Guess what? I've just learned about the Arkansas Research Center at Conway, which has compiled a mountain of Arkansas-information on this very subject.
Neal Gibson, director of the state center, writes with some links to the subjects and also a link to a recent report on wages and Arkansas college graduates. Short version: Grades seem to be more reliable indicators of college success in Arkansas, too.
Concerning your post, “Grades mean more than college test scores, study finds,” I thought you might be interested in looking at data I prepared for a report to the Legislature.
The first I did about 18 months ago, when our office was just getting started. At the time, I did not have enough data to actually look at on-time graduation rates, so I had to limit it to First Year College GPA. If you look at the CorrleationMatrix spreadsheet, what it shows is which measures are more highly correlated with FYC GPA among ACT Scores, High School End of Course (EOC) exams, and HS GPA. In general, you square the value of the correlation to determine the “power” of the relationship. HS GPA wins hands down.
The second spreadsheet was created more recently, as our office has had the time to put together data sets from multiple agencies. These are actual percentages of on-time graduation rates based on ACT composite score on the X axis and HS GPA on the Y. The general idea here is “which tells you more, X or Y”? You can see quite high percentages for students that have composite ACT scores well below the magic “19,” if they happen to have a GPA of 3 or higher. Conversely, you can see quite low percentages for students with very high ACT composite scores if their HS GPA is less than 3. You don’t really need a fancy statistical model to show you what’s going on—the actual results speaks pretty clearly.
We’re very new, and we’re still trying to figure out how to propagate some of our research. Our latest report concerns the wage outcomes for Arkansas’ college graduates going back 10 years: