“…The point estimates from our main results are substantively unchanged, although the standard errors for these lagged estimates are large, given fewer posttreatment years. For all estimates, we cluster standard errors at the state (treatment) level to control serial correlation in outcomes (Cameron & Miller, 2015), particularly given the salience of state policy and geography in shaping outcomes for students and institutions (González Canché, 2014, 2017, 2018; Odle, 2021; Odle & Delaney, 2022; Odle & Monday, 2021).…”