So are Pollsters under sampling Hispanics a major issue?

[swf541]

Been thinking about this a lot this morning, with the Latino targeted polling over the last couple of weeks showing such a massive Clinton victory and a higher interest in voting, are national and some state polls under sampling Hispanic turnout?
for example we had the odd Florida poll with what was it 7% or so Hispanic and I also noticed the Reuters tracking poll is only 7% African American with Clinton only getting 61% which doesnt seem to make much sense.

[Classic Conservative]

Well, with Florida, I would think that would be Trump's best state with Hispanics.

[swf541]

Well, with Florida, I would think that would be Trump's best state with Hispanics.

Thats an argument but I am more arguing about the percentage of them than the poll than whom they are voting for.

[Coolface Sock #42069]

Well, with Florida, I would think that would be Trump's best state with Hispanics.

You're forgetting Texas.

[afleitch]

Yes. Latino focused polling was very accurate in 2012. If it picks up high turnout it 'locks' in otherwise swing states for Clinton which the state polling might otherwise miss.

[DINGO Joe]

Well, with Florida, I would think that would be Trump's best state with Hispanics.

You're forgetting Texas.

He's saying that Florida will be Trump's best performance with Hispanic voters, thanks to the Cubans

[ProudModerate2]

In my opinion, it's not a "major" issue, but probably a minor problem in both national and some state polling.

[Xing]

In some states, yes, but it's not just about undersampling Latinos. In some states, they specifically undersample Latinos who speak Spanish as their native language by only conducting interviews in English.

[dspNY]

Yes, especially in states with a significant Latino population

[Virginiá]

Been thinking about this a lot this morning, with the Latino targeted polling over the last couple of weeks showing such a massive Clinton victory and a higher interest in voting, are national and some state polls under sampling Hispanic turnout?
for example we had the odd Florida poll with what was it 7% or so Hispanic and I also noticed the Reuters tracking poll is only 7% African American with Clinton only getting 61% which doesnt seem to make much sense.

Well yes, if a Florida poll severely undersamples minorities like that, then its results will probably be very different than the actual election result. In Florida, African Americans made up 13% of the electorate in 2008/2012. Hispanics made up 14% in 2008 and 17% in 2012. Further, the share of the electorate that is white college educated has been going up and white non-college going down, consistent with national trends.

I haven't seen that poll but if you oversample demographics favorable to Republicans then you'll get a skewed Republican result. Just like it being skewed for Democrats if they were to use a sample that was 20% African American and 30% Hispanic, and so on.

Further, the Hispanic share of the FL electorate is going to be more than in 2012, again, consistent with historical trends.. So that needs to be taken into account, in addition to accurately polling that demographic - which means not only using landline-using English-speaking Hispanic respondents.

=

Projected electorate demographics of Florida (2012):

Non-college whites:           33.7%
College-educated whites:  32.7.%
African American:              12.9%
Hispanic:                            16.8%
Asian:                                 4%

Projected electorate demographics of Florida (2016):

Non-college whites:           30% (-3.7)
College-educated whites:  33.9% (+1.2)
African American:              13.4% (+0.5)
Hispanic:                            18.3% (+1.5)
Asian:                                 4.4% (+0.4)


So the makeup of the electorate matters substantially and people should account for changes in the electorate.

[SillyAmerican]

Been thinking about this a lot this morning, with the Latino targeted polling over the last couple of weeks showing such a massive Clinton victory and a higher interest in voting, are national and some state polls under sampling Hispanic turnout?
for example we had the odd Florida poll with what was it 7% or so Hispanic and I also noticed the Reuters tracking poll is only 7% African American with Clinton only getting 61% which doesnt seem to make much sense.

Well yes, if a Florida poll severely undersamples minorities like that, then its results will probably be very different than the actual election result. In Florida, African Americans made up 13% of the electorate in 2008/2012. Hispanics made up 14% in 2008 and 17% in 2012. Further, the share of the electorate that is white college educated has been going up and white non-college going down, consistent with national trends.

I haven't seen that poll but if you oversample demographics favorable to Republicans then you'll get a skewed Republican result. Just like it being skewed for Democrats if they were to use a sample that was 20% African American and 30% Hispanic, and so on.

Further, the Hispanic share of the FL electorate is going to be more than in 2012, again, consistent with historical trends.. So that needs to be taken into account, in addition to accurately polling that demographic - which means not only using landline-using English-speaking Hispanic respondents.

=

Projected electorate demographics of Florida (2012):

Non-college whites:           33.7%
College-educated whites:  32.7.%
African American:              12.9%
Hispanic:                            16.8%
Asian:                                 4%

Projected electorate demographics of Florida (2016):

Non-college whites:           30% (-3.7)
College-educated whites:  33.9% (+1.2)
African American:              13.4% (+0.5)
Hispanic:                            18.3% (+1.5)
Asian:                                 4.4% (+0.4)


So the makeup of the electorate matters substantially and people should account for changes in the electorate.

Yes, like Virginia says, if Florida polls are under sampling minorities, then the poll results will probably be very different than the actual election results.

That said, perhaps Dem strategists are (or should be) a bit concerned about Trump’s 19-percentage-point margin of victory over hispanic home grown candidate Marco Rubio during the GOP primary. Trump gathered up votes from dormant Republicans casting first time ballots, and he managed to grow the ranks of the Republican Party of Florida by the tens of thousands. Again, he managed to do this while running against a home state senator. Trump's victory was solid: Rubio lost every county except Miami-Dade. Will Hillary Clinton have a better chance of winning Florida than Rubio? Will she manage to do better with hispanics than a hispanic? We'll see, but I have to say, I have my doubts...

And yes, I understand that the primary results were only Republicans. But the registration numbers for 2016 as reported by the Florida Department of State are as follows:

  Republicans:   4,396,609
  Democrats:   4,659,801
  Other:      3,222,170

[Adam Griffin]

Is it a "major problem" in terms of producing fundamentally different results in polling when compared to the overall election result (EC)? No. The vast majority of Latinos live in safe D/R states and therefore won't be driving swings large enough to upset the overall result. Those in swing states (with the exception of FL) are relatively small in number and such a discrepancy would only impact the most narrow of elections. Even in states like CO & NV, the polling discrepancy would only impact things if the election there was within 1-3 points.

So, it matters in FL and might matter in AZ, but otherwise...I don't think it matters.

It's a major problem in the long-term, though: pollsters need to figure out how to sample this demographic or their polls will increasingly become inaccurate even if they manage to navigate the internet/cell phone/land-line hurdles.

[ProudModerate2]

Been thinking about this a lot this morning, with the Latino targeted polling over the last couple of weeks showing such a massive Clinton victory and a higher interest in voting, are national and some state polls under sampling Hispanic turnout?
for example we had the odd Florida poll with what was it 7% or so Hispanic and I also noticed the Reuters tracking poll is only 7% African American with Clinton only getting 61% which doesnt seem to make much sense.

Well yes, if a Florida poll severely undersamples minorities like that, then its results will probably be very different than the actual election result. In Florida, African Americans made up 13% of the electorate in 2008/2012. Hispanics made up 14% in 2008 and 17% in 2012. Further, the share of the electorate that is white college educated has been going up and white non-college going down, consistent with national trends.

I haven't seen that poll but if you oversample demographics favorable to Republicans then you'll get a skewed Republican result. Just like it being skewed for Democrats if they were to use a sample that was 20% African American and 30% Hispanic, and so on.

Further, the Hispanic share of the FL electorate is going to be more than in 2012, again, consistent with historical trends.. So that needs to be taken into account, in addition to accurately polling that demographic - which means not only using landline-using English-speaking Hispanic respondents.

=

Projected electorate demographics of Florida (2012):

Non-college whites:           33.7%
College-educated whites:  32.7.%
African American:              12.9%
Hispanic:                            16.8%
Asian:                                 4%

Projected electorate demographics of Florida (2016):

Non-college whites:           30% (-3.7)
College-educated whites:  33.9% (+1.2)
African American:              13.4% (+0.5)
Hispanic:                            18.3% (+1.5)
Asian:                                 4.4% (+0.4)


So the makeup of the electorate matters substantially and people should account for changes in the electorate.

Yes, like Virginia says, if Florida polls are under sampling minorities, then the poll results will probably be very different than the actual election results.

That said, perhaps Dem strategists are (or should be) a bit concerned about Trump’s 19-percentage-point margin of victory over hispanic home grown candidate Marco Rubio during the GOP primary. Trump gathered up votes from dormant Republicans casting first time ballots, and he managed to grow the ranks of the Republican Party of Florida by the tens of thousands. Again, he managed to do this while running against a home state senator. Trump's victory was solid: Rubio lost every county except Miami-Dade. Will Hillary Clinton have a better chance of winning Florida than Rubio? Will she manage to do better with hispanics than a hispanic? We'll see, but I have to say, I have my doubts...

Amusing to see that trump supporters still need to quote and rely on Republican primary results, to give themselves some assurance or (desperate) hope of possible victory.
It's like a "security blanket" that a 4-year old continues to cling-to, and refuses to release.

[SillyAmerican]

Amusing to see that trump supporters still need to quote and rely on Republican primary results, to give themselves some assurance or (desperate) hope of possible victory.
It's like a "security blanket" that a 4-year old continues to cling-to, and refuses to release.

Yup, thanks for pointing that out; I added the following to my post:

And yes, I understand that the primary results were only Republicans. But the registration numbers for 2016 as reported by the Florida Department of State are as follows:

  Republicans:   4,396,609
  Democrats:   4,659,801
  Other:      3,222,170

[swf541]

Amusing to see that trump supporters still need to quote and rely on Republican primary results, to give themselves some assurance or (desperate) hope of possible victory.
It's like a "security blanket" that a 4-year old continues to cling-to, and refuses to release.

Yup, thanks for pointing that out; I added the following to my post:

And yes, I understand that the primary results were only Republicans. But the registration numbers for 2016 as reported by the Florida Department of State are as follows:

  Republicans:   4,396,609
  Democrats:   4,659,801
  Other:      3,222,170

Yep, my issue with Trump supporters claiming that Trump is expanding the base is (especially in the south/ Appalachia) the people switching parties are those who'd vote GOP even if they were Dem.

I believe articles have also shown that many of Trump's supporters are not new voters, they tend to vote in the general but not the primary.

Anyways if Clinton can win Florida its pretty much over.

[Coolface Sock #42069]

Well, with Florida, I would think that would be Trump's best state with Hispanics.

You're forgetting Texas.

He's saying that Florida will be Trump's best performance with Hispanic voters, thanks to the Cubans

I understand, but Texas Hispanics are also very conservative by Hispanic standards.