How Do Policy Memes Spread? An Analysis of the UK House of Commons Debates

Abstract

Scholars still know very little about the diffusion and evolution of policies within legislatures over time. This paper presents a first step towards a computational model that captures how policies are discussed by political actors and diffuse among them. We introduce the concept of ‘political meme’, that is, a set of tokens that refers to a specific policy issue. Drawing on research in information diffusion theory, we develop a framework to account for the diffusion policy memes. We use a novel dataset of legislative speeches from the UK House of Commons that includes every debate from 1935 to 2013. Focusing on the last five legislative sessions we extract the most frequent bigrams and trigrams in the texts, and identify a sample of policy memes. Next we then derive their cumulative frequency distributions and test whether their diffusion growth curves can be modeled as a logistic function. Our findings suggest that not all political memes follow an ‘S-shape’ curve. Indeed, very idiosyncratic memes referring to sudden events or policy issues register steep increases in attention, while others appear to be discussed at constant rates, therefore being better described by a linear function. Further research will integrate our findings with network science, investigating the structure of the network of speakers revolving around pre-defined political memes and extending our analysis to include full range of data.