In each game, students are randomly paired to compete in one or more elections, the first of which may be preceded by one or more polls. For each election or poll, candidates simultaneously choose a policy, represented by a position on the line from 0 to 100. Each voter has an ideal policy (also represented on the line) which remains constant across the game. A voter chooses the candidate whose policy is closest to his own, choosing randomly if indifferent. Candidates' chosen policies and the number of voters choosing each candidate are revealed after each poll or election.
Students gain firsthand experience with a standard location-based (spatial) voting model.
In the standard one-dimensional spatial voting model with two candidates, in equilibrium both candidates choose the policy most preferred by the median voter.
Most pairs will converge to each choosing policies near the Median Voter Position by the third poll or election. If you adjust the Median Voter Position between games, altering students to the possibility of a change in voter sentiment will facilitate convergence to the new equilibrium.
By default, in each Period, each pair first chooses policies in # of Polls=2 polls, followed by # of Elections=1 election. The only difference between the two is that while elections factor into student payoffs, polls do not.
In each poll or election, each voter (51 by default) truthfully chooses the candidate whose policy is closest to his own. Half of indifferent voters are allocated to each candidate, with the last odd-voter is assigned randomly. To avoid a tied election (in which the winner is chosen at random), choose an odd number for Number of Voters.
By default, the Median Voter Position is 50. One half of voters will be uniformly drawn from 0 to the Median Voter Position, and one half of voters will be uniformly drawn from the Median Voter Position to 100.
In a particular period, the exact voter distribution differs across pairs, but a particular pair faces the same voter distribution across all polls and elections in a single period. If you set Periods>1, pairs remain fixed across all periods of a game, but the exact distribution of voters differs across rounds. If you choose to replay a game, choosing Fixed keeps pairs together, while Random assigns new partners at random.
A player earns Winner Payoff=100 each time she wins an election. With Payoff Type=Winner Takes All, this is the entirety of player earnings.
With Payoff Type=Ideal Point, each candidate has a most-preferred policy (Ideal Point), and after each election each player receives a payoff equal to 50 minus the distance between her ideal point and the policy chosen by the winning candidate. (This is in addition to the payoff that the winner receives.) Because this payoff depends on the winning policy (as opposed to the policy chosen by the candidate), it only alters a candidate’s best response when there are multiple policies that result in a candidate winning the election. This does not alter the Nash equilibrium of each player choosing the median voter’s policy, but the extra consideration might slow down convergence.
Likewise, decreasing the Number of Voters decreases the informativeness of small policy changes and may slow down equilibrium convergence.
The results we present show, for each period, the extent to which play across all pairs converged to the equilibrium prediction of choosing the median voter’s policy. Use the Go To: dropdown menu to switch between periods.
For each period, we present a graph (see Figure 1) summarizing choices in each poll/election. Use the checkboxes in the legend to choose which of the following series are displayed:
Location of the Median Voter;
Most preferred policy for Candidate 1 and Candidate 2 (if Payoff Type=Ideal Point);
A marker for each Candidate 1 position and a marker for each Candidate 2 position; and
The average leftmost and rightmost policy chosen in each poll/election.