Adam Carlson Profile picture
Recovering political pollster, current market researcher, and election nerd. Spreadsheets help me relax. Views are my own.
5 subscribers
Nov 9 11 tweets 4 min read
Based on race calls from AP & DDHQ (plus a candidate concession), here’s where the House stands:

🔴 Republican: 216 (+2 tilting R)
🔵 Democrats: 209 (+4 tilting D)

Ds need an inside straight

Let’s go district by district for the 10 races that are still too close to call:

🧵 Image #AKAL

No update, with Begich (R) leading Peltola (D) by just over 10K votes (or 4.1 pts) with 76% reporting.

A lot of the remaining vote appears to be from rural areas, which should help Peltola given its high Alaska Native population.

Probably Tilt R but a lot left to count
Nov 5 10 tweets 4 min read
When you see results starting to come in tomorrow night (& beyond) context is everything

Sometimes mail and/or early ballots are counted first (more D), then Election Day ballots are counted (more R) — causing a “blue mirage”

Sometimes there’s a red mirage

Here’s a guide:

🧵 Sources:

NBC News
nbcnews.com/politics/2024-…

FiveThirtyEight
projects.fivethirtyeight.com/when-will-we-k…

Bookmark both of these (FiveThirtyEight’s breaks down all 50 states, which will be helpful for downballot races)
Oct 31 18 tweets 4 min read
We had a flurry of new polls in PA today that net out to an effective tie.

Those who know me well know I’m not much of an optimist.

But gather ‘round the campfire. Let me explain why I’m bullish on Harris in PA — and it has nothing to do with polls or early/mail voting.

🧵 Let’s start with what happened in 2020.

In its highest turnout election since 1992 (76.5%) — due largely to COVID/mail voting — Biden beat Trump by 1.2 points (50.0 - 48.8), or 80,555 votes.

He racked up huge margins in Philly and its suburbs & Allegheny (home to Pittsburgh). Image
Oct 25 21 tweets 4 min read
Ok let’s dig in, shall we?

The final NYT/Siena National poll (10/20-23, n=2,516 likely voters) shows:

Head-to-head — tied (unrounded)
🔵 Harris 48%
🔴 Trump 48%

Full field — tied (unrounded)

🔴 Trump 47%
🔵 Harris 46%
🟢 Stein 2%
🟣 Others: 2%

What should we make of it?

🧵 The first reaction I’m seeing: “not great for Harris!”

I get that. And to some extent I agree with it.

Biden won by 4.5 pts in 2020 & won the Electoral College by the skin of his teeth (tipping point state, WI, was D+0.6).

Clinton won the popular vote in 2016 but lost the EC.
Oct 23 4 tweets 1 min read
You may not like every decision she’s made or stance she’s taken, but it’s worth acknowledging how remarkable Harris’ campaign has been

Campaign continuity, ground game, raising her net favorability from -16 to -1, gaining 5-7 pts in every swing state — all in less than 100 days Image
Image
Like you (and I) would be lying if you thought her campaign operation would be anywhere near this good even 4 months ago — especially after her 2019 presidential campaign

It was high-risk high-reward swapping out the presumptive nominee so late, and it ended up being the latter
Oct 22 4 tweets 2 min read
New poll among swing state Latinos from the phenomenal @carlosodio & @EquisResearch

Fielded Oct 14-18, n=1,298 registered Latino voters in AZ/GA, NC, NV, MI, PA, WI

Harris leads 55-38

Harris has narrowly taken the lead on being “better for the economy”

weareequis.us/research/latin…Image The gender gap is *very* stark

Harris leads 61-31 among Latinas and is tied 47-47 among Latinos. Harris wins 82% among the youngest Latinas.

Echoing what we’ve seen in other polling of Latinos, there’s a sharp gender divide in who is seen as a stronger leader. Image
Oct 19 6 tweets 2 min read
It’s November 6 at 4am.

GA & NC have been called for Trump.

MI & PA have been called for Harris.

WI is too close to call & we wait on mail ballot drops in razor thin AZ & NV

You think: “shit maybe I should’ve actually volunteered instead of obsessing over tiny polling shifts” Image I get that we all want to know what will happen in advance, and we (mistakenly) look to polling to give us some level of certainty around that.

I get that it’s easier to obsess about polling shifts online than to spend time knocking or donors or making phone calls.
Oct 16 8 tweets 2 min read
From Oct 1-15, we’ve had 16 non-partisan aligned high-quality (2.0+ / 3.0⭐️ per 538) pollsters field national polls (combined total n=28,171)

I’ve aggregated the crosstabs by subgroup here, and compared them to 2020 estimates from Catalist/Pew/AP VoteCast

Some observations 🧵 Image By margin, Trump has made the most gains relative to 2020 among the following voter groups:

• Black (+16)
• <$50K household income (+11)
• Independents w/o leaners (+10)
• Bachelor’s degree (+9)
• Independents w/ leaners (+8)
• Men (+5)
• Non-College Grads (+4)
Oct 13 9 tweets 3 min read
Newly released oversample of Latino voters from the NYT/Siena’s most recent national poll (9/29-10/6, n=902 LVs)

🔵Harris 56%
🔴Trump 37%

Per an avg of post-election estimates from Catalist/Pew/AP VoteCast, Trump’s matching his 2020 vote share but Harris is lagging Biden’s by 5 I’ve been aggregating crosstabs of high-quality non-partisan national polls since Nov 2023

Similar to Black voters, Harris has improved decently on Biden’s vote share among Latinos (55-58% range)

Trump ranges from 36-43%

The high water mark for an modern R is Bush 2004 (44%) Image
Oct 12 9 tweets 2 min read
Here are the updated polling averages from the major aggregators.

Basically zero movement outside of NV (which inched less than a half point to the right due to WSJ poll showing Trump +6).

6 of the 7 swing states are within 1 point on balance, and the 7th (AZ) is at Trump +1.3. Image Basically: the truly bizarre amounts of Dem panic and anxiety from today — exacerbated by media pieces citing anxious D chattering class & vaguely sourced stories of internal polling — were completely out of touch w/ the empirical reality of very steady averages across the board
Sep 26 6 tweets 2 min read
Y’all are looking for polling to be a crystal ball like it was in 2008, 2012 & handful of midterms

But that’s really not polling’s purpose

It’s a blunt instrument designed to measure moments in time, and is inherently prone to sampling & non-sampling error.

Embrace uncertainty Polling’s primary purposes are to:

• Gauge *directional* movement overall and among key subgroups

• Test which messaging is most effective among key segments of the electorate to help inform campaign strategy/decision making

• Inform campaigns’ strategic resource allocation
Sep 26 6 tweets 1 min read
No matter which polling avg you look at, all seven swing states are within a ~2 pt margin in either direction — all well within the range of an historically normal polling error of flipping to the trailing candidate

Which way each state leans is truly in the eyes of the beholder Conversely, we’re also well within the range of an historically normal polling error for the candidate that’s currently leading each state to win it fairly comfortably.
Sep 7 6 tweets 2 min read
I have yet to hear a compelling argument that this is not the current state of play 59 days out:

🟣 President: Toss-Up
🔴 Senate: Lean/Likely R
🟣 House: Toss-Up President:

High quality polling averages all have all seven swing states within three pts

You can argue that the polls will consistently miss in one direction (plausible!), but it’s purely speculative/wishcasting to declare I’m which direction it will miss prior to the election Image
Sep 3 5 tweets 2 min read
I just signed up for ActiVote via their app.

I said my name was Gregor Mendel and that I live in Arizona. I put Phoenix City Hall’s address as my home address.

There was zero verification of any of these things.

My candidate preferences will we reflected in their next AZ poll.

Image
Image
Image
I’m now going to delete my account on their app. This was just to prove a point.

This is why none of ActiVote’s polls should be taken seriously.

Here’s a link to their methodology page:

activote.net/polling-method…
Aug 8 9 tweets 2 min read
New Marquette Law School National Poll

538 Rating: 3.0 / 3.0 ⭐️
Field dates: 7/24 - 8/1
Among n=683 likely voters
Margin of error: ±4.7%

*Full Field*
🟦 Harris: 50%
🟥 Trump: 42%
🟨 Kennedy: 6%
🟩 Stein: 1%
🟪 Oliver: 1%
🟨 West: 0%

*Head-to-Head*
🟦 Harris: 53%
🟥 Trump: 47% Among n=879 registered voters
Margin of error: ±4.1%

*Full Field*
🟦 Harris: 47%
🟥 Trump: 41%
🟨 Kennedy: 9%
🟩 Stein: 2%
🟪 Oliver: 1%
🟨 West: 1%

*Head-to-Head*
🟦 Harris: 52%
🟥 Trump: 48%
Aug 6 8 tweets 2 min read
Some parting thoughts on the veepstakes whirlwind. I just want to get these thoughts off my chest, I genuinely don't care if you agree/disagree with them:

1) Walz skyrocketing from relative obscurity (Minnesotans & election nerds) to Harris' running mate in ~2 weeks is bonkers 2) People are describing picking Walz as a "keep the good vibes going" pick, but you'd be foolish to think that Harris' campaign didn't conduct polls/focus groups on these potential picks

3) We have no idea how the interviews with Harris went, and personal chemistry is critical
Jul 28 4 tweets 1 min read
Tim Walz went from being only known by Minnesotans & mega election nerds to becoming one of the most beloved Democratic governors in the country and a legitimate VP contender in less than one week.

I can’t think of anything else that’s remotely comparable to that speed & scale. Even if he’s not selected as Harris’ running mate (more likely than not), his simple, direct message that “Trump/Republicans are weird” has spread like wildfire in Democratic messaging this past week — including that of the Harris campaign.

He’s left his mark no matter what.
Jul 24 7 tweets 2 min read
My opinion on this doesn’t really matter at all, but here’s where my head’s at on the “big four” potential running mates for Harris, in order of who I want the most to least:

1) Gov. Roy Cooper (NC) Image 2) Sen. Mark Kelly (AZ) Image
Jul 15 20 tweets 5 min read
It seems like a political eon ago, but here are the aggregated crosstabs from well-regarded non-partisan national polls from June 2024

First up — the head-to-head matchup between Biden & Trump, juxtaposed w/a flat avg of post-election estimates from Catalist/Pew/AP VoteCast

🧵 Image Here are the same aggregated crosstabs based on the two-party vote — which eliminates undecideds & third-party votes in the polling, as well as minor party votes for the 2020 results.

This allows for a cleaner, more apples-to-apples comparison between June 2024 & 2020 estimates. Image
Jun 10 4 tweets 1 min read
1) We are a normal polling error away from a decently comfortable Biden win

2) The election is almost 5 months away

3) I’ve never understood what regular people openly panicking actually accomplishes, other than making other regular people more likely to openly panic “How was your day, honey?”

“Pretty productive! I used my reach to be publicly super anxious about Biden’s chances on Twitter, so I definitely did my part to help his re-election prospects & save this country from pseudo-autocracy. Way easier than donating or volunteering. You?”
Jun 2 6 tweets 2 min read
Based on a flat average of the latest ratings from @CookPolitical, @InsideElections, @Center4Politics & @SplitTicket_ here are the 47 House races that are the most likely to flip in November (ranked by most likely to flip).

A star denotes an incumbent.
Image Here’s the full list of races that are deemed to be in play (or likely to flip) by those four race rating/forecasting organizations: Image