Ability, preparation, family background can all be expected to influence how one performs in college. These variables are all largely predetermined before college. So obviously things that get decided while in college such as how many courses to take or whether to do an internship may also be important.
How do all these various factors influence one’s academic performance? What is the contribution of each? This is the main question of this post.
Brugiavini, Carraro and Kovacic (2014) examine these issues. Their sample is all the students (minus a few exceptions) who graduate from the University of Venice between 2004 and 2012. This covers around 13,000 people.
They measure academic performance by two main variables: GPA and time to degree (i.e. how quickly a student graduates). Drop-outs are excluded from the sample.
Let’s look at the determinants of GPA first, in order of importance (ordered by me, not the authors). To interpret these results, note that Italian college GPAs are measured on a 18-30 scale.
Ability and preparation. A one standard deviation increase in final high school grade increases GPA by .72. Having attended a more academic-oriented as opposed to vocational school increases GPA by .43 to .84 depending on the exact school type. Clearly, higher ability students and students with better preparation have an advantage.
Courses taken. A one standard deviation increase in credits taken in the first year (an increase of 14 credits) is predicted to increase GPA by .57 points. However, a similar increase in second year credits leads to a mere .17 increase in GPA. A one standard deviation increase in the number of exams taken during the first year (an additional 3 exams) decreases GPA by .28. Interestingly though, taking more exams helps in the second year; but the effects are quantitatively small (.08 rise in GPA only).
Altogether, these results imply that taking few but hard courses (low number of exams, but high number of credits) in the first year is beneficial. While in the second year, loading up on hard courses may not be so helpful. This may be because hard courses in the first year provide a strong foundation for studies later on, which increases GPA in the longer run.
A last thing to mention in this category is that class attendance has a large positive effect on GPA as well. Those whose attendance is above 75% have on average a .21 point higher GPA.
Working and internships. These activities generally have a negative impact on GPA. Working regularly decreases GPA by .25 points, doing an internship outside school decreases it by .10 points. Occasional jobs and internships at school have a much more negligible effect (the latter being statistically insignificant).
Family background. Mother’s education is the most important determinant in this group. If a student’s mother has a college degree, their GPA is expected to be .12 points higher. Father’s education has a negligible and much less robust effect. Working class students on average have a .08 point higher GPA, middle class students a .05 higher one, when compared to upper class students. So after controlling for ability, coming from a poorer background actually increases GPA. This may be because these students are more motivated to do well in school (given their worse background).
So as one can see, after controlling for ability family background is the least important determinant of GPA. This is a desirable thing, but family background still has a significant effect, so this is not to say that it can be ignored. It is still nice to see that other variables (over which students have higher control) are more important. Specifically, courses taken, which is 100% within a students control while in college comes in second and is a very important predictor of success.
Another interesting result that comes out of this analysis is the role of gender. While gender is not important at all in the whole sample, males seem to have a slight advantage in the top 10%. This is consistent with other studies that have shown that males tend to have larger variance in their abilities, and thus there are relatively more male under- and overachievers. We could therefore expect that males would perform worse than females in the bottom 10% of the grade distribution. Unfortunately, the authors do not carry out the analysis for that subgroup, so this hypothesis cannot be tested.
Now that we’ve looked at the determinants of GPA, let’s check out the results on time to degree. The authors estimate the probability of being on-track (i.e. on schedule) with one’s studies. Let us look at all factors influencing this variable in order of importance (again, ranked by me and not the authors).
Courses taken. Not surprisingly, the amount of credits taken during the first and second years have the largest effect on the time of graduation. Clearly, taking more credits will allow one to graduate earlier. Taking fewer but larger exams (especially in the second year) seems to be beneficial for early graduation. This goes somewhat against what we saw for GPA. Class attendance is still important and has a positive effect on being on-track.
Work and internships. Working regularly has the greatest negative effect on time to degree in this category. Occasional jobs tend to delay graduation as well, whereas internships do not have a negative effect. This may be because internships can count towards graduation in some programs.
Ability and preparation. Preparation (i.e. having attended a rigorous high school) can help students stay on-track, but ability loses much of its importance. Still both preparation and ability are statistically significant, and have a positive effect on being on-track.
Family background. This category has almost no effect on time to degree. Parents’ education doesn’t matter anymore. There is a slight delay in graduation for working class students compared to others. And students who are awarded a (need-based) scholarship tend to graduate faster. But the sizes of these effects are small.
An interesting variable that is hard to fit into the above categories is unemployment within one’s major. Students tend to graduate later when the unemployment rate is higher. An increase in unemployment thus decreases the probability of being on-track. The size of this decrease is roughly equivalent to that of a one standard deviation increase in high school grade, but obviously opposite in direction.
Finally, let’s combine GPA and time to degree. We can expect GPA and time to degree to affect each other directly as well. The authors thus specify their last, perhaps most complete model, which accounts for this interdependency. In other words, GPA is now allowed to depend on time to degree and vice versa. The results presented above still largely hold (both qualitatively and quantitatively). In addition to this, we can also see that being on-track as opposed to off-track increases GPA by a mere .08 points. This is statistically significant, but it is quantitatively small. Being on-track is thus not really an important determinant of GPA.
GPA also positively influences timely graduation. Increasing GPA by one standard deviation (2 points) increases the probability of being on-track by roughly the same amount as a one standard deviation (16 credit) increase in second-year credits does. So a good GPA seems to be important for timely graduation.
To sum up, I would say that the recipe to succeed in college includes proper preparation, good planning (of what courses to take and when), and staying focused on one’s studies whenever possible (e.g. avoid working, attend classes).
We also learned that if a policymaker wants to decrease the average time to degree, they would need to restrict the freedom of students’ course choices. This could result in more students following regular academic tracks and thus graduating on time.
I suppose this shows that free course choice is one of those freedoms that can be really useful for students who know what they’re doing (and choose courses effectively), but it can cause delays and problems for those who are not responsible or forward-looking enough. The average student will probably fall into the latter category making restricting course choice a potentially beneficial policy. At least if we want to decrease time to degree.