NH-UMass Lowell: Tied

[SirMuxALot]

The first one completly wrong Party reg (D+1), while the second one a little bit R-friendly. (R+)

It's R+2.8% party reg. But Indis way too big in both.
http://sos.nh.gov/NamesHistory.aspx

This puzzled me as well.  Are we sure they weren't actually asking two party-affiliation *questions*...i.e., first ask the respondent what party they identify with, then secondly ask them which party they are registered with?  (Both have self-reporting error/bias, of course.)

Because that's the only explanation for those registration numbers differing.

[SirMuxALot]

Lol, people here haven't learned anything from 2012 and 2014, have they?

Too many people here have only had lessons from 2012/2014 (and 2008).

Lots of folks who haven't learned or remembered the lesson of 1980.  And many who know the lesson, but think it couldn't happen again anyway.

Not saying it will happen this year.   But there will be, in most of our lifetimes, another election that is a massive polling miss.

[SirMuxALot]

1980 wasn't a massive polling miss.

Oh, not this crap again.  Do you guys ever actually go back and pull the raw information yourselves rather than just reading some leftist website talking points?

It was a big miss.  Final result was R+10.  Gallup was R+3 in their final poll.  If we had an RCP average back then, it would have shown about R+2 to R+4 at best, with the handful of public polls we had back in those days.

Gallup was regarded by the public as the only "professional" pollsters back then, and their next to last poll was Carter +8, less than two weeks to election.

And the point isn't about how the pollster's Tuesday morning election polls performed.  The point is about how high a probability of error exists for the polls one week out (about where we are now).

It's worth remembering how Pat Caddell describes the Carter internal polls in 1980.  "Saturday [before election day] we were tied.  Sunday we were down 5.  Monday we were down 10.  The bottom fell out that fast."  (Paraphrased quote from a podcast interview I heard him on yesterday.)

That is my very point.  At this exact point in 1980, we would have all been saying "way too close to call" and "Carter firewall will hold".  And this board would have had its hair on fire about how big a surprise the actual outcome of R+10 was.

[SirMuxALot]

Carter trailed for basically the entire race.  Reagan gained in the final week.

Perhaps I'm not stating my point well.  I'm not claiming that Carter had a lead and a win Reagan win shocked everyone.  I'm claiming that only the absolute final 1980 polls (literally the weekend conducted and Mon/Tuesday released polls) caught wind of what was about to happen.  There was a broad consensus that election night 1980 was going to be a long one, a nail-biter.  It turned out we all went to bed fairly early that night.

Let me try to state it this way:

Lots of people here have only learned the lessons of 2008 & 2012.  That is, the polls one or two weeks out are stable and fairly accurate.  The implied assumption being that a late break of 5-10 points just doesn't happen (anymore).

Don't mistake my point as being that Carter was ahead and lost.  I am not claiming that.  My point is that Carter and Reagan were very close, within the margin of error through all of Sept/Oct.  "Too close to call" was the wide consensus through the entire fall of 1980.

Yet a late break led to a R+10 result.

My point being, plenty of people here don't know what a late break like 1980 looks like.  It's made more difficult by the change in polling methodology and the sheer quantity of publicly available polls now (that we didn't have in 1980).

It *might* (emphasis on might) be that a late break in the modern polling era looks a lot like what we are having now.  Big volatility in polls, "outliers" galore, and relatively wildly divergent models showing us irreconcilable differences like we are getting this week.  This is conjecture on my part, I readily concede that.  But I lived through 1980 and remember how shocking the R+10 result was on election night.  Not so much that Reagan won, but that he won so big.

I'm not proposing Trump can win by 10 here.  I'm am proposing that 1980 race shows us that Clinton +2 to +4 right now may not be as safe as it seems.

We'll know in only days.