Baumgartner & Leech (2001: 1200) have proposed the concept of "policy bandwagons" to capture policy issues that become the "object of veritable lobbying extravaganzas". They analyzed the distribution of lobbying activities, at the US federal level, across a random selection of 137 policy issues: 5% of these issues attract more than 45% of all lobbyists' attention. By contrast to these policy bandwagons, 50% of issues do not mobilize more than 3% of all lobbying activities (i.e. "interest niches"). Halpin (2011) replicated the US study for 1'691 public consultations organized by the Scottish government (1982-2007) and he also observed very few policy bandwagons co-existing with many "quiet corners". If the skewed pattern of interest groups mobilization is observable in different institutional venues, political systems and time periods, what are the potential explanations of this phenomenon? The proposed paper analyzes the dynamics of policy bandwagons in five California policy-making processes. Beyond providing an empirical test of the research hypotheses suggested by Baumgartner et al. and Halpin (size and scope of the policy issue, mimicry and attention cascades), the main added-value of this article is (1) to consider simultaneously all institutional venues of a policy making-process, including the administrative, parliamentary, judicial and direct democracy venue; (2) to add new empirical material with the comparative analysis of five California case studies with obvious policy bandwagons; and (3) to opt for an innovative research method (Markov transition models) to study the policy bandwagons dynamics. Thus, our contribution is to offer a first step towards Jones and Baumgartner's (2005:142) recommendation to rely on Markov switching models (which rely on a Markov transition model as component) to grasp the consequences of attention cascades. To the best of our knowledge, our proposed paper is the first to apply this methodological approach in the study of interest groups.
