NYT LIVE POLL THREAD:

[Florida Man for Crime]

My opinion is that even if Hurd holds on this year, he has a good chance of being knocked off in 2020.

Yeah, he is more likely to lose with Presidential turnout. That was always going to be the case with certain districts (especially ones where the Dem base is more prone to drop off in midterms, especially Hispanics). Remember that in 2008, for example, Dems picked up additional seats that they failed to get in 2006. Although both 2006 and 2008 were good Dem years, this was in no small part due to differences in turnout patterns between Presidential years and Midterm years. If 2020 is a good Dem year, as seems fairly probable with Trump presumably on the ballot, there is a good chance that we see this same sort of 2006-->2008 phenomenon.

[Zaybay]

I know this thread was kinda derailed, partially by me, but Im looking at the TX-23 poll, and I noticed the fact that around 90% of the voters are coming from the suburbs of San Antonio. Is this district really that unevenly populated?

[wbrocks67]

Hm, idk. Primary turnout is usually *somewhat* indicative in ways, and the primary turnout in TX-23 was pretty lopsided (45k Dems vs. 31k Republicans) -- that's a substantial lead for Dems, so I don't really buy that this is a low turnout district in this case.

I'll wait for more to come in, but Hurd barely won in 2014 and 2016. There's no reason to believe that he suddenly became more popular in the past 2 years.

[Holy Unifying Centrist]

Hm, idk. Primary turnout is usually *somewhat* indicative in ways, and the primary turnout in TX-23 was pretty lopsided (45k Dems vs. 31k Republicans) -- that's a substantial lead for Dems, so I don't really buy that this is a low turnout district in this case.

I'll wait for more to come in, but Hurd barely won in 2014 and 2016. There's no reason to believe that he suddenly became more popular in the past 2 years.

He was a first term congressman in 2016. He beat an incumbent in 2014.

Hurd has had multiple high profile attacks on Trump over immigration, health care, and Russia. He had a highly publicized road trip to DC with Beto.

Why can't atlas admit that Hurd has become a lot more popular? Is it because he's black?

[Florida Man for Crime]

Dear Zaybay, you posted this response 6-7 minutes after I submitted my post.

That was most certainly nothing like enough time to carefully read my (long) post, check/read the NYT/Siena methodology descriptions in light of my post, and type this response.

There is such a thing as truth and falsity, and you are continuing to say things that are simply factually correct. Do your self a favor and take the time to read and understand, research, and double check that there is a  before you simply repeat the same thing you said before.

Alright, a bit long, but you still misunderstand what I mean.

I will say it as clearly as possible, so we are all on the same page.

My problem with these polls is threefold
1. The polls have a rather low sample of 500 voters

This has nothing to do with 2014, nor is sampling error something that is somehow unique to Siena/NYT. Non-sequiter. Good luck getting lots of house polls with huge samples. For any poll, regardless of who does it, you should always interpret it in line with the sample size/MOE.

You must be logged in to read this quote.

This is just straight up false (and also totally different from the 2014 issue, stop changing the subject again). You have not taken the time to carefully read and understand their methodology. Do so.

You must be logged in to read this quote.

Again, this is simply false. When you repeat something that is false, that doesn't suddenly make it true. If you want to comment intelligently on these polls, go read and understand the articles on the NYT site explaining the methodology please.

[Florida Man for Crime]

I know this thread was kinda derailed, partially by me, but Im looking at the TX-23 poll, and I noticed the fact that around 90% of the voters are coming from the suburbs of San Antonio. Is this district really that unevenly populated?

Welcome to TX-23 (although it is an exaggeration to say 90%, only 46% of the sample is from Bexar county so far).

[Zaybay]

Again, dont want to clutter

Well, its mostly because, as I said in my post, you misunderstood me.Most of the post is about how they call people who vote in 2014 and those who have not, which has nothing to do with what Im saying. I am not saying that they are calling people that voted in 2014, or a similar number. Im talking proportions. As I said, the polls use almost the exact same voter proportions as the 2014 elections, which was an R favoured year.

The outline for these polls clearly states this, that

You must be logged in to read this quote.

And if you think I am misinterpreting this, Im not, for the NY polls, done by Siena, are doing exactly what Ive been saying.

Take the NY-22 poll, as an example, what was the sample used, lets say, in a partisan sense. It was 32% D, 47% R, and 19% I. Guess what it was in 2014? It was 31%D, 49% R, and 18% I. And thats not even touching the race and gender demographics.

And this isnt cherrypicking. Hopping over to Dana Balter's district, we get 34% Democrats, 38% Republicans, and 24% Indies. What were the turnout demos in 2014? Almost the exact same, and, in fact, more D heavy, with 35% D, 36% R, and 25% I.

I could go on, but this is the truth, the pollster Siena does use 2014 to guide its baseline, and, as I said before, thats my problem with this.

[Zaybay]

I know this thread was kinda derailed, partially by me, but Im looking at the TX-23 poll, and I noticed the fact that around 90% of the voters are coming from the suburbs of San Antonio. Is this district really that unevenly populated?

Welcome to TX-23 (although it is an exaggeration to say 90%, only 46% of the sample is from Bexar county so far).

Makes sense, that is the border of TX for ya. Though I do believe that a district not including the SA suburbs and just the rurals would be likely/safe D, though I forget how to draw it.

[Dr. Arch]

WI-01, LET'S GO!

[Zaybay]

WI-01, LET'S GO!

Interesting, Im going to say the R leading by around 6-8

[GeorgiaModerate]

Complaining that the sample size for a single CD is around half of what we'd get for a statewide or national poll is... an interesting media strategy.

Like, Nate Silver understands basic error band calculation, right? Is he just being opaquely misleading for his dumber followers?

Anyway, these polls all seem about near expectations so far. I don't know what the big fuss is.

I think Nate's just trying to point out that these CD polls are inherently going to be more error prone than ones from larger jurisdictions typically are.

[Antonio the Sixth]

Complaining that the sample size for a single CD is around half of what we'd get for a statewide or national poll is... an interesting media strategy.

Like, Nate Silver understands basic error band calculation, right? Is he just being opaquely misleading for his dumber followers?

Anyway, these polls all seem about near expectations so far. I don't know what the big fuss is.

I think Nate's just trying to point out that these CD polls are inherently going to be more error prone than ones from larger jurisdictions typically are.

But that's not how statistical error works. A smaller population naturally demands a smaller sample size for a given error bound. He's being deliberately misleading by implying that the relatively-smaller absolute sample size lessens the significance, unless he's really much less versed in his own industry than we've all assumed. (I won't rule out the latter possibility. He doesn't impress me.)

You're the one who doesn't get how statistical error works. The predictive power of sample size has nothing to do with the size of the target population. A 1000-person poll of the entire world population would be more accurate than a 500-person poll of a mid-sized city, provided that sampling is truly random (of course, sampling is not actually random, but that's equally true of House polls as it is of State polls - and weighting tends to make smaller sample sizes even more noisy if anything).

[Florida Man for Crime]

A smaller population naturally demands a smaller sample size for a given error bound.

To some extent. But once you reach large population sizes, the effect of increasing the population size further begins to level off quite substantially.

For example, a 500-sized sample of a population of 100,000 has an MOE of 4.37.

And a 500-sized sample of a population of 10,000,000 (100 times higher population) has an MOE of 4.38.

Hardly any difference between a 4.38 and 4.37 MOE, though technically of course there is a small difference.

[GeorgiaModerate]

Complaining that the sample size for a single CD is around half of what we'd get for a statewide or national poll is... an interesting media strategy.

Like, Nate Silver understands basic error band calculation, right? Is he just being opaquely misleading for his dumber followers?

Anyway, these polls all seem about near expectations so far. I don't know what the big fuss is.

I think Nate's just trying to point out that these CD polls are inherently going to be more error prone than ones from larger jurisdictions typically are.

But that's not how statistical error works. A smaller population naturally demands a smaller sample size for a given error bound. He's being deliberately misleading by implying that the relatively-smaller absolute sample size lessens the significance, unless he's really much less versed in his own industry than we've all assumed. (I won't rule out the latter possibility. He doesn't impress me.)

Not at the population sizes we're talking about.  The average CD has what, 700K people?  Let's guess that there are 300K voters.  A sample size of 250 is going to have about the same MoE for a population that size or one much bigger.  The absolute sample size DOES make the difference.  With a sample size of 250, MoE is about +/-6.2% whether the population is 300K or 30 million.  With sample size of 500, MoE is about 4.4% for either population.  With sample size of 1000, it's about 3.1%.  The difference in MoE created by the population size differences is insignificant compared to that created by the sample size differences.

[Florida Man for Crime]

Complaining that the sample size for a single CD is around half of what we'd get for a statewide or national poll is... an interesting media strategy.

Like, Nate Silver understands basic error band calculation, right? Is he just being opaquely misleading for his dumber followers?

Anyway, these polls all seem about near expectations so far. I don't know what the big fuss is.

I think Nate's just trying to point out that these CD polls are inherently going to be more error prone than ones from larger jurisdictions typically are.

But that's not how statistical error works. A smaller population naturally demands a smaller sample size for a given error bound. He's being deliberately misleading by implying that the relatively-smaller absolute sample size lessens the significance, unless he's really much less versed in his own industry than we've all assumed. (I won't rule out the latter possibility. He doesn't impress me.)

You're the one who doesn't get how statistical error works. The predictive power of sample size has nothing to do with the size of the target population. A 1000-person poll of the entire world population would be more accurate than a 500-person poll of a mid-sized city, provided that sampling is truly random (of course, sampling is not actually random, but that's equally true of House polls as it is of State polls - and weighting tends to make smaller sample sizes even more noisy if anything).

The size of the population does have an effect, just a fairly small one when you get up to large populations. You can use this calculator:

https://americanresearchgroup.com/moe.html

To see a significant effect, compare using relatively small populations.

For example, a sample of 500 with a population of 1,000 is 3.1 MOE.

Whereas a sample of 500 with a population of 10,000 is 4.27 MOE.

https://americanresearchgroup.com/moe.html

[Antonio the Sixth]

Complaining that the sample size for a single CD is around half of what we'd get for a statewide or national poll is... an interesting media strategy.

Like, Nate Silver understands basic error band calculation, right? Is he just being opaquely misleading for his dumber followers?

Anyway, these polls all seem about near expectations so far. I don't know what the big fuss is.

I think Nate's just trying to point out that these CD polls are inherently going to be more error prone than ones from larger jurisdictions typically are.

But that's not how statistical error works. A smaller population naturally demands a smaller sample size for a given error bound. He's being deliberately misleading by implying that the relatively-smaller absolute sample size lessens the significance, unless he's really much less versed in his own industry than we've all assumed. (I won't rule out the latter possibility. He doesn't impress me.)

You're the one who doesn't get how statistical error works. The predictive power of sample size has nothing to do with the size of the target population. A 1000-person poll of the entire world population would be more accurate than a 500-person poll of a mid-sized city, provided that sampling is truly random (of course, sampling is not actually random, but that's equally true of House polls as it is of State polls - and weighting tends to make smaller sample sizes even more noisy if anything).

The size of the population does have an effect, just a fairly small one when you get up to large populations. You can use this calculator:

https://americanresearchgroup.com/moe.html

To see a significant effect, compare using relatively small populations.

For example, a sample of 500 with a population of 1,000 is 3.1 MOE.

Whereas a sample of 500 with a population of 10,000 is 4.27 MOE.

https://americanresearchgroup.com/moe.html

What's the theory behind it? That doesn't match what I learned in class about the frequentist approach to stats and the central limit theorem, but maybe pollsters are using a different framework?

[Florida Man for Crime]

Well, its mostly because, as I said in my post, you misunderstood me.Most of the post is about how they call people who vote in 2014 and those who have not, which has nothing to do with what Im saying. I am not saying that they are calling people that voted in 2014, or a similar number. Im talking proportions. As I said, the polls use almost the exact same voter proportions as the 2014 elections, which was an R favoured year.

The outline for these polls clearly states this, that

You must be logged in to read this quote.

And if you think I am misinterpreting this, Im not, for the NY polls, done by Siena, are doing exactly what Ive been saying.

Read the wikipedia article I linked to earlier on stratification. You seem not to understand what stratification is.

Also, previous Siena polls of NY are not using the same methodology and are pretty much irrelevant to these polls.

And re: comparing samples of the polls composition by various demographics, if you want to make a comparison, you need to post a link. And no, a poll from 2014 is not the same thing as the actual demographics of the 2014 electorate.

Because they are using the voter file, they can see exact and accurate data on the demographics (at least some demographics such as age/gender/party registration, while other data such as race often has to be modeled and is therefore somewhat less accurate) of the 2014 electorate which you do not have access to because you do not have a copy of the relevant voter file.

So your statements seem to continue to be based upon falsities and misunderstanding.

[Florida Man for Crime]

What's the theory behind it? That doesn't match what I learned in class about the frequentist approach to stats and the central limit theorem, but maybe pollsters are using a different framework?

https://en.wikipedia.org/wiki/Margin_of_error#Effect_of_population_size

Sometimes an approximation whereby the population is assumed to be the same as though it were infinitely large is used. That is a reasonably good approximation in some circumstances, but only when the population is indeed quite large (see the example numbers I posted above where it makes relatively little difference for large populations, but relatively more difference for small populations)

Frequentist vs. Bayesian methodology wouldn't really make much difference here, I don't think.

[Antonio the Sixth]

What's the theory behind it? That doesn't match what I learned in class about the frequentist approach to stats and the central limit theorem, but maybe pollsters are using a different framework?

https://en.wikipedia.org/wiki/Margin_of_error#Effect_of_population_size

Sometimes an approximation whereby the population is assumed to be the same as though it were infinitely large is used. That is a reasonably good approximation in some circumstances, but only when the population is indeed quite large (see the example numbers I posted above where it makes relatively little difference for large populations, but relatively more difference for small populations)

Frequentist vs. Bayesian methodology wouldn't really make much difference here, I don't think.

Oh, I see. That makes sense. Still, making this assumption could only bias the margin of error upwards, which means it's a fairly reasonable conservative assumption to make. And the gain in precision from estimating a finite population is so small that I'm not sure why pollsters even bother (or why AMB1996 thinks it's so damning that Silver didn't mention it). The effect of the non-randomness of samples, and subsequent need for weighting, is far, far more significant.

[Zaybay]

Well, its mostly because, as I said in my post, you misunderstood me.Most of the post is about how they call people who vote in 2014 and those who have not, which has nothing to do with what Im saying. I am not saying that they are calling people that voted in 2014, or a similar number. Im talking proportions. As I said, the polls use almost the exact same voter proportions as the 2014 elections, which was an R favoured year.

The outline for these polls clearly states this, that

You must be logged in to read this quote.

And if you think I am misinterpreting this, Im not, for the NY polls, done by Siena, are doing exactly what Ive been saying.

Read the wikipedia article I linked to earlier on stratification. You seem not to understand what stratification is.

Also, previous Siena polls of NY are not using the same methodology and are pretty much irrelevant to these polls.

And re: comparing samples of the polls composition by various demographics, if you want to make a comparison, you need to post a link. And no, a poll from 2014 is not the same thing as the actual demographics of the 2014 electorate.

Because they are using the voter file, they can see exact and accurate data on the demographics (at least some demographics such as age/gender/party registration, while other data such as race often has to be modeled and is therefore somewhat less accurate) of the 2014 electorate which you do not have access to because you do not have a copy of the relevant voter file.

So your statements seem to continue to be based upon falsities and misunderstanding.

Alright, Im sick of this, its clear you wont listen to what Im saying. The thread has become too polluted because of this,and I would rather this thread talk about the NY polls, not some dumb debate on methology.

I will back off.

[IceSpear]

Ugh, why are they wasting time with a safe R district like WI-01? I don't care how badly the Iron Deadbeat Jailbird gets BTFO by Paul Ryan's mini me.

[Politician]

Ugh, why are they wasting time with a safe R state like Alabama? I don't care how badly Doug Jones gets BTFO by Jeff Sessions' mini me.

Also, only IceSpear would believe candidate quality is unimportant and fundamentals are king, then think a nominee being arrested for DUI makes him "unelectable". George W Bush had been arrested for DUI.

[IceSpear]

Ugh, why are they wasting time with a safe R state like Alabama? I don't care how badly Doug Jones gets BTFO by Jeff Sessions' mini me.

Also, only IceSpear would believe candidate quality is unimportant and fundamentals are king, then think a nominee being arrested for DUI makes him "unelectable". George W Bush had been arrested for DUI.

Wow, that was quite an accusation against Jeff Sessions.

A fundamental of this district is that it is extremely Republican. The Republican Supreme Court candidate carried it even while he was losing statewide by double digits, and Paul Ryan is very popular there. The Iron Deadbeat Jailbird's "colorful" past only takes matters from bad to worse.

Also, the last minute DUI revelation was the only reason 2000 was close.

[Florida Man for Crime]

Also, the last minute DUI revelation was the only reason 2000 was close.

Nah, it is because Al Gore started wearing earth tones and made his populist💓 pitch.

http://www.cnn.com/ALLPOLITICS/time/2000/08/28/fight.html

[Holy Unifying Centrist]

Also, the last minute DUI revelation was the only reason 2000 was close.

Nah, it is because Al Gore started wearing earth tones and made his populist💓 pitch.

http://www.cnn.com/ALLPOLITICS/time/2000/08/28/fight.html

Wow, politics was so incredibly lame back then.