The sample size of the poll in New Hampshire was 255 people from Nate Silver, I was unaware of this.

[Demsrule86]

I am sure the Warren campaign is breathing a sigh of relief in hearing this.

Link to tweet

[cwydro]

1. I pay no attention to any of these polls at this early date.

And I doubt I will at any later date.

We all saw what happened in 2016.

[TidalWave46]

4. Polls in 16 were overall really good.

Damn near spot on in the last week of the election.

[redqueen]

11. Closer to election days yes

Further out not so much

A good portion of every poll is 'undecided' or 'don't know' - but of course highlighting the fact that things could change significantly and have in the past doesn't get clicks and shares.

[TidalWave46]

13. Your subject line is accurate.

Your second sentence is opinion with no backing.

[RainCaster]

2. My opinion of our candidates is not conditioned on surveys

I listen to them and read their statements.

[Demsrule86]

6. Polls and surveys give us feel for who is ahead however...thus I find them useful...but unlike

you, I am not undecided...I am a Biden supporter all the way...so I have but one candidate in the primary. Not that there is anything wrong with examining 'our' candidates in order to decide your preference. I have just moved beyond this in the primary. I will vote blue no matter who of course.

[LongtimeAZDem]

3. Best reply tweet:

"That’s like half of New Hampshire, though."

[Demsrule86]

5. Hahah...clearly a majority!

[Sloumeau]

9. LOL. :D

[ritapria]

7. It is a really a bad poll

The CNN/UNH NH Poll is much better …. Their sample is a much better reflection of the state's electorate …….Everybody understands that Pete is surging in NH ...We'll have to see the results of better polling than St. Anselm's work product to determine it's full extent ….

[Demsrule86]

17. I find it hard to believe that Pete will win in New Hampshire and if he doesn't one has to wonder

is this an instance of the Bradley effect? I believe Sanders will win New Hampshire with Warren and Biden quite close. I have no idea how close Buttigieg will be.

[RicROC]

8. small poll sizes should be ignored

I'm no statistician but seems to me that polls involving 255 people should be ignored, even if it does favor my choice, Mayor Pete.

Let's consider polls involving no less than 2000? people.

[thesquanderer]

12. 255 is still statistically significant in this case. The number to pay attention to here is MoE.

Please see my post #10.

[Demsrule86]

15. I don't believe it is significant for the reason detailed below...I expect either Warren or Sanders

to win or tie in New Hampshire with Biden being competitive...I believe Iowa and New Hampshire will not be conclusive this year. Nevada and South Carolina combined with super Tuesday will tell the tale...

[Demsrule86]

16. I agree and selection is important too...I don't like paid participant surveys...although

some have good ratings these last few years...too much room for bias in my opinion.

[thesquanderer]

10. Not really important, since sample size is reflected in the Margin of Error figure...

...which is more important than knowing the sample size per se.

It would be wrong for people to dismiss the poll because "its 255 people." That is baked into the MoE, and that sample size is probably the main reason the poll's MoE is on the high side of about 6. "255" tells you nothing, "6.1 MoE" is how you turn it into something meaningful. A 10 point lead with a MoE of 6 is still good news for Buttigieg (it means there's a 95% chance that Buttigieg is leading by at least 4 points).

[Demsrule86]

14. I personally would have failed my statistics classes had I used a 255 person sample

and then attempted to use the MOE to justify it...doesn't work...still a very small sample size that can not be addressed in the MOE.

[thesquanderer]

20. MoE doesn't *justify* small sample, but rather *adjusting* for it is part of its purpose.

About 250k people voted in the 2016 Dem primary. If that's the target population being sampled, than roughly 250 people out of 250 thousand is one tenth of one percent. That is a significant enough sample size to be valid for this size population. How valid you ask? That's what the MoE tells you.

The 255 number itself is meaningless without comparing it to the total population being sampled. 255 would be pointless in forecasting the general election popular vote with its 130 million voters, but it is enough to get some sense of what a total population of 250 thousand is thinking. This simply isn't an area where absolute numbers by themselves are very significant.

[boomer_wv]

18. No it doesnt.

MOE are two tailed. If you have 25 in the poll with a 6 MOE it means they are 95% sure you will be within 6 points of that, so the ci is 19-31. If 2nd place is at 15, that ci is 9-21. So a result of 19-21 is within the ci.

[Demsrule86]

19. Nicely done...

[thesquanderer]

21. True, I allowed for lowering Pete by 6, but didn't similarly raise Biden/Warren by 6. Still...

...even in that "worst case scenario," where Pete loses the full 6 of the MoE and Biden/Warren gain that full 6, Pete is still close. Thanks for the correction, but there's still no way this isn't good news for Pete.

And keep in mind that MoE is a curve. Not all figures within the 6 point MoE are equally likely, even if they all fall within 95% probability. Results closer to the reported number are more likely to be right than figures that are farther from the reported number. So the odds that two candidates both get the full six point swing in opposite directions would be one of the less likely results.

[boomer_wv]

24. Sure.

There is also a 1/20 chance the numbers are outside of the ci.

This is a good poll for Pete, but if your prior is that pete is rising in NH, but the field is still tight at the top, I dont think this should change that.

There is a loose group in this electorate and they flow around. Pete looks to be gaining among this group who was previously congregating around Warren.

I think Pete is rising now but I question, as I did with Warren, if he can maintain this support or if he reverts to his mean.

[thesquanderer]

23. Right, I addressed that...

in my post #21.