Note: if you want to read the data yourself, the main report is here while the detailed riding tables can be found here.
This image was being shared on my social media feeds yesterday, and I imagine similar ones were being shared in so-called swing ridings across the country. It was created by Vote Together, a project created with the explicit goal of defeating Stephen Harper’s bid to once again become Prime Minister.
The basic idea is to identify places where the number of non-Conservative voters is high enough that they could, hypothetically, elect a non-Conservative candidate, but the vote is “split” among people voting for the Liberals, NDP, and Greens, allowing Conservatives to take the riding without receiving more than 50% of the vote. Vote Together aims to allow so-called strategic voters the ability to see which candidate has the best chance to defeat the Conservative, and then galvanize the vote accordingly.
Cariboo-Prince George has been identified as one of those swing ridings and, according to a basic glance at the information, Trent Derrick of the NDP is the candidate best positioned to beat Todd Doherty of the Conservatives. Accordingly, there have already been numerous media stories and social media posts calling Derrick the front-runner.
The truth, as ever, is more complicated.
Beware riding-level data
First of all, I should tell you that I do not know a ton about statistics, other than I know they can easily be misread, misleading, and generally misunderstood. I also know to beware of polls commissioned by an organization with an agenda.
So to start, I fired off an email to Éric Grenier of ThreeHundredEight.com, a blog that does high-level poll tracking to try and get a snapshot of the federal election as well as riding-level projections. I outlined the stated methodology and asked if anything set off alarm bells. His reply was, “I don’t think there is anything suspect about the polling, though riding polls are always more difficult to do. So that is something to think about.” If you want to know more about the difficulty of getting riding-level data, see here, but basically it’s time-consuming and costly, especially as people become more mobile and landlines fall out of favour.
Also for what it’s worth, Grenier currently has the riding going Conservative, though he ranks it as “marginal” for NDP gains (meaning not impossible, but not super likely, basically).
Always note the margin of error
With that in mind, I called up Environics Research, the company commissioned by Lead Now/Vote Together to conduct this poll so I could better understand what I was seeing.
First things first was the margin of error: for the Cariboo-Prince George it’s sitting at 4.4.
Basically what this means is the level of support for the parties/candidates listed below could move up or down by 4.4, so long as the final combined number adds up to 100 (as in, 100% of the people surveyed). So what does that mean for Trent Derrick leading by 6 points?
Well, he could be leading by 6 points. But, it could be 4 points less than that — bringing him down to 32. Give 2 of those points to the Conservative and 2 to the Liberals and your new projection is NDP: 32, Conservatives: 32, Liberals: 31. Not nearly as much of a lead.
Or the NDP could be down 4, the Conservatives steady, and the Liberals up 4. Then you have Liberals 33, NDP, 32, and Conservatives 30. Now the Liberals are best-positioned to win.
Check the questions being asked
After getting that out of the way, I started drilling into the actual questions being asked (which you can find here). Here they are:
Q1. The first question is, are you over 18 years of age, a Canadian Citizen, and resident of Canada for the last 6 months?
This is just to establish the person being surveyed is an eligible voter in the riding.
Q2. If the Federal election were held tomorrow, which party’s candidate would you vote for here in the riding of Cariboo Prince George?
This one is a bit more interesting. It’s fairly straightforward but the report indicates (and I had it confirmed to me) that people were then read a list of the parties and the candidates standing for election.
This is all well and good, but it only covers the Conservatives, Liberals, NDP, and Green. Cariboo-Prince George also has a candidate from the Christian Heritage Party and two independents. Feasibly, people who would vote for them could fall into the “undecided” category, but just as feasibly they would choose one of the other parties, not hearing their preferred candidate listed. Historically speaking, non-party candidates don’t get huge chunks of the vote, but they can still be a factor. So at this point the results start to get a little less accurate.
The next thing you’ll notice is there is a question directed at the voters who said they were undecided:
Q3. Even though you are undecided, is there a party’s candidate that you are leaning towards?
And of the 50 people who said they were undecided, 39% continued to say they had no idea where they were leaning, so at this point they were removed from the survey.
So that Trent Derrick has 36% support number is actually “Trent Derrick has 36% support among voters who have either decided how they are voting or say they are leaning in a certain direction” — and the sample size is reduced to 465. If you include those who say they have no idea the numbers become Trent Derrick/NDP 33%, Todd Doherty/Conservative 28% and Tracy Calogheros/Liberal 27%. The NDP still leads, but the gap between them the and Conservatives is reduced from 6 to 5 points, and the gap between the NDP and Liberals falls from 7 to 6.
And it’s worth bearing in mind this is including undecided voters. If you go back to just those who have actually made up their mind, the numbers are NDP 31%, Conservative 26% and Liberal 25% — again, the parties are in the same order, but the gaps are narrowed.
Now we get into the really interesting stuff.
Q4. If local polling showed that another candidate had the best chance of defeating the Conservative candidate in your riding, would you be willing to change your vote in order to defeat the Conservative, or would you stick with your current vote regardless?
On this one, we see that 57% of the people voting for someone other than the Conservatives indicated that yes, they would change their vote if they thought it would lead to a Conservative defeat. A full 71% of those who said they would vote for the NDP would change their vote, while just 44% of Liberal supporters would do the same. This indicates a higher level of fluidity amongst the NDP voters. This may not be a big deal if they stick with the assumption that the NDP candidate has the best chance of winning, but if for whatever reason there’s a sense the tide is going Liberal, it could lead to a shift.
The next question is about who people voted for in the last election, which I find interesting but doesn’t really tell us much about this election. Here it is excluding those who either didn’t vote or didn’t remember how they voted.
A couple of things I will note here. First of all, more than any other party, the Conservatives are relying on repeat voters. 89% of those who say they’ll vote Conservative this time voted Conservative last time, while just 69% of those who are voting NDP this time did so last time and only 25% those voting Liberal this time voted Liberal last time. That means those other two parties are siphoning off votes from elsewhere. For the NDP, new supporters are primarily coming from the Conservatives and Greens (12% each) while the 34% of those who say they are voting Liberal this time voted NDP last time, and 31% who are voting Liberal this time voted Conservative last time.
I’m not really sure what that means, although I would suggest that it might not go well for the Liberals to be sitting in third locally, because people who previously voted Conservative might go back to their party to try and ensure a Conservative win while those who previously voted NDP might go back to that party in order to try and ensure a Conservative defeat. Or not. Who knows?
Anyways, the last question before the “how old are/what gender are you/what level of education do you have?” is the money:
Q6. In this election, would you say you are voting to re-elect Stephen Harper and the Conservatives, or are you voting to defeat the Harper Conservatives and elect a different government?
This one is interesting because it sheds some light on the “Anyone But Conservatives” question, ie is this really a referendum on if we like the Conservative Party or not?
The entire premise of this poll is that people are voting to defeat Harper, so it stands to reason they would want this question asked, and the results seem to bear that out: 62% say they are voting to defeat Stephen Harper.
However, when you drill down into that it gets a bit weird. 5% of the people voting for the Conservatives are doing so to defeat Stephen Harper, which makes no sense in my mind, and 5% of those voting for the NDP are doing so to re-elect Harper, which makes equally as little sense unless they somehow think the Liberals are on their way to forming government, and the NDP have a better chance of beating the Liberals locally, so by voting for the local NDP candidate they will deny the Liberals a crucial seat, helping Harper to victory which seems like way too much strategic voting for anyone to be engaging in.
The person I talked to said I should probably chalk this up to poll error; I’d also suggest it shows the problem with asking an either/or question to something as complicated as “why are you voting?” People could be doing it because they like the local candidate and don’t care about the leaders, or because they know the person running, or because they are lifelong members of a party or whatever. Anyways, weird results, I wouldn’t put much stock into it.
Conclusion: we have no idea who’s going to win until the election is over
Just like the federal polls, this one shows a three-way race that could go any number of directions based on any number of factors, including surprises from other candidates who aren’t even accounted for in the way this survey was conducted. The only thing I’ll say is after a lifetime of Cariboo-Prince George being considered “safe” for the Reform/Conservatives, it’s a heck of a lot more interesting being in a swing riding.
