SAT scores well in predicting college success

katonahmom · January 12, 2007, 8:45am

Article in the Baltimore Sun reports that the much-maligned but ever-influential SAT received positive marks when it comes to the test’s power to predict long-term college success in Maryland. According to an analysis of recent student data prepared for the Maryland Higher Education Commission, the SAT is also an accurate predictor of retention and graduation rates at all of the state’s four-year colleges and universities.

"“The higher the SAT scores of students, the greater the likelihood that they not only returned for a second year of study but eventually earned a baccalaureate as well,” the report said.

The study comes a month after Salisbury University, joining a national trend, became Maryland’s first public four-year college to allow some prospective freshmen to apply for admission without submitting standardized test scores.

The University of Baltimore, Bowie State University and Frostburg State University also have said they are considering test-optional policies in an attempt to attract more qualified students who might be dissuaded from applying because of low standardized test scores.

Though the commission’s analysis confirmed the SAT’s long-term predictive strength, the purpose of the study was not to test the test but rather to try to fairly evaluate Maryland’s public colleges by isolating a variable common to them all – SAT scores – and then determining which institutions were most successful…

Salisbury officials said yesterday that while they agreed that the SAT was a good predictor of college success, they believed high school grades and the rigor of a high school curriculum were better predictors.

“We never negated the value of the SAT as one tool in the toolbox,” said Ellen Neufeldt, Salisbury’s vice president of student affairs.

Salisbury makes the SAT optional only for prospective freshmen whose high school grade average on a 4.0 scale is 3.5 or better.

Arguments remain
The Maryland analysis does not weaken arguments against the SAT, said Robert Schaeffer, public education director of FairTest, an advocacy group critical of the way standardized tests are used.

“It’s well known and nobody has ever denied that there is a relationship between SAT scores and persistence, on average,” Schaeffer said. “Indeed, there is the same relationship with retention and outcomes based on grades and based on family income. The point is that SAT scores add little useful information, and in some cases contradictory information.”

A spokeswoman for the College Board, the New York-based nonprofit that administers the test, said the SAT is prized by colleges because it is a standard measure and not subject to variability among high school grading practices."

<a href=“http://www.baltimoresun.com/news/education/bal-md.sat11jan11,0,5307321.story?coll=bal-education-top[/url]”>http://www.baltimoresun.com/news/education/bal-md.sat11jan11,0,5307321.story?coll=bal-education-top</a>

northeastmom · January 12, 2007, 9:02am

I don’t know why they are considering waiving the SAT for admission since they believe it is a good predictor of success. Towson’s graduation is unimpressive as it stands now.

Chedva · January 12, 2007, 9:07am

Well, isn’t persistence also correlated with success in college, and with graduation rates? Therefore, assuming that this is true, the SAT would logically be a good tool for measuring persistence.

ophiolite · January 12, 2007, 9:53am

I really don’t believe in the whole SAT and college success prediction, thus, I think all schools should have an optional option. For example, I got a SAT score slightly above the national average (1120 in 1999) and ended up having wild success in college and am now in a top PhD program. Conversely, my fiance had well above a 1200 on the SAT (can’t recall exact number, same year) and ended up flunking out of college within 2 years at the same university (due to a maturity issue). SAT’s will never be able to predict the internal drive and maturity of students and the system hurts students like myself who are hard-workers but it doesn’t show on the standardized exams.

Furthermore, I believe all high schools should be required to publish class rank on their transcripts so colleges can properly evaluate the student’s GPA. Now, I know this isn’t a correct-all, as schools with very small student bodies can be at a disadvantage.

I think a great alternative to SATs would be a Regents-style exam requirement for all students across the country at the end of each school year. This way you are standardizing schools across the country on a single subject and it gives a way for a university to evaluate if the student really understood the material to earn that ‘A’ in the course.

northeastmom · January 12, 2007, 10:07am

ophiolite, I disagree. My son went to a very competitive public high school. He is successful so far in college. His SAT reflected his abilities, but he may have easily been in the middle or even below the middle of the pack in his high school. I do believe that the SAT really does help to equalize grades between high schools. His hs just does not give out As like candy, and a good deal of the students get tutoring after school, summers, and frankly at times don’t even see daylight. Also, APs are not open to all students. I think that the SATs helped him in admissions, and in the area of merit aid. I think that the strength of his hs was actually harmful in admissions to some degree, and definitely kept him out of some merit aid. BTW, his hs does not rank, because they do not think ranking is fair to their students.

weenie · January 12, 2007, 10:14am

^ I agree with northeastmom. I used to be really “anti-standardized exam,” until I saw the grade inflation at our local public schools. And then I realized that basing college admission on just GPA wasn’t fair either.

marite · January 12, 2007, 10:16am

What a strange argument. Exams, whether SATS, ACT, APs, or college tests, are not supposed to evaluate personality! They’re supposed to evaluate how academically prepared for the next level–in this case, college. Exchanging the SAT for Regents-style exams will not make it any easier for colleges to predict whether students will flunk out of college due to “maturity issues.”

ophiolite · January 12, 2007, 10:29am

northeastmom and weenie:

I never said that I agree with GPA calculations. What I believe in is class rank in conjunction with standardized exams at the end of the year, such as a Regents exam to standardize across schools, could help universities better evaluate a school’s performance better than with SAT scores (which only test math and English). If there was a standardized exam at the end of the year that tested students working knowledge of say, chemistry, that each student across the country had to take for a score to evaluate the strength of the class they took that year, it would be much more useful to an admissions committee than a silly SAT score.

At my high school there was NO grade inflation. I graduated with under a 3.0, but was still in the upper third of my class (never saw my final transcript but probably ended up in 25-30% range). Also, APs were only open to students recommended by a teacher to take that particular AP class. For example, the only AP classes open to a student like myself were AP Bio and AP Stat, as those were the ones recommended to me. I ended up taking no AP classes and it didn’t hurt me in the long run either (in part due to the difficulty of my high school). I do not know how these factors affected me regarding admissions, as I only applied to three schools, but it did affect my ability to get aid.

Exams such as the SAT are coachable, a well thought-out exam for an individual subject to evaluate in depth understanding of a particular concept is not as coachable. Therefore, the exams I propose, I believe would be more indicative of whether a student got a solid enough education to be adequately prepared for college. I’ve run into so many students at the universities I’ve been affiliated with who had great scores, but frankly didn’t absorb any material there were supposed to know going into college.

edit:

marite:

Yes, I concede, I didn’t present my argument in a logical manner. Nothing can evaluate maturity other than recommendations or personal essays.

garland · January 12, 2007, 10:35am

^so, great, we’re back to judging students by the school you went to. The flip side of your situation is the very bright students going to a less than optimal high school. I guess that once they didn’t get a “solid enough education” they should be doomed for life, and to heck with how smart they might be. If SATs were so easy to scam, everyone would have 1600/2400.

Plenty of kids from subpar high schools excel at selective schools, and the SAT is a way to show that they can.

Papa_Chicken · January 12, 2007, 10:40am

There is certainly much academic research on this topic, and I think many articles can be found on the College Board’s web site.

I have read some research papers by Jesse Rothstein, who has been studying the University of California database. My short take on his theme is that the SATs can be viewed as a surrogate measure to family demographic and originating high school variables, and if an admissions committee takes into consideration those demographic and school variables in their assessment of an applicants potential for academic success more directly, then individual SAT scores from the applicant don’t add anything new to the evaluation. [I have probably butchered his meanings, but you can read for yourself.]

Here are a few Rothstein references. (Ophiolite…I think some of this hits on your points of other measures more indicative of college success, plus, to your point about individual SAT scores, have a look at the last quote saying that the originating high school’s SAT average is more indicative of college success that the individual’s deviation from that average.)

presentation abstract from a 2001 SAT conference, title “Assessing the informational content of the SAT scores”:
<a href=“http://senate.ucsb.edu/meetings/townhall/sat/abstracts/5a-3.pdf[/url]”>http://senate.ucsb.edu/meetings/townhall/sat/abstracts/5a-3.pdf</a>
here’s the conference web site w/ links to other abstracts on the subject FYI:
<a href=“http://senate.ucsb.edu/meetings/townhall/sat/[/url]”>http://senate.ucsb.edu/meetings/townhall/sat/</a>

abstract

Previous research shows that SAT scores are consistently predictive of college performance as measured by freshman GPA (FGPA). Moreover, the SAT score retains substantial predictive power after controlling for students’ high school GPA (HSGPA). Thus, any admissions office that considers only HSGPA is passing up information about the expected performance of admitted students. However, college admissions offices have much more information about applicants than just their HSGPA and SAT scores, and the existing literature provides little guidance about the contribution of the SAT score in the presence of this other information. This is crucial to assessing the utility of the SAT score in admissions: How much does knowledge of an applicant’s SAT score add to our ability to predict his/her college grades based on all the other available information?
This paper investigates one subset of the variables available to admissions offices: measures of the demographics and resources of applicants’ high schools. These measures are strongly correlated with both SAT scores and FGPA, raising the possibility that the predictive power of the SAT is an artifact of the exclusion of school-level variables from prediction equations. Moreover, the school variables are reliable and readily available–the California Department of Education collects and publishes them annually as part of the API school evaluation program, and many other states publish similar measures. Colleges could certainly use the school measures to predict applicants’ performance, but most do not. It is found that much of the apparent predictive power of the SAT in simple models is due to the SAT?s correlation with omitted school-level variables. The SAT’s contribution to prediction is much reduced–though not eliminated–when the school-level variables are included in the prediction model. This result is found to be robust to changes in the outcome variable and to nonlinearities of arbitrary form in the prediction model. It seems more difficult to justify conducting admissions on the basis of demographic characteristics than on the basis of individual performance, though both are predictive of future performance. The demonstration that SAT scores function partly as demographic measures thus raises questions about the appropriate role of the SAT. These questions are considered, and an effort is made to describe the implications of different university policies for the conduct of admissions. As an application, the different prediction models are used to forecast the performance of students admitted under the University of California?s new class-rank-based ?Four Percent Plan.?

paper on “SAT scores, High Schools, and Collegiate Predictions” circa ~2005
<a href=“http://www.princeton.edu/~jrothst/rothstein_CBvolume.pdf[/url]”>http://www.princeton.edu/~jrothst/rothstein_CBvolume.pdf</a>

abstract

Family background is correlated with collegiate performance, as is the SAT. Moreover, SAT scores are themselves highly correlated with family background. Validity research has rarely taken account of these connections, though the extent to which the SAT?s predictive validity derives from its correlation with background is crucial to the interpretation of prediction models. Some of the most powerful proxies for family background are measures of the demographic characteristics and test performance of the high school attended, perhaps because parents who are involved with choosing good schools for their children tend also to be involved with their children?s education in other ways. Using data from the University of California, I examine the role of high-school-level characteristics in SAT validity models. I find that the school average SAT score is a substantially more powerful predictor of collegiate performance than is the individual deviation from that average. A portion, but by no means all, of the school average SAT effect can be attributed to its correlation with demographic characteristics of the school, particularly the racial composition. Validity models that use the unadjusted SAT score without demographic controls thus overstate the direct contribution of the individual SAT to prediction, attributing to it substantial variation that is better attributed to readily-observed school characteristics.

northeastmom · January 12, 2007, 10:42am

BTW, I know of some large high schools that offer SAT coaching sessions (prep classes) on week ends. They are at a nominal charge, or free of charge. Nobody ever brings this up. They only talk about how the wealthy have access to tutoring for the SAT, but in my experience this is not the case. It should be available to all students in those high schools. The same is true for the SATII.

weenie · January 12, 2007, 10:43am

Well, clearly no one parameter should be weighed too heavily…

Even class rank doesn’t tell a very complete picture. For one thing many schools won’t rank anymore. The argument my sons’ HS used when they dropped rank quite a while ago was that their top 25% kid was still better than the top 5% kid across town…Where does that argument end???

I guess schools that can still try to evaluate the whole kid are going to have a clearer picture of who they have accepted, and schools that are relying heavily on numbers will probably be less clear. In the end though a huge portion of the kids will succeed or fail based on their efforts and, yes, maturity.

marite · January 12, 2007, 10:45am

Re: post 10: This is probably correct in the aggregate, but colleges do not admit in the aggregate. They admit individual students. 
Dont’ those of us who have more than one child know that our children are not clones of one another? That they can differ in their SAT scores as well as their academic interests, class performance, etc… Mine did. Same parents with same levels of education and same SES. They went to the same schools, had the same teachers in many cases, and performed quite differently both in class and on the SATs.

northeastmom · January 12, 2007, 10:52am

weenie, I strongly agree with the reasoning your high school makes about ranking. I think that the same reasoning can be applied to our high school.

Chedva · January 12, 2007, 10:53am

Good luck with getting one of those for each “concept” across the country! It’s hard enough to get it to happen in a single state. And you don’t think there would be an increasing amount of “teaching to the test”??

In order to implement a national subject test, you’d have to get each individual school board and each state to give up its autonomy to set its own curriculum. Again, good luck with that - we still have school boards debating creationism v. evolution, and whether Harry Potter promotes witchcraft!

And at what level to you set the test? I’m sorry, but a kid who goes to public school in inner-city Washington doesn’t have the same opportunities as the kid who goes to Andover Prep, but you want them to reach the same level of knowledge in order to proceed to college. Gee, that sounds fair. Or, do we set it to a minimum level? In that case, it doesn’t really help to differentiate the top students.

northeastmom · January 12, 2007, 10:55am

Chedva, Very good points. Also, we already have SAT IIs for those that opt to sit for them.

chaoses · January 12, 2007, 10:58am

I think SAT score DOES NOT reflect college performance in general.

what if ur not a math type of person and do really poorly on the math section. When you go to college you major in dance or something not math-related and do really well?

what if ur a foreigner who have trouble in english, do really bad in both sections, does that mean you can’t learn english in college and can’t do well?

mkm56 · January 12, 2007, 11:15am

IMO work ethic is probably the biggest predictor of success in college—but of course how could that be measured? Certainly not just by gpa, as posters have pointed out, this is determined by the difficulty/competitivenss of the high school. I have seen students “squeak” in to colleges where they are are the low end of acceptance statistics and do extremely well. I have seen other with near perfect SAT’s who go to school and bomb out the first year.

And of course I’ve seen the students who fit the stereotype of high SAT and high college performance and the reverse. The one consistent factor among the successful students whether high or lower SAT is the determination to work and succeed—any test for that?

MarathonMan88 · January 12, 2007, 11:26am

One way to think about the roles of SATs in predicting college success is to ask how much additional predictive power the SAT gives in particular admissions processes. Some colleges have models that factor in (formally or sometimes more informally) school rigor, school resouces, public/private HS, parochial/nonparochial, gender, rank-in-clas, HS GPA, the adcom’s academic rating, recs, essays, and more. As you factor in lots of variables, the incremental contribution to predicting academic success of the SAT becomes smaller–so small, in some cases, that some institutions decide that they can live without it. It’s hard to imagine a cash-strapped public college implementing a multi-variable model like this, because it’s expensive, and for all it’s faults, the SAT-required route probably still works just as well or better, and more efficiently.

Maybe there’s no one-size fits all answer re the utility and predictive power of SATs.

CountingDown · January 12, 2007, 11:26am

Our school system offers SAT prep classes as an elective during the regular school day, in addition to classes on the weekends for a nominal fee. They’ve also signed a contract with CB for their online SAT prep class so that students can take it for free. They are big into increasing the number of kids from all socioeconomic brackets who take the SAT while keeping their test averages very high.

SAT scores well in predicting college success

Related topics

CONNECT WITH US