern Political Discourse. Although I was not involved in the recent discussions or decision on reinstating the SAT, these past experiences give me a fairly comprehensive knowledge of MIT’s complex and unusual educational ecology.

MIT differs from most other American colleges and universities, including most other selective institutions, in that its programs in undergraduate majors begin in the second year rather than in the third. From the first semester of their sophomore year until graduation, students are expected to devote 75 percent of their coursework to intensive majors in science, engineering, economics and management that all require an expertise in calculus and, in most, knowledge of physics, chemistry and biology. Consequently, because calculus and physics are essential to so many disciplines, the coverage in the first year is comprehensive and the speed of instruction intense. Each entering student takes two semesters of both calculus and physics, each two-semester class being the equivalent of three or four semesters at many other universities, and one semester each of chemistry and biology.

Quite simply, if students are not ready for such concentrated and rapid learning, they will not make it through their first year at MIT. But how does MIT ascertain a student’s capability? The only assessments available for all entering students are SAT and ACT scores.

Most students who attend MIT have very high SAT math scores. Over 75 percent of entering first-year students have SAT math scores of between 780 and 800, the top score. Moreover, fewer than 10 percent of applicants with SAT math scores of 750 or above are accepted. Therefore, the test is not used to distinguish among applicants with similarly high scores. The test, however, can be used to identify candidates with lower scores who almost certainly lack the necessary mathematical preparation, and that is its primary use at MIT. There are, however, better predictive tests that provide MIT with much more relevant information, such as the College Board’s own Advanced Placement Calculus tests and similar tests as part of the International Baccalaureate program. But many students do not have access to these Advanced Placement and International Baccalaureate programs. Teacher preparation is expensive, as is the scoring of the tests by a team of high school AP teachers and college faculty who teach the college-level equivalent of these courses. (At this moment, the Educational Testing Service is still advertising for faculty to evaluate AP tests during this summer’s grading sessions.) While AP and IB courses and tests are abundant in wealthy public school districts and private schools, access is either severely limited or nonexistent in poorer school districts.

These facts explain MIT’s motivation in reinstituting the SAT. Rather than being an instrument to distinguish among highly qualified students who have access to advanced coursework, it is primarily used to identify students who do not have access to more advanced programs and who would nevertheless succeed in MIT’s curriculum. What would happen to students who have strong academic records but no advanced coursework, but who did not take the SAT either because it was either optional or completely abolished? Absent the better predictors of success, the AP Calculus and the IB math tests, the SAT math score is the best available tool to distinguish between those students who would succeed at MIT and those who would not.

I had an experience 30 years ago that sheds light on this issue. I was assigned an advisee from a working-class area, who had not taken AP math classes because they were not offered at his school but had an SAT math score in the mid-600s, extremely high grades, top class ranking and superb teacher recommendations. He also excelled as a classicist, being almost completely fluent in reading Greek and Latin. Unfortunately, he was unable to successfully complete both first-year mathematics and physics. MIT provided him with a generous scholarship. He never finished his degree at MIT.

The SAT is a poor test that is not necessary for evaluating applicants at most or maybe any other American college or university. The three-hour multiple-choice SATs are the McDonald’s of educational assessment: they are mass-produced, fast, cheap and have little nutritive educational value. There are many problems with the SAT, but two central and related ones are its form and who constructs it. The four-item multiple-choice format of most of the SAT encourages gaming strategies, such as learning how to quickly eliminate two of the four choices and then performing the minimum necessary calculations to determine which of the two remaining choices is correct. This peculiar form privileges agility in answering multiple-choice questions quickly rather than performing the kind of mathematical computation and analysis that is the focus of much college math. It is also what makes test-preparation companies so effective and profitable.

On the other hand, while the AP Calculus test contains a large section of five-item multiple choice questions, it also includes substantial free-response sections where students are required to derive the answer and show their work. The IB Calculus tests consist entirely of free-response questions, and students are required to show all the steps in their calculations.

Moreover, a multiple-choice test like the SAT can be graded quickly by a machine at almost no cost, while the free-response sections of the AP Calculus and IB math exams require time-consuming and expensive grading by humans. Also, in a very simplistic sense, the multiple-choice machine-readable questions are more reliable. An answer will always receive the same score. Humans’ scores may fluctuate.

This brings us to the second main problem with the SAT. Unlike the AP and IB tests, the SAT tests are developed primarily by statisticians with little input from educators in the specific academic fields being evaluated. Statisticians love multiple choice because of its reliability and ease of constructing statistically similar new tests. They care less about validity—that the test is assessing what abilities it purports to assess—and more about reliability and consistency, that student scores will largely be stable over different administrations of the test. Moreover, it allows evaluating noncredit “candidate” items that do not comprise part of a student’s score but can be used to construct statistically equivalent new tests.

However, because exceptionally high scores on the SAT math section offer the only standardized indication for all applicants of the extremely advanced mathematical ability that is necessary for success at MIT, it serves a narrow and very limited function for MIT admissions that is unnecessary at most, if not all, other colleges and universities.

But retaining the SAT for such a limited use is not a suitable long-term fix. One solution would be to provide federal funding for AP or IB classes and testing at all American public high schools. Such programs would greatly improve high school instruction for promising students throughout the country. While many colleges and universities may not need to use AP or IB test results for admission decisions, such a program would help level the playing field for acceptance to highly selective institutions, including MIT, and probably increase overall student preparedness.

What is common to both of these tests is that they require extensive expense in training teachers, substantial classroom time, usually in the form of separate classes, and the expenses involved in the evaluation of long essays or complex problem sets by high school expert teachers and university faculty that usually take days or even weeks to complete. While not every college or university may need this kind of advanced, standardized coursework and test results to make good admissions decisions, providing the universal opportunity to enroll in such programs will help them—and, more importantly, their students—by making educational opportunities more equitable. And it might finally allow us to see the end of mass-market standardized testing.