Ontario 2018 election

[Filuwaúrdjan]

EPP, using their "gut feeling" methodology has called all but two seats: Sault Ste Marie and Ottawa South. 

Here's their Liberal holds (9 seats):

Ottawa-Vanier
Ottawa Centre
Don Valley East
Don Valley West
Eglinton-Lawrence
Scarborough-Guildwood
St. Paul's
Toronto Centre
Vaughan-Woodbridge

Mostly EPP has a bias towards 'conservative' outcomes and the re-election of incumbents if there's any doubt. Sometimes this works out well (let's just say that a 91% success rate is rather better than you'd have got from almost any British political pundit last June), sometimes - as in the past two Canadian federal elections: getting roughly one in four seats wrong is a bit *ouch* - it's a bit of a fiasco. But always fascinating and I'm glad it's still going in this age of dreary DATA driven projections (which mostly are no better).

[Filuwaúrdjan]

BWV 208

[Filuwaúrdjan]

There's even a recent and highly relevant example from Ontario as to why you shouldn't automatically write off a party with a tradition behind it in a Canadian context: the NDP were reduced to an irrelevant rump - a sad ghost of a party - at the 1999 election, and now look.

[Filuwaúrdjan]

I've come to think of said "dreary DATA driven projections" as a little like the psephological equivalent of talking GPS: it may "get you there" (seemingly), but it's no match for well-tuned navigational skills--or what London taxi operators call "The Knowledge".  It's an easy crutch; that's it.

Yes. And the thing is, when an election is fluid as this (especially given the possibility of all sorts of freakish results on low poll shares...) if the data-driven projections are right it won't be because of their inherent accuracy, but luck...