Fetterman oz debate

Perhaps, but its never clear what a perfect sample would be. This same argument was central to the old “unskewed polls” fiasco from the Obama days, which didn’t work out for the unskewers.

Not that I am saying it’s a good poll. I think we should be skeptical of it, like all single samples, but also we should be skeptical of the criticism of it.

 

It is, however, clear that a sample that wildly off on demographics isn't a good one.

 

Im probably on the skeptical end of empiricism thinkers, but I would say maybe. The gold standard currently is randomness, and this sample was a random one. There is an argument for adjustment, as you are suggesting, but it does come at the trade off of this randomness.

The best approach, perhaps, is that if 538, which adds in each poll weighted by date and prior record. I don’t know if 538 is tracking this race, but if so, I bet this poll factored in to their model as a small fraction. And I think that’s a valid way to deal with it, somewhere between determinative and meaningless.

 

Oz strikes me as slimy and I like Fetterman a lot and have been following him since I saw a segment on CBS Sunday Morning about 5 years ago. BUT, you have to admit the donkey comment from Desantis was funny.

 

It is. I think you're fine when you look at aggregations. The point I'm making is when you analyze a single poll, it's quite significant if the demographics in terms of age are miles off from the usual electorate in that state, particularly when it would cause the poll to shift in a certain direction.

 

I love guys who openly wish to empty jails and protect murderers over citizens. Great guy. At least his medical condition doesn’t make him even more unstable.

 

biden??????

 

I would consider DeSantis’s old donkey comment to be funny if I wasn’t aware that DeSantis is a heartless, angry human being with a superiority complex.

 

I definitely know what you’re saying, and I understand it’s logic. I am just not as sold on the demographics making that much of a difference.

I found 538’s page for this race, and they are including the insider advantage poll. They also added a more recent poll of 1,000 likely voters by Wick that found nearly the same result: 48% to 46% in favor of Oz. Looks like a close one.

Pennsylvania : U.S. Senate : 2022 Polls

 

Funny thing about the Wick poll is they had Oz up by 5% earlier in October (the only poll on 538 to have Oz in the lead since May at that time). In other words, their poll shifted 3% towards Fetterman after the debate.

Demographics in polling matters a lot. If you have a poll that oversamples white people or undersamples young people, you'll likely get a result that is biased towards the GOP. If you have a poll that oversamples Black people or undersamples people in rural communities, you're likely to get a poll that is biased towards the Democrats. It's why pollsters work so hard to try and model these things correctly. (Because demographics vary from election to election, it's hard to do. But when it's wildly off, you have a problem.)

 

And as I pointed out earlier, that particular poll also has independents favoring Mastriano 55% to 30% over Shapiro in the PA governor race. That seems like a really high share of independents that favor the extreme right Mastriano. The most recent CBS/YouGov polls shows Shapiro at 64% and Mastriano at 36% with independents.

 

I maybe worded my last message poorly. I do take your point about over and under-sampling demographics. My reticence to take this data as fatal to a poll is basically because I don’t really trust any poll that much to begin with. There are so many factors that lead to polling error that adding in a perceived demographic bias doesn’t change much. It’s like accidentally spilling paint on a picture made by a toddler. Sure it changes the picture, but it’s not necessarily worse, given where it started.

 

Another poll has found Oz up by 2+. Tough to make a call on this one.

PA SEN 10.28

 

NYT/Siena has Fetterman up by 5.
Times/Siena Polls Show Democrats Slightly Ahead in Key Senate Races

 

It seems one was surveyed from Oct 24-26, so I’m not sure how much post debate sentiment it is capturing. I assume this Oz bump will fade, but not sure by how much.

 

Like most mid term elections, voter turnout is lower, so there is often a wave that the polls miss. In the primary season it was a dem/moderate wave. Polls recently ending a rising republican wave.
We will see which is right this year, but the polls are close enough and we are close enough to the election that unless there is a big change fast, analyzing minor poll variances won’t mean much.

 

Your source for the quote is incorrect. It was spoken by Michael Corleone in Godfather III. One minute mark.

 

Problem is PA polling has been skewed Dem heavy in the past decade or more and the real results often shift the red way. In other words, Fetterman is not ahead.

 

Each election is different and this one could swing the pubs way, but the 2020 RCP average of the final polls hit the PA number to the tenth of a percent.

https://www.realclearpolitics.com/epolls/2020/president/pa/pennsylvania_trump_vs_biden-6861.html