Wednesday, March 20, 2019

nanos clearly has the best data. but, it's also widely misunderstood - and i've been over this before, but let's do it again. this has everything that is useful in the tracking poll, and it actually says that there's really been minimal movement lately.


you'll note the little blurb there, explaining that this is a rolling sample. the gigo models want you to believe that you're better off average everything out, which is actually just going to produce middling data; the best predictor of an election is to look at a snapshot poll conducted as close to the date of the election as is possible, as it prevents old and out of date data from corrupting the sample.

what nanos does is the exact opposite of that, and it's not some kind of accident, but he does it this way because he's not trying to predict an election but rather trying to measure party branding, which is going to fluctuate a little week over week. nik is basically trying to smooth the data out, which is arguably more useful in determining the effects of a scandal on a party brand, as intended, but isn't going to tell you who is going to win the election tomorrow. and, there isn't an election tomorrow. but, we have to understand what we're actually looking at, here.

if you look at the line for the liberals across the top, you can see that they're down by some amount, pretty much across the board. so, the liberals have lost some support, clearly.

but, if you look at the results for the conservatives and the ndp, you see a different story.

the conservatives went down a little over the course of the week, but this is just noise relative to the margin of error, which is actually pretty big. the conservatives are in fact down from the year's highs, but mostly moving sideways. there is no evidence here of the conservatives taking advantage of the decrease in liberal support, at all.

the ndp are up over the week, but that is also just noise in the long count. while the trendline is moving in one direction recently, it would be wrong to conclude meaningful movement, relative to the margin and they, too, are down from their year high. there is not any convincing evidence that they are taking advantage of this, either.

so, where is the support from the liberal party actually going?

i previously surmised that maybe the greens were picking something up, but they don't appear to be, either. nor is the bloc.

so, this is very strange - how can the liberals be down if nobody else is up? and, the answer must be that people are disengaging.

we saw this in 2011, when liberals neither wanted to vote for ignatieff nor for his opponents, and the conservatives won a majority by accident.

it's way too early to make predictions, but if the apathy sets in long term, trudeau's primary opponent may end up being reheated pizza, easy access to pot and still-stupid reruns of friends.

"i was going to get up and vote, but then i got high."