President Biden: The NEW one and only politics thread

We used to run SJ political polls. I guess not this year?

 

Jack Nicklaus actually produced mid ass golfers.

 

Like LeBron. And MJ.

 

It is interesting since 2008 (starting a consecutive run where Republicans have lost the national vote and three of four electoral votes) it’s the GOP claiming the polls are wrong and don’t trust them. We had Newt ranting on Fox that they had the election results wrong claiming the early returns weren’t accurate. The only time the GOP was truly right was in 16 when Hillary lost. You could argue 20 was closer than polls indicated, but they were basically right.

Now we have polls showing a tight race and Trump leading in the states that matter. And here we are coming up with reasons for the polls to be wrong or gleaming to potential outliers as proof.

Have pollsters over corrected? God knows. Are women afraid of someone hearing they like Harris so they say Trump? Who knows.

What I know of this is a race that should be having Harris waving to the crowd as she jogs to the finish line with Trump shitting himself on the side line. Instead, she’s sprinting and counting on a lot of no-way-to-prove-it-true arguments the polls are wrong.

Just how angry are women? I wish we had an accurate projection. Iowa could be a clue or an outlier. Unable to determine returns in Georgia. Tight polls in Michigan that are being adjusted.

It’s necessary we brace for Trump win. With very little concrete things to point to, we have to. This is going to be 2016 all over again where we melt down because we really thought Trump was going to lose. Not to say he for sure will win but we really need to be ready for it.

 

too much hate too much

 

My coping mechanism in times of acute stress is to suck up as much information as I can about the situation in order to feel like I have some control, of my emotions if nothing else.

Also I took a couple of statistical analysis classes for my marketing degree that got pretty into the weeds about how polls work, the math behind them that shows how a sample of just a few hundred people can produce a valid (though not ironclad) understanding of opinions, the necessary steps for constructing a rigorous poll, etc.

They aren’t perfect Lord knows and you have to discern which pollsters are actually doing quality work. And they are undoubtedly overreported compared to the actual issues and consequences surrounding an election. But they are far from the medicine show hokum so many here like to claim without much understanding the how and why behind them.

 

Predicting that...

...CNN's Harry Enten becomes the star of Election Night coverage this year like Kornacki did a few years ago.

 

No, just killed violently, apparently.

 

Like all the people who fill out Nielsen books saying they watch lots of PBS to look sophisticated.

 

I don't know where I read it or I'd give attribution, but the best thing I saw today was "right now, you could use a random number generator to create any integer between 46 and 49 and be as accurate as 99 percent of the polls."

It's sort of the Marcel the Monkey version of political projections. ZiPs, Pecota and Steamer have very complex algorithms in an attempt to predict future performance, but Szymborski admits baseball players are way more likely to have a 20th or 80th percentile season than all of them being exactly 50th.

Marcel, by just taking the past three seasons of statistics, assigning a weighted average for recency bias and dividing by 3 is almost as successful.

Right now, everybody's trying to put all their resources and knowledge to predict what in essence is no more than a coin flip. When it comes down to probability error exceeding projection, polling can only guess heads or tails.

That's why I think most of this is just noise. And covering their asses in case it's really not that close at all. Is Selzer onto something? We won't know until Tuesday night.

 

She has go pedal to the metal through Tuesday. Too many Dems remember Hillary didn't in 2016.

 

Yeah, not sure "not statistically significant" is what you're going for there. Maybe, "not statistically reliable/defensible."