I am pleased to share a new co-authored article on the symbolic politics of the EU flag. It's open access, so please read and share freely. www.tandfonline.com/doi/full/10....

I plan to post research-related material this week for APSA, but first, a reminder that they are coming for all of us soon. www.nytimes.com/2025/09/10/u...

Vive les républiques. . . www.youtube.com/watch?v=HM-E...

I got distracted thinking about turnout in US presidential elections. Here is a graph showing the total presidential votes from 1824-2024 as a percentage of eligible voters and as a percentage of total population.

#no-tryants!

With this thread in mind it’s a good time to praise current IAS professor Alondra Nelson for doing right by this legacy: time.com/7285045/resi...

Never. There, I said it.

To be clear, Canadians writing about Canada think this is always true.

The final results are in, per CBC, for vote share (and seats). LPC 43.7 (169). CPC 41.3 (144). Bloc: 6.3 (22). NDP 6.3 (7). Green 1.3 (1). Using these vote shares, the seats are high end of the forecast ranges for the CPC (126-148), Bloc (12-25) and Green (0-2).

The votes are still being counted but it appears the LPC is headed for a minority government with 168 seats. Based on the preliminary popular vote, my model over-predicted their seat count, but the CPC (144) is within the predicted range (albeit on the very high end).

CBC just called it for the Liberals.

Peak Canada

RE: Trump's statement on the Canadian election, endorsing Pierre Poilievre without actually naming him, saying PP will help Canada become a state.

Once the results are in (probably tomorrow), I’ll run the model with the actual vote share to see how the model did this time around. My guesstimate is the LPC with about 185 seats with the NPD ending up with around 7.

My forecasts based on the final CBC polling averages. The LPC are on track for a majority government with 190-191 seats with the CPC in opposition (132-133 seats). NDP (1-2), Bloc (17-18), and Greens (0-1) are far behind, but the NDP will probably do better at the expense of the LPC.

Urbanistas out there will enjoy this book on 'Four lost cities' by @annaleen.bsky.social ...a fun read with lots of fascinating detail about Çatalhöyük (in present-day Turkey), Pompeii (in present-day Italy), Angkor (in present-day Cambodia), and Cahokia (in present-day USA)

The model forecasts a 100% chance of a Liberal majority government, but a successful NDP defence of its stronghold ridings could conceivably push the LPC into minority territory. The hints of a slight national swing to the NDP would be less relevant than focused attention on key districts. 2/2

Polls still don’t show any major national swings. The CBC average has the LPC (42.5%) leading the CPC (38.7%) by about 4 percentage points, and 8 of the 10 most recent polls has them leading by at least 3 points. Today’s average forecasts 190-191 seats for the LPC and 130-131 for the CPC. 1/2

Although the absolute support for the LPC is lower in these two polls, it gives them the highest average seat count: 194.5, and a substantially worse result for the Conservatives (124 seats). These polls are also marginally better for the NDP (3.5 seats) and worse for the Bloc (18 seats). 2/2

No student has shown up to my office hours, so I ran one more set of simulations, with the Nanos Research / CTV-Globe and Mail. This poll – along with one from Liaison Strategies / NEPMCC – has the largest difference between the LPC and CPC (43% to 37%). 1/2 newsinteractives.cbc.ca/elections/po...

The 40-40 tie from the Mainstreet poll still forecasts a Liberal victory, but with an average of only 181 seats and the CPC in opposition with 140-141 seats. Some simulations show a Liberal minority, but less than 1% of the time. As in 2021, the FPTP system helps the Liberals substantially. 4/4

The Angus Reid results have the Liberals averaging 190 seats and the Conservatives 129, with no simulation producing a minority government. The forecast has the Bloc with a solid 21-22 seats share and the now-familiar dire results for the NDP ( 1 seat or less).3/4

I ran the simulations with polls at the extremes. Angus Reid has the LPC at 44% and the CPC at 39% while Mainstreet Research has both at 40%. Caveats: the former uses a much larger sample size and the CBC ranks it as more reliable. 2/4

The first set of post-debate polls are out and on average they don’t show much change. There is perhaps a 0.5% move in favour of the Conservatives relative to the Liberal Party, but not enough to change the forecast of a Liberal majority. 1/4

Advanced voting is in progress.

There were no dramatic moments in the debates – other than the Canadiens qualifying for the playoffs (Go Habs Go!) – so it seems unlikely that we will see much movement in response to them. At this point, it is the Liberal’s election to lose. 3/3

The model continues to forecast a comfortable majority for the Liberals, averaging 198 seats, with a 20% chance of cracking the 200-seat ceiling, and no chance of falling below a majority. 2/3

The polls have narrowed *slightly* in the past week, but basically they are all within the margins of error of prior ones. Nonetheless, it is a good time to lay down a marker before the effects (if any) of the debates show up. 1/3

The current polls + the 2025 boundaries give the LPC a favourable seat-vote ratio (1.33); the CPC has a slightly negative one (0.98). It’s much worse for smaller parties. Even with the adjustment, the FPTP district system favours the Liberals; redistricting was not a Conservative gerrymander. 6/6

That is not to say that the Commissions created partisan gerrymanders in favour of the CPC. The seat-vote ratio in 2021 created a massive advantage for the LPC – they got many more seats with fewer votes than the CPC. 5/6

Under the current boundaries, the LPC averages 200 and the CPC 125. In short, the 2022-2023 adjustment favours the Conservatives, but not enough to give them a majority. The old boundaries are slightly better for the other parties as well, but the effects are not large. 4/6

I ran the model using the 2021 boundaries (338 constituencies) with today’s CBC polling averages (10 April) to see what the likely outcome would be under the old system. The LPC averages 207 seats while the CPC averages 111. 3/6

Boundary Commissions adjusted the borders of constituencies in every province in response to these seat additions and to population change (as measured by the 2021 Census). Although the Commissions are strictly non-partisan, these changes could have partisan effects. 2/6

The polls in Canada are not moving much so it’s a good time to talk about the effects of the 2022-2023 redistribution that modified Federal constituency boundaries. Parliament added five seats to the Commons, raising it from 338 to 343. Alberta got 3 more, while Ontario and BC got one each. 1/6

There has been very little movement in the polling averages since 2 April, which is good news for the Liberals because they continue to hold a commanding lead in seats. The average count across the simulations is 201 seats, with CPC at 125.

Based on the Hockey Night commercials, both the LPC and CPC are leaning heavily into "patriotism" for their campaigns. The only advert that really made it work, however, was Molson.

In contrast, the model shows a more typical majority for the LPC when using the Mainstreet numbers, with an average of 187 seats, a high of less than 200, and one simulation (out of 10,000) even showing them one seat below a majority. 4/4

Anyway. . . The Nanos poll show the LPC with a commanding seat lead. The average is 210, and there is a small chance (<1%) of more than 220 seats. Similarly, there is <1% chance of fewer than 200 seats, which would be an unprecedented majority for them.

Both polls have a margin of error of 2.5% so their estimates actually overlap. But it's a good illustration of how the model responds to polling figures, and a good time to mention the "garbage-in, garbage-out" caveat. No model will work with inaccurate polling data. 2/

The CBC polling average has barely changed in the last few days, so I ran the model using two individual polls, both taken on 1-3 April. The Nanos poll is better for the LPC (46% versus CPC's 36%), while the Mainstreet poll is better for the CPC (40% versus the LPC's 43%). 1/

Given that the NDP can also read polls, I expect that they will focus resources on their strongest ridings. My model doesn't make adjustments for party activity, incumbency, or district-level socio-demographics so I suspect that they will do much better than the 1-2 seats the model is predicting.

The LPC is up about 1.5% and the NDP down about 0.7% in CBC polling average compared to 31 March, but this is enough to move the forecast. The Liberals are now averaging 198 seats and the model shows a 15% chance of zero NDP seats.

A stunning new poll from Poisson d'Avril shows the election now as a two-party race. The Greens look to dominate the next government.

There are not a lot of changes from the 29 March results. The Liberals and NDP are up slightly, and the Bloc is down a bit in polls. The Liberals are maintaining their (forecasted) majority (average of 194 seats) and things are *slightly* less dire for the NDP (but still only averaging 2.5 seats).

This graph provides useful background for election projections. It shows the relationship between the national vote and seats won from 1945 to 2021. The bias usually works in favour of parties getting more than about 32% of the national vote. The 2021 bias favouring the Liberals was unusually large.

My forecast based on the CBC polling average for 29 March. The Liberals have extended their lead in the polls, which translates into a projected majority government in all simulations. The model has the LPC averaging 189 seats. Things look worse for the NDP, with zero seats in 5% of the simulations.

My updated forecast using the 26 March CBC polling averages has the Liberal Party winning a majority in almost every simulation, and an average of 184 seats across the 10,000 simulations. There are some dire outcomes for the NDP, with some simulations resulting in zero seats.