Evaluating the Median Voter Model’s Explanatory Power

We match individual senators’ voting behavior on legislative proposals with 24 real referenda decisions on exactly the same issues with identical wording. This setting allows us to evaluate the median voter model’s quality with revealed constituents’ preferences. Results indicate a limited explanatory power of the median voter model: It explains 17.6 percentage points more than random voting and a senator’s probability to accept a proposal in parliament increases on average by 8.4 percentage points when the district median voter accepts the proposal.