Search News


Browse Archives

News

The SAT 'at War With Itself'

April 16, 2009

Share This Story

FREE Daily News Alerts

Advertisement

SAN DIEGO -- When American high school students take an SAT that is an hour longer than it used to be, and that includes a writing test many top colleges ignore, Richard Atkinson may be the man they have to thank.

Atkinson is the former president of the University of California. When he announced in 2001 that he was recommending that the university system stop requiring the SAT of applicants, he got the attention of the College Board in a way that other critics of the test never could. The prospect of losing all of those University of California applicants led to all kinds of changes in the SAT (and succeeded in keeping the university among the institutions requiring the test).

In a speech here Wednesday at the annual meeting of the American Educational Research Association, Atkinson said that while the changes had resulted in "significant" improvements, he was decidedly unimpressed with the results. He said that the essential criticism he made in 2001 -- that colleges need measures of achievement and knowledge, not some sense of students' aptitude -- was still valid.

He argued that high school grades and the SAT II (the subject matter tests) could give admissions offices the information they need far more than the SAT I (the main exam, which most people just call the SAT). Atkinson was particularly critical of the "critical reading" portion of the SAT -- and said that it appeared to serve no real purpose, and that it was "remarkable" (and not in a good way) that adding a new test and an hour's time to the SAT had failed to improve its validity.

The remarks were significant for all kinds of reasons. Not only was Atkinson the person who set off the changes in the SAT, but the University of California is now moving in the opposite direction that he endorsed -- eliminating the subject test requirement and keeping the SAT. Atkinson's criticism was also notable because of his background and perspective. Some critics of the College Board are generally opposed to standardized testing. But Atkinson said he strongly believes there is a need for standardized testing. And he's spent years studying the subject, in research on cognition and as past chair of the Board on Testing and Assessment of the National Academy of Sciences.

Atkinson's talk was based on a larger paper he co-wrote with Saul Geiser, who has conducted extensive research for the university system on admissions issues. The paper, "Reflections on a Century of College Admissions Tests," is available on the Web site of the Center for Studies in Higher Education.

In his talk, one of Atkinson's themes was that the underlying flaw of the SAT is that it was designed to measure student aptitude, and remains so, long after the College Board removed "aptitude" from its name. Atkinson said that there is a much higher validity to tests based on actual knowledge learned in courses, and that -- grade inflation being what it is in high schools -- admissions officers genuinely benefit from a national tool to compare students boasting A's in calculus, chemistry or French at high schools where an A may mean different things. (Despite those concerns, Atkinson stressed his view that grades in college preparatory courses are the single best way to predict college success.)

Prior the reforms of this decade, Atkinson said, the SAT was known for "trickery" and "esoteric analogies" that encouraged students to try to learn test-taking skills. He applauded the elimination of the much-mocked analogies section and applauded especially the addition of the writing test to the SAT. Not only does it test something that students actually need to do in college, Atkinson said, but it sends a message to high schools to take writing instruction seriously.

But while backing these changes, he said that the SAT was "at war with itself" because it tells different things to different people. Atkinson noted that the writing test -- the part of the SAT of which he is most fond -- is ignored by many colleges, where educators don't believe it encourages the best student writing. So while the College Board tells Atkinson and like-minded people to focus on the writing test, the College Board tells admissions officers elsewhere that they still have the "old SAT."

And even with the improvements, he said that the SAT remains designed to theoretically measure potential, which he said is very difficult to do well, rather than what students have actually learned. (The ACT, the College Board's main competitor, tends to boast that it is more focused on classroom instruction, and Atkinson applauded its roots in that belief, but offered what for him was clearly a damning criticism in that he said that the ACT has become more like the SAT over the years.)

Fundamentally, Atkinson said, the College Board "wants to have its cake and eat it, too," producing tests that really measure learning (he mentioned Advanced Placement tests and the SAT II tests), while still encouraging millions of students to prepare for an aptitude test of limited value. "The fundamental question is: what is the SAT measuring?" he said. The answer (aside from writing) is that it is testing things that are "remote from what students encounter in the classroom."

Atkinson also directly challenged the spin that the College Board placed on last year's validity studies of the new SAT. The studies found that the new test was as successful as the old test, and the College Board declared that a major success. Atkinson said that if you overhaul major features and add time to a test, success would be increased validity, not the same validity.

The irony in Atkinson's presentation was that the University of California is moving in the opposite direction. The university is dropping the SAT II tests as requirements, while keeping the SAT I. The move is designed in part based on the belief that more black and Latino students will apply and enroll, although Asian American leaders are urging the university to reconsider its plans because of projections that Asian American enrollment will drop.

In his speech, Atkinson did not mention the university system's actions. But an audience member asked him about it and he noted that he is no longer president, and suggested that those interested in the university's current position on the issue consult the university's Web site.

A spokeswoman for the College Board said that the organization was "gratified that Dr. Atkinson enthusiastically supports our Advanced Placement program and the SAT Subject Tests." But she also defended the SAT I. "Years of independent research and thousands of studies consistently show that combining SAT scores and high school grades is the best indicator of college success," she said. "Virtually every college that does not have an open door admissions policy uses multiple measures when evaluating students. The College Board has always vigorously recommended that SAT scores be used in conjunction with grades, the rigor of courses and other measures. Ultimately, no single measure or test fully reflects a student's readiness for college."

She also said that Atkinson was citing studies of the old SAT and that studies of the new SAT have found that it is in fact "more predictive" than Atkinson gives it credit for. "Colleges can depend on SAT scores as fair measures of college preparedness," she said.

While audience members appeared to have mixed feelings about standardized testing, one college president questioned Atkinson for not going far enough.

Robert Weisbuch, the president of Drew University, said that minority applications surged when the institution ended its SAT requirement. If the SAT reliability remains "so low" and it provides "such a minor help" in admissions, while it "mis-shaped kids' lives" by teaching them about education in a "very mechanistic way," Weisbuch said, why tinker with the SAT at all? "Why don't we just blow the whole thing up?"

See all postings »
Advertisement
Advertisement

Matching Jobs

Comments on The SAT 'at War With Itself'

  • What about WICS and other "21st Century Skills Assessments"?
  • Posted by Joe Beckmann on April 16, 2009 at 8:30am EDT
  • While Atkinson's observations are certainly piquant, there's been considerable discussion about 21st Century Skills assessments and measures like Sternberg's WICS (Wisdom-Intelligence-Creativity-Syntheses) both in identifying more substantial skills and predicting later college - and hypothetically even later career - success. Any discussion of these measures inevitably effects how we approach NCLB measurement of schools themselves, of classroom and other teaching, and of a host of other what might be considered "reflective" rather than "projective" measures or imputtions. In fact, we might better distinguish between projective (how well they'll do in the future, in either school or career) and reflective or reactive (how well they did in the past) than between the more traditional formative and summative assessments which have become so muddled that those distinctions are now largely meaningless.

    In fact, the distinctions between instruction and assessment are also moving toward the kind of trivial debates more like angels on the heads of a pin. A high quality assessment should inspire the kind of reflection in the student that any other high quality instructional opportunity inspires, just like measuring that inspiration - either in terms of a number, a grade, or a light bulb flashing in a kids' eye - is as valuable and as capturable (whether on paper or via a cell phone picture posted on YouTube). With new and innovative institutions like Olin College of Engineering moving to behavioral assessments for admission, "frankly my dear, I don't give a damn" might be the appropriate college admissions assessment of almost all these assessments.    

  • Posted by David Kane on April 16, 2009 at 10:00am EDT
  • Is the above comment from Joe Beckmann a parody of the sort of anti-test, anti-IQ attitude that was wrecked so much damage in education over the last 40 years? I think so! Here is the give-away:

    "With new and innovative institutions like Olin College of Engineering moving to behavioral assessments for admission"

    Sure! That's what Olin does. That's why its 25th percentile SAT scores are 700 reading and 740 math.

    http://www.olin.edu/DataSets/documents/CDS_C_Admission_freshmen.pdf

    It just happens that the SAT scores are about the highest in the country even though its admissions system uses "behavioral assessments."

  • Challenge assertions about writing test
  • Posted by Merilee Griffin on April 16, 2009 at 10:00am EDT
  • I hope readers will not leave this article with unquestioned faith in the SAT writing test. Any timed writing test that isolates students from the real world of the Internet, library resources, the glut of unreliable sources, conflicting opinions, multiple viewpoints, ambiguous data, and general intellectual messiness of reality has limited value in assessing students' ability to think and to write. It will test largely three rather low-level competencies: fluency in English (which is largely a result of growing up in an educated, English-speaking household); ability to organize ideas at the level of the five-paragraph essay (roughly, an 8th-grade skill); and the ability to marshall whatever evidence comes to mind to support whatever opinion the student holds when the test begins.

    Unfortunately, the importance of doing well on such writing tests has forced much K-2 writing instruction to focus on these simple skills and to ignore the more complex one students need to master in most first-year composition courses, let alone in real life.

  • A knock on high schools?
  • Posted by RG on April 16, 2009 at 10:00am EDT
  • RE: "Atkinson said that ... grade inflation being what it is in high schools -- admissions officers genuinely benefit from a national tool to compare students boasting A's in calculus, chemistry or French at high schools where an A may mean different things."

    Come on, now ... why continue to look down our noses at high schools? Is there no grade inflation in higher education? Is there not an A that might mean different things at different colleges and universities? And what about the difference in grading rigor possible (and maybe probable) in the same course taught by different instructors at the same college or university? Might not these issues affect higher education as well, perhaps when a student wants to transfer from one institution to another?

  • in the absence of evidence...
  • Posted by mathprof on April 16, 2009 at 1:30pm EDT
  • Nobody knows the answer to the question "Since everybody can't get into a first-rate college, who should the society rationally and ethically choose?"  So what we need is a filter, something that will quantitatively rank-order the applicants to these colleges.  The SAT, ACT, and grade-point averages, have historically provided this filter.

    We can argue about which filtering system will populate the colleges with the "best" students, the "most diverse student body," the "student body that will stimulate national economic growth," or the "most satisfying students to teach."  But we need to remain aware that the debate over these filters is really a debate over how to fairly distribute a scarce good -- admission to the nation's top colleges and universities -- in a situation where there is little consensus about what "fairly" means.

    Whatever is decided by an individual college today -- and for my money, high school grades and whether the student has made the most of whatever academic resources her or his particular high school offers are the best criteria -- at least we're doing something different than what was done before the 1950s, when belonging to a rich, white, Protestant family was what got students into Harvard or Yale.

  • Being fair to Atkinson...
  • Posted by WS , Instructor of Latin at St. Andrews Sewanee on April 16, 2009 at 1:30pm EDT
  • "RG"'s post above (titled "A Knock on High Schools") asks a good question: might not universities be held to the same standard, regarding grade inflation? As interesting as that question might be, however, let's be fair to Atkinson. He is discussing the SAT, not the GRE. While the same grade-inflation problem might exist for both institutions (secondary as well as university), the object of his discussion was secondary education, so Atkinson's mention of high schools and not universities is reasonable. He's not looking down his nose at high schools, he's being critical of SAT measures. Ad hominem?

  • Being fair ...
  • Posted by RG , former higher education official on April 16, 2009 at 7:15pm EDT
  • Yes, WS makes a good points about being fair to Atkinson because his focus -- the SAT -- is tied to high schools. My comments about looking down on high schools were meant to cover wider territory. I find it tiresome that high schools suffer ongoing criticism, even if it is a relative side note here, from a community that has many of the same issues regarding grade inflation and consistent academic rigor that also arise when a student transfers from one institution to another, or from one higher education level to another.

  • The Real Issue
  • Posted by Gary Moss at Focus On Learning on April 23, 2009 at 11:15pm EDT
  • With reference to the original article, I think it’s important to focus on the key issue. What is measured by college admissions tests and grade inflation (on any level) is secondary (no pun intended) to the real issue which is what predicts success on the college level. The NACAC “White Paper” on College Admission Testing(Feb2007,www.nacacnet.org/PublicationsResources/Research/Documents/TestingWhitePaper.pdf) addresses this, but still leaves the primary responsibility to the individual colleges and universities. In particular, every institution must investigate:
    1. The predictive validity of all inputs into the admissions decision. This would include prediction of freshmen grades, final grades, retention and graduation.
    2. Whether any of the inputs (including college admissions tests) is biased? I.e., is there any systematic under or over prediction for any demographic group?
    Because of their possible effects on validity, the effects of coaching on college admissions tests need to be more objectively measured. Studies that fail to find effects of coaching do not prove that coaching can not increase scores, but only that the coaching conducted by the researchers could not increase them. (These researchers probably couldn’t run a mile in 4 minutes, but that doesn’t mean someone else couldn’t.) The only fair and objective way to evaluate the effects of coaching would be to sponsor open competition. Pitting the best coaches against each other in a rigorously controlled study would undoubtedly find the significant improvement students can achieve with these scores that can be directly attributable to preparation and coaching.
    Once the improvement in scores attributable to coaching and preparation is openly recognized, then the effects of coaching on the validity of college admissions tests could be investigated.

  • Not to be or to be?
  • Posted by Gerald G on May 22, 2009 at 2:15pm EDT
  • I think that in some way the SAT is a must, because the SAT measure student ability to solve hard problems and also challege his/her to think more faster and advants. But the SAT testes you on how well you can take a different exam, not on what you learn in class. And keep in mind that kids who have wealth parents have a better chances to do better becaues they can pay to take the SAT over and over again and get private tutiors( most of this kids are white decents). Lower income school can't prepared students for SAT because there fouce is to pass student on states test on in order to get money for the school. Student are taught how to pass states testes not the SAT. I thinks college take every event a kid do in high school and use it to measure his or her succes in college, that mean grades, finals exams, balance work and/or sports with school, voluteer, class rank, states tests ect...

     

  • To David Kane: Tut Tut, Coincidence is not Cause
  • Posted by Joe Beckmann on May 22, 2009 at 8:15pm EDT
  • There are many measures of intelligence, not the least of which are SAT's, and not the only of which are WICS metrics of carefully analysed, validated, and standardized rubrics of wisdom, intelligence, creativity and the synergy between them. Sure, Olin has high SAT's, it's a very prestigious school of engineering, just finishing its first decade. Generating a high initial standard means they have high scores on many standards. Yet tests that go "beyond the bubble" need not sacrifice all that poor Mr. Kane bemoans in my critique of the SAT.

    The year before mine at Columbia, decades ago, a long lost Admissions Dean used ONLY the SAT to create the class of 1964. That class was memorable in many ways, not the least of which was that that dean fled to work for the College Board the year his cohort started. So much for objectivity! 

  • The SAT and High School
  • Posted by ARG , Student on May 23, 2009 at 3:45pm EDT
  • In regards to the SAT and high school:

    The entirety of the test asks student to think in a way that is not encouraged in class nor is recommendable for college preparations. For example, many respectable study books for the SAT such as The Princeton Review explicitly state that "Just because the SAT features math and reading problems, it does not mean that the test reflects what you learned in high school... The test writers claim that the test measures "reasoning ability", but actually, all the SAT measures is how well you take the SAT." In essence, it should only be used as a test-taking resource, not a tool to measure aptitude or achievement. The high school curriculum does not focus as much on preparing students to take a test as it does understanding subject material and applying it to related situations. Thus, it would only be proper to alter the SAT so that it only assesses skills required for college and contains questions that must be solved as they would be in the classroom. As for the writing part of the test, I believe it should be taken out completely. The 10th grade writing test is sufficient for testing writing ability. The SAT version does nothing to improve the validity of that portion of the test.