Data show that GPA is far more predictive of college success than standardized test scores. AOs are facile at understanding transcripts, grades, and rigor…they are not going to offer admission to students who can’t succeed.
Remember, all of these schools (with the exception of MIT and CalTech) enroll students every year with lower than average stats, such as some proportion of the recruited athletes. And the vast majority of those students succeed and ultimately graduate at these top colleges.
Edited to add: Look at the common data sets of the top schools…you will see a couple of matriculants with ACT test scores in the 18-23 range and/or at least one section of the SAT in the 500 range every year.
I think test optional or better yet no test admissions is a wonderful development. Kids have 4 years to show their educational and intellectual prowess and curiosity. Subject tests are being discontinued as well. In addition standard tests can be prepped for which reduces their usefulness in predicting college success. The idea that only kids with a 1600 SAT can succeed at top schools and everyone is written off seems unfounded. A student with less stellar math SAT or ACT scores who wants to pursue CS can show math proficiency in courses and rigor. Additionally they will quickly discover if they have the chops once they’re at school and can pivot to another area of study if CS doesn’t work out for them, or they simply find it’s not what they want to study. My nephew began as CS and ended up with applied math. Another started in CS and ended up as an econ major. That’s the beauty of the US system.
Without any test, isn’t it more difficult to compare applicants? Sometimes someone could have a little bit higher of a GPA but could have benefited from some grade inflation while someone else did not.
Possibly but many students prep for the standardized tests which makes them less standardized. In addition, colleges know the rigor, grade inflation/deflation of high schools, and how students from those schools have fared at their institutions. The decisions aren’t made in a vacuum, there’s ample information for the admissions committees to draw from in making their decisions. JMO
It’s possible for there to be grade inflation/deflation within the same school.
At larger high schools, there could be multiple teachers who teach different sections of the same course. Within the same course at the same school, some teachers could grade easier/tougher than others.
True, it’s not a perfect system but neither is standardized testing. And I do believe that the overall reputation the high school and the institution’s experience with prior students plays varying degrees of importance. That’s why some high schools are feeders to elite schools and some high schools have success getting their kids accepted to certain schools and struggle with others.
As a current high school senior I can tell you that it was impossible to get a SAT test last year. Because of covid they were canceled and rescheduled and canceled again. I assume your “top schools” are taking pity on kids who couldn’t travel to another state to take standardized testing.
Lots of highly selective colleges have been test optional for years. There is never a significant difference in graduation rate between test submitters and test optional kids. I’ve literally never heard of any highly selective college with a significant difference. ’
Few if any colleges publish this. However, one can make some rough estimations based on other available information. For example, the Harvard lawsuit study found that nearly all of the admits with the lowest stats were recruited athletes, so it’s extremely likely that the kids with the lowest ACT score in the CDS were recruited athletes. Under Ivy League athletic conference rules, athletes are expected to have stats ~1 standard deviation below non-athletes – a very significant difference in the overall average between athletes and non-athletes, even though certain individual athletes have higher scores. Harvard 6-year graduation rates for the 2 groups are below:
Harvard 6 Year Graduation Rate (only GSR is available on NCAA site)
2019: Athletes = 99%, Non-athletes = 97%
2018: Athletes = 100%, Non-athletes = 98%
SAT/ACT score in isolation is a poor predictor of graduation rate. It typically explains less than 10% of variance in graduation rate. Some studies have found SAT/ACT explains as little as 4% of variance in graduation rate. At Harvard, the contribution may be even less since Harvard students almost never fail out. I expect the few who fail to graduate from Harvard almost always have other reasons besides their score.
It’s nice that the combination of SAT in isolation + GPA in isolation is a better predictor than either alone. However, I expect that’s not how admission works at any of the “top schools” referred to in the thread title, so it’s not especially relevant.
Instead “top schools” consider the full transcript. Did the student take rigorous AP/honors/… courses? Which courses had lower grades and how relevant are they to planned field of study? Upward/downward trend? How harshly does the HS grade? They also consider many factors besides just stats – LORs, essays, ECs, awards, personal strengths/drive, personal background context, interview, etc. The more factors you consider, the less predictive ability you lose, if one of those factors is removed.
And? As you noted, every additional piece of data is valuable, exactly as I pointed out.
Since you agree that additional data is valuable, why do you appear to be advocating that one specific piece of data that serves as better predictor of college success be ignored.
Since Ithaca factors legacy into admissions, and the links your provided don’t work, can I assume they proved that being born to a graduate inherently improves your graduation chances and it’s a valuable input into the admissions process? What was its positive predictive value?
Ithaca still considers standardized test scores, per their CDS, so this study doesn’t seem to have convinced them that the data is useless.
That’s not what I wrote. My post implied that little predictive ability is lost when SAT is removed beyond all the rest of the application. This rest of the application includes more than just admitting based on highest GPA in isolation, without considering anything about rigor of courses, harshness of grading, LORs, essays, ECs, interview, passion/drive, etc. You could probably make a similar statement about several other parts of the application. For example, I doubt much information is lost between applicants who submit an optional 3rd LOR and applicants who do not. However, the point of the thread is not whether admits who don’t submit an optional 3rd LOR will be as successful as those who do. It is whether admit who don’t submit optional test scores will be as successful as those who do.
I edited the link to include an older version of the wesbite from archive.org that works for me. The only available demographic variables were gender, race, and first gen, which explained 8% of variance compared to 43% with the rest + test optional (no SAT) and 44% with the rest + test submitter. A summary is below: The point is that the study suggested that going test optional would result in a negligible 1% loss in predictive ability. Upon seeing these results, Ithaca chose to go test optional and has remained test optional since then.
Gender, Race, First Gen – Explains 8% of variance in cumullative GPA
Above + GPA + Strength of Schedule + AP Hours – Explains 43% of variance in cum. GPA
Above + SAT – Explains 44% of variance in cum. GPA
Highly selective schools generally report that the lower stat students graduate at similar rates to the overall rates. I expect that some of those students use above average resources though, and we do know that many athletic teams, for example, have dedicated tutors for the team.
You’re right, U Chicago doesn’t complete a CDS, but they do divulge some info. Class of 2024 admitted students test ranges were ACT 20-36, and SAT 1020-1600. UChicago was already TO last year, so that means applicants applied with the scores on the low end. Of course we don’t know how many admitted students there were at the low end of these ranges, but it must be very few because the mid 50% range is so high:
Over the years schools have done analyses of academic success and grad rates for both test submitters and non-submitters. DePaul’s data showed no differences, and their former VP Enrollment, Jon Boeckenstedt is an esteemed leader in the industry, now at Oregon State.
Edited to add link to data on test optional outcomes: