I challenge Strong’s (2022) findings that student employment is not related to attendance. I argue that the original analysis is guilty of controlling for a post-treatment variable. As a result, the coefficients in the regression model do not show how employment causes changes in attendance. I show that employment likely has a negative effect on attendance even given severe confounding. Academics should, if asked, tell students that their attendance will likely suffer the more paid work they do.
|Publication status||Accepted/In press - 6 Oct 2022|