Voting system affects voter behavior and outcome, simulation finds

This is a CBC News (Canadian Broadcasting Co.) story datelined 17 October 2011 in Toronto. No author-name was stated. I (Warren D. Smith) have simply copied that story to this web page, except that I have edited the story somewhat to

1. Change the terminology to what is more common in the USA,
2. Correct some errors and misleading wording,
3. Alter formatting and add some hyperlinks,
4. Add a few more numbers I computed.

The main lesson of this study in my opinion is: it clarifies how large "favorite betrayal" is in different voting systems, and makes it clear that Instant Runoff voters are using strategy. Specifically (reading from the pink row of the first table), if the PR voters are regarded as "honest" while the FPTP voters are regarded as partially "strategically lying" about their top-rank preferences, then in this election the IRV voters appear to be strategically lying approximately half as often about their top-preference votes, as the FPTP voters.

A simulated election that examined different voting methods during last week's Ontario election shows that for at least some voters, how they cast their ballot depends on the voting system presented to them.

Entitled Three Ontario Votes, the project allowed voters to cast fictional ballots online ahead of the Oct. 6 Ontario election using three different electoral systems. More than 9,000 people visited the site since its launch on Sept. 17.

The intention of the project was to inform voters about the differences of the three systems studied: first-past-the-post (FPTP) also called plurality voting, instant runoff (IRV), also called "the alternative vote," and proportional representation (PR).

See table III below for an explanation of how each of the three different voting systems work. The FPTP system is the one currently in use in all provincial and federal elections in Canada.

Using the online tool, visitors to the site were asked to vote three times, using each of the three voting systems. They cast their anonymous votes using a list of real candidates in their riding.

The researchers and academics behind Three Ontario Votes published the results of data collected during the study and also calculated how the actual results from last week's election would be different in each of the voting systems studied.

First, the study examined how site visitors cast their ballots in each of the three systems.

Table I: Votes cast under three different electoral systems. (Pink row arises as the unweighted average of the 5 preceding rows, each renormalized so their final entry is 100%.)

Those who conducted the study concede the results in the graph above are not representative of the actual election results.

Most participants kept their choices consistent regardless of the voting system they were using. However, many switched their votes when using different systems. For example, 12% of voters ranked a different party as their top choice in the IRV system than the one they voted for in the FPTP election. A total of 18% supported a different party under the PR system than they did when voting under an FPTP system.

Laura Stephenson, a professor at the University of Western Ontario and one of the researchers behind Three Ontario Votes, said the results show that voters are aware of how their votes would be translated in each of the different systems.

"It shows that incentives exist in each voting system and that people are aware of them and respond to them," she said.

Translating actual results from the Oct. 6 election shows how the IRV system would shift a few percentage points of support from the Liberal and Progressive Conservative parties to the NDP and Green Parties (see table below, but note that the NDP and Green parties would not actually win any more seats under IRV, so this "extra support" was somewhat illusory). A PR system would also benefit the NDP, Green Party and fringe parties at the expense of the Liberal and PC parties, but the results are more dramatic.

Table II: Actual FPTP Election Outcome and Weighted IRV and PR Vote Results

One aspect of the IRV system that sets it apart from other methods of voting is the option for voters to rank parties second or third after their first choice. A total of 67 per cent of those who ranked the Liberals as their first option ranked the NDP second. A total of 56 per cent of those who ranked the NDP first ranked the Liberals second.

Seat breakdowns vary widely from actual vote under PR

Perhaps most interesting is how a PR system would translate into seats. In last week's provincial election under FPTP system, the Liberals came just one seat shy of a majority, winning 53 seats while the PCs finished with 37 seats, the NDP 17 and the Green Party with none.

Under a PR system, the Liberals would win only 36 seats while both the PCs and NDP would wind up with 31 seats each. The Green Party would fare much better if last week's election used the PR system. Instead of no seats, the Greens would come away with eight.

Table III: Actual FPTP seat distribution and estimated PR seat distribution

Supporters of the PR system often argue that it results in governments that more accurately reflect voters' intentions. A common argument against PR however is that it too often leads to minority governments.

Return to main page