and, this is what i expected to see:
those are the actual numbers.
the survey companies then assume that the 14% of people that answered in some non-standard way are equally likely to vote for any of the parties and reweight the outcome, accordingly. but this is always the primary source of error in polling, in canada. i keep yelling at them to stop doing that, and they keep doing it.
the undecideds are never truly randomly distributed, in the end. the data companies need to be neutral, so it's up to somebody like myself to try to read the polls and figure out what they actually say.
the balance of the data - however sparse it is - suggests that there are two major swing blocks in bc right now:
1) rural voters trying to choose between the liberals and conservatives. they may rather vote conservative, but i suspect they'll mostly actually vote liberal.
2) ndp/green fence voters that want to vote green but don't want the liberals to win. if the headlines are telling them an ndp majority is obvious, they may be more likely to vote green. oops.
there is a third, smaller block:
3) upper class voters in the downtown core may opt for the ndp instead, this time, instead of their preferred liberals. this will likely have an effect on the popular vote, without affecting the outcome much.
so, i strongly suspect that that 42% is about where the ndp will end up, in the end - a modest bump. they may get 43, 44 - something like that. suggestions that they're running over 45% do not appear to be grounded in meaningful evidence.
further, that 14% should distribute partly to the greens and mostly to the liberals, who will likely end up somewhere above 35.
so, i'm going to produce a wide popular vote projection, after all:
ndp: 40-45
liberals: 35-40
greens: 15-20
and, as stated, trying to predict a majority or minority from that is foolish.