2016 Postmortem
Related: About this forumIt breaks my heart as a statistical analyst to do this, but the polling. The polling done be
Fucked up in Michigan, and there are a lot of pollsters out there that have taken a major reputation hit tonight. Hell, 538 is already defending itself with the fact that their probabilities are based off polling, and, if the polls are wrong, obviously their results are going to be wrong.
So what can we conclude analytically from tonight?
Pollsters have no idea how to weigh this year in open primaries (or caucuses) but are much closer to reality in closed primaries. This means there are some real fundamental weaknesses for those that polled Michigan, because closed primaries are considered easy to predict (Much easier to match ID to voting rolls, easier to follow patterns, etc).
It completely calls into question polling in any state outside of closed primaries or areas with heavy favorites (SEC vs Blue Sky, for example). And, this is as a Clinton supporter now, it is clear Kos stupidly jumped the gun with his GE footing post over at DKos. Bernie, whether winning or losing in a squeaker in Michigan is not going anywhere, now.
And finally, and most sadly from my point of view, that, while delegate math is still easy to follow, the people who have repeatedly insisted polling means Jack shit this time around have an exceptionally valid point. Polling, as an industry whole, is going to have to address some really hard questions about sampling, population parameters and reaching people who will show up to vote but not answer their phones for a pollster when this cycle is finished.
Congrats to Bernie and his followers on Michigan, regardless of where it ends up he clearly outperformed expectations. I think it is safe to say no one has any idea how states like Ohio are now going to play out.
vdogg
(1,384 posts)Every open primary this year has been closer than expected ( though this one takes the cake). It makes it hard to trust their future polling. I think from now on I'm going to take polls in an open primary state with a grain of salt.
TM99
(8,352 posts)as long as they meet the deadlines to re-register as either a Democrat or a Republican. I have done so in my state.
Stallion
(6,474 posts)I think a lot of it is that they don't know who is going to show up at the Democratic primary in open primary
MADem
(135,425 posts)Good grief.
It's going to be a long night.
I was just about ready to throw in the towel. I still don't see anything from Detroit proper reporting. Think I'm gonna go to bed. Can't stay up for this one.
MADem
(135,425 posts)votes were coming from. They're still saying TCTC with Sanders up 34K.
It's a proportional state so both candidates, no matter who wins, will take a pile of delegates out of there.
vdogg
(1,384 posts)Warren DeMontague
(80,708 posts)winter is coming
(11,785 posts)ibegurpard
(16,685 posts)From someone who has a stated preference. I salute you.
tex-wyo-dem
(3,190 posts)I think the uncertainty in polling is coming partially from a lot of crossover voters (independent and moderate republicans) which are difficult to poll. Bernie is attracting a lot of these voters with his populist message. Registered Democrats I think tend to go with the safe choice, which they (IMO wrongly) consider to be Clinton.
vdogg
(1,384 posts)They called it for Obama when he was down 5. Cuyahoga came in big for Obama but really late and all at once. Polls suddenly flipped. We might see something similar tonight.
jeff47
(26,549 posts)and it is already 75% in. Clinton's winning it roughly 60/40, and that isn't enough votes to flip it.
ismnotwasm
(41,976 posts)Depressing, but I'll deal with it. There were a couple polls showing something like this, but the main polling was horribly off this is a huge gain for Sanders.
Hiraeth
(4,805 posts)spelled momentum
Uglystick
(88 posts)Make it a very hard lesson to learn.
Now you know Clinton has no momentum left, and she's going to be losing a lot of March 15th states because the polling is all completely wrong?
You and Nate can share the same dirty boat, thank you.
Godhumor
(6,437 posts)Kalidurga
(14,177 posts)that was uncalled for. If I was the alerting type...
reformist2
(9,841 posts)madfloridian
(88,117 posts)Things are different this time. The pollsters can't reach the youngest and newest voters. They were not prepared for the independent influx.
DirkGently
(12,151 posts)POLLSTER SAMPLE WEIGHT LEADER
CLINTON
SANDERS
MAR. 6 Mitchell Research & Communications 475 LV
0.82
Clinton +37
66%
29%
MAR. 3-6 Monmouth University 302 LV
0.66
Clinton +13
55%
42%
MAR. 2-4 YouGov 597 LV
0.50
Clinton +11
55%
44%
MAR. 1-3 Marist College 546 LV
0.50
Clinton +17
57%
40%
MAR. 4-5 American Research Group 400 LV
0.48
Clinton +24
60%
36%
FEB. 29-MAR. 1 EPIC-MRA 400 LV
0.18
Clinton +25
56%
31%
MAR. 2-3 Mitchell Research & Communications 616 LV
0.10
Clinton +18
55%
37%
FEB. 22-27 Marketing Resource Group (MRG) 218 LV
0.04
Clinton +20
56%
36%
JAN. 25-MAR. 3 Michigan State University 262 LV
0.03
Clinton +5
52%
47%
MAR. 1 Mitchell Research & Communications
http://projects.fivethirtyeight.com/election-2016/primary-forecast/michigan-democratic/
wilsonbooks
(972 posts)continue to be embarrassed.
TM99
(8,352 posts)You and I have had some dust-ups about this.
As I said in another thread, math can not predict human behavior. It can only make approximations. After that, all it takes are slight human variable to come into play that will throw it completely away from what was originally predicted.
I am willing to bet that in the case of Michigan it was Clinton's artful smear of Sanders around the auto-bailout. That may have made the difference in how voters actually turned out and voted. I also think that there is a big difference in the AA communities in red states versus blue states which is ignored when one sees them as a monolithic voting bloc and a statistical entry on a spreadsheet. It will be interesting to see what happens as we go forward.
I would say the race is very much still on.
Lucky Luciano
(11,253 posts)I am much more of a math guy than a stats guy (I know you know the distinction!) - the art/science of polling has definitely changed. From the Greek referendums/elections to Jeremy Corbyn, the massive win for Trudeau, etc, the polls have been off the mark.
I wonder if the closed primaries are also esier to predict because the ages will probably skew older for people registered in a party - thereby making the under 40 crowd less relevant and older polling methods more valid. Who knows.
superkona
(21 posts)likely voters
vdogg
(1,384 posts)Some pollsters will have a lot to answer for tomorrow.
vdogg
(1,384 posts)That got 2% of the vote.
vdogg
(1,384 posts)"Fifteen states Connecticut, Delaware, Florida, Kentucky, Maine, Nebraska, Nevada, New Jersey, New Mexico, New York, Oklahoma, Ohio, Oregon, Pennsylvania have closed primaries. Semi-closed. As in closed primaries, registered party members can vote only in their own party's primary."
Some of these have already voted, and were fairly close to their polling. This will probably continue to be the case.
BeyondGeography
(39,370 posts)At least she won somewhere.
Godhumor
(6,437 posts)BeyondGeography
(39,370 posts)He not only jumped the gun he was obnoxious about it.
longship
(40,416 posts)Where media is not just televisionary sets, and telephone lines no longer have any telephone lines!
So how is one going to measure public opinion under those conditions? One would presume that Nate Silver and 538 will go to telegraph wires next. Ya never know. It could be the next big thing, the big secret, the crystal ball. Or maybe the idiot just got lucky and is cashing in on it.
garbage in. garbage out.
retrowire
(10,345 posts)JackRiddler
(24,979 posts)Turnouts are small and local motivation on the day of the vote is everything. Silver's too far up the bunghole of his own successes to admit this even to himself, it seems. When you data-driven guys have everyone wired via brain chip - I'm sure some of you are looking forward to that - you might manage this feat.
Meanwhile every other post on this site is oooooh, 98 point lead for Hillary in State X! Stop the fight! We see what the effects of this voter suppression strategy have been: higher turnouts for the Republicans, who are eating up all the coverage with their shitshow. (It probably won't work out for them in the end, for obvious reasons, dick size, etc. We will learn if there is a bottom to underestimating the intelligence of the American public...)
applegrove
(118,634 posts)may not have taken it into account that Michigan was an open primary where independents could vote. The polling was definitely off.