Panel Paper:
The Effects of No Pass, No Drive Policies on High School Education
*Names in bold indicate Presenter
I categorize No Pass, No Drive policies into three groups, differing in the group of students targeted by the policy. Enrollment-based NPND policies, in 20 states, will revoke a driver’s license from a student who is habitually absent from school or who has dropped out of school. This group of policies has negligible effects on annual dropout rates, but decreases the Averaged Freshman Graduation Rate (AFGR), a popular graduation rate estimate over the past 10 years, by more than one percentage point. However, this lower graduation rate does not stem from a true reduction in the likelihood a student will graduate, but rather from students delaying their dropout decisions. In turn, these students are retained in the ninth and tenth grades, increasing ninth grade enrollment by 2.8 percent, relative to eighth grade enrollment the year prior. This underestimation of the AFGR is not benign, however, as the AFGR was the primary graduation rate estimate used under the No Child Left Behind Act from 2004-2012 in determining Adequate Yearly Progress. Truant-based NPND policies, in 4 states, will revoke the license of truant students – those who are enrolled in school and habitually absent. Importantly, these policies do not target dropouts; a truant student losing his or her license under one of these policies can drop out of school and have their license reinstated. These policies cause a dramatic increase in the annual event dropout rate – approximately 1.2 percentage points, or a 27.8% increase over the mean. Additionally, these policies reduce 11th grade enrollment by 2.65 percent, relative to 10th grade enrollment. Behavior-based polices, in 3 states, had no significant effects. The validity of these results was tested using a series of Placebo tests, testing for type I error. Placebo testing showed that the enrollment-based results were valid, but the truant-based NPND policies faced severe type I error. I then corrected for this error using the restricted wild clustered bootstrap-t proposed in Cameron, Gelbach, and Miller (2008). Overall, I show that NPND policies have important unintended, and previously unobserved, consequences, and a reevaluation of the intended outcomes, language, and implementation of these policies is needed.