Surprise, surprise (CO): Latinos much more Dem than exit polls showed

[Adam Griffin]

Clinton beat Trump by 70 points in Colorado, according to Harvard statistician: exit polls showed anemic 37-point win for Clinton among CO Latinos

I have no reason to believe that similar discrepancies don't exist in many other states as well. I know some analysis has been done in other places: feel free to link to any other state analyses.

You must be logged in to read this quote.

[Antonio the Sixth]

A colleague of mine came up with very similar findings on New York.

[(Still) muted by Kalwejt until March 31]

Heaviest Hispanic precincts≠Hispanics as a whole. Hispanics vote differently when they're around different people.

Look at the 2012 #s for comparison.

Also, the most Hispanic precinct in that scatterplot is closer to 20% DJT than 14%.

[(Still) muted by Kalwejt until March 31]

I'm sure if you looked at precinct data nationally you would find that Clinton cleaned up among college educated voters as well.

-Nonswing is making a brilliant point, but he almost certainly doesn't realize it.

[RI]

Why does everyone always use this technique? It doesn't prove anything other than how Hispanics in downtown Denver voted. Even then, the modeling assumes a high degree of linearity at the upper levels that I'm not sure is really born out by the data they're presenting.

[Adam Griffin]

Heaviest Hispanic precincts≠Hispanics as a whole. Hispanics vote differently when they're around different people.

Look at the 2012 #s for comparison.

Also, the most Hispanic precinct in that scatterplot is closer to 20% DJT than 14%.

It didn't look at just majority-Latino precincts...? It's not a case of "the most heavily Latino precincts were 80%+ Clinton therefore the Latino vote in CO was 80%+ Clinton", unless I'm missing something here.

Why does everyone always use this technique? It doesn't prove anything other than how Hispanics in downtown Denver voted. Even then, the modeling assumes a high degree of linearity at the upper levels that I'm not sure is really born out by the data they're presenting.

Elaborate, please.

[(Still) muted by Kalwejt until March 31]

Heaviest Hispanic precincts≠Hispanics as a whole. Hispanics vote differently when they're around different people.

Look at the 2012 #s for comparison.

Also, the most Hispanic precinct in that scatterplot is closer to 20% DJT than 14%.

It didn't look at just majority-Latino precincts...? It's not a case of "the most heavily Latino precincts were 80%+ Clinton therefore the Latino vote in CO was 80%+ Clinton", unless I'm missing something here.

Why does everyone always use this technique? It doesn't prove anything other than how Hispanics in downtown Denver voted. Even then, the modeling assumes a high degree of linearity at the upper levels that I'm not sure is really born out by the data they're presenting.

Elaborate, please.

-Griffin, you're a numbers guy, so I expect you to be better than this.

[mgop]

colorado will be red in 2020, when all illegals from state gets deported and liberal shenanigans from 2016 will not be possible.

[krazen1211]

It's possible this is the case, but that would of course mean Trump did better than expected with Colorado white voters. That is very bad for the Democrat party!

[Virginiá]

I think it raises some important questions that perhaps the next round of exit polling should consider. The Upshot already made a compelling case that EPs are under-sampling white working class voters, so to me it hardly seems like a far-out suggestion that they over-sampled more R-friendly Hispanic voters, especially considering that polling Hispanic voters already has some known issues. I don't believe this is an isolated study either. There have been others.

[(Still) muted by Kalwejt until March 31]

colorado will be red in 2020, when all illegals from state gets deported and liberal shenanigans from 2016 will not be possible.

-Wrong. Trump would have still lost CO with 2004 demographics.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Even if the survey wasn't just majority Hispanic precincts, it seems to have focused on those precincts. Plus, even if it has some validity, questioning the exit poll is not an endeavor that is the least bit worthwhile, so let's just not do it.

[vote for pedro]

This is the same guy who claims that all the exit polls were wrong and the Latino Decisions prediction was actually correct.  Bryan Wilcox-Archuleta. 

https://uselectionatlas.org/FORUM/index.php?topic=253876.50

[(Still) muted by Kalwejt until March 31]

It's possible this is the case, but that would of course mean Trump did better than expected with Colorado white voters. That is very bad for the Democrat party!

Yes, it would be terrible if Democrats did better among a growing voter group than initially thought at the expense of a shrinking voter group.  Your posts are becoming more and more epic.

-White voters are not magically a shrinking group in Colorado.

[hopper]

It's possible this is the case, but that would of course mean Trump did better than expected with Colorado white voters. That is very bad for the Democrat party!

Yes, it would be terrible if Democrats did better among a growing voter group than initially thought at the expense of a shrinking voter group.  Your posts are becoming more and more epic.

-White voters are not magically a shrinking group in Colorado.

True College Educated Whites are nearly as responsible for Dems success in the state as Hispanics are.

[(Still) muted by Kalwejt until March 31]

I'm not sure ecological inference is necessarily the best method to calculate the Hispanic vote share (since as mentioned, not all Hispanics live in heavily Hispanic areas, and their voting patterns are quite variable).

The Hill posted an article about the Latino Vote in Florida being calculated by ecological inference, and the results were absolutely absurd.

You must be logged in to read this quote.

I have a very hard time believing these numbers, Romney and Trump would have had to get absolutely crushed among White voters in Miami-Dade to make the math of these findings work. When you look at the remainder of Florida outside of Miami-Dade, the study got extreme results in the other direction:

You must be logged in to read this quote.

Trump would have had to basically run the table among Osceloa Whites to make this figure achievable, just to pick one example, when the results at the precinct level don't seem to indicate that's what happened at all.

Overall, I tend to trust the NYT Upshot estimates of Clinton winning the national Hispanic vote 68-26, with Clinton generally beating her stated support among Latinos in each state by a few points, but no more than that.

Umm - you do realize that "White voters in Miami-Dade" are largely Jewish, Gay, Northeastern Transplants, and/or European Transplants, right?  Also the numbers aren't surprising because the Cuban vote is the GOP base in Miami (or at least was) and Democrats do well with whites there as they would in any other big city... also there's a large black population that props up Democratic numbers as well.

-What about Osceola?

[NOVA Green]

So why hasn't anyone actually run any real precinct numbers on this one?

From what I have observed in more heavily Latino Precincts in the Pacific Northwest, and to some extent in Cali, it appears that there was a significant defection to 3rd Party Candidates rather than increase in support for Clinton vs Obama....

It appears to be pretty clear that in heavily Latino areas in the cities of Texas, that there was a major Democratic bump (Harris County jumps out)....

If we look at Florida, it's pretty clear that Cuban-Americans swung towards Clinton, and "angry" Puerto Ricans voted overwhelmingly Democratic.

What appears to be more of an item of debate are rural and small-town Latinos....

Regardless, it's not good news for the Democratic Party either way, since obviously an increasing Latino Democratic vote in certain counties, precincts, and states, means that even if the exit polls are off, the Anglos swung much more heavily Republican in these areas....

[krazen1211]

It's possible this is the case, but that would of course mean Trump did better than expected with Colorado white voters. That is very bad for the Democrat party!

Yes, it would be terrible if Democrats did better among a growing voter group than initially thought at the expense of a shrinking voter group.  Your posts are becoming more and more epic.

-White voters are not magically a shrinking group in Colorado.

They are also not a particularly shrinking group in most of the vulnerable Senate seats in 2018 that the Democrats will lose bigly.