Good Predictor of Achievement: GPA?

            GPA easily condenses four years of college into one number. But what does that number mean? Intelligence? Work ethic? Potential for success? As an undergraduate, I worry about maintaining a good GPA, partly because of my impression that I need a good GPA to succeed professionally, and partly because I feel pride over good grades and shame over poor grades. However, I have never believed that grades alone disclose students’ aptitudes or foretell their future successes. A number of recent events have motivated me to investigate what college GPA really means.

  1. My Microbiology 201 Honors professor gives an A or A-minus to every student in the top half of my 23-person class; my friend’s Microbiology 201 professor grades on a standard “A = 93 to 100, A-minus = 90 to 93” scale. How can you compare the grades of students from these two microbiology classes?
  2. The moderators of a seminar about scholarships that I attended said outright that if you are a senior with a 4.0, you have probably never dared to challenge yourself, or else you have spent all of your time studying and missed everything else about college. How does a GPA indicate experience and critical thinking?
  3. At the annual student awards ceremony, Penn State seniors in the top 0.5% of their class (3.99 GPA and above) were honored. Additionally, senior Chris Rae was honored for winning a Gates-Cambridge scholarship, and senior Ryan Henrici was honored for winning a Goldwater Scholarship, a Marshall Scholarship, and an Astronaut Scholarship. Neither student was in the top 0.5%. Is a student with a 4.0 and no scholarships more successful than a student who won a Gates-Cambridge scholarship with a slightly lower GPA?

Why is GPA important?

According to Roth and Bobko, there are two main reasons to use GPA: it is easy and cheap to obtain and interpret, and it is generally believed to reflect student aptitude and work ethic. Thus Caulkins, Larkey, and Wei say that colleges consider GPA in awarding financial aid, ranking students, and determining whether students have performed well enough to earn a degree. Employers use GPA to screen interviewees; indeed, students may associate GPA with their self-worth. Certain majors at Penn State, such as Mechanical Engineering, also mandate that entrants have a 3.0 cumulative GPA. GPA thus has a potentially lifelong impact on career, income, and satisfaction.

What are the flaws of GPA?

            The rationale of using GPA is that it indicates student aptitude and efficacy, according to Roth and Bobko. This is partly rational, as GPA is weakly correlated with general intelligence. However, the correlation constant (r) is somewhere between 0.3 and 0.7, which means that r2, which tells how much of the variance in GPA is due to variance in intelligence, could be between 0.09 and 0.49; that is, at most, only half of the difference in GPA between two students can be attributed to difference in intelligence. At the least: only 9%. The authors said that student conscientiousness also affects GPA (r = 0.34, r2 = 0.12), but that external factors, such as differences in grading scales, course difficulty, biased professors, and other time commitments like part-time employment, impact GPA.

Caulkins et al. illustrate that external factors in grading invalidate student-to-student GPA comparisons, especially across majors. For example, in a study of seven colleges, the average math class GPA in 1991 was 2.53, while the average English class GPA was 3.12—a difference of 0.59. Are English students that much smarter? No, say Caulkins et al.: the math classes drew higher achieving students but had more stringent grading standards, making it impossible, without adjustment, to compare the GPAs of students with substantially different course work. That GPA contains numerous factors besides intelligence and conscientiousness, such as the difficulty of classes taken, argues against using it alone to rate students.

Then what, if not GPA?

Caulkins et al. argue that no single number can accurately describe student performance. However, though GPA can never truly be accurate, it can be much more so than it is now. They evaluate five ways to adjust grades class by class to make GPAs more accurate:

  1. Calculate the difference between a student’s raw grade and the average grade in each course, and then average those differences to calculate GPA. GPA = average(student grade – course average). This method, however, does not consider that grades in some courses are clustered, while grades in other courses are widely spread.
  2. Calculate the z score of the raw grade a student received in each class, then average the z scores to find the GPA. GPA = average((student grade – course average) / course standard deviation). While this method corrects the clustered/spread problem, it does not consider that an average student will earn a lower GPA by taking classes with high-achieving students than by taking classes with low-achieving students.
  3. Instead of comparing students to the average scores of their classmates, compare students to the average historic grade of every student who has taken the class, by subtracting the average historic grade from each student’s grade. GPA = average(student grade – average historic grade).
  4. Instead of subtracting, divide each student’s grade by the average historic grade. GPA = average(student grade / average historic grade). This method and method 3, however, do not consider that in some classes, the lowest- and highest-performing students receive grades within a narrow range, while in other classes, grades are spread widely.
  5. Combine methods 3 and 4: first divide by an average historic grade, then subtract another value to adjust the grade, then average these adjusted grades to calculate GPA. GPA = average((student grade / average historic grade) – adjusting value). This method corrects for the variability in grade distribution.
  6. The Item Response Theory (IRT) method is an already-developed method that has successfully adjusted GPA to better match student aptitude.

Which method is best? Caulkey et al. studied whether these GPA-adjusting methods were better correlated with three pre-college-admission variables—high school GPA, math SAT, and verbal SAT—among Carnegie Mellon undergraduates. They found that for natural science courses, method 3 GPA correlated best with high school GPA and math SAT, and IRT GPA correlated best with verbal SAT, although not statistically significantly so. Thus the authors argue that since GPA correcting formulae are simple and more effective than raw GPA, colleges should all employ them. However, they concede that phasing in adjusted GPAs should be gradual, as a sudden change would upset colleges and employers, who use GPA so widely.

GPA inevitably has limitations

Still, a single number, even an adjusted GPA, cannot encode all that a student is or can become. Importantly, we must recognize what GPAs mean. Not only are they affected by intelligence and motivation, but also by professors, classes, classmates, extracurriculars, and dumb luck. Students should not inextricably associate these inherently flawed statistics with their self-worth, potential, or any parameter mentioned already.

Ernest William Goodpasture got a D in Latin at Johns Hopkins. Later, he pioneered growing flu vaccines in chicken eggs, a method that has saved millions of lives up to this day.

Leave a Reply