We address this empirical challenge by exploiting unexpected and transitory shocks to voting costs due to precipitation on election day. Following previous studies documenting that rain decreases turnout (Gomez et al. 2007; Hansford and Gomez 2010; Fraga and Hersh 2011), our test for habit formation amounts to asking whether election-day precipitation decreases voter turnout not only in the current election but also during future elections. Because precipitation on election day affects future turnout only through its impact on current voting, we are able to isolate the effect of habit formation. Our main estimates imply that a 1 percentage point decrease in past turnout lowers current turnout by 0.7-0.9 percentage points.
--Thomas Fujiwara, Kyle Meng, and Tom Vogl, "Estimating Habit Formation in Voting," on the virtuous cycle of doing your civic duty