Liberals keep the lead, but province wide numbers are very close

Coup d'oeil sur cet article

Even though the Conservatives have had better polls recently (see Forum and Ipsos) after a small decline, the regional numbers just do not bold well for this party. They seem to be up in Toronto (416), but since they are stable/down province-wide, it means they are down somewhere elese. According to the polls, they are down in the North/Central Ontario and in the South West. Even though my model doesn't use regional numbers directly, I did modify my proejctions slightly to take that into account. Being up in the 416 is nice but won't translate into many seats, while being down in the North nd the SW could be costly.


And the detailed projections are here.

It's really surprising to see how close a majority the Liberals of Kathleen Wynne currently are. Especially since they didn't get one last time and are actually below their 2011 results. But the first thing to notice is that pretty much everyone is below its 2011 numbers. The Tories are down and the NDP is stable/down. So currently, only the Green are projected to experience a positive swing (and starting now, I'm adjusting the poll numbers for the Green as I believe polls overestimate small parties. It's not nothing major, but it explains why I have them at "only" 6%). So it might not matter that you are decreasing if everybody else also is. On top of that, as mentioned above, the regional swings just don't look optimal for the Tories (in other words: they are up when it won't matter much for them).

The regional adjustments move the race from a 8-9 seats lead for the OLP to what you have presented here. I'll keep monitoring this closely. I usually don't like to adjust my numbers based on regional poll numbers, even average, becuse these numbers are based on small samples and it's not always clear which region is which (or which riding is in which region according to the pollsters). Still, my numbers were just too far off in Toronto and the North.

The probabilites of a Liberal majority are 25%. It's not super high but it definitely shows that this is a possibility. Of course, this 25% includes the uncertainty due to the polls. If you assume the poll average is spot on and exactly what each party will get on election night, there is a lot less unknowns as it only leaves the effects of the electoral system. Below, I have 6 graph, two for each party. They show the possible seat outcomes with the double uncertainty (polls+electoral system) or just  the electoral system.

Possible seat outcomes for the Liberals, double uncertainty:

Lib, double


Possible seat outcomes for the Conservatives, double uncertainty:

PC double

Possible seat outcomes for the NDP, double uncertainty:

NDP double

And now the distributions if we only account for the uncertainty due to the electoral system (or, if you prefer, the vote efficiency)

Possible seat outcomes for the Liberals, uncertainty of the electoral system only:

Lib, single

Possible seat outcomes for the Conservatives, uncertainty of the electoral system only:

PC, single

Possible seat outcomes for the NDP, uncertainty of the electoral system only:

NDP, single

By the way, if we assume that the poll average is right, the chances of winning are 99% in favor of the Liberals and the chances of a Liberal majority are only 2%. In other words, if you believe the polls (or at least the average) are right, then a Liberal minority is the most likely outcome by far.

In conclusion, there remains enough uncertainty in this race to call it that, a race. However, make no mistake, it isn't a 50-50 contest right now, the Liberals are heavily favourite to win the most seats on June 12th. On top of the advantage the Grits have with the electoral system, there are also some fundamentals that lead us to believe the Liberals are closer to a majority than to losing the election. One of them is the approval ratings of the various leaders. Tim Hudak just isn't that popular. Even among PC supporters, he's scoring lower than Wynne among Liberals or Horwarth among the NDP voters. And Hudak as abysmal numbers among the Liberals and NDP. It probably means that while the Tries can count on a motivated base (that will turn out to vote), they can't expect many swing voters to change their mind in the last minute. Thus, an underestimation in the poll isn't very likely. Really, the only thing that could indicate otherwise is the fact that likely voters )or very comitted ones) lean PC. This is quite obvious in the Ipsos numbers (but they are based on very few observations and I'm not 100% convinced of their definition of likely voetrs), a little bit less with the adjustments from Abacus (that actually helped the Liberals in their most recent polls). It'll be interesting to see the numbers from Ekos as soon as this firm publishes its adjustments for likely voters.

2 commentaire(s)

Thierry Soucie dit :
30 mai 2014 à 13 h 51 min

J'obtiens, selon les chiffres ci-haut:

46 OLP 41 PC 20 NDP

T Orwell dit :
3 juin 2014 à 0 h 19 min

Bonjour, Merci pour l'article sur l'Ontario (et en anglais!). C'est beaucoup mieux que d'autres fusions de sondages disponibles.

Deux petites corrections: dans l'avant-dernier paragraphe: "chances of a Liberal majority are only 2%. In other words, if you believe the polls (or at least the average) is right, then a Liberal majority is the most likely outcome by far." Je pense que le duexieme "majority" devrait être "minority". Aussi the chef du NDP est "Horwath" sans "c". "Horwatch" sonne comme "regarde pute" en anglais (un petit mais malencontreuse erreur).

Merci encore pour le travail impressionnant!