Missing the (Grade) Point

A new study about grade inflation begins by quoting an 1894 report of a Harvard University committee that was distressed by grade inflation. Apparently at Harvard at the end of the 19th century, students were earning As for "work of not very high merit" and Bs "for work not far above mediocrity."

But after establishing that complaints about grade inflation have been rampant for well over 100 years, the study goes on the question the phenomenon – as well as the significance of the evidence cited in many studies of grade inflation, namely changes in the average grade-point average. One author on grade inflation is charging that this new paper is flat-out wrong, in its numbers and analysis.

A better way to measure grade inflation, the new study argues, is to look at the "signaling" power of grades for employment (landing prestigious jobs and higher salaries). To the extent the relationship between earning high grades and doing better after college is unchanged – and that’s what the study finds – the "value" of grades can be presumed to have held its ground, not eroded.

The study "Is the Sky Falling? Grade Inflation and the Signaling Power of Grades" (abstract available here), has just been published in Educational Researcher. The authors are Evangeleen Pattison, a graduate student in sociology at the University of Texas at Austin; Eric Grodsky, associate professor of sociology and educational policy studies at the University of Wisconsin at Madison; and Chandra Muller, a professor of sociology at UT Austin.

The study used three long-term federal data sets that tracked students and their academic performance.  These data sets were built on, among other things, transcript analysis so as to avoid the problem of students or graduates inflating their grades when reporting them to researchers.

While the study found that high school grades have gone up, on average, since the 1970s, it found no evidence for grade inflation at four-year colleges. In fact it found the average G.P.A. decreasing at colleges, including selective colleges. The study used data on the average G.P.A.s in 1972, 1982 and 1992 – covering periods of time that some past researchers have said that grade inflation was taking off. (It should be noted that while grade inflation is constantly debated in academe, researchers have been divided on it – with some seeing it as rampant and others doubting it.)

Average Four-Year College G.P.A.

The study goes on to examine "signaling." For graduates of four-year colleges, the study found a consistent correlation between higher grades and the odds of entering a prestigious profession. While the link between grades and pay declined slightly over the decades studied, it was not a statistically significant decline.

Looking at both G.P.A.s and post-college employment, the study concludes by saying that there is "virtually no support" for the idea of grade inflation across higher education.

Stuart Rojstaczer, a retired professor at Duke University who has written extensively about grade inflation (arguing that it is a real problem), said his data and other studies show a different story. The period in which the authors argue for grade deflation was one in which "every published paper" has found grade inflation, Rojstaczer said. He said that he doesn't "know where the error is," but believes there is a "huge error" in the data. Rojstaczer said via e-mail that he believes the new paper has "no value."

When Inside Higher Ed relayed Rojstaczer's criticisms to the authors of the new study, they rechecked their numbers and said that they stand behind them.

Pattison explained in an e-mail that their numbers differ from others for several reasons. One is that some previous studies have included all institutions of higher education, while the new study includes only four-year institutions.

In addition, one earlier study looked only at those who completed 10 credits, a limitation the new study did not use. Pattison said that the differences in the findings actually reinforce an important point about the limitations of judging grade inflation by G.P.A. averages. "Mean G.P.A. is very sensitive to the way the population of interest is defined," she said. "It is because of this that we argue in our paper that absent other changes in the distribution of grades, mean shifts do not themselves imply devaluation."