Recent studies reveal surprising results. In a randomized field experiment, Olken (2010) finds no difference in the policy outcome between direct democratic and representative decision making. Yet participants of the experiment regard the outcomes of the former to be more legitimate, although they are in substantive terms the same in both cases as only the procedure is distinct. This result indicates that direct democracy matters for the perception of the process itself, and not for the outcome of the decision making process. Empirical findings on the US, using advanced survey models, come to a similar conclusion. Lax and Phillips (2009, 2010) show that congruence between the median voter’s preferences and the enacted policies are not greater in states with more direct democratic institutions. This is puzzling, as it suggests that institutionalized participation of the electorate does not enhance the general democratic performance defined within the concept of congruence.
Measuring policy preferences is challenging and survey questions rarely cover a large set of policy fields which would allow to generate a general and robust measure of the median voter's preference. But to evaluate institutional factors which may or may not contribute to policy congruence it is paramount that one has a battery of questions which encompasses a large number of relevant policies. We present a novel data set with almost twenty specific policy questions which allows us to generate district specific measures of policy preferences and compare that to the policies in place.
We apply fruitfully multilevel regression with post-stratification (MrP) to generate district specific estimates of policy preferences. This allows us to analyze whether more direct democratic provisions actually increase policy congruence. Further, this study also illustrates which institutional elements may be conducive to the impact of direct democracy. Institutions such as all party governments may change the strength of the direct democratic impact as well as the size and strength of organized interest. This point is even more important as empirical work often assesses the impact of direct democracy by regressing some measure of congruence on a measure of direct democracy. But since direct democracy requires organizational power (members to collect signatures, funds for campaigns, …) one has to also emphasize the role of agency when looking at the structure. We show in this paper how the strength of organized interest mediates the institutional effects.
