Evidence on voter preferences from unrestricted choice referendums

From 1939 to 1968 Florida used a unique referendum system to set property tax rates for public school operating expenditures at the median millage rate selected by voters. These referendums revealed the entire distribution of voter preferences, which is not possible in the standard up or down referendum. We are the first to use the Florida referendum data. The form of the ballot played an important role in how people voted. Voting machine elections were much more likely than paper ballots to result in rejections of the recommendations of school boards, and produced much greater dispersion of expressed preferences.