Sorry I get very excited when I see something I actually know! I really know little about football or basketball beyond being a fan and watching.
It's hard to know too much by just looking at the graph in the Dartmouth report. They leave out the statistics thout would be definitive, so I am inferring a lot.
But what I see is the standardized test predicting first year Dartmouth GPA really well. The two parallel lines on the graph would suggest overprediction of GPA for the low SES group if you mixed the two groups together (the way that standardized tests are usually used - without separating out SES by group).
In other words, if your goal is to predict first year GPA, the standardized test is NOT BIASED against the lower SES group, If anything, it is biased against the higher SES group.
I happen to be somebody who has looked again and again and again at standardized cognitive tests in employment. And it's the same story. I've never once seen standardized test scores show bias that disfavors minorities when predicting job performance. Commonly, the tests will overpredict performance.
Bottom line - the graph and paper suggest no bias in predicting GPA. The bigger question, as
@chito_and_leon has pointed out, is what your "criterion" or goal is as a university. Is it to get a student population that gets the best grades possible? If so, nothing is better than the SAT or ACT. It works much better than high school GPA, and it's not biased for that purpose. However, if you want other things, like well adjusted, hard working students, it doesn't tell you much. If you value diversity as an end in itself, obviously SAT and ACT work against that objective - unless maybe you mess around with how it's being used (which maybe Dartmouth is doing) and not just say "higher is better."