well, a few days ago, i was talking about how the way that you interpolate the undecideds is key to predicting the outcome. you'll notice something different about this poll - it adds up to 97%. most of the polls released over the last few months add up to around 90%. and, the difference went entirely to trump.
to be clear....
clinton is about flat.
johnson is about flat.
stein is about flat.
but, trump made up 6-7 points as the undecideds fell to the margins.
a few days ago, i asked the question: what does undecided actually mean? we had data indicating that both candidates were unpopular, but no way to really piece anything causal together. does it mean you hate both candidates? and will vote third party? or will stay home? so, will it distribute?
what this poll seems to suggest is that undecideds may be leaning heavily towards trump.
i find that very hard to believe, given his numbers. if you believe this poll, he's running twenty points ahead of his approval ratings.
note that the labour day weekend is always perilous to poll over and is known for producing outliers. but, we'll have to see what happens.
also note that while the nbc poll is more in line with the trendlines, the methodology is garbage and it should be ignored.
there may be a...
this idea is articulated a lot of ways. but, the idea is that you may get some bias in polling around issues of race.
given the media coverage, i would acknowledge the plausibility that a subset of people may be embarrassed to admit they plan to vote for trump. they may be defaulting to undecided.
i don't want to call that a "shy conservative effect", or a reverse anything effect. it's just people not wanting to admit it.
to suggest that this is the entire undecided vote, however, strikes me as too much.
so, even if this poll is revealing something that has to this point been hidden, i'm still more likely to point to the labour day weekend as an exaggerating factor. if trump voters are lower income and less educated, they're more likely to be home and ready to pick up the phone this weekend.
and, why did the undecideds come down this week, all of a sudden? perhaps they're more affluent.
i'm not going to make predictions. again: i don't think it's possible to predict the outcome of elections. i'm just pointing out reasons to be careful with this poll, and reasons it may end up in right field. or, may be illuminating, too.
at least note that it is a real poll.