Damien Bol, King’s College London
André Blais, University of Montreal
Simon Labbé St-Vincent, University of Montreal
What is the story?
Lab experiments are increasingly popular to study elections. In a recent book published at Springer, we present the variety of voting experiments, in the lab and on the field, in showing their respective contributions to research in the domain.
In lab experiments, researchers have the possibility to manipulate the factors that may influence the outcome of an election, such as the electoral system, the distribution of preferences or the party platforms. In doing so, they can to isolate the causes of this outcome and the mechanism behind it.
In a recent paper published in Political Science Research and Methods, we report the results of a one-of-a-kind experiment where we put parties and voters into the lab. We hope this will pave the way for new voting experiments that consider how interactions between multiple actors shape the electoral outcome.
A one-of-a-kind experiment
For each experimental session, we organised 4 series of 5 elections between 17 subjects at experimental lab CIRANO in Montreal. We randomly assigned the 17 subjects to a role: 6 were designated as parties and 11 were designated as voters, and to a position on a scale ranging from 0 to 10 (see Table 1 below, parties are represented in letters, voters are represented Roman numbers). The roles and positions were randomly reshuffled after each series of 5 elections.
Table 1: Positions of voters and parties on the 11-point scale
| Scale | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| Parties | A | B | C | D | E | F | Voters | I | II | III | IV | V | VI | VII | VIII | IX | X | XI |
The elections had two stages. First, the parties decided to form an alliance or not with their paired party (the pairs are A/B, C/D and D/E). Then, the voters saw on their screen which parties are participating and had to decide for which party to vote among them.
After the election, the party that received the most votes was declared the winner. If this winning party was not in an alliance with her paired-party, she got 40 points. If there was an alliance the 40 points were divided between the two partners. The distribution of gains between them constitutes an experimental condition. Depending on the series, it is either equal (each party receives 20 points) or unequal (the party that enters the election receives 30 points, while the other receives only 10 points).
Findings
One of the advantages of our experiment compared to observational studies is that we know how many parties could have potentially participated in the elections, that is 6 parties. In reality, researchers can hardly determine how many parties envisioned participating in an election before the campaign. We can thus calculate how much the Effective Number of Parties (ENEP) actually observed in our experimental elections deviates from this theoretical maximum, and the part of reduction that is due to parties forming alliances and to voters voting strategically, that is deserting the party closest to their position if this party has no chance of winning.
Table 2 reports these results. It shows that on average voters reduce 38% of the fragmentation they could theoretically by voting strategically, and parties 60% by forming alliances. Also, we observe that the degree of reduction by parties is larger when the distribution of gains within alliances is equal.
Table 2: Degree of ENEP reduction by voters and parties
| Gains Equal | Gains Unequal | Total | |
| Reduction by voters | 36% | 40% | 38% |
| Reduction by parties | 64% | 57% | 60% |
| ENEP | 3.10 | 3.18 | 3.14 |
Our conclusion is that the contribution made by party strategic exit is greater than that due to strategic voting. We explain this difference by the nature of the coordination problem and the amount of gains at stake (party coordination is much easier to achieve than voter coordination). Also, in our experiment just as in real elections, parties have much more to gain if they win, and have thus more incentives to behave strategically than voters.
For more information, see:
Damien Bol, André Blais, and Simon Labbé St-Vincent (Forthcoming.) Which Matters Most: Party Strategic Exit or Voter Strategic Voting? A Laboratory Experiment. Political Science Research and Methods
