Latino Decisions: Week 11. Obama 73, Romney 21

[King]

As I noted in my thread, the Romney debate surge has been polls showing movement to 40-45% Latino support for Mittens.  Without that figure, Romney is still down 3-5 points nationally.  Indeed, the only two polls that haven't shown that movement: RAND and TIPP have been giving this race to Obama for awhile now.

If Latino Decisions is actually the one that is right here (along with RAND and TIPP) and not all the other pollsters, we're heading straight into another Dewey Defeats Truman moment and the polling problems with Hispanics that gave Republicans false hope will probably be talked about for years to come.

In truth, I think the real answer is somewhere in the middle.

[Ty440]

As I noted in my thread, the Romney debate surge has been polls showing movement to 40-45% Latino support for Mittens.  Without that figure, Romney is still down 3-5 points nationally.  Indeed, the only two polls that haven't shown that movement: RAND and TIPP have been giving this race to Obama for awhile now.

If Latino Decisions is actually the one that is right here (along with RAND and TIPP) and not all the other pollsters, we're heading straight into another Dewey Defeats Truman moment and the polling problems with Hispanics that gave Republicans false hope will probably be talked about for years to come.

In truth, I think the real answer is somewhere in the middle.

Who would you put your money  on, actual Hispanic pollsters  who know the culture and speak Spanish or some American pollsters who probably only know two words  in Spanish, hola and gracias.

I would trust Latino Decisions, I believe this is their  expertise.

[J. J.]

As I noted in my thread, the Romney debate surge has been polls showing movement to 40-45% Latino support for Mittens.  Without that figure, Romney is still down 3-5 points nationally.  Indeed, the only two polls that haven't shown that movement: RAND and TIPP have been giving this race to Obama for awhile now.

If Latino Decisions is actually the one that is right here (along with RAND and TIPP) and not all the other pollsters, we're heading straight into another Dewey Defeats Truman moment and the polling problems with Hispanics that gave Republicans false hope will probably be talked about for years to come.

In truth, I think the real answer is somewhere in the middle.

Who would you put your money  on, actual Hispanic pollsters  who know the culture and speak Spanish or some American pollsters who probably only know two words  in Spanish hola and gracias.

I would trust Latino Decisions, I believe this is their  expertise.

I trust their ability to speak Spanish, but that doesn't make them good pollsters. 

[sobo]

The only 2008 poll I could find from them was released early October 2008, but appears to have been conducted mostly in September.

http://www.latinodecisions.com/blog/2008/10/07/new-poll-suggests-latino-voters-may-make-the-difference-in-four-key-states/

They called Nevada 67-20 (actual 76-22)
Colorado 71-18 (actual 61-38)
New Mexico 67-23 (actual 69-30)
Florida 45-43 McCain (actual 57-42 Obama)

So they did pretty well in Nevada and New Mexico, but pretty poorly in Colorado and Florida, though they missed in opposite directions so their national number probably would have turned out okay.

[J. J.]

Not a great pollster, then. 

[Bacon King]

The only 2008 poll I could find from them was released early October 2008, but appears to have been conducted mostly in September.

http://www.latinodecisions.com/blog/2008/10/07/new-poll-suggests-latino-voters-may-make-the-difference-in-four-key-states/

They called Nevada 67-20 (actual 76-22)
Colorado 71-18 (actual 61-38)
New Mexico 67-23 (actual 69-30)
Florida 45-43 McCain (actual 57-42 Obama)

So they did pretty well in Nevada and New Mexico, but pretty poorly in Colorado and Florida, though they missed in opposite directions so their national number probably would have turned out okay.

It's worth pointing out that you're comparing their results with an exit poll's results in a single state of a group that makes up like 15-20% of respondents in the state. There could also be considerable error here due to margin of error of the exit poll's subsamples.

For the record, Latino Decisions also did a national poll in 2008, which showed a 37 point margin of Obama over McCain, extremely close to the 36 point margin showed in the national exit poll.

[J. J.]

It's worth pointing out that you're comparing their results with an exit poll's results in a single state of a group that makes up like 15-20% of respondents in the state. There could also be considerable error here due to margin of error of the exit poll's subsamples.

For the record, Latino Decisions also did a national poll in 2008, which showed a 37 point margin of Obama over McCain, extremely close to the 36 point margin showed in the national exit poll.

That poll was taken in June 2008. 

[ajb]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

[J. J.]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

Well, the did the opposite in CO, so it isn't a bias issue. 

[ajb]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

Well, the did the opposite in CO, so it isn't a bias issue. 

Not suggesting it is. But it might be worth remembering that their 2008 FL numbers look like a lot of pollsters' 2012 FL numbers for Hispanics. There's precedent, in other words, for pollsters underestimating how Obama would do with Florida Hispanics.

[J. J.]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

Well, the did the opposite in CO, so it isn't a bias issue. 

Not suggesting it is. But it might be worth remembering that their 2008 FL numbers look like a lot of pollsters' 2012 FL numbers for Hispanics. There's precedent, in other words, for pollsters underestimating how Obama would do with Florida Hispanics.

Yes, but that doesn't answer if LD is a good pollster. 

[opebo]

Not a great pollster, then. 

In fairness, they're attempting probably the most difficult task - tracking Hispanics.  How do other pollsters fare with this?  Perhaps worse than Latino Decisions.

[afleitch]

Not a great pollster, then. 

In fairness, they're attempting probably the most difficult task - tracking Hispanics.  How do other pollsters fare with this?  Perhaps worse than Latino Decisions.

The Pew Hispanic Center http://www.pewhispanic.org/ had Obama at 69% to 21% for Romney.

Gallup's October 1 - October 21 RV sample that had Obama up 1 in Registered Voters overall had the Hispanic vote at 69-27 for Obama.

So leads of 48 and 42 respectively

[Fmr President & Senator Polnut]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

Well, the did the opposite in CO, so it isn't a bias issue. 

Not suggesting it is. But it might be worth remembering that their 2008 FL numbers look like a lot of pollsters' 2012 FL numbers for Hispanics. There's precedent, in other words, for pollsters underestimating how Obama would do with Florida Hispanics.

Yes, but that doesn't answer if LD is a good pollster. 

Only one way to find out...

[J. J.]

Not a great pollster, then. 

If so, mostly because they substantially underestimated Hispanic support for the Democrats in Florida.

Well, the did the opposite in CO, so it isn't a bias issue. 

Not suggesting it is. But it might be worth remembering that their 2008 FL numbers look like a lot of pollsters' 2012 FL numbers for Hispanics. There's precedent, in other words, for pollsters underestimating how Obama would do with Florida Hispanics.

Yes, but that doesn't answer if LD is a good pollster. 

Only one way to find out...

Which is what we're asking about.  With any poll, we should be looking at their track record.