They seem to control for day of week:

Importantly, this study design permits adjustment for potential confounding factors that could be related to both album release timing and traffic fatality risk (e.g., albums tend to be released on Fridays) while still taking advantage of the otherwise exogenous variation in album release timing.

Edit: Direct link to full paper: https://www.nber.org/system/files/working_papers/w34866/w34866.pdf