Here's my final update to this model of California recall polls. I'm calculating an aggregate that adjusts polls based on whether they use partisanship in their weighting schemes, and draws different trends for adjusted v unadjusted data. Newsom +18 +/- 10
gist.github.com/elliottmorris/…
The point of this project was to illustrate the different methods we can use to aggregate polling data — esp in how to improve existing popular averages that don't peer under the hood of how pollsters are processing their data, an increasingly important aspect of public polling.
So, note two things:

1. Popular averages magnify unlikely trends in public opinion by being whipsawed by data that is subject to higher standard errors than a decade ago (when the models were first made). Weighting by party flattens trends by decreasing nonresponse
2. The uncertainty of the model increases as we adjust for more factors. The uncertainty may seem so large as to make the aggregate useless, but that's a little silly. Instead, I think these intervals better represent the total survey error in polls — that 538 etc have ignored.
Will probably blog more about this modeling later. I think it's important and represents an innovation over decades-old averaging techniques w/ ad-hoc adjustments for past accuracy that don't capture what's actually happening in the data-generating process gelliottmorris.substack.com

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with G. Elliott Morris

G. Elliott Morris Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @gelliottmorris

15 Sep
Aaand there it is folks! Early vote-by-mail results in California's recall election are way ahead of tied-race benchmarks and signal an imminent victory for Governor Gavin Newsom, possibly by high double digits. I'm going to bed early tonight livevoterturnout.com/sandiegoca/Liv… ImageImage
Listen to the (good) polls, folks!
where is my award for beating dave by one hundred and twenty thousand microseconds
Read 4 tweets
14 Sep
At this point, given the CA recall polling and VBM data, all we're looking for in the early vote tonight is confirmation of projected partisan distributions. Returned ballots are sufficiently Dem that we just need to assess loyalty + turnout. LA county VBM +25 Newsom would do it.
The very fuzzy math here is that early mail ballots statewide (+40) were about 10 points more pro-Biden than the final results (+30) in California in 2020, and LA County (+45) was about 15 pts more Biden than the whole state. So +55 early LA = +30 CA-wide in the end. +25 = tied.
You don't call an election just based on one county, of course. But Los Angeles cast 25% of CA's votes in 2020 so it's a good guide.

Other VBM benchmarks could be:

Recall +11 in Orange County (+19D 2020 VBM = 30 overall :: -11D VBM = 0)
Keep +3 in SD (+33D VBM = 30 :: 3D = 0)
Read 4 tweets
14 Sep
Blog post: Media coverage of California's recall election highlights big issues with popular poll aggregation models

gelliottmorris.substack.com/p/polls-of-cal…

FiveThirtyEight and RealClearPolitics are unable to capture modern outliers or the impacts of non-response & different weighting schemes
Expanding on what I've been saying here and elsewhere
Regardless of what happens in the election, the bottom line is that modern statistical methods can provide better analyses of polling data than the stuff available ten years ago, especially (but not only) because of recent problems in the polling industry. Time to do better.
Read 4 tweets
9 Sep
The thing about Rasmussen (and to a lesser extent, some other right-leaning pollsters and aggregators) is that the conservative information ecosystem has provided a top-dollar audience for confirmation bias, and there's not much AAPOR or good political journos can do about it.
538 could start by banning them, though
Read 4 tweets
8 Sep
Two of the enduring patterns of polling over the past 20 years are (a) that pre-election polls tend to underestimate the dominant party in a given state, especially lopsided ones (like California) & (b) that polls underestimate the status quo option on referenda and recalls. So..
Don't worry, I have receipts!

For ballot initiatives/referenda:

And see the graph from @MHeidemanns for potus polling errors Image
@MHeidemanns Also, given that the crosstabs for Newsom v Cox pretty much match perfectly with the 2018 results in the state, I'm inclined to believe the media was hoodwinked by partisan non-response and outliers (esp from lower-quality polls) over the last month.

Read 5 tweets
7 Sep
Maybe Newsom was always ahead (as everyone’s prior prediction should have been all along) and people were just overreacting to outliers and temporary partisan non-response that got magnified by over-aggressive polling averages?
I do think maybe people have been conditioned to rely on basic poll averages (like 538’s) too much at the expense of studying the underlying data and empirical dynamics of historical polling on recall elections (big “yes” bias), both of which call the average into question
Anyway, maybe I will be wrong next week, but the underlying conditions of this contest have been pretty consistent — despite what a couple overhyped self-reported likely voter polls have shown

Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!

:(