Are high school GPAs better predictors of college performance than the ACT?

Not in Kentucky!

There is an effort underway across the country and in Kentucky to reduce emphasis on and attention to the fairly disappointing scores from the ACT college entrance test. The argument being made is that high school grade point averages (GPA) are better predictors of college performance, so emphasis on the ACT is misplaced.

But, is that really true?

Before answering, keep in mind, there are absolutely no statewide standards for course grading in Kentucky, and grading is a rather subjective area under even the best of conditions. Furthermore, the KEES scholarship program has placed considerable pressure on grading in Kentucky, at least in some districts and schools, because the amount of KEES money awarded to each student is dependent upon that student’s GPA. Clearly, there is plenty of potential for GPAs to not be very accurate, if not purposefully inflated.

But, is there any data that shows what is happening?

The answer is data is available to investigate the issue of ACT versus high school GPAs as indicators of college performance. Clues to that came in a new report from the Kentucky Center for Statistics (formerly the Kentucky Center for Education and Workforce Statistics) that indicated that Center was indeed collecting data on GPAs and ACT scores, though the information was not all collected in the same spreadsheets and appeared to cover all graduates for the high school data but only had college data for those students who attended a Kentucky postsecondary school.

So, I contacted the Center to see if they could assemble a spreadsheet that only included high school data for those students who went on to college in Kentucky. That restriction was necessary because the Center has the college freshman GPAs from in-state schools but not for those out of state.

The Center came through beautifully with a series of spreadsheets for the Kentucky high school graduating classes in 2014, 2015 and 2016.So far, I have looked at the 2014 and 2015 high school graduates’ data for each Kentucky public high school. This includes the overall high school and college freshman year average GPAs for each school and each school’s average ACT Composite Score. The messages in the data for these two graduating classes are very consistent:

• High school GPA was a weaker predictor of college performance than the ACT for both the Kentucky High School Class of 2014 and 2015.

• As I found in an earlier blog, high school grading across the state varies significantly, with some districts grading notably harder than the statewide average and others grading much easier.

• Large variations in grading have important implications for the state’s KEES scholarship program because the amount of KEES money each student receives is determined by high school GPAs.

• For both the Kentucky High School Classes of 2014 and 2015, the correlations between the ACT Composite Scores and freshman college GPAs for each high school are notably higher than the correlations between the school’s average high school GPAs and the college freshman GPAs. This indicates that the ACT, not high school GPA, provides a better prediction of college performance, though neither statistic is guaranteed to be accurate for all students.

• Some schools grade pretty much in line with the way colleges grade, but others have massive grading inflation. For example, in 2014 the Harlan High School’s (Harlan Independent District) graduates who went on to a Kentucky college posted a high school GPA of 3.31. In their freshman year of college, they posted a very similar GPA of 3.11. Harlan did just as well in 2015, by the way. At the other end of the 2014 spectrum, the Fulton County High School’s graduates who went on to a Kentucky college posted a high school GPA of 3.23. However, at the end of their freshman college year, these same students only posted a college freshman year GPA of 1.38. Keep in mind, students must maintain a 2.00 minimum to remain on Kentucky public university campuses.

So, those who would have parents and the public disregard the message from the ACT are barking up the wrong tree. This test does provide important information.

Also, depending upon the school, parents would be well advised to be cautious about grading. Some schools do it well, but that is far from universally true in the commonwealth.

I put together a spreadsheet that looks at the 2014 and 2015 graduates’ data in several different ways. One worksheet for 2014 (2014 HS GPA Vs College GPA Diff) and another with a similar title for 2015 graduates looks at the difference between the high school GPA and the college freshman GPA for each Kentucky high school that had data (if a school had data for fewer than 10 graduates, the data was suppressed for privacy reasons). You will see there is enormous variation in grading on this spreadsheet depending upon the high school.

The second set of worksheets (e.g. “2014 HS GPA Vs College GPA”) looks at the GPA information in a different way. Each school has a ranking for its high school GPA and a separate ranking for its college freshman GPA. Schools are then ranked by the difference in these two rankings. A school grading about on average will have a small difference in the rankings. A large positive difference shows a school tends to grade much harder than average while a large negative difference shows a school that grades way too easy. Those easy graders may well be doing that so their students get larger KEES scholarship awards.

The last set of worksheets (e.g. “2014 ACT Vs College GPA”) uses a difference in ranking scheme with the ACT Composite Score versus the freshman college GPA. The ACT scores here are from 11th grade testing conducted by Kentucky with all students and therefor are from the year prior to the high school graduation year.

The correlation between the GPAs and between the ACT and the college freshman GPA are found at the bottom of each worksheet and are summarized on the first tab of the spreadsheet package. The ACT and GPA data come from custom Excel spreadsheets assembled for me by the Kentucky Center for Statistics. Other calculations are mine.

As mentioned earlier, the data from the Kentucky Center for Statistics isn’t online, but the Center can be contacted for this information, or I will certainly forward the spreadsheet to requestors.

One more point. I got a heads up from a good stats guy that the GPAs might work better on a subject-by-subject basis, so I have started looking into that with English, math and science grades from high school and college and the ACT scores for those subjects. So far, I’ve only looked at 2014 high school graduates’ data, but the spoiler alert is that so far ACT is going to be a winner over grades even at the individual subject level. Stay tuned.