Decades of research have offered strong evidence for policy diffusion, whereby one government's adoption of a new policy influences subsequent governments' enactment of the same innovation. But most of this rich research has narrowly focused on the spread of statutes in the legislative arena, neglecting the myriad other venues where policy change occurs. And even when scholars have taken note of policies adopted via multiple forums, they have typically employed binary models to estimate enactment without accounting for inter-venue dynamics that might affect policy diffusion. In addition, nearly all diffusion studies fall prey to selection bias, explaining the transfer of innovations that have knowingly diffused, omitting from the models those policies that failed to spread. What is more, most of this research has focused on the transmission of the policy itself, overlooking the potential diffusion of alternative aspects of the policymaking process. This dissertation addresses these omissions and capitalizes on existing opportunities in the policy diffusion literature. First, to better understand the spread of policies beyond the legislative context, I mapped the diffusion of a large sample of ballot measures across U.S. states from 1902 - 2016, and both anti- and pro-gay marriage policies via multiple venues from 1993 - 2015. I offer evidence of policy diffusion via state legislatures, legislative referenda, citizen initiatives, state courts, and federal courts. While the results reinforce much of our current understanding of policy diffusion, they also help refine the precise nature of this dynamic process across varying institutional arrangements. Second, I used an established but underutilized modeling strategy-multinomial logistic regression-to better account for the transfer of innovative ideas via multiple competing arenas. This approach allows me to simultaneously recognize each factor's contribution to policy adoption in the respective venues and uncover inter-venue dynamics. Third, to address the persistent selection bias in diffusion studies, I relied on the same large sample of ballot measures pursued across U.S. states from 1902 - 2016. I find that nearly half of the ballot measures did not diffuse to other states, and almost three-quarters of the measures were enacted by less than a handful of states. Moreover, when I reran the models omitting policies that did not diffuse or only narrowly spread, policy learning's effect on adoption was twice as large when compared to the full set. This suggests that policy scholars may be overstating the rate of policy diffusion and inflating fundamental mechanisms' effect on the process.Finally, fusing the policy-diffusion and venue-shopping literatures, I investigated whether policy actors' choice of venue to press for anti- or pro-gay marriage policies in one state influenced subsequent states' actors to pick the same forum, a process I term venue diffusion. I posit that policy advocates look to and learn from others, purposively seeking a solution to their shared problem (i.e., policy learning) and how best to achieve that solution (i.e., political learning). By incorporating political learning into my models, I am better able to explain the dynamics of policy diffusion and offer evidence of venue diffusion, at least in the context of a salient morality policy. States are more likely to pick the venue that other, especially similarly-situated, states have chosen to enact the policy successfully. The interdependence between the American laboratories of democracy appears to go beyond merely the copying of a policy idea to emulating a fundamental input of the policymaking process.
Read
