This is a non-partisan projection of the outcome of the Green Party of Canada’s 2020 leadership race. The projection is based on a statistical model, developed over the past year, that was designed specifically to project ranked-ballot leadership races in Canada. Using this model, 100,000 simulations were run and the results analyzed.
Projection
Annamie Paul wins the leadership in the 7th round.
She defeats Dimitri Lascaris 64.5% to 35.5%.
Expected Round-by-Round Results
The table below shows the expected standings in each round, along with eliminations and vote transfers. Values are rounded to the nearest 0.1%.
| HADD | HOWA | KUTT | LASC | MERN | MURR | PAUL | WEST | |
|---|---|---|---|---|---|---|---|---|
| Rnd 1 | 7.5% | 13.9% | 8.1% | 23.5% | 6.6% | 8.5% | 29.1% | 2.9% |
| +0.2% | +0.4% | +0.3% | +0.1% | +0.5% | +1.2% | +0.3% | elim. | |
| Rnd 2 | 7.7% | 14.3% | 8.4% | 23.6% | 7.1% | 9.7% | 29.4% | |
| +0.5% | +1.7% | +1.6% | +1.1% | elim. | +0.7% | +1.5% | ||
| Rnd 3 | 8.2% | 16.0% | 10.0% | 24.7% | 10.3% | 30.8% | ||
| elim. | +1.1% | +3.5% | +1.7% | +0.4% | +1.4% | |||
| Rnd 4 | 17.2% | 13.5% | 26.5% | 10.7% | 32.3% | |||
| +3.3% | +2.2% | +0.2% | elim. | +5.1% | ||||
| Rnd 5 | 20.4% | 15.7% | 26.7% | 37.3% | ||||
| +5.5% | elim. | +3.3% | +6.8% | |||||
| Rnd 6 | 26.0% | 29.9% | 44.2% | |||||
| elim. | +5.6% | +20.4% | ||||||
| Rnd 7 | 35.5% | 64.5% | ||||||
| defeat | win |
The same data, condensed into a straightforward ranking, looks like this:
- Annamie Paul
- Dimitri Lascaris
- Courtney Howard
- Amita Kuttner
- Glen Murray
- Meryam Haddad
- David Merner
- Andrew West
Accuracy and Factors to Consider
The model used for this projection showed very good accuracy when applied to data from the 2017 NDP and 2020 Conservative leadership races. (For details on the model, see below.)
The Green Party leadership race is a more challenging one to forecast for several reasons, including the pandemic, much less data available compared to larger parties, and several irregularities that occurred during the campaign.
A question that many supporters may have is, assuming this projection is accurate, is there still time for these standings to change? Can a person’s favoured candidate, in this final week, improve upon their standing?
The most important factor remaining at this time is the effectiveness of each campaign’s “get-out-the-vote” operation. If all campaigns do an equally good job of realizing their potential through organized GOTV, the final outcome is likely to be similar to this projection. If any campaign has a particular strong or weak GOTV, standings could shift as a result.
Candidate-by-Candidate Analysis
Annamie Paul
Annamie is expected to place first in initial voting. Maintaining her lead through the early rounds, Annamie is projected to pick up substantially more transferred votes than other candidates in the later rounds. In particular, Annamie should receive the majority of Courtney Howard’s votes after Courtney is eliminated in Round 6. This is projected to propel Annamie to a 65-35 win over Dimitri Lascaris in the seventh and final round.
BEST/WORST OUTCOME: Win in the seventh round.
Dimitri Lascaris
Dimitri is projected to have a strong first-round result, placing almost 10% ahead of Courtney Howard and just under 6% behind Annamie Paul. However, Dimitri is likely to be hamstrung by weak vote transfers from eliminated candidates, especially in later rounds. Any hope for a come-from-behind victory over Annamie would require strong transfers, and as a result, is not projected to happen.
BEST OUTCOME: Defeated in the seventh round, approximately 60-40.
WORST OUTCOME: If Courtney Howard’s performance is stronger than projected and Dimitri’s is weaker, there is a slim chance of Dimitri finishing third. (3% of simulations.)
Courtney Howard
Courtney is expected to receive the third-most votes in the first round, but will be almost 10% behind Dimitri. She is projected to receive strong transfers in most rounds, allowing her to maintain a solid hold on third place, but the gap between her and Dimitri is almost certainly too large to be made up. In 96% of simulations, Courtney is eliminated in Round 6.
BEST OUTCOME: Courtney has a very slim chance (3% of simulations) to take second place rather than Dimitri.
WORST OUTCOME: In a very small number of simulations (1%), Courtney is eliminated in the fifth round due to an unexpectedly strong performane by Amita Kuttner.
Amita Kuttner
Of all candidates, Amita has the widest set of possible outcomes. Their most likely result (seen in 60.9% of simulations) is elimination in Round 5, a modest upset over Glen Murray, who is expected to start with slightly more votes in the first round. Amita has the distinction that, out of all candidates, they are projected to draw the most second-choice votes relative to their overall ranking. In the majority of projections, this allows them to pass Glen by Round 4.
BEST OUTCOME: In a very small number of simulations (1%), Amita reaches the sixth round instead of Courtney, placing third overall behind Dimitri.
WORST OUTCOME: Amita could be eliminated as early as the second round (15% of simulations) or third round (14% of simulations).
Glen Murray
Glen is one of three candidates projected to be eliminated between the second to fourth rounds. Despite a reasonable 8.5% projected first-round result, Glen is expected to have mediocre vote transfers in subsequent rounds, limiting his ability to continue beyond the fourth round.
BEST OUTCOME: Elimination in the fifth round (14% of simulations).
WORST OUTCOME: Elimination in the second round (3% of simulations).
Meryam Haddad
Propelled by a late surge of support, then surviving an eleventh-hour expulsion and reinstatement, Meryam is the candidate most likely to defy accurate projection. Her expected first-round result will put her in the middle of a pack with Glen, David Merner, and Amita. Her expected transfers from eliminated candidates should be strong, but her projected result is the most sensitive to the order of elimination of other candidates.
BEST OUTCOME: Elimination in the fifth round (14% of simulations).
WORST OUTCOME: Elimination in the second round (35% of simulations).
David Merner
David is second only to Amita in his projected ability to draw votes from eliminated candidates. However, David’s first-round result is expected to be weaker than Meryam, Amita, and Glen, making him the most likely to be eliminated early. That said, David still reaches the fifth round in almost 10% of simulations.
BEST OUTCOME: Elimination in the fifth round (10% of simulations).
WORST OUTCOME: Elimination in the second round (46% of simulations).
Andrew West
Andrew is projected to be eliminated first, as his expected first-round result of 2.7% is well behind the other candidates. To advance beyond the first round would require an unexpectedly strong performance by Andrew, coupled with a very weak performance by another candidate. This situation did not materialize in any of the simulated runs of the model.
BEST/WORST OUTCOME: Elimination in the first round.
About the Projection Model
The model that was used to create this projection is based on objective data gathered from social media, traditional media, and other sources. The model does not incorporate polling or any sort of subjective “judgment call”. The gathered data is used to project, for each candidates, their first-round vote totals and their ability to attract transferred votes from eliminated candidates.
These two inputs are then used to run many simulations, each one with slight random variations. (This is known as the Monte Carlo method.) The results of these simulations are analyzed to determine each candidate’s spread of possible outcomes and the factors most likely to influence their result. For example, running numerous simulations can reveal whether a particular candidate has a wide or narrow spread of possible finishes, or how strongly their finish depends on the performance of others.
When tested on two past leadership races (NDP 2017 and Conservative 2020), the model correctly projected the final outcome, the number of rounds, and the ranking of candidates in each round for both races. Often (though not always), the vote percentages were accurate within +/- 2 percentage points. As discussed earlier, it remains to be seen what level of accuracy the model can achieve in the unique circumstances of the Green Party leadership race.
Disclosure
I began developing the model used in this projection in late 2019. From February to July of 2020, I was part of a campaign team taking part in the leadership race. This projection is intended to be non-partisan, and was created and published as a way to test the model under real-world circumstances.
Saul Bottcher 