"No, the SAT is not Required." More Colleges Join Test-Optional Train

[Data10]

<p>

The data from the research shows the direct relationship between SAT scores and graduation rates. This is important from the aspect of college admissions because high SAT correlates well with high graduation rates. Low SAT also correlates well with low graduation rates. Another reason why SAT is important for ranking purposes.

When adding controls, the study found only a ~2% difference in graduation rate for a test score 1 SD away from the mean of the college. If the average student at that college had a 72% chance of graduating, the 1 SD low SAT scoring student would have a 70% chance of graduating. I certainly wouldn’t call this slight difference a strong correlation. The author of the study goes so far as to say that the correlation is too small to be statistically significant, as quoted below:</p>

<p>" The ATT indicates that there could be a small SAT effect on graduation (2.2 percentage points for a standard deviation increase in college SAT), but this does not reach statistical significance. The ATU is much smaller in magnitude and is not significantly different from zero."</p>

<p>

For instance, Bowdoin has a mean SAT of about 1400 for submitters and about 1300 for non-submitters. Based upon the graph one would not see that much of a difference in graduation rates from submitter and non-submitter . Some how Bowdoin is able to select relatively high SAT scoring students from the non-submitters. Had Bowdoin selected high HSGPA students with average SAT scores from the non-submitters, it is clear from the data that Bowdoin would see marked differences in graduation rates of non-submitters vs submitters…

Note that the study compared the rate of graduation for students with test scores +/ # SDs from the mean of the college, not the mean of the overall population. It’s essentially looking at whether the low scoring students at a particular college have a reduced chance of graduating compared to the average or high scoring students at that particular college, when controls for the individual are held constant (similar HS GPA, similar HS curriculum, similar race, …). That table does not consider how low the test score is an absolute sense, only how the test score compares to other students at that college, sort of like what percentile the test score is at that college.</p>

<p>It’s also not a big mystery how Bowdoin is selecting non-submitters with high test scores compared to the overall population. Test scores add relatively little to the prediction of academic success beyond other sections of the application because they are correlated with other sections of the application. You rarely find students who excel in GPA, curriculum, LORs, essays, ECs, and other areas compared to the Bowdoin applicant pool; but totally bomb the SAT and get scores near the US mean of 1000/1600. Instead most of non-submitters have test scores that are somewhat lower than the average submitter at Bowdoin, but still have far stronger test scores than the typical US student.</p>