The Devil Shift and Brokerage Roles: Comparing Policy Networks in Four Countries

Misperceiving the influence of political opponents is a common phenomenon in the policy process. Observing this general tendency, Sabatier et al. (1987) firstly formulated it as the devil shift, which is defined as perceiving political opponents as being more powerful than in reality and also more evil than they actually are. Later, the opposite phenomenon, the angel shift, which refers to inflating the influence of one’s own coalition partners, was identified by Leach and Sabatier (2005). Since the proposal of these concepts, their applicability to empirical cases have been tested but significant effects are difficult to find probably because the effects of each “shift” can cancel out each other (Fischer and Sciarini 2015). We argue that discerning the effects of the devil and the angel shift has been hampered by a lack of proper analytic measures. We propose a way to measure the degree of overestimation of influence by utilizing the chi-square test and differentiate the devil shift and angel shift phenomenon by applying a perspective of fuzzy set in which the degree of devil and angel shifts are seen as a continuum instead of being discrete phenomena. This way it is possible to measure the degree of the devil and the angel shift simultaneously. Furthermore, previous research has not taken full notice of how network structures contribute to the devil shift phenomenon. We argue that conceptualizing the different roles that brokers can occupy is essential for explaining when the devil shift takes place. The ideal condition for the devil shift is when an actor is an opponent who is connected to other opponents with whom ego is not connected with. Such a relation represents gatekeeper type of brokerage. Our suggestion is that the phenomena of the devil and the angel shift can only be understand with the help of the brokerage typology proposed by Gould and Fernandez (1989). We operationalized and tested these ideas in the context of climate policy network surveys collected in four countries: Finland, Sweden, Germany and Japan. The respondents were the most important organizational actors involved in the domestic climate policymaking process and the respondents were asked to indicate who they consider to be influential. We created a composite variable indicating respondents’ policy positions on climate change and took the absolute difference between each pair of actors in order to see which actors oppose each other’s policy positions. After calculating the devil shift and angel shift scores, we set it as a dependent variable in QAP regression. As the main independent variable, we set the brokerage typology by counting how many times alters show-up in a brokerage position from the perspective of respondents. Across all four countries, we found a significant and positive effect for the devil shift in the case of the gatekeeper type of broker relation. Previous research has treated the devil and the angel shift as dyadic relations but our results indicate that they are instead triadic phenomena.