What the hell happened in Michigan?

[Lyin' Steve]

I've seen arguments that Michigan polling was particular difficult because in 2008 there was no contest, while in 2004 and 2000 Kerry and Gore had already sealed the deal by the time the primaries happened.  If this is the case it would mean the models were simply off in Michigan, but less likely to be off in Ohio and Illinois.  I'm not putting faith in anything until March 16th though.

[Averroës Nix]

Aren't most pollsters including cell phones in their samples by now? I thought that many were even weighting their sample for cell users alongside weighting demographic characteristics.

[muon2]

Aren't most pollsters including cell phones in their samples by now? I thought that many were even weighting their sample for cell users alongside weighting demographic characteristics.

Some use weighting and some try to identify respondents that match general demographic characteristics. Even when cell phones are included they aren't as easy to place at a specific location as a land line. Location matters as well if the pollster is trying to find representative voters for their sample.

[Landslide Lyndon]

They say that the problem was the companies stopped polling before Sunday's debate but I can't take this excuse seriously. Almost everybody, even foreign media, said that Clinton had a strong performance.
But even if they were wrong it would take a Sara-Palin-with-Katie-Couric type of meltdown to lose a double digits lead in one day.

[DemPGH]

What happened in Michigan was just about the best thing that could have possibly happened across the board, and when I was really getting ready to functionally tune this election out, I'm now peeking in. I hope it keeps up, and I hope it encourages more people outside the party cliques and platforms to get involved to seek office. And, we're seeing that people who come in from the outside can be doing more than just attention whoring (Ventura, at this point) or just being nutty (assorted third party types or extremist libertarians), and as I said, I hope this starts a new trend.

[SillyAmerican]

Considering it was an open election, pollsters only ask "likely" (strictly Dem + older) voters, a large part of the disparity could've been accounted for, and the rest was up to voter motivation/luck.

Polling continues to be a problem, carrying over from 2012. Good polling is based in large part on identifying a representative sample of likely voters across all demographic groups. The decline of land line phone and the decreasing willingness of people to take surveys over the phone make it hard to assemble a useful sample. Pollsters have also seen that the historic qualities that identify likely voters from registered voters don't predict voter behavior as well as they once did.

What's most interesting to me is the fact that instead of believing what the polls were saying, voters are going about casting ballots anyway. Which is to say, no indications of a "foregone conclusion" response from the public. Recent polls showed Clinton having a double digit lead. So pollsters are wasting their time, as it looks like the electorate isn't interested in playing along.

[Bojack Horseman]

Probably Sanders' late ad blitz. While I've been out of state since Monday, Bernie Sanders ads flooded the airwaves nonstop, and just watching the evening news on network television the ads were pretty much 3-1 Sanders vs Clinton.

[SingingAnalyst]

A few things happened in MI:

(1) Lots of crossover voting. Last night was Feb. 22, 2000; Hillary Clinton was George W. Bush; and Bernie Sanders was John McCain. Sanders upset Clinton in an open primary just as McCain had upset Bush 16 years earlier. Sanders did well in many traditionally GOP areas, getting 59% in fast-growing Livingston Co., 57% in Hillsdale Co. (seat of the college with the same name), and 64% in heavily GOP, Christian-Reformed Ottawa Co.  Sanders also did well in university counties such as Washtenaw, Ingham, Calhoun, and Kalamazoo.

(2) Youth enthusiasm. Almost to a girl, every young voter I know went to Sanders, and were very enthusiastic about their choice.

(3) Early voting. Trump may have won with 37%, but that means 63% wanted someone else. I believe much of Trump's votes were cast early, before some of the more recent debate performances. Also, the 63% voted at different times as well: there were times Rubio, Cruz, and Kasich all seemed to be surging. Had these 63% coalesced around one candidate, they could have beaten Trump.

[Senator Spark]

No freaking clue. The question is, is this a sign of a greater trend or is this isolated to MI? I imagine Sanders will now win OH, MO and make it close in IL. He will carry this fight to the convention. There will be no early healing or no dropout on March 16. This is the worst possible scenario for Clinton at this point in the game.

[SingingAnalyst]

No freaking clue. The question is, is this a sign of a greater trend or is this isolated to MI? I imagine Sanders will now win OH, MO and make it close in IL. He will carry this fight to the convention. There will be no early healing or no dropout on March 16. This is the worst possible scenario for Clinton at this point in the game.

It could happen in any state with an open primary.

[Ebowed]

They say that the problem was the companies stopped polling before Sunday's debate but I can't take this excuse seriously. Almost everybody, even foreign media, said that Clinton had a strong performance.

Did you watch the debate, though?  Sanders did quite well in it and Clinton's performance was below average, with several poor answers that highlight her inability to give straightforward responses.  With the debate in mind, the numbers in Flint make a lot of sense.

[This account no longer in use.]

They say that the problem was the companies stopped polling before Sunday's debate but I can't take this excuse seriously. Almost everybody, even foreign media, said that Clinton had a strong performance.

Did you watch the debate, though?  Sanders did quite well in it and Clinton's performance was below average, with several poor answers that highlight her inability to give straightforward responses.  With the debate in mind, the numbers in Flint make a lot of sense.

It's Landslide Lyndon. Clinton could quite literally say that Hitler was right on the debate stage, and Landslide Lyndon would still praise it as an 'excellent performance by Clinton'.

[Antonio the Sixth]

I'm a Bernie supporter, and I thought Hillary won the debate.

[Landslide Lyndon]

They say that the problem was the companies stopped polling before Sunday's debate but I can't take this excuse seriously. Almost everybody, even foreign media, said that Clinton had a strong performance.

Did you watch the debate, though?  Sanders did quite well in it and Clinton's performance was below average, with several poor answers that highlight her inability to give straightforward responses.  With the debate in mind, the numbers in Flint make a lot of sense.

Did you read the rest of my post? Unless she had a nervous breakdown or said the that the voters of Michigan are morons there is no way a mediocre, or even downright awful, debate performance to wipe out in a day a double digit lead.

[IceSpear]

I've seen a lot of theories floated. Most likely it's a combination of multiple factors:

1) She was never up 20+. Those polls were junk. She may have been up in the low double digits like Monmouth showed, or perhaps they were a bit off too and it was actually high single digits. In addition, no polls besides junky Mitchell were out in the field late, which was a similar problem that happened in Iowa this year. Polls were also wonky because there hasn't been a competitive Michigan primary in a long time, so they were unsure how to model it.

2) Bernie outspending her paid off.

3) Hillary supporters got complacent, and either decided to stay home or cross over to vote in the Republican primary because she already had it "locked up."

4) Bernie got a debate bump. Most people didn't think it would make much of a difference, but perhaps the voters thought differently. The trade portions in particular may have hurt Hillary.

5) Youth and independent turnout was much larger than expected.

Likely some or all of these conspired together to create the perfect storm.

[Sir Mohamed]

The pollsters were dead wrong. Dead wrong.

[Likely Voter]

WaPo thinks they figured it and and apparently it is all Howard Dean's fault...
https://www.washingtonpost.com/news/the-fix/wp/2016/03/09/why-were-the-polls-in-michigan-so-far-off/


Basically the likely voter screens of pollsters were using models based on the weird 2008 race and were underestimating the number of men and young voters.  So in theory MI is an anomolie because of their issues with the 2008 election, which shouldn't apply to polling in other states like OH.

[SingingAnalyst]

I've seen a lot of theories floated. Most likely it's a combination of multiple factors:

1) She was never up 20+. Those polls were junk. She may have been up in the low double digits like Monmouth showed, or perhaps they were a bit off too and it was actually high single digits. In addition, no polls besides junky Mitchell were out in the field late, which was a similar problem that happened in Iowa this year. Polls were also wonky because there hasn't been a competitive Michigan primary in a long time, so they were unsure how to model it.

2) Bernie outspending her paid off.

3) Hillary supporters got complacent, and either decided to stay home or cross over to vote in the Republican primary because she already had it "locked up."

4) Bernie got a debate bump. Most people didn't think it would make much of a difference, but perhaps the voters thought differently. The trade portions in particular may have hurt Hillary.

5) Youth and independent turnout was much larger than expected.

Likely some or all of these conspired together to create the perfect storm.

3 and 5. As for 5, an unusually high 11% of MI Dem voters were under 25, as opposed to only 6% of MI GOP voters. The under 25 Dems went for Sanders 85-15. If their proportion of the total was 6% rather than 11%, Clinton would have won.

As for 3, take Huntington Woods, a middle-class, extremely highly educated, liberal-leaning suburb of Detroit. Fully 33% of Tuesday's vote was cast on the GOP side, which is a higher fraction than the GOP has won in a GE since 1988. Kasich won a whopping 58%, while on the Dem side Clinton won just 51-47, even though the community is a natural fit for Clinton. I take it a few hundred Clinton supporters crossed over to vote Kasich, to try to stop Trump.

[Landslide Lyndon]

Also, for those saying that Clinton was hurt by her debate performance.
She actually won late deciders 51/44. Sanders' margin of victory came from those that had decided a month or more ago.

[○∙◄☻¥tπ[╪AV┼cVê└]

WaPo thinks they figured it and and apparently it is all Howard Dean's fault...
https://www.washingtonpost.com/news/the-fix/wp/2016/03/09/why-were-the-polls-in-michigan-so-far-off/


Basically the likely voter screens of pollsters were using models based on the weird 2008 race and were underestimating the number of men and young voters.  So in theory MI is an anomolie because of their issues with the 2008 election, which shouldn't apply to polling in other states like OH.

Lets see, the newspaper that recently ran 16 hit pieces against Bernie in just 16 hours is trying to convince people that some election 8 years ago screwed up all the polls in Michigan, and none of the other 49 states could be used as models? What a joke!

[Landslide Lyndon]

What a joke!

Yes, you are.

[This account no longer in use.]

What a joke!

Yes, you are.

Are you seriously trying to deny that the Washington Post ran 16 negative Sanders stories in sixteen hours? Because there's proof right here.

[🦀🎂🦀🎂]

What a joke!

Yes, you are.

Are you seriously trying to deny that the Washington Post ran 16 negative Sanders stories in sixteen hours? Because there's proof right here.

Oh no! Our candidate is being covered like any other candidate would be! #conspiracy #Illuminati

[emailking]

What a joke!

Yes, you are.

Are you seriously trying to deny that the Washington Post ran 16 negative Sanders stories in sixteen hours? Because there's proof right here.

Regardless, the likely voter screen explanation makes a lot more sense than the other theories people are floating, which are either explanations for why Bernie did well (not why the polls were off) or are things that can happen in any election and often do but don't lead to 20+ point mishaps.

[This account no longer in use.]

What a joke!

Yes, you are.

Are you seriously trying to deny that the Washington Post ran 16 negative Sanders stories in sixteen hours? Because there's proof right here.

Oh no! Our candidate is being covered like any other candidate would be! #conspiracy #Illuminati

Give me an example of the Washington Post running 16 negative stories on another candidate, or for that matter, anybody else in as many hours.