Wednesday, October 14, 2015

#Elxn42: Strategic Voting....Oooooooooooh Scary!


At least that's the story according to the-above-the-fold front page headline in today's Vancouver Sun.

To wit:

Analysis: Strategic voting can be confusing and can easily backfire.


As you might expect, the actual story by Peter O'Neil is not nearly so 'scary'.

But still there is this bit, wherein Mr. O'Neil focuses on one of the most muddled ridings in all of British Columbia:

...The new riding of Vancouver Granville, for example, might turn out to be the subject of an academic thesis on the risks of strategic voting.

An organization called Leadnow had two polls done by the firm Environics, one in August and one in September, suggesting that the NDP’s Mira Oreck had a modest lead and was in a better position than Liberal Jody Wilson-Raybould to defeat the Conservative candidate.

For months Leadnow has done door-to-door canvassing, while providing voters with that data.

But things became murky on Tuesday when Leadnow issued a new Environics poll for that riding showing the Liberals at 35 per cent, the NDP at 33 per cent, the Conservatives under candidate Erinn Broshko at 28 per cent and the Greens’ Michael Barkusky at four per cent. That poll of 505 respondents, done Oct. 9 through 11 and using automated voice technology, had an error margin of 4.4 percentage points, putting the Liberals and NDP at a statistical tie.

But another poll, sponsored by Friends of Canadian broadcasting and conducted by Mainstreet Research on Oct. 8, had the Liberals way ahead there. It used similar technology but with a larger sample size (665) and resulting lower error margin (3.7 percentage points). That poll put 44 per cent of decided and leaning voters with the Liberals, 28 per cent with the NDP, 20 per cent with the Conservatives, and nine per cent with the Greens...

Mr. O'Neil then goes on to confuse the issue further by treating aggreators and riding-specific polls equally. The problem with that is that even the best of the aggegators, Eric Grenier, makes it clear that the aggregators become, essentially, only marginally better than chance when things are really close (scroll down).

And, as we noted yesterday, Vancouver Granville is one of the true exceptions in that it really is just too close to call for either the Dipper or the Liberal at this point. Regardless, it is still very, very clear that a vote for the Green candidate in that riding is a vote for the Stealth-Con.

Of course,  the thing that Mr. O'Neil does not delve into in any depth is the fact that there are numerous  other ridings in British Columbia where the Strategic Voting choice is already pretty much crystal clear, as we and others have also pointed out recently.


What is this really all about Alfie?

I mean, honestly, are the fine folks running the VSun actually suggesting that the seventy percent of us who want our country back should just roll over and play dead?


You know...

Paying attention is just too damned hard or something.


Interestingly, in his kicker, Mr. O'Neil quotes political scientist Richard Johnston as saying that no one should trust polling info that is more than a few days old when making their ultimate decision...I agree with that, especially in ridings where things are close...Perhaps more interestingly Mr. O'Neil failed to mention how a riding specific-poll in very close race helped save Esquimalt from the Cons in 2011...Imagine that!



Anonymous said...

With strategic voting, one should look at the big picture as well as the riding. For example, using totally fictitious numbers, suppose the ridings break down as follows:

Cons - 100
Libs - 100
NDP - 98

Now suppose in the last riding, the NDP are slightly ahead of the Libs, with the Cons more or less out of it. Voting strategically at the riding level you might vote NDP, but this would put the Cons and Libs tied for the lead. However, voting strategically while looking at the big picture, you would vote Lib to put the Libs out front.

So by voting strategically at the riding level, you may split the ridings and let the Cons go up the middle.

RossK said...


Interesting point.

But how are we to know this info before the polls close when we are actually voting?

(I can really see these types of scenarios becoming important if we ever get to proportional rep where your vote will count, at least partially, no matter what)


Anonymous said...


You don't know. But, for example, the Liberals are leading now so if you have a choice of voting strategically in your riding between NDP and Liberal, vote Liberal. This will add to the Liberal seat count instead of the NDP seat count and prevent the Cons from coming up the middle. If the NDP should happen to come back, then you would vote NDP.

However, I am feeling more and more comfortable that the Cons won't get back in. I really would have liked to see the NDP have a shot at running the country, though. I know that Mulcair has taken some criticism, but I think he's a very smart guy and would have made a good PM.

Anonymous said...

Just to be clear (did I really just say that?) regarding strategic voting at the big picture level. If at the riding level, there is any chance that the Cons might win, forget about the big picture and vote for whoever has the best change of defeating the Cons in the riding. That is the number one priority. Only if the Cons are out of it at the riding level should you start looking at the big picture.

RossK said...


I really, really think that is a very dangerous call to make given the overall volatility (and the fact that the database- and unfair elections act- assisted shenanigans on election day could net the Cons a point or three in critical ridings that they already know they need to target).

Regardless, it's important not to get locked into thinking that all ridings are like the example used by Mr. O'Neil where vote changing could swing things between the Liberals and the Dippers (In addition to Vancouver Granville it looks like Delta could still be like that also).

Point being that if a riding is too close to call you should vote your first choice.

However, when the one Non-Con that can beat the Con is clear, then I believe you should still vote strategically.

And, finally, if the CPC really is swooning I would like to see them driven below a potential balance of power position (such that they could NOT be resurrected by a minority Liberal group on down the line if, say, Mr. Harper were to first resign).


Anonymous said...

RossK - Well, if you are going to vote strategically, you need to use whatever data is at hand to make your decision as to how to vote. Most likely, that data will be polls done at the riding level and the national level. Therefore, it is best to leave your decision making until very close to Oct 19 so you have the most recent data. Even still, if that data are bad and you base your decision on that data, then you will probably make a mistake with your strategic vote.

I'm not advocating that everybody vote strategically. That is a personal choice. I'm just saying that if you are voting strategically, you should consider your strategic vote at both the riding level and, if relevant, at the national level.