Predicting the outcomes of policy diffusion from U.S. states to federal law

TL;DR

This study develops predictive models to determine which state policies in the U.S. are likely to become national laws, emphasizing the importance of policy adoption count over state traits.

Contribution

It introduces a forecasting approach that predicts policy diffusion timing using state-level adoption data, challenging assumptions about state traits influencing national policy adoption.

Findings

Traits of initiating states are not systematically linked to policy becoming national.

Number of state adoptions predicts the timing of national policy adoption.

Forecasts for policies like marijuana legalization and 'stand your ground' laws are provided.

Abstract

In the United States, national policies often begin as state laws, which then spread from state to state until they gain momentum to become enacted as a national policy. However, not every state policy reaches the national level. Previous work has suggested that state-level policies are more likely to become national policies depending on their geographic origin, their category of legislation, or some characteristic of their initiating states, such as wealth, urbanicity, or ideological liberalism. Here, we tested these hypotheses by divorcing the set of traits from the states' identities and building predictive forecasting models of state policies becoming national policies. Using a large, longitudinal data set of state level policies and their traits, we train models to predict (i) whether policies become national policy, and (ii) how many states must pass a given policy before it becomes national. Using these models as components, we then develop a logistic growth model to forecast when a currently spreading state-level policy is likely to pass at the national level. Our results indicate that traits of initiating states are not systematically correlated with becoming national policy and they predict neither how many states must enact a policy before it becomes national nor whether it ultimately becomes a national law. In contrast, the cumulative number of state-level adoptions of a policy is reasonably predictive of when a policy becomes national. For the policies of same sex marriage and methamphetamine precursor laws, we investigate how well the logistic growth model could forecast the probable time horizon for true national action. We close with a data-driven forecast of when marijuana legalization and "stand your ground" laws will become national policy.