We have unconfirmed reports that the Liberal blogosphere has just exploded, however some eyewitnesses are claiming that it was more of an implosion rather than an explosion.
The cause of the event is universally agreed to be the recent poll which was released by the Strategic Council (available here) which put the Liblogers beloved party down in the polls (gasp!) and the Conservatives up in the polls (double gasp!).
Now, information is sketchy at best at this time, but we have one eyewitness describing Libloger Scott Tribe writing:
“I don’t need to remind you about their Harper majority government poll numbers last year, do I? 2ndly, despite the aforementioned scepticism [sic] of any poll commissioned by SC, the poll still shows a neck and neck when the MOE is factored in. That would put it in line with the Decima and SES polls which showed a virtual tie between the Liberals and the Tories last week.” (source)
We have other eyewitnesses describing blogger Stephen Downes as writing:
“It is irresponsible journalism to merely repeat poll results with no analysis or criticism. This is especially the case when the response to the question “How would Canadians vote if an election were held today?” is misinterpreted to read “How would _you_ vote if an election were held today?”" (source)
Moreover, we have further reports of the king of the Libloggers Jason Cherniak has written about “Polls Schmoles” (source).
We’re going to turn this over now to our Senior Asian Reporter Trisha Takanawa who is interviewing Paul from Paulitics: Paul’s Socialist Investigations.
Trisha: Paul, thanks for speaking with us, today.
Paul: My pleasure.
Trisha: Paul can you talk to us a little bit about Scott Tribe’s attack on the Strategic Council’s methodology.
Paul: Well, Scott actually makes two accusations against Strategic. The first is that since they were in his opinion inaccurate in last year’s election, they ought not to be taken as seriously as other polling firms. As I discussed on my Polling Resource Page, it is true that SES had by far the closest predictions, Strategic actually came in second (behind SES) of the four polling firms that I analyzed in terms of accuracy. So I don’t know if it’s entirely reasonable to just write their methodology off.
The second reason for disregarding this poll that Scott makes is that “This poll still shows a neck and neck when the MOE is factored in.” Now, I have to admit that I’m not a polling expert. My background is in Canadian politics, political theory, media studies and Marxist philosophy. However, I did take a stats course in my undergraduate degree and I do still have my old textbook kicking around. So, I may be missing some aspect of this or I may be confusing some aspect of this (for instance, multi-directional or uni-directional z-scores always tripped me up) and if I am, then I’ll be the first to admit it. But here’s what I found when I examined Scott’s second claim.
What Scott’s claiming is that, since the Margin of Error (MOE) of this poll is +/-3.1%, and since the Liberals are at 29% and the Conservatives are at 34%, then they’re only 5% apart. Hence, since (3.1) x (2) = 6.2 > 5, then the Liberals and Conservatives are actually tied. This is highly problematic reasoning.
I’ve brought along some diagrams with me, do you mind if we show them to your audience Trisha?
Trisha: No, not at all, Paul.
Well here we see two normal distributions which illustrate the probabilities that we’re looking at. I’ll explain the math later, but let’s go ahead and take a look at these right now.
Since, there are 5% points separating the Liberals and Conservatives, the Liberals would have to be 2.5% higher and the Conservatives would ALSO have to be 2.5% lower than the Strategic poll predicted. Therefore, what we’re really looking at is the probability that the Liberal number is beyond the point I’ve drawn all the while simultaneously the Conservatives are below the point I’ve drawn.
Since probability is the area under the curve, it looks like there’s a pretty good chance that Scott’s right, doesn’t it Trisha?
Trisha: It sure does Paul.
Paul: That’s right, at first glance it does. But let’s take a closer look.
The MOE for this poll was 3.1% 19 times out of 20. In other words at the confidence interval of .95, we have a MOE of 3.1. Therefore, if my undergrad stats class didn’t fail me, then we can find the standard deviation like this:
Since 0.95 = a z-score of 1.96
Therefore, then standard deviation (s) = 3.1/1.96 = 1.582%
Therefore, using this standard deviation, we can find out that:
2.5% = a z-score of 2.5/1.582 = 1.5806.
Thus, the probability that EITHER the Liberals OR the Conservatives are 2.5% off from what Strategic predicted is 5.71% (to find out how I got this number, just look up a standard z-score chart).
Trisha: Well that doesn’t sound so low Paul.
Paul: Well, just a second there Trisha. 5.71% is the probability that EITHER the Liberals OR the Conservatives are 2.5% off from what Strategic predicted. But in order for Scott’s contention that the Liberals and Conservatives are neck-and-neck, they BOTH need to be 2.5% off. The way we calculate that is by multiplying the probabilities together.
(probability the liberals are off by 2.5%) x (probability the conservatives are off by 2.5%) = the probability that Scott Tribe’s right.
Thus the probability that Scott Tribe’s right(0.0571) x (0.0571) = 0.0033 = 0.33%
Trisha: That’s a pretty small chance that’s Scott’s right, Paul.
Paul: That’s right Trisha. But then again, I’m not a statistician, so you’d have to consult somebody who’s more knowledgeable on his subject to check my numbers.
Trisha: But what about Stephen Downes’s loud claims?
Paul: Well I really don’t really fully understand what Downes is getting at, but if you go to Strategic’s actual report and not the Globe and Mail article, you’ll see exactly which question they ask and it seems pretty normal to me (and I have worked for the Decima polling firm in the past, so I’ve personally conducted polls like this one.)
“If the federal election was being held tomorrow, do you think you’d be supporting the (ROTATE LIST) Liberal candidate in your area, Conservative candidate in your area, the NDP candidate in your area, or the Green Party candidate in your area or (QUEBEC ONLY) Bloc Quebecois candidate in your area?”
Trisha: But Paul, what do you make of Jason Cherniak’s dismissal.
Paul: Well Trisha, I’ve been told that if you can’t bring the Yiddish, that you shouldn’t try, so I wouldn’t want to attempt anything with regards to Cherniak.
What I can say though is that all three of these bloggers are hardcore partisans who simply don’t like the prospect that their beloved party is down in the polls. I don’t see them complaining when the Conservatives, NDP or Greens take a “nosedive” in a given poll, so it seems like here, in my opinion, we have a case of ’the lady doth protest too much methinks’.
Me, on the other hand, I hate both of these parties equally and really couldn’t care less which one is ahead in the polls.