College Board released scores almost two weeks ago from its March 5th National New SAT test date and the public school weekday New SAT test administrations. A flurry of articles, reports, and inflamed letters has appeared in response.
The highlights of what we have learned so far include:
• Score Inflation: New SAT scores are “inflated” compared to the old SAT scores. For the vast majority of students, New SAT scores are roughly 60 to 80 points higher for corresponding sections of the Old SAT.
• Percentile inflation: College Board is now reporting a newly defined Percentile called the National Percentile. This percentile uses a new definition and a new sampled reference group, “raising” percentiles by 6 to 8 points across the middle score ranges, and as high as 10 percentile points over part of the scale.
• Substantially decreased Benchmark scores: In 2015, College Board defined college readiness as achieving a 1550 SAT score, which roughly 42% of students met. In the new SAT score reporting, the Composite benchmark is now a 1010 score, which equates to a 1370 on the Old SAT (37th percentile). The New Readiness standard has also been redefined with a lower GPA and a higher predicted likelihood.
• Concordance Tables released after only one New SAT test: The last time the SAT changed its test in 2005 College Board waited a full year before producing Concordance Tables. Since many juniors were counseled to either take the ACT or the Old SAT early, it seems likely that these Concordance Tables are based on a non-representative student sample. It is highly likely that they will change significantly over time.
• ACT CEO is NOT in agreement with College Board’s published Concordance of the New SAT to the ACT: CEO Roorda wrote two inflammatory letters explaining that the ACT was not consulted in forming the Concordance table and that we need a full year’s worth of data to create a “full and fair” sample.
The Old SAT had an average (roughly 50th percentile marks) Composite score of 1500, comprised of roughly 500 scores on each of the three sections. The New SAT has an average that is closer to 1090, a full 90 points higher. The test has not been “dumbed down”, nor have students suddenly become brilliant. Instead, the test changes have created some of this discrepancy: 1) there are only 4 answer choices instead of 5, 2) there is no Wrong Answer Penalty, 3) challenging and arcane vocabulary has been eliminated, and 4) the test questions are supposed to be more closely aligned with real school work.
So the New and Old tests are different in content and scoring. More comparisons across a variety of score categories comparing New SAT scores to Composite (CR and Math) Old scores show consistent inflation in the 40-70 point range 1:
o A New SAT Composite score of 1200 corresponds to an Old 1130.
o A New 1300 score corresponds to an Old 1230 score.
o A New 1400 score corresponds to an old 1340 score.
o A New 1500 score corresponds to an old 1460 score.
o A New 1600 score is the same 1600 of the Old test.
College Board prominently reports a National Percentile and further down in its report states a User Percentile (similar in both SAT and PSAT reporting). Students and families need to understand that the Percentiles stated in Old SAT reports were similar to the User Percentile, not the National Percentile. National Percentiles are based on a definitional change and a new sampled reference group. The Percentile definition has been changed from “the percentage of students scoring below you” to “the percentage of students scoring at or below your score”. (This new definition is in line with the definition used by the ACT). This definition change inflated percentiles by roughly 2 to 6 points on the PSAT.
The National group is reported to be a more stable population of students (All juniors), but has no relevance to a pool of students who are taking the SAT or ACT. This group is formed from a national sample, and it is unclear how it was derived, how accurately it reflects the national pool of students, or when or if it will be modified. These National percentiles cannot be compared to any prior data and by definition are inflated above the reported User percentiles. User percentiles (juniors who took the New SAT) are based on a changing population based on who is taking the test. Since different states are in fluctuation in deciding whether to contract with the ACT or SAT for public school diploma testing requirements, this population will continue to change over time.
One executive from College Board states his dislike for the User Percentile for just this reason – it is not a “stable” population. However, it seems unfair to the uneducated student to not provide more detail that explains that the User Percentiles are the only way to compare to prior years.This is why the Concordance Tables are necessary, and they have just been released.
Why did College Board inflate scores and percentiles?
Kate Dalby, a prominent SAT/ACT tutor in Pittsburgh, called these New scores (both PSAT and SAT) “The College Board Bait and Switch?” 2 Given the inflated scores and percentiles from the PSAT last fall, many students opted to take the New SAT instead of the ACT. How many of these students would have opted for the ACT had they understood the inflation? Dan Edmonds of Noodle Education speculates the College Board may be intentionally inflating scores to attract more students. 3 Certainly, College Board has been under intense competitive pressure with the ACT since in 2012 the number of ACT test-takers surpassed that of the SAT.
Substantially Decreased Benchmark Scores
Composite Benchmark scores that College Board defines as “college readiness” were substantially decreased from a 1550 (old SAT) in 2015 to a 1010 New SAT, which is equivalent to a 1370 old SAT. Percentiles for these scores have dropped from 42nd to 37th for these benchmark scores.
Looking deeper into the components of the Benchmarks, the Math section appeared virtually unchanged; however, the English-Language Arts benchmarks were reduced by at least 110 points. Almost half of the National Representative sophomores would miss the old benchmark, yet only 10% of National Representative sophomores miss the new benchmark. It is difficult to understand the rationale behind these changes.
College Board also redefined its Benchmark or College Readiness Standard to read, “The college and career readiness benchmarks for the SAT predict a 75 percent likelihood of achieving at least a C in a set of first-year, credit-bearing college courses.” The Old Standard had read, ’The college readiness benchmark was calculated as the SAT score associated with a 65 percent probability of earning a first-year GPA of 2.67 (B-) or higher.”
The change to the new Readiness standard brings the New SAT more in line with the ACT wording. The definitions did not contribute to the substantially decreased benchmark scores. Concordance Tables released after only one New SAT test In 2005, the last time that College Board did a comprehensive overhaul of its SAT, CB waited an entire year to create Concordance tables to equate the New and Old tests. This time, after only the initial new test, College Board produced its Concordance Tables. The majority of juniors were counseled to avoid taking the first two New SATs, so it is highly unlikely that these test results come from a typical representative sampling of juniors. It is unknown when or how often these Concordance Tables will be revised.
ACT CEO is NOT in agreement with College Board’s published Concordance of the New SAT to the ACT
Further, ACT CEO Roorda has published two letters 4 (first letter is printed in Dalby’s blog) expressing his anger and disagreement with the Concordance tables that attempt to equate to the ACT. First, the ACT was never contacted to collaborate in the production of these tables. The last time the SAT and ACT collaborated was 10 years ago. Roorda feels that in order to create a rigorous concordance, meaningful data must be collected for roughly a year. He tells students and their families that “Until then, we urge you not to use the SAT Score Converter. And not to listen to messages suggesting the old SAT and the new SAT, or even the ACT, are comparable. For me, that’s unequivocal…”
Stay tuned, because it’s clear that we’ll be hearing more about comparing New SAT-Old SAT-ACT scores as the data continues to be collected and analyzed….