Does "Vote Splitting" Have a Large Impact on Canadian Elections?

TL;DR: I made this visualization about vote splitting in the 2011 Canadian election.

If you're a Canadian and haven't been hiding under a rock, you're probably aware there's an election coming up. When we're not hearing about candidates peeing in mugs or how Bill C-51 reflects the teachings of Jesus, we're knee-deep in speculation over who's going to be in power later this fall.

Unless you support the Conservative party (full disclosure: I do not), there's a good chance you've lamented the "vote splitting" of the left, and the tactical/strategic voting that seems necessary to prevent Conservative (CPC) rule.

The spectre of vote splitting is invoked countless times in the run-up to each election, but we rarely get a chance to see how it actually impacts the results. Meanwhile, the poll numbers we see in the press are almost always the nation-wide popular vote, which give no insight into what's happening at the riding level. Since our first-past-the-post voting system means our elections are really hundreds of separate contests, it's complicated and counterintuitive to see how changes in popular vote translate to movement in seats.

I took an interest in this topic and built a data visualization that allows you to modify the popular vote in the 2011 election and see how it would have affected the results. You can also zoom in and click on ridings to see how each of them would have played out with your changes, or look at the absolute best case of strategic voting in 2011.

Some pointers on using the visualization:

Warnings and Caveats

Before drawing any conclusions, it's important to stress that its data is limited to the 2011 election. This year's election has 30 new ridings to consider, many existing boundary lines redrawn, and polls that look very different than they did four years ago. In summary, you really can't extrapolate too confidently from the 2011 results to today. However, many of the trends and party strongholds remain the same in 2015, and gaining a better understanding of past results is still useful in revealing the dynamics of our democracy.

The Greens

The Green party is the party most frequently accused of siphoning off votes that could be used to defeat the Conservatives. With the visualization we can see that even if all 2011 Green voters voted for the second-place NDP, or voted perfectly strategically, we would have still ended up with a Conservative government, and still a majority in the first case. This lends some credence to Elizabeth May's rebuttal that the Greens did not split the vote appreciably. But could it be different this year with the top three parties closer in the polls?

The Liberals and NDP

If you played around with some of the Green vote scenarios above, you may have noticed that transfering their votes to the Liberals instead would have cost the CPC an extra five seats. This highlights a pattern: even though the NDP finished second in seats in 2011, the Liberals were actually in more close contests with the Conservatives and may stand to steal more seats from them. This may partially explain why most of the CPC's attack ads to date have focused on Trudeau and not Mulcair, despite the NDP's strength in the polls for much of the campaign.

There were actually only a handful of ridings where the Conservatives and NDP were in close competition with each other in 2011. That's why if you are looking to vote strategically, choosing the top non-CPC party in the popular vote numbers may be counterproductive; you're much better off looking at riding-by-riding polls and supporting whoever has the best chance of knocking off the Conservatives in your district.

The Conservatives

One thing that's clear from running through different scenarios is that if the Conservatives lost support to the Liberals as current polls indicate may be happening, they'd be in trouble in many swing ridings across the country.

At the same time, their strongholds in the prairies appear to be very safeĀ­ in some ridings they collected upwards of 80% of the vote in 2011. There has been speculation that the NDP's victory in the Alberta provincial election may presage a shift in the federal election, but it seems unlikely and the current polls don't support that theory either.

Takeaways for Progressive Voters

From trying out various scenarios, it seems there are three major classes of ridings for the would-be strategic voter to consider:

If you were to go back in time to re-cast your ballot in 2011, it seems clear that the majority of ridings are of type A or B. If you lived in one of them, you wouldn't need to worry about vote splitting at all, and should vote for the party that best aligns with your values. However, in the ridings that are shaping up to be battlegrounds, you may want to vote strategically, with the help of VoteTogether and the current polls in your riding.

A few more things that deserve attention for October 2015:

On a final note, the source code for the visualization is available. Please file an issue on Github if you have any suggestions for improvement.