Policy & Politics Journal Blog

Daniel J. Mallinson

Since the 1960s, political scientists from across the globe have been studying how and why policies spread. This substantial body of research begs the question, what have we learned? My project aims to answer that question, at least in part. It finds both substantial growth in the literature and gaps that remain to be filled.

I conducted a meta-review of policy diffusion studies that focus on the American states. By casting a wide net using Google Scholar and Web of Science, I identified all (to my knowledge) studies published between 1990 and 2018 that referred to “policy diffusion” and “berry and berry.” Berry and Berry are important because their 1990 study of state lotteries introduced the unified model of policy diffusion. Essentially, this model combined the internal characteristics of states with influences external to the states to explain policy adoption. Over time, scholars also recognized that the attributes of the policy innovations themselves condition how far and how quickly they spread.

Figure 1. An Updated and Simplified Unified Model of Policy Diffusion

Figure source: Mallinson 2015

The literature search yielded over 2,700 articles, with 185 of them including 507 statistical models of policy innovation adoption that capture at least some of the components of Figure 1. From the resulting Policy Diffusion Results database (paper, data), I found the following:

• Not all policy topics are represented in the existing studies. Some are more commonly studied (e.g., law, health, administrative, and civil rights policies) and others barely register (e.g., social welfare, public resource management, local government, and agriculture).
• Studies best explain diffusion from roughly 1970 to 2010.
• Many diffusion models are highly contextualized, meaning they include multiple variables specific to the innovation (e.g., electricity prices in a model of electricity deregulation policy). Other variables that are expected to have more general effects across different types of policies are not always included (e.g., ideology and legislative professionalism).
• A meta-analysis of the results from these studies supports the presence of a positive effect of neighbour state adoptions, the negative effect of ideological distance between a state and past adopters, a positive effect of government liberalism, and negative effects of Republican control of state government and divided government on innovation adoption.
• The meta-analysis results vary across different policy types. Regulatory policies appear driven by citizen liberalism and inhibited by Republican control and divided government. Morality policies show strong external influences from geographic neighbours and ideologically similar states. Governance policies are positively influenced by neighbour state adoptions and the availability of the citizen initiative, but negatively by federal intervention.
• There appears to be null results bias in the cases of neighbour adoptions and ideological distance, which suggests there may be a “file drawer” problem in the published corpus.

There is much left to be done to explore the contours of the existing literature on policy diffusion, but this study points to clear gaps that can be filled by future work. By illuminating important biases in policy diffusion research, the article makes recommendations for addressing those biases as well as increasing international collaboration on policy innovation research.

You can read the original research in Policy & Politics:

Mallinson, Daniel J. (2021) ‘Growth and gaps: a meta-review of policy diffusion studies in the American states’,  Policy & Politics, DOI: https://doi.org/10.1332/030557321X16119271286848

If you enjoyed this blog post, you may also be interested to read:

1st May – 31st July 2021 highlights collection on policy diffusion