New Iowa Poll....

[boomer_wv]

There is a new Iowa poll out from Neighborhood Research and Media. It's problematic for a few reasons that I will get into.

First let's get the top line results out of the way. It has Biden leading with 23% followed by Buttigieg (17), Warren (15), Klobuchar (11), and Sanders (10).

The first sign that there might be a problem with the poll is that Donald Trump is in 6th place at 5%....in a Democratic Caucus poll of people who claim to be definite voters.

2nd is the pollster and publication. The firm is run by Rick Shaftan who is a name you might recognize from some other GOP strategy (read propaganda) firms and his history of being racist and getting fired from campaigns when people notice his racism. And the publication...Beitbart, needs no introduction.

3rd is just reading the release of the poll. It reads like something a college freshman would write after looking at data in their first Poly Sci class, and that's being generous. Essentially the poll shows no understanding of basic statistics. One example, the author describes Biden as in "Freefall" because on the first night of polling he was at about 30% and on the following night was at 20%. However, looking at n for each night, you can see both samples are pretty small, so the confidence intervals are pretty huge. More likely the first night was just, by chance, made up for a higher number of Biden supporters and the 2nd, by chance, was a little lower. This is why we don't see reputable pollsters talking about how their samples broke down by the day they sampled, because those samples are too small to be reliable. Even the total n for this poll, 300, is borderline too low to be useful.

This doesn't even delve into their sampling methods, poll design, weighting etc.

Of course, this also assumes that the poll isn't just entirely made up and given the sources is entirely possible.

I just want to post this information before anyone else sees the poll topline and starts a discussion without looking a little deeper into it. If you want to take the topline and include it in an average, I don't think that's going to cause a lot of harm, but I wouldn't look at the poll and see it as anything indicative of the actual race.

[Gothmog]

1. Joe Biden Leads by 6 in New Iowa Poll!

Link to tweet

[Thekaspervote]

4. +1000

[judeling]

2. Yes I saw it

and had the same reaction. While 538 has them rated as (B/C) take very little of it at face value.

[boomer_wv]

5. Yeah,

but that grade is based off of one poll.

[RandySF]

3. Sanders 10%

I'm not sure I buy it.

[peggysue2]

13. I'm not sure I buy that either but . . .

if internal polls show Sanders slipping then the flurry of recent attacks make a certain kind of sense.

[Iliyah]

6. The Gold Standard is . . .

Des Moines Newspaper when it comes to Iowa's polling so I understand it. That said, seems like the polling is up and down regarding NH and IA, which means there is a race between at least 4 Democratic candidates.

[boomer_wv]

12. I think

this is a best take.

We see results in good polls that don't agree with each other, we can assume the race is close.

[getagrip_already]

7. adding trump may actually make the remaining stats more reliable...

Because a lot of trump supporters end up being polled, despite the pollster best efforts. So by including trump, you draw them out and throw away their responses.

Otherwise, they might just give a ratfuck response and say bernie.

Bernie at 10%. It's possible. His dustup with warren did not win huim any new supporters, and may have cost him some.

The real poll is coming. These won't influence anyone.

[boomer_wv]

9. It doesn't work like that though.

This is a caucus, so if you show up your vote is going to count to somebody and it won't be Trump because he's not an option and he won't get 15% anywhere, so they would have to go to somebody else.

That means, at best, the 5% of Trump voters here needs to be reallocated to somebody. Or, worse, their LV screen is really badly broken and letting people in who aren't going to be voting.

[getagrip_already]

10. I was referring to the poll, not the caucus.....

The pollster added trumps name to the list of who the person supports. If that wasn't there, they would just pick someone they want to see win the primary.

These aren't likely dem caucus voters. They just find themselves on call lists.

[boomer_wv]

11. But they are supposed to be

likely democratic caucus voters. That's just it, the poll is using an LV screen than allows 1/20 of the respondents to name somebody who isn't even on the ballot. Why aren't those filtered out?



I'm just saying, despite a really good topline number for my candidate, that I wouldn't put much stock into anything this poll says unless we suddenly start seeing other polls showing the same thing.

[OilemFirchen]

8. You are correct.

Methodology notwithstanding, sometimes the messenger deserves to be shot. This one's cut and dried.