A basic tenet of democracy is that policy should correspond to citizens’ preferences. While there is general correspondence between public opinion and policymaking, occurrences of policy-opinion incongruence are frequent. Direct democracy is widely thought to improve representation. Such a hope can also rely on game theoretic reasoning which suggests that it not only has an influence on policy through successful popular votes but also through the behavior of forward looking legislators who factor the possibility of initiatives into their decisions. However this mechanism fails if actors possess insufficient information. I posit that the initiative works for simple and salient policy issues but fails to work on all and every issue. This could explain divergent findings in the literature that direct democracy improves congruence for individual issues but has no effect on more general measures of policy. I test my hypotheses with data on policy and opinion in US states.
