Lies, Damn Lies, and Pollsters
A Selected Dump of Recent Incoming Election Analysis FWIW
Many have asked me for personal prognostications - I have none, in large measure because I remain highly skeptical of polling and to some degree because too many problems in our elections process remain unchanged in places where narrow margins can make all the difference.
Nonetheless, attached (below) is a solid, comprehensive guide and analysis based on publicly available data. It was compiled by someone serious who does this each election and it is being circulated via DC insiders.
Here also are some reactions to it from a very smart and sophisticated (and humble) data analysis friend. I agree with most of his observations, fwiw:
___
Except for some minor insignificant matters, I agree with all of this analysis though the downstream ballot stuff is beyond my ken. Just on the national and swing state presidential polls:
Not all polls are created equal
Not all pollsters are incentivized equally
Major media polls generally have “no skin in the game” and their reputation has been sunk anyway; their only incentive is to be about where the other weak media polls and to play to their audience, to keep their audience of partisans
RCP average of polls is noise, I recommend de-emphasizing it
What would this look like if you only took subset of polls that performed closest to national results in 2016 and 2020?
Check out the demos on Trafalgar poll swing states… Trafalgar has a lot of $$$ on the line, and they should be in the poll subset to watch… more important, Trafalgar’s multi-tiered approach (though imperfect, is in top tier) is now *overweighting* registered Dems. So they are being conservative in those swing states IMO.
This analysis doesn’t include objective stock market analysis really smart folks have made; market movers have access to the most & diverse triangulated data and analysis and the markets from what I’ve heard from people who follow have priced in, and continue to price in and move, like a DJT win. And the market movers (Blackrock, Citadel, Cap Group +++) care about making $$$, and certainty, not partisanship. Their constituencies are their investors.
To me, the only issue is whether this is a hairline thin Trump win or a far bigger win for DJT than imagined, a blow-out. I’d place a lot of $ on a close win for DJT as reflected in the betting markets. To me the Iowa rogue poll is a reveal of why there will be massive poll misses, state and national, in this election. All kinds of misses. Demographics of the US are not the same as 2020 given outmigration of GOP-leaning young voters from liberal cities like San Fran & NY to places like… rural MI, PA, and FLA. This makes modelling and “correcting” the broken polling instruments from 2020 guesswork.
All this said, the Trump “more diverse bubble” vs Harris “ever-narrowing bubble” is the best explanation. We hear what we want to hear with confirmation bias… and, of course, I could be wrong, very wrong. The only thing I know is that to predict something with certainty with such complex dynamism and noise is a sign of a lack of humility and self-awareness and expertise.
___
“To predict something with certainty with such complex dynamism and noise is a sign of a lack of humility and self-awareness and expertise.”
Amen,
Jeff
