Thread by @Peoples_Pundit

All I hear today, is mediates and Democrats doubting the polls.

1. General election polls at this point have ZERO predictive value.

1a. As I laid out in 2016, the argument that candidates who are trailing in August/September cannot win, was a flimsy, at best.

2. Statewide polls in 2018 were NOT accurate. Big media is pretending coming somewhat close to the predicting the national vote means they fixed their problem.

They have not.

It is simply not good enough. Elections for POTUS are won at the state level, in the Electoral College.

3. At least @BretBaier had the fortitude to tell viewers how that poll faired in Florida in 2018. Not very well, folks. They didn't fair well anywhere in 2016. They missed the mark and margin in every single state.

Here's what I'd say to people looking at polls in Florida.

4. First, look at the margin among Hispanics. Go back, and do some checking. Look at the projected vote share among Hispanics for Republicans in Florida, according to the polls, juxtaposed to their vote on Election Day.

You'll notice a large disparity, and here's 1 reason why.

5. For starters, the sample size, though by no means is that typically the only issue. It's often far too small to collect sufficient responses from Hispanics in Florida.

Notice there is often no Hispanic detail Q. They wouldn't be able to have one with the usual sample sizes.

6. We always look to the pipe/branch/logic for surveys of Fla. Hispanics. What does that mean? If a respondent chooses Hispanic in the race/ethnicity Q, they're piped off to a Latino detail.

Are you Puerto Rican, Cuban, Mexico, Venezuelan, etc, etc?

It is very, very important.

7. It's also very important to know exactly where they are in the state AND that ethnicity detail. You'll notice that almost never happens, maybe there's a branch for Cubans, but maybe not.

Not good enough.

But I'm done giving away tips. Lest they catch on to step one.