I think we can trust Latino Decisions to capture the mood of Latino voters accurately.
Why?
Perhaps we can, but mind you that Latino Decisions argues the existence of a systemic error in exit polls (et por cause, in every other poll). They claim other pollsters under-represent hispanic voters turnout and their level of support for Democratic candidates. Maybe they're right, maybe they're wrong -it's noteworthy that Latino Decisions is associated with Democratic leaning interest groups, namely the La Raza Council, that pays for many of their polls, including this one- but it's important to keep in mind they have different priors than all the other pollsters.
If this poll is true, so much for the GOP making progress in attracting Latino voters.
Seems like they are, at least in Colorado.
Comparing apples to apples (ie using other Latino Decisions polls as baselines):
2014 Udall +57 (allocating undecideds)
2012 Obama +77
http://www.latinovote2012.com/app/#all-co-all2010 Bennet +62
http://www.latinodecisions.com/blog/wp-content/uploads/2010/12/co_nov2.pdfAnyway, as often happens with LD, this poll has a large number of 'message testing' questions before the horserace ones. Personally I think this is bad methodology and tend to disregard polls that use it.
