“Why are Liberals higher with a live interviewer than with the robot?”
you really gotta figure that out. i’ve provided some ideas. but, you really need to work that through.
i just want to add that the comments i’ve posted are not meant to be generalized. the dynamics of this specific election – where 2/3rds of the population sees the two opposition parties as interchangeable – is not likely to repeat itself.
but i would also point out that you were lowballing the ndp when they were ahead, too.
whatever explanation you get should have to explain both of these things. mine can do that.
www.ekospolitics.com/index.php/2015/10/deadlock-broken-liberals-surging/#comment-60049
Jeremy
It is interesting. I have heard some say the “shy Conservatives” are more likely to hang up the phone (based on UK studies). Without declaring my own support – I know I hang up the phone anytime I hear a robo-dialler without waiting to here what it is about.
deathtokoalas
the thing with biases like this is that they’re constantly shifting, constantly cancelled out, constantly complicated…
i’ve posted here a few times. i think he’s picking up a “dithering progressive” reaction over ivr, where liberal-ndp swing voters are having a harder time committing to a machine, but are reflecting social bias on the phone.
you can transfer the mainstreet results down from 38/33 (decided only) to 34/30 (including undecideds). it’s harder to do that with ekos’ data because he’s mixing sample, which i think is really the core of the inconsistencies. but, i would suspect that those results would scale down consistently – at least with the conservative vote.
he may actually simply be correctly picking up that the polls putting the liberals over 35, and heading to 40, are picking up a lot of very soft liberal support – people that would tell a live interviewer they’re voting liberal, and legitimately will probably vote liberal if they vote at all, but are, in their heart, truly conflicted.
if a polling company is meant to predict results, it’s a crappy result. if it’s meant to understand what people are actually thinking, there’s perhaps some usefulness in what he’s picking up.
ways to know frank was picking something up:
1) an unexpected last minute low turnout leads to a higher conservative outcome. this may happen if the ditherers can’t pull the trigger and stay home.
2) there’s an unexpected surge in ndp support. i know he’s not picking that up. but he seems to be picking up huge amounts of indecisiveness (that maybe broke at the last minute).
caveat: i’m interpreting these results in ways the pollster has not, and likely would not (yet) endorse. but, i’m mildly worried about (1), if not particularly worried about (2).
Raunch
Yeah……the polls for the National Post (Postmedia), Global News, and the Globe and Mail (who all have the Libs in front) are trying to support the Libs? Seriously?
3 weeks ago, when Nanos, Ekos, Angus Reid, etc. were showing the Cons taking a lead over the other two parties, the Con-friendly commentors on here thought these polls were legit.
Now, because these polls show different results than what you want to see….you think they are garbage. Well, good luck with your way of thinking. Seems a little hypocritical, don’t you think?
deathtokoalas
i’ve been wondering for a while if there might be a kind of scare tactic at play, to rev up the base.
but, at the least, you don’t expect nik to play around with this. he nails elections. you can do just about anything you want with an online panel at this point, then blame it on untested technology. but, nik has a reputation, and if he’s putting those numbers out, it’s because he believes in them.
--
that liberal swing in quebec is going to elect bloc mps.
i’m plugging in:
libs: 160+, but not 170. 90+ seats in ontario.
cons: ~100, probably a seat or two less. -25 seats in ontario.
ndp: ~45. only ~15 seats in quebec.
bloc: ~35, by coming up the middle on the ndp–>liberal shift. i also think that they get a last minute boost around quebec city to stop the conservatives there.