Measuring Policy Diffusion in Federal Systems: The Case of Legalizing Cannabis in Canada under Time Constraints

Abstract Existing studies of policy diffusion rely on quantitative or qualitative methods depending on the number of cases and the policy at hand. Studies of diffusion in Canada, for instance, almost exclusively use qualitative techniques due to the limited number of subnational units. In this article, we explore whether machine learning techniques can complement qualitative approaches in these contexts. In 2015, the Canadian federal government decided to impose the legalization of cannabis and gave the provinces and territories a short time frame to develop and implement legislation. Previous qualitative research on this case found that within-province policy development was more salient than interprovincial diffusion. Using a plagiarism detection software, we find limited evidence of exact matches between provincial legislation, but a cosine score approach reveals significant similarities across provinces. These results suggest that computational and qualitative techniques together should be used where possible to identify and analyze policy diffusion in certain contexts.