Today's results: polls vs. actual

[paulthompson]

I find it interesting that the Democratic primary polls continue to be staggerly inaccurate sometimes, and often Bernie Sanders does much better than expected. (All the data here comes from RealClearPolitics.)

In Arizona, there was a poll a month ago where Clinton led Sanders by 56% to 22% - a 34 point lead. Then there was a poll ten days ago where Clinton led Sanders 50% to 24% - a 26 point lead. The actual election results were 58% Clinton to 40% Sanders - an 18 point lead. That's still not a good result for Sanders, but it's markedly better than the 30 point average difference in the polls.

In Idaho, there was only one small poll done a month ago, which had Sanders leading by 47% to 45%, a 2 point lead. The actual result? 78% Sanders to 21% Clinton. That's a 2 point lead to a 57 point lead, a difference of 55 points!

In Utah, there was a poll a month ago that had Clinton leading 51% to 44% - a seven point lead for her. Then there was a poll last week than had Sanders leading 52% to 44% - an eight point lead for him. The actual result? 79% Sanders to 20% Clinton. That's a difference of 51 points from the latest poll!

In Michigan, Sanders had possibly the biggest primary upset of all time, beating the poll results by about 20%. Apparently, there was only one presidential primary (New Hampshire in 1984) that was that far off.

Now, admittedly, there was very little polling data in either Idaho or Utah, and both of those were caucuses, which are hard to predict. But still, to be off by more than 50 points for both states is incredible! Has there EVER been primary season poll results that far off from the actual election count?! I don't know, but I'd like to find out.

In my opinion, we can see two trends here. One is that, at least outside the South, Sanders usually starts way behind and then dramatically gains support as more voters get to know him. The problem he has is when primaries come so fast in different states that he doesn't get enough time for the voters to learn about him. There's usually only a week to campaign before the next primaries come up.

The other trend is that polls sometimes are remarkably inaccurate this primary season, and often underestimate Sanders' actual vote counts. Note for instance other states like Kansas, where there was only one poll from a week prior which had Clinton up by ten, and Sanders ended up winning by 25 points, a difference of 35 points. Even last week, which was very dissapointing for Sanders, he still had some big surges, for instance going from 37 points and 42 points down in two polls in Illinois to losing the state by 2 points about a week later.

The odds are still great that Clinton will win the nomination, barring some dramatic event (which may well happen - her potential indictment is hanging in the air). But Sanders has the potential to make things a lot more interesting than many people expect.

For Sanders to beat the polls by over FIFTY POINTS in two states on the same night is absolutely crazy, even if they were caucus states!

[bobbobbins01]

1. Yet the MSNBC Headline is...

"Clinton, Sanders, Trump, Cruz all win" ...with a big picture of Hillary smiling.

[november3rd]

3. NBC's

owner and Board of Directors are going to be peeved if Sanders gets the nomination. Why? Because he is in favor of more public control of the airwaves: more regulation in the public interest. He's also in favor of publicly financed elections, which will rapidly let the air out of the media money-in-politics balloon, which has been soaring to stratospheric heights since Citizens United.

Finally, the oil companies, the chemical companies and the pharmaceutical industry, television's biggest sponsors, all hate Sanders, too.

[bobbobbins01]

5. At least they're being exposed...people are waking up.

They'll all go out of business if they don't evolve quickly.

[Hortensis]

9. Bernie spent a fortune CAMPAIGNING in

these states right before the election. Of course, poll numbers are going to change. If campaigning made no difference, why would anyone do it?

[pat_k]

2. Think you meant MI. (But I do hope WI is a big upset for Sanders on 4/5!)

It's another open primary, so who knows?

[paulthompson]

4. Thanks

I fixed it.

[Ghost Dog]

6. You highlight important points, paulthompson. The pollsters

tend to have MSM connections, do they not?

(nb. In your second paragraph, I make 56% to 22% a 34 point difference.)

[paulthompson]

19. Thanks

I fixed it.

[paulthompson]

7. CNN

I was just looking at CNN's analysis article about the election results. Here's all it has to say about Sanders' huge wins in Utah and Idaho:

"All along, Sanders has promised his fortunes would change when the Democratic contest moved West, and he easily won the Idaho and Utah caucuses Tuesday night."

That's it. One sentence. It doesn't even mention the actual numbers of how big the margins were.

http://www.cnn.com/2016/03/23/politics/primary-election-2016-takeaways/index.html

[paulthompson]

8. The main CNN story

Oh, and the main CNN story on the election results is even more misleading, saying even less about Sanders' two huge wins:

"The Arizona wins handed enough delegates to Trump and Clinton to help them maintain their leads in the delegate count despite victories from their primary competitors. Bernie Sanders won morale-boosting Democratic victories in Utah and Idaho while Ted Cruz came out on top in the Utah Republican caucuses."

http://www.cnn.com/2016/03/22/politics/election-2016-arizona-utah-idaho-primaries-caucuses-highlights/index.html

From that, one would get the impression the delegate count stayed the same, when in fact it appears Sanders will gain ten or more delegates over Clinton on the night.

[noretreatnosurrender]

10. You Have to Work

awful hard to come up with a misleading piece like that. The corporate media sure are living up to our expectations, aren't they?

[Motown_Johnny]

11. The poll are slanted in Hillary's favor.

Bernie isn't picking up support at the last minute. His support is being misrepresented all along.

[paulthompson]

12. Disagree

I think the polls actually aren't that bad and he is picking up a lot of support at the last minute. Take Illinois. The polls one week out had him down 40 points, but the polls taken a day or two out had him down only a couple of points. In the end, he lost by two points.

The polls may be a little off in Clinton's favor, because Sanders has more new voters who are hard to poll, and maybe more voting enthusiasm, but they're not anywhere close to 50 points off.

I think most of what's happening is that most people don't know that much about Sanders, and Clinton has high negatives of about 55%. Nobody has ever won the presidency with negatives that high. Then, a few days before the primary in any given state, Sanders has huge rallies, runs ads, gets some media coverage, people tune in to debates and town halls, and so on. A lot of voters had been planning on reluctantly voting for Clinton as a kind of loyal Democrat default position, due to her much greater name recognition and so on, but when they get to know Sanders and his positions, many of them switch.

The problem is, he's running out of time. If it were two or three weeks between primaries instead of one week, I'm sure he'd be winning more states and by bigger margins (except probably in the South). The Democratic primaries are deliberately designed to help the establishment candidate win by having lots of big states vote in early to mid-March, so the candidate with more money and name recognition will win most of them. There were states where Sanders didn't even appear once or run any ads (mostly in the South), because there just wasn't enough time or money.

But now things are slowing down and Sanders will have more time to make his case to the remaining states. I think he'll do better as a result. (Unfortunately, it's probably still not enough for him to fully catch up though, unless some dramatic news changes the race. But I think this is a crazy election all around and dramatic developments may well happen. We'll see.)

[firebrand80]

13. These states were hardly polled at all nt

[paulthompson]

14. Yes, but....

Yes, but they were polled. And you can't say that being off by over 50 points means nothing. Clearly, a lot of people don't know much about Sanders, and when they find out more, many want to vote for him. Such huge swings in support in such a short time are truly remarkable. It may not be enough for Sanders to win, but it exposes some weaknesses that should have Clinton concerned. She's just not generating much enthusiasm.

[firebrand80]

15. We don't know that there was a "swing in support"

There's not nearly enough data to pick up such a trend. It could have just been bad polling.

There is a reason why it's better to rely on polling averages than single polls.

[paulthompson]

16. We do know!

For instance, in Utah, there actually were seven polls going back to the start of the primary election. All of them had Clinton in the lead, except the most recent one with Sanders with a narrow lead:

http://projects.fivethirtyeight.com/election-2016/primary-forecast/utah-democratic/

So at some point, you have to admit there was a dramatic shift of support towards Sanders. Either that, or polls have no meaning at all. We're not quibbling over a few points. This is over FIFTY points!

You can see the same thing in state after state. Except for the states in the South where Clinton crushed (and Sanders never visited once or ran any ads, like Alabama, Mississippi, etc), Clinton always starts out with a massive lead in the polls and then Sanders steadily narrows the gap. Sometimes he surges into the lead and sometimes he falls short, but the pattern is the same: the more people hear about Sanders, the more they want to vote for him.

[NJCher]

17. interesting analysis, paulthompson

Cher

[paulthompson]

18. Thanks

I'm surprised that not even many Sanders supporters are commenting about this. Exceeding poll results by over 50 points in two states on the same day could very well be the biggest upset in the history of presidental primary polling. It far exceeds the Michigan primary upset.