Thread by @Peoples_Pundit

Think some in the media should probably stop forecasting and polling Florida.

Normally, if people want to make complete fools of themselves, over and over, I'd say, "Do you."

But you mislead a lot of people, including those we all know you agree with and are trying to help.

Florida is a (slightly pink) battleground state, one decided by low single digits. That not only has been the case with actual outcomes, but our SBDP polls never showed a contender with more than a 4-point lead in Florida. And they won't get one if both candidates are acceptable.

Btw, that was Rick Scott and Andrew Gillum, as you all know. Neither ever had more than a 4-point lead, at best.

Bill Nelson never got better than even. Never. Think, folks. He just started to return fire. He was able to close it. But he never led Rick Scott.

Sorry. Never. Led.

In a landslide election, Barack Obama defeated John McCain in the state of Florida, 51.03% to 48.22%. Until Donald Trump, who beat Clinton by just over 1%, that was the largest margin.

This is state is not decided by 7 or 13 points. Quinnipiac and CNN, now I'm talking to you.

CNN gets another chance and I'm going to tell you why.

At least they tried after their 2016 blunder. They fired their old pollster and hired another one. I understand, it must've been a lot of pressure for them in their first year, and we're going to give them a break.

But...

I can't tell you how many webinars, seminars and whatevernars I've been to since the failure in 2016. There are people in this industry, some who might surprise you, who understand the problem. But legacy pollsters don't want it to evolve. Why? 1) The Club and 2) The Money.

If it's not the perks of being in the Cool Club, it's the money. Pure and simple.

To make a very long story short, innovators lost the debate and the legacy pollsters won it, largely due to forecasters cheering them on and the media declaring what is and is not accurate.

Now look at you. Embarrassed. Inaccurate.

And for what? The Cool Club and the money? Sad.

Someone pry the random digital dialer and call centers from their cold, dead, inaccurate hands and tell them to move into the 21st century.

Let me let you all in on a little secret Big Media pollster don't want you to know. Voters don't participate in interviews equally across all modes of data collection. And even if you tailor modes to them, which we do, some are still more likely than others to participate.

How does that have a practical impact on polling? Take @FoxNewsPoll with their 14-point miss (Donnelly+7) in Indiana. There were multiple reasons for that miss, the first is their data collection mode. They can't reach a representative sample, not regionally, not ideologically.

But that's Indiana, and I could do a thread on that, alone. Florida is a whole other animal. There are too many polls and not enough pollsters.

Let's start with the Quinnipiac University Poll in Florida...

Bill Nelson and Andrew Gillum led Rick Scott and Ron DeSantis by 7%.

They sampled 1,142 "self-identified likely voters."

Really, are they?

How the hell do they know? They self-identified, which indicates Quinnipiac 1) did not use a voter file and 2) B/C they don't use it, they can't even verify if those are in fact voters.

But let's pretend...

We have minimum sample sizes for each state. The 1,142 Q collected would be adequate for Florida. But let's look at the demographic breakdown.

Q collected appx 100-200 Hispanic responses BUT treated them all the same. No Hispanic detail question.

Nada!

Why does it matter?

A major problem we see w/ Florida polls is how badly they survey Hispanics. We have minimum response requirements for Hispanic subgroups/detail, not just the overall Hispanic population.

Look at Quinnipiac's result among Hispanics. That's a data source & collection mode issue.

I could go on. But statewide, polling sucked again & glorified poll-readers have a vested interest in continuing on w/ this lie.

If someone reading this is tired of being wrong and embarrassed come Wednesdays, they can privately DM me and can privately talk about this more.

Sorry, one more thing. Look at the margin among white voters. Again, that's an issue with their data collection mode.

How many white voters in say, Micanopy, did Quinnipiac reach via live-caller on their cellphone?

Uh huh.

Let me guess... They wouldn't talk to you.

Hi! Click.