*thread* Polling. I'll begin by apologizing in advance for the inevitable grammar/spelling/autocomplete errors in this thread. For the new folks, I've worked on dozens of campaigns since 1992. I've managed 6 US House races, worked on dozen of other Federal and statewide races.

In my career I have directly commissioned dozens of polls, so I'm not your typical Twitter *expert*. There's a reason why multiple people have asked me to run a presidential campaign *humble brag*.

You need to understand a few important things when talking about polls. You need to know data vs the story that data tells (which is the art.) You need to understand internal, private polls vs public polls, and you need to understand "likely voter vs registered voter."

First off, data collection is relatively simple. A polling firm hires a phone center (though many firms also have in-house operators) who dial a certain number of potential respondents until a goal is met. A decent poll is going to have roughly 400-500 respondents to be credible

That should give you a margin of error below 5%, which is usually the magic number (many other factors go into MoE but number of replies matters.) That poll should cost you around $15-25,000

Every campaign I've worked on starts off with what we call a "baseline" poll, which is pretty much where we start. Name ID of the candidate, who someone is considering voting for, their stance on certain, key issues, etc. We build the vote model from that poll so it matters.

Now, any math nerd can import the raw responses into an Excel file and email that to the campaign. That's basically what you see on sites like 538, etc. But you have to interpret the data yourself. The true art of a poll, and what separates pollsters, is understanding the story

See, the numbers all tell a story if you understand how to see the patterns. And a good pollster can extrapolate that from the numbers, as well as understand the questions that need to be asked in order to get a clear picture. It's an artform and few are good at it

Now, understand that there are true, professional pollsters who focus on serving the client and telling the truth no matter where the numbers lead, and there are pollsters who can make the numbers tell the story that the client wants to hear. It's a subtle difference

Also, understand that there are polls that you commision to share with the media, the client and donors and polls you commission to help the campaign properly dedicate it's resources and that you may not share with the client or donors. Especially because donors leak shit

The polls that you often see Trump share and brag about fall into the first category: polls you commision purely to make the client and donors feel better, or to make the media write about your campaign. Theyre a joke and no professional takes them serious. But, Trump

Also, when you see morons and Trumpers brag about "fake polls" or some bullshit and claim that said poll doesn't match actual voter registration, then youve just spotted a fucking idiot who doesnt understand LV vs RV. Lemme explain:

"registered voter" is exactly what it sounds like and is self explanatory. For example, in FLorida there are 13,731,883 registered voters. It breaks down as 4,851,116 GOP (35%), 5,127,920 DEM (37%) and 3,752,847 other and NPA (27%)

So, if you were commissioning a poll of RV's (which you wouldn't on a campaign) you would have your polling firm dial as many numbers as possible until your poll had a response from 35% Republicans, 37% Democrats, etc. Make sense?

The "likely voter" part is what gets everyone into trouble and which few Trumpers actually understand. Like a lot of things, but I digress

Predicting LV's is an art form. You're literally trying to predict in advance what the electorate will look like on Election Day amd that's really hard to do. It's like predicting which spirit I'm going to drink next. Shit, I don't even know

So, do pollsters get that part wrong? Duh, of course they do. How good are you at predicting world events tomorrow, let alone five months from now. So does that mean a poll is "wrong" or "fake" in hindsight? Smart people understand that it doesnt.

DO I need to spend more time on LV vs RV or do yall get the point? Its really critical because it's the party most people don't understand

The problem is that RV polls are cheap and easy to do where LV polls are not. But polling, like plastic surgery, is something where you often get what you pay for

Here's where I get to some interesting stuff and why I genuinely think Biden will win in an utter landslide. It's all about feelings. And how voters feel is everything in politics.

Read this new poll from AP (and, if you've bee npaying attention, here is the methodology ( http://www.apnorc.org/PDFs/AP-NORC%20June%202020/methodology.pdf): http://www.apnorc.org/projects/Pages/As-Pessimism-about-the-Country%E2%80%99s-Outlook-Spreads,-Few-View-Trump-as-a-Unifying-Force.aspx

I'll give ya'll a minute

Here's where we start, and it's a killer:

You can correctly make the argument that this poll is skewed towards minorities, hence the results on race relations. But keep in mind the impact this has on the presidential race: 90% of African American voters vote Democrat.

One pollster would go to Trump and say, "Sir, the first line doesnt matter because 90% of *blah blah blah* so just ignore that part."

You with me so far?

Ok, back to that first graphic before going on. Take a closer look at it. What jumps out at you?

For me is it starts with the "leadership" questions.

"Approve/Disapprove on race, coronavirus, healthcare, economy." This probably seems pedantic but it matters: those are the only questions that matter come Nov. And Trump is significantly upside down on each one.

Now, back to the LV/RV thing, here's why that would matter: if you were doing a strict RV poll, then the numbers would look slightly better because most Republicans would say they approve on those issues. So the poll would be weighted in that direction. Make sense?

So, a skilled (vs a good) pollster would be able to build a sample that heavily used RVs to create a more positive picture. Whereas LVs (which would tilt Dem in a presidential election year and also include indies) would be worse for Trump. Hope that makes sense

Back to my earlier point: elections are all about emotions. At least going back to my first one in 92 (and Im sure much further) the winning candidate is the one who makes you feel better, happier or less anxious about things.

In Mast '16 it was easy for us because we were contrasting a war hero who loved America, went to Harvard and was very strong on the environment vs an angry white guy who raged at everyone and knew shit about the local community. In short, you were excited to vote for Brian

In '16 Trump was far better at generating positive emotions in voters than Hillary was. He may be a total fucking moron but he has an extraordinary emotional IQ and understands manipulation on a genius level. Hillary, well, uh, yeah. Easy contrast.

Shit, this is gonna be longer than the Burt story. Sorry.

As my man @LenCollins (good dude, good follow) correctly pointed out, those numbers are fatal for an incumbent. Think about it this way: when you are unhappy about things do you make a change or decide to double down on another four years. Those things go beyond partisanship

And when people are in the midst of a storm the first thing they want is dry, safe land. If Trump were running against Hillary again, or Bernie, Warren (sorry, @MadameButtons) or Harris, I'd say he had a good chance of re-elect because those candidates aren't safe land in a storm

But Joe Biden has literally been an elected official since before I was born. He doesn't really believe in anything. He's an empty vessel. And he comes off to most people as an honest, decent man. Safe land in a storm? Yup.

Basically the only way incumbents are re-lelected is if things are trending positively, a majority of voters are happy, and the alternative is too scary. None of those things are true in 2020.

Let's move on to each question in more detail. Here's the one on race relations and it's EXTRAORDINARY

Before I go further (I heard those groans) if you really want insight into numbers than @steveschale is your man. I'm not a great numbers guy. My strength is comms and overall management and leadership. Steve can really drill down into this stuff.

Back to that race question. I don't know the national breakdown of party reg off top of my head, so I'll use Florida as my baseline

Let's pretend America mirrors Florida with a 35-37-27 R-D-I voter makeup, what jumps out at you? Look at the very first question and then look at the voter ID again

32% of Americans approve of Trump on race relations. Contrast that with the fictional 35% R voter ID and what do you see?

Now look at the next one. 7% of African Americans approve, yet only 10% AA vote Republican. So that means a significant number (such as it is) of Black Republicans disapprove of Trump's handling.

A good Pollster (and I'm pretending to be one tonight for purpose of this discussion) sees the story that number is telling. In fact, a good pollster would use that number to commission a follow up poll to drill down deeper. But Trump doesn't have a good pollster.

Line #4: 32% of Republicans disapprove of Trumps handling of race relations. If I were running Biden (and they have way, way, way, WAY smarter people than me) I would fucking run on that all day. That's a mic drop moment

I feel like I'm getting into the weeds a bit, Are ya'll still with me?

So, again, a pollster who wanted to keep getting money from Trump could build a sample where they could show Republicans generally approve of how Trump is handling race relations. Let's move to the next question. Spoiler: it isn't good for Trump

Hooooooleeeeee shit. "Are we more divided" is just another (are you happier than you were four years ago" question, and to me it's a bombshell.

But the bigger bombshell is on the "Republican" line. Take a close look at that and tell me what stands out. I'll take a sip of whiskey and wait for you.

Are ya'll there? Did you catch the "37% of Republicans think Trump has made Americans more divided?"

If I were smart I would deem this poll as one actually about voter enthusiasm. And what I see here are a shit ton of Republicans who are deeply concerned about the direction that our country is headed in. There are cracks in the foundation, folks.

And my earlier point stands: when people are panicking and see the boat filling with water, they're not going to run to the person they aren't convinced won't put more water into the boat. They want the safe choice. And that aint Trump. Sorry.

I'll skip to #5 to drive my point home. Here it is:

This is the money question. What does this tell you?

Point #1: there is an increase of 9% of Americans who think America is headed in wrong direction. Nine. Percent.

lemme restate: The trend is downward at NINE. PERCENT.

*whispers* theres a bigger holy shit moment. Did you see it? I'll wait.

*sips whiskey*

(hint) check out the red line

I feel like this is a good place to end things. I'm open to general questions about methodology or the mechanics of polling if you have any. Otherwise, if you're still here I'm genuinely flattered and hope everyone is well. Be safe out there.