With regard to Illinois and other state polls

[BernieforPres2016]

Why do we still pay attention to them after Michigan, particularly in any state with open primaries?

In Michigan, Bernie won 71-28 among independents and Hillary won 57-41 with Democrats. Bernie has been winning by huge margins among under 30 voters and Hillary has been winning by huge margins among African Americans, but only a little over 2:1 in Michigan versus 5:1 and greater in the deep South.

So unless a pollster can predict voter turnout, or at least the percentage of voters that will fall into each of these groups that go heavily toward one candidate or the other, how can they possibly produce an accurate poll?

Bernie was on This Week (ABC) this morning. He said that if voter turnout was large, he expects to do very well on Tuesday in Illinois, Ohio, Missouri and North Carolina. He said Florida will be a tougher state. Florida is a closed primary and no independents is bad for Bernie. And it skews old, which is a Hillary demographic.

I am hoping that Illinois could not just be a victory, but a decisive victory for Bernie. I think Hillary's linkage with the enormously unpopular Chicago governor Rahm Emanuel could really hurt Hillary with African American voters where she needs to win big.

[revbones]

1. At least Bernie is pulling ahead in Illinois now

[BernieforPres2016]

5. My point is that the polls in open primary states aren't going to be accurate

Unless the pollsters can predict the mix of voters. And if they can't predict the total number of voters, how can they possibly get the mix right? Bernie keeps talking about needing large turnouts because first time voters (who will be underrepresented in polls of likely voters) will go heavily in his favor.

[Metric System]

2. The odds of every poll in every other state being as off as Michigan are slim.

[Kalidurga]

3. I don't think closed primaries hurt Bernie as much as conventional wisdom says.

All states have provisions for people to switch parties. All states allow people to register to vote for the first time at some point before the election. People who are activists for Bernie know the ins and outs of their state and each one is different, they act accordingly. Bernie is getting new voters end stop No matter what state the voting is taking place in.

[Attorney in Texas]

4. The polls AREN'T all over the map wrong. They are either correct or wrong in favor of Hillary

outside of the Bible Belt (in the Bible Belt, the polls are all over the map). As a result, the polls outside of the Bible Belt do tend to set a fairly accurate ceiling on Hillary's support and a floor on Sanders' support.

[BernieforPres2016]

7. The most accurate poll in Michigan had Bernie losing by 5% instead of winning by 2%

Every other poll had Bernie losing by double digits, many by over 20%. I will try to assemble the latest state polls for Florida, North Carolina, Ohio, Illinois and Missouri tomorrow going into Tuesday primaries and let's see how accurate they turn out to be. My guess is, not very.

[Attorney in Texas]

9. Right. That poll projected Hillary's support at 52% with a 4% margin of error and so it correctly

predicted her ceiling.

[BernieforPres2016]

10. That was one poll of 7 or 8

Every other poll had Hillary in the high 50's to mid 60's.

[Attorney in Texas]

12. Right. What I'm seeing is that if you look at Hillary's polling in the last weeks before the vote,

she rarely does better on election day than her range of results set by the margin of error for her worst poll (outside of the Bible Belt, Hillary's worst polls seem to set a ceiling on her election day performance).

In contrast, Sanders usually performs within the margin of error of the best polls for him. That is, polling outside of the Bible Belt taken in the last two weeks before the contest seems to set a floor on Sanders' election day support.

[BernieforPres2016]

13. I suspect they're underestimating the number of independents and how heavily they go to Bernie

And also underestimating the turnout of young voters.

A comparison of the polls for the next 5 states versus the results will tell us if the pollsters have caught up with what's going on. I don't know that it can be captured.

[Attorney in Texas]

14. The polls are (1) over-estimating Hillary's support because the lack of enthusiasm is hurting her

election day performance as compared to her polling performance, (2) over-estimating the relationship between the views of landline phone poll participants and the election day results because those who still use a landline are unnaturally resistant to change and therefore are more inclined toward Hillary's establishment status quo message over Sanders' transformative change message, (3) under-estimating how many "conservative Democrats" are deciding to vote in the Republican primary and these would be Hillary supporters if they had voted in our primary, (4) under-estimating first-time-voting Democrats who are breaking for Sanders in a huge way, and (5) under-estimating independent voting in favor of Sanders.

The pollsters will not catch up; they are committed to a model which they won't revist until after the primary. But we can note which polls are consistently wrong by a mile (for example, ALL of the robo-call pollsters) and just stop paying attention to those pollsters.

[Gwhittey]

6. Most of the polls base call sheets off Voters form 2008

So too many of lists focus on middle to old age range. In past Elections the younger crowd has been a no factor so polls now are not even bothering trying to correct for this. But Sanders has gotten the younger people fired up to vote, and add in the fact that for first time in decades the young voting block now outnumbers to boomers.

[BernieforPres2016]

8. And that brings in the issue of people without land lines, people who screen calls with call waiting

My guess is that it's a lot easier to reach older people (though not a lot of them like me) than younger people. They can work harder to reach enough young people to get the percentages they want, but the percentages depend on what they think the age mix of voters will be and that is a guess.

[lasttrip]

11. Polls are fun!

?w=604
Peace.
LT