With the redesign of the SAT in 2016, scores went up, which made it hard to predict what kind of scores highly selective colleges would want. With the release of data for the Class of 2017, we now know that at many schools average SAT scores effectively dropped. Check out how 97 highly selective schools reacted to the new SAT in the table at the end of this post.
It’s been a trying couple of years in testing and admissions, thanks to the redesign of the SAT in 2016. Leaving aside the optional essay, the experimental section, double-percentiles, the trouble College Board had with writing good math questions, and other woes, students, families, and school counselors have had to deal with the confusion sown by the rise of all scores on the new SAT. Where once a student applying to a highly competitive school had to break 700 on each section, the concordance College Board released with the new exam suggested that 730 is the new 700. The question was whether admissions offices would go along with this inflation, expecting higher scores than ever before.
The answer is that there is no simple answer, but we think the news is pretty good for students.
We’re about to get into the weeds, so feel free to scroll all the way to the bottom where you can see the 2016 and 2017 middle fifty percent for 97 highly selective schools.
The challenge in figuring out how colleges dealt with the redesigned SAT is that we are comparing apples to oranges, although apples to pears might be more apt. The old SAT and new SAT, after all, have a lot in common. They are both scored on the same 200-800 point scale, but the old one had three section scores. The content on the two tests is pretty similar, although on the new test the reading comprehension and multiple-choice writing sections are combined to make one 200-800 Evidence-Based Reading and Writing (ERW) section score, and on the old test the essay was factored into a 200-800 Writing section score. The essay on the new SAT is optional and does not factor into the ERW score.
The final, particularly important distinction between the old and new version of the SAT is that the redesigned exam no longer penalizes students a quarter-point for wrong answers. The effect of this change is that it made it harder to get a very low score on the SAT. Simply filling in (A) for all the answer choices would get a student a score in the high-200 to low-300 range. As a result of this change, everyone’s scores on the test effectively shifted higher, since they were no longer being penalized for getting questions wrong and were now encouraged to guess rather than leave questions blank. The average combined SAT score went from about 1000 on the old SAT to 1060 on the new version. The ACT, it should be noted, has the same issue. It’s average is also not the middle of the scale (18), because it is very hard to score in the single digits on the ACT.
What this change meant is that a 700 on the Math section of the new exam was not equivalent to a 700 on the old exam. According to the concordance released by the College Board, a 700 on the new SAT was the equivalent of a 670 on the old test. A 700 on the old test was more like a 730 on the new one. Admissions officers looking at the roughly 128,000 students from the Class of 2017 who only submitted scores from the old SAT were encouraged to use the concordance to compare their scores to the more than 1.7 million students who submitted new test scores to make sure every one was on equal footing.
So, here come the weeds. That table is still down at the bottom. No problem if you want to skip to it.
The recently released Common Data Set–a collaboration among the College Board, U.S. News and World Report, and Peterson’s–for the Class of 2017 lets us compare the new SAT averages to the old SAT averages among students admitted at schools that expect highly competitive exam scores. Colleges and Universities report the middle fifty percent of students who matriculated, and that is the metric we are using here. They also report the percentage of students attending by score ranges. We’ll follow up with a post on that in the future. The important thing to note is that these are the percentages for the students that began attending in 2016 or 2017.
We began with about 120 schools with admit rates below 40%. Because we wanted to be able to use the ACT as a point of comparison, we eliminated schools that did not report ACT scores. We also removed a few schools that had clearly reported erroneous data. In the end, we were left with 97 schools.
If we compare the Class of 2016’s middle fifty percent for SAT Math to the Class of 2017’s middle fifty percent for SAT Math without concording the scores, the scores go up for the majority of schools, as you would expect. Even then, however, there were a surprising number of the 97 highly-selective schools in the cohort where the unconcorded scores stayed the same or even went down. At about 43% of the schools, the 75th percentile on the Math stayed the same or went down from the previous year.
|SAT Math 25th Percentile for Unconcorded Scores||SAT Math 75th Percentile for Unconcorded Scores|
We could not compare the unconcorded ERW scores of the Class of 2016 to the ERW scores of the Class of 2017, because of the differences in scoring. The College Board Concordance does however concord the sum of the Reading and Writing sections of the old exam to the redesigned ERW section.
After concording the scores from the Class of 2016 to the Redesigned SAT, we can see that the changes were actually smaller than we might expect. In fact, at a majority of schools the scores effectively went down!
|SAT Math 25th Percentile for Concorded Scores||SAT Math 75th Percentile for Concorded Scores||SAT ERW 25th Percentile for Concorded Scores||SAT ERW 75th Percentile for Concorded Scores||ACT Composite 25th Percentile||ACT Composite 75th Percentile|
The first three rows count the number of schools in three categories. The single largest group is the number of schools, 66 out of 97 (68%), where the 75th percentile ERW score went down. The smallest groups were schools where the 25th and 75th percentile went up, just 6% in each category. In Math, more than 54% of schools saw their middle 50 percent drop.
What makes these drops in the scores notable is that the ACT middle fifty percent held or went up at 89% of the colleges and universities. The average change for each SAT category went down, most significantly for the 25th percentile ERW score, while it went up for the ACT.
The natural question, of course, is why? Why did ACT scores creep up at more than a quarter of schools while SAT scores did not move nearly as much? Even the 25th percentile Math score, which went up at 23% of schools, went down at 56%, more than twice as many.
There is probably no single answer to those questions. We do know that there were many more–between 40,000 and 50,000 more–students scoring in the 700s on Math or EWR in the Class of 2017 than there were scoring in that range in the Class of 2016 on Math, Reading, or Writing.
|Year||Math 700-800||ERW 700-800||Reading 700-800||Writing 700-800|
What we might have expected is that colleges would have reached up higher into the score range to find students, thus pushing their middle fifty percent even higher. We do see that occurring in many cases, but the increase is not as large as the concordance would lead us to expect. Our fear was that, if the concordance was to be believed, and 730 was the new 700, that we would see many more schools were students needed to break 750 on each section in order to get serious consideration. That, mercifully, is not what happened.
Perhaps with more students in the 700s, admissions officers at many schools actually felt freer to put a little less emphasis on scores, and students on the “lower” end of the scale benefited. “Lower” is definitely relative in this context, something more like a score in the high 600s or low 700s, but with more students scoring in that range on the SAT there was a larger pool to draw from, possibly leading to slightly lower scores at many schools.
To be sure, the scores that highly selective schools are looking for from applicants remain high and the changes in the middle fifty percent are largely ten to twenty point decreases, although they were over forty points at some schools. It would be rash to say that it has gotten any easier to get into any of these schools. The good news is that the new SAT probably has not made it harder. Fingers crossed for the future.
You made it. It’s TABLE TIME!
This table compares the middle fifty percent of concorded scores for the students who entered freshman year in 2016 (or the high school Class of 2016) to those who entered in 2017 (HS Class of 2017). If a score increased it is marked in red. If it stayed the same, it is black. If it went down, it is marked in blue.
|School||Concorded 2016 25%-75% Math||2017 25%-75% Math||Concorded 2016 25%-75% ERW||2017 25%-75% ERW||2016 25%-75% ACT||2017 25%-75% ACT|
|Bryn Mawr College||630-750||660-770||670-750||650-730||28-32||29-33|
|California Institute of Technology||780-800||780-800||750-800||750-790||34-36||34-35|
|Carnegie Mellon University||750-800||730-800||700-770||700-760||31-34||32-35|
|Case Western Reserve University||710-780||690-780||660-740||650-740||30-34||30-33|
|Claremont McKenna College||700-770||680-770||700-760||660-740||31-33||30–34|
|College of the Holy Cross||640-720||640-710||650-720||630-700||28-31||28-31|
|College of William and Mary||640-760||640-740||670-750||660-740||28-33||29-33|
|Franklin and Marshall College||650-750||640-720||n/a-n/a||620-700||28-31||28-32|
|Franklin W. Olin College of Engineering||750-800||740-800||730-790||710-770||32-35||33-35|
|Harvey Mudd College||760-800||750-800||710-780||720-770||32-35||33-35|
|Johns Hopkins University||740-790||740-800||720-770||720-780||32-34||33-35|
|Massachusetts Institute of Technology||780-800||770-800||730-790||720-770||33-35||33-35|
|New York University||650-780||640-760||n/a-n/a||650-730||29-33||29-33|
|Stevens Institute of Technology||690-770||680-760||n/a||640-710||29-33||29-33|
|Texas Christian University||560-670||560-670||580-680||570-660||25-30||25-30|
|The George Washington University||620-730||600-700||650-720||580-690||27-31||27-32|
|The University of Chicago||750-800||750-800||740-790||730-780||32-35||32-35|
|The University of North Carolina at Chapel Hill||620-740||620-720||640-730||640-720||27-32||27-32|
|The University of Texas at Austin||600-750||610–740||610-710||620-720||26-32||26-33|
|The University of Tulsa||570-730||560-720||590-710||590-720||26-33||25-32|
|United States Air Force Academy||640-750||640-710||n/a||630-700||27-33||29-33|
|United States Coast Guard Academy||630-720||610-690||620-700||570-660||26-31||26-31|
|United States Military Academy||620-730||600-710||620-710||590-690||26-31||23-28|
|University of California, Davis||560-730||570–710||570-680||550-650||24-30||25-31|
|University of California, Los Angeles||600-760||610-760||620-740||630-730||25-33||27-33|
|University of California–Berkeley||660-780||640-770||660-760||610-730||29-34||29-34|
|University of California–San Diego||650-780||590-720||640-720||550-660||27-32||26-32|
|University of California–Santa Barbara||600-750||620-760||620-710||620-710||25-31||26-32|
|University of Michigan–Ann Arbor||700-780||670-770||690-750||660-730||29-33||30-33|
|University of Minnesota–Twin Cities Campus||640-760||650-760||620-730||620-720||26-31||26-31|
|University of Notre Dame||710-790||690-770||700-770||680-750||32-35||32-34|
|University of Pennsylvania||730-790||720-790||720-780||700-770||31-34||32-35|
|University of Richmond||640-750||640-750||660-730||630-710||29-32||29-32|
|University of Rochester||660-780||650-770||660-740||600-720||29-33||29-33|
|University of Southern California||670-780||650-770||680-760||650-730||30-33||30-34|
|University of Virginia||640-760||640-740||660-750||650-730||29-33||29-33|
|Washington University in St. Louis||740-800||750-800||720-780||720-770||32-34||32-34|