Gravis NH Poll: Sanders leads field in NH with 21% support

[LincolnRossiter]

Biden, Warren follow with 15 and 12% respectively.

[link:http://orlando-politics.com/wp-content/uploads/2019/08/New-Hampshire-August-8-2019-v2.pdf|

Obviously NH should be Bernie's strongest early state. Be interesting to see if he can get a head of steam by pulling it out in Iowa and then holding on to win in NH. I think if Biden wins in Iowa and keeps it close in NH, he'll be poised to head out of SC with all the momentum. And obviously Warren is going to be a major factor in both early contests as well.

Any idea why were aren't seeing more polling out of Nevada? I think Nate Silver asked a couple of days ago as well.

Edit: It is worth noting that this poll stands in contrast to other recent NH polls, which typically show Biden with small-modest leads. In this poll, Bernie also gets 2/3 of the black vote, which while probably inconsequential in a state with NH"s demographics, contravenes the conventional wisdom that nationally black folks are Biden's demo.

Starting to get excited about the horserace here.

[book_worm]

1. We'll have to see if this is an outliar since nearly every other poll of NH

has either Biden ahead with Warren surging. Gravis isn't considered one of the top pollers, either.

[LincolnRossiter]

2. No disagreement here...

I take every individual poll with a grain of salt and try to dig into the numbers to see what could produce anomalies. In this case the 67% black support stood out.

But I also assiduously try to avoid falling into the "fake polls" trap that so many of my idiot friends on the right are caught in. At the end of the day, a single poll is a small set of data points capturing one snapshot in time. The aggregates and trends matter much more.

[OKNancy]

3. C+ poll rating - online polling n/t

[bluewater]

9. Same as the C+ online Harris Poll that shows Biden leading nationally? Yes.

DU-ers constantly report online poll results from relatively low rated pollsters.

Harris C+ online

MorningConsult B- online

YouGov B online

SurveyMonkey D- online

Change Research C+ online

These online polls survey the same pool of voters over and over and over, introducing a systemic bias into their results.

They are not truly random polls of the public at large.

When I found this out about polls like these, I stopped posting their results and started giving them much less credence.

Not surprisingly, people often seem to point out a poll being low rated and online only when it's not good news for a candidate they prefer.

[LincolnRossiter]

15. You seem pretty interested in pollster ratings, so you should know that they aren't strictly based..

Upon their predictive accuracy. They also account for methodological soundness and integrity.

Now gravis is a relatively middling C+ rated poll and has only been accurately called races 67% of the time.

If you scan the list though, you'll see an F-rated poll (Strategic Vision) that was accurate 88% of the time.

Survey Monkey is a D- rated poll that was accurate 83% of the time.

Monmouth U is an A+ rated poll that was accurate (wait for it) 80% of the time.

I don't thinks these ratings mean what you think.

[link:https://projects.fivethirtyeight.com/pollster-ratings/|

[bluewater]

17. 538 Ratings are based on the historical accuracy and methodology of each firm's polls.

That is at the top of 538's page on pollster ratings.

https://projects.fivethirtyeight.com/pollster-ratings/

The 538 Pollster ratings are 538's opinion of the relative merits of the pollsters' accuracy and methodology.

THAT is what I think the ratings mean, it's what 538 thinks their ratings mean.




If you think 538 isn't taking into account a pollster's "accuracy" enough, feel free to take it up with 538.

Pollsters' accuracy is often inflated when pollsters "herd" their final results to match that of other pollsters.

538 wrote an article about this:

Here’s Proof Some Pollsters Are Putting A Thumb On The Scale
By Nate Silver

[snip]
Perhaps you could construct some rationale, apart from herding, for why the polls behaved this way. Maybe it became easier to predict who was going to vote and that made methodological differences between polling firms matter less. As a more technical matter, the volume of polling increased as the election approached; this presents some complications, which I address in the footnotes.10

But there are two dead giveaways that herding happened. One is the unusual shape of the curve. Rather than abiding by a linear progression, it suddenly veers toward zero in the final week or so of the campaign.

https://fivethirtyeight.com/features/heres-proof-some-pollsters-are-putting-a-thumb-on-the-scale/

[LincolnRossiter]

18. Well that's a good explanation...

Pollsters' accuracy is often inflated when pollsters "herd" their final results to match that of other pollsters

I did just want to point out to you that a pollster's rating isn't strictly a function of accuracy in calling races. I even mentioned that Gravis (the poll we've been chewing over) isn't especially well rated nor is it especially accurate. But there are poorly-rated polls that boast high accuracy (herding almost certainly been a factor for most) and there are well-rated polls, like ABC-WAPO (A+), that boast pretty mediocre predictive results (78% in their case).

My broader point is that one shouldn't just jump into a thread that simply displays the results of a poll with the banal "crappy poll, doesn't count" remarks. That's Trumpian. Nate still incorporates all polls, save the F ones, because he considers them useful in capturing the big picture.

And I consider the matter closed. It's a poll. It might paint an accurate picture of where the race stands in NH at this time, or it might not. They do hit and miss--even the A-rated ones.

[qazplm135]

4. it's early

the next 2, 3, 4 debates will move the polling significantly. It could create a large Biden lead, or it could push Warren or Sanders into the lead, or it could bring Harris back up into 2nd or 3rd, etc.

I am not really excited until I see where the polls are in early November.

[LincolnRossiter]

7. I'm not so sure about that.

the next 2, 3, 4 debates will move the polling significantly

In the age of 24hour news (which is mostly political news) and social media, I think we may be according these debates a good bit more weight than they actually merit. Politically-attuned voters like us don't shift our votes for such superficial displays. I thought Biden did horribly in the first debate and just okay in the second. I though Warren, Castro, and Buttigieg were outstanding in both and that Harris was outstanding in the first. I didn't randomly change my vote, however, just because some candidates presented better in some 10-man WWE-style "debate."

I bet most people who are tuning in one-year out from the election feel the same way. And on the other hand, most casual voters will be barely tuned in by September of next year. So they aren't shifting around much either.

[Thekaspervote]

11. Agree..don't think the debates will have earth trembling moves in the polls. Maybe some shift

But not Enough to truly change the dynamics

[qazplm135]

14. no one is paying attention right now

except political nerds like the folks on here.

They will start paying more attention as we get closer to the primaries.

Where Obama was in June July 07 was not where he was in Nov 07.

Most people don't have hard thoughts on who they support, again unlike the political nerds we are on here.

There support is soft for the most part, whether it is Biden, Warren, Bernie or anyone else.

I get it, if no one changes, Biden wins, and that's what you want...and maybe that happens.
But it's about as likely that there are plenty of changes...and Biden could still win in that scenario...or not.
We don't know yet.

[11cents]

5. Gravis is crap.

There's been an explosion of low-quality polling operations, especially at the state level.

[crazytown]

6. The Gravis (C+) poll is an outlier

but I wouldn't give too much credence to anyone of them - particularly Morning Consult, and HarrisX with their chained online panels that report the same results week after week.

[LincolnRossiter]

8. Well if you don't pay attention to or place any stock in polls, I suppose that's your right...

Just don't be the guy or gal who cheers every poll that you like and then throws up 538's somewhat arbitrary grading of the ones you don't. It's just one data point. Maybe it's horribly off. Maybe it's not. At the end of the day the only polls that matter are elections. But let's not join the "fake polls" crowd.

[crazytown]

10. I said TOO much attention

To my knowledge the Gravis poll is the first since VP Biden's announcement that has Sanders in the lead in any State. As for Morning Consult & HarrisX, they do report the same results (+/- 1) every week. Sanders has polled 19% (+/-1) since May which contrasts strongly with polls that don't employ chained panels.

[LincolnRossiter]

12. At the risk of sounding argumentative, what's the right amount of attention?

You know just the right amount. Not too much, not too little. Because I think my posts were very qualified.

[crazytown]

16. I go by the RCP Averages for the nation

as do 538 - as a ready reckoner.

RCP (8/13) JB 30.8, EW 18.0, BS 16.8, KH 8.3, PB 6.3

And A rated polls for individual states

Eg. A+ Monmouth IA (8/8) JB 28, EW 19, KH 11, BS 9, PB 8

As I said, the Gravis poll is an outlier: To my knowledge, it is the first poll, of any rating, of any State since VP Biden's announcement that has Sanders in the lead.

[jimmy the one]

19. Rasmussen is a c.r.a.p. poll which pads trump

I go by the RCP Averages for the nation as do 538 - as a ready reckoner.

Not sure I follow you on real clear politics (RCP) averages equating with 538 - I'm presuming poll averages of all recent polls (yes or no?), including disreputable polls like rasmussen, which strategically pads its polls to the right. 538 is not a poll, it is an 'analysis' of polls, like an oddsboard at a racetrack - 3/5 hillary, 4-1 trump, field 1000 - 1; 4-1 shots come in regularly at tracks.
Coincidentally I just wrote of this last night:

.... the real clear politics (RCP) average includes rasmussen, which strategically pads its trump approval polls, by plus 5 - 8 points it seems.
This skews the RCP average in two ways. By the padding itself, and because rasmussen's poll is (apparently) included in every single RCP poll average, while other reputable pollsters are subject to about a 2 to 3 week dropback and then dropped from the up to date poll average. This in itself gives rasmussen much more influence in the RCP average than reputable live pollsters which poll only once a month or longer.

Rasmussen is a daily tracking poll which averages 3 consecutive days of its polls into a rolling avg, updating the next day while dropping the last day. Rass is C.R.A.P. poll = computer responsive automated polling - ring ring, hello, push one for gore, two for bush, thank you bye. AKA IVR.
Rass does not post a daily poll on saturday or sunday but even so it is included in RCP avg every day, since the weekend does not make it drop out, due the 2 - 3 week dropback.

Note below in the RCP link the expanded view going back months, how Quinnipiac is about to drop off after 3 weeks, Gallup is not listed in the current rolling average, nor is pbs, nbc or PPP (nor fox!). Do not be fooled that rasmussen is not mentioned in the monthlies dropback either, it has some sort of special RCP privilege because it is up every day.
Well, RCP is pretty much right wing, but how they justify including right wing republican scott rasmussen's strategery is beyond fairness & reason. (I think scott left, but I still smell a residual stink in rasmussen).
(Rass strategically pads by counter padding some republicans less, or some dems more, generally in foregone races but not always, so as to try to leave a net zero to pretend to be 'unbiased').
Upshot is, the RCP average needs be tweaked by subtracting 0.5 to 1.0 pts, to counter the 'Rasmussen Affect'.
https://www.realclearpolitics.com/epolls/other/president_trump_job_approval-6179.html

Note: Harris, which is somehow a rasmussen affiliate, it seems polls about every 3 weeks, so it will appear on almost every RCP avg as well. I think this is a planned strategy. I think some dem leaning polls are aware of this strategy as well. Ras & Harris have identical ratings, C+ which is below par on 538, & both with a 1.5 republican bias.
Survey monkey is at D- & a 4.5 or so dem bias, but survey monkey does not appear on RCP poll of polls.

[Thekaspervote]

13. It's not surprising or earth shaking that sanders is now polling 1st in NH. He won the primary there

In 2016. One would be really surprised if he didn’t, Warren too being from the NE

[NYMinute]

21. You mean NE and not NW right?

As in North-East or New England?

[Thekaspervote]

22. Oh duh!! Thx for pointing that out!

[NYMinute]

20. Point, click and vote lol