i'm going to vote for noise, right now.

the conservatives are running near the high end of their error bar, but it's normal to inflate conservative support a little bit in the numbers - it's a sampling bias and everybody knows it. i've repeatedly presented the number of 30% as a workable zero for conservative support, and it's been hovering near there, now, for years. is 32%, 33% conservative support believable? well, that's still absurdly low, really. but the difference between 34 and 30 on a single poll is negligible, given the importance of that number 30. that's noise. it's going to have to stay put a while to be believable.

the liberals are running around historically normal levels, which is about 40%, maybe just under it. again, 37% is at the bottom of the error bar, but, right now, it's just noise.

...except that something that looks to be a little bit more convincing is a steady uptick in ndp support. the noise in the conservative numbers could be underestimating the distance between the conservatives and the liberals, and presenting a false deduction regarding the lower liberal numbers, which are really bleeding to the ndp.

the other option is that they're bleeding on both sides. but that doesn't seem to be consistent with the data, which is presenting a gap in popularity between the conservative party and it's leader. so, i'm going to vote for noise in the conservative data, for right now - and maybe a signal with the ndp.

http://www.nanosresearch.com/tickers/PDF/20180112%20Political%20Package%20Eng.pdf

jagmeet singh must cut his beard.