Somewhere between 1.7% and 43.5% of success is just showing up


Inspired by this quickie analysis in which Fabio Rojas finds that 27% of variance in exam grades is explained simply by class attendance, and also by the recent end of my semester, I decided to take a look at how attendance affects grades in my big undergrad class. His class is on social theory, and I assume it’s just one thing; mine, by contrast, is a graphic design class made up of split but related lecture and lab components. I have a formal attendance score for lab, but only a proxy for lecture — how many quizzes they took.

The results? Pre-midterm quiz-taking has no significant relationship with scores on the midterm, which covers only lecture material; it accounts for only 1.7% of variance. On the other hand, lab attendance is hugely predictive of total points earned on lab projects, accounting for 43.5% of variance. I’m not terribly surprised by this, since lecture material is much easier to get outside of class than lab material is (primarily from the textbook), and because quiz-taking is an imperfect proxy than only takes into account who showed up at the beginning of class on quiz days. The crummy nature of the variable is somewhat confirmed by the fact that lab attendance is a significant predictor of midterm score, accounting for 7.2% of variance.

Share this post:

    Tags: ,