Cory McCartan
@corymccartan.com
580 followers 220 following 120 posts
Asst. Prof. of Statistics & Political Science at Penn State. I study stats methods, gerrymandering, & elections. Bayesian. Founder of UGSDW and proud alum of HGSU-UAW L. 5118. corymccartan.com
Posts Media Videos Starter Packs
Reposted by Cory McCartan
andycraig.bsky.social
It is egregious, but this is the normal framing for a normal minority party making normal policy demands in a shutdown standoff, which is how Democrats have chosen to present it. The whole ACA focus has made it easier, not harder, to push a "blame Dems for the shutdown" narrative.
joshuajfriedman.com
Egregious framing from @apnews.com:

"Democrats are making good on their threat to close the government if President Donald Trump and Republicans won’t accede to their health care demands."
apnews.com
BREAKING: Democrats vote down a GOP bill to keep the government open, putting it on track for shutdown after midnight.
corymccartan.com
Yes, conditioning on covariates leads to much weaker assumptions than EI of yore! Akin to the difference between estimating causal effects with a difference in means vs. a regression adjustment.

In our application, we control for geography at a fine scale, & basically recover Freedman's nbhd model
corymccartan.com
The best news is that all of this is implemented in new software we've developed!

I will write more about this next week. I've worked hard to make it ergonomic & efficient!

corymccartan.com/seine/
corymccartan.com
This also leads naturally to a sensitivity analysis that lets researchers evaluate how violations of the key identifying assumption affect their inferences. (Heavily based off of the "Long Story Short" paper by Chernozhukov et al)

We apply this to an air pollution example in the paper
corymccartan.com
How to do estimation with many covariates? Building on DML and Riesz regression, we propose a semiparametrically efficient series estimator that estimates nuisance functions within a restricted partially linear function space. You can get a root-n rate on your main estimand under ~weak conditions!
corymccartan.com
It's the "given covariates" that is so important (and analogous to SOO in causal inference). If you can collect many covariates, EI can be more plausible!

Existing methods (eg King's EI) do not usually leverage covariates, and so implicitly make a very strong assumption that is probably not met.
corymccartan.com
Take estimating vote choice by race, based on precinct data that have vote % and race % (but not both jointly).

We (@shirokuriwaki.bsky.social and I) show that to do EI, you have to believe that each group's preferences are independent of the racial makeup of the precinct, given covariates
corymccartan.com
Very excited to share this week a new paper on EI that is two years in the making! What is EI? It's everywhere! EI is when you try to learn about individual relationships from aggregate data.

We formalize identification & propose an efficient, assumption-lean estimator!
Reposted by Cory McCartan
adambonica.bsky.social
I’m starting to notice a trend in the polling data…

—Top Public Worry: Corruption

—Biggest problem in Fed Gov: Corruption

—Top fear: Corruption

—What one word would you use to describe American government?: “Corrupt”

It’s almost like voters are trying to tell us something.
Top 10 American Fears of 2024 (Chapman Survey)
Horizontal bar chart ranking the top fears of Americans (percentage “afraid” or “very afraid”):
	1.	Corrupt government officials (65.2%, top fear for years).
	2.	Loved ones becoming seriously ill (58.4%).
	3.	Cyberterrorism (58.3%).
	4.	Loved ones dying (57.8%).
	5.	Russia using nuclear weapons (55.8%).
	6.	Not having enough money for the future (55.7%).
	7.	U.S. becoming involved in another world war (55.0%).
	8.	North Korea using nuclear weapons (55.0%).
	9.	Terrorist attack (52.7%).
	10.	Biological warfare (52.5%).
Red bars display percentages; small arrows indicate change from 2023 rankings. Top Public Worries in the U.S. (Yale & GMU poll, May 2025)
Stacked bar chart of worries among U.S. adults. Categories ranked by share “very worried”:
	•	Government corruption (54% very worried, top issue).
	•	Other leading concerns: cost of living (48%), the economy (47%), state of democracy (44%), disruption of federal services (44%), cultural/social divisions (36%), treatment of immigrants (35%), global warming (29%), crime (26%).
	•	Lower worries include job security (17%), health (16%), and being targeted because of identity/beliefs (15%).
Green shades show “very/somewhat worried,” yellow/orange shades show “not very/not at all worried. Perceptions of Federal Government Problems (AP-NORC poll)
Bar chart showing the percentage of U.S. adults who consider various issues in the federal government to be a major problem, minor problem, or not a problem.
	•	Corruption: Overall 70% major, 22% minor, 7% not a problem. Higher among Republicans (78%) than Democrats (63%).
	•	Inefficiency: 65% major overall, with Republicans (81%) much higher than Democrats (55%).
	•	Red tape (bureaucracy): 59% major overall, with Republicans (73%) higher than Democrats (47%).
	•	Civil servants unwilling to implement president’s agenda: More partisan split—Republicans 56% major problem, Democrats 20% major problem; overall 34% major, 36% minor, 28% not a problem.
Title: “Majority of the public believe corruption, inefficiency, and red tape are major problems in the federal government. Word Cloud of How People Describe American Government (Berkeley Democracy Policy Lab)
Large central word: “Corrupt.” Other prominent words: Broken, Chaotic, Dysfunctional, Shit, Clueless, Divided, Inefficient, Crooked, Hijacked, Justice, Woke, Bloated, Untrustworthy, Hopeless, Frustrated, Disastrous, Messy, Sneaky, Turmoil, Delusional. Smaller scattered words include both negative and neutral terms such as Crap, Important, Poder, Resilient, Unfocused, Needs Help. Visual emphasizes “Corrupt” as the dominant public perception.
Reposted by Cory McCartan
anthonylfisher.bsky.social
This is it.

Cancel culture, but for real. The Twitter Files, but for real. Actual, unadulterated collusion between the government and media-owning corporations to censor Americans.

Shame on the MAGA libertarians and fake-heterodox civility cops who helped pave this road to authoritarianism.
corymccartan.com
Error bc the precinct adj graph that redist_map builds is not connected?

You'll need to add edges yourself and pass in the adj graph to the argument in redist_map(). Manual is best, based on state criteria, but you can use geomander for quick & dirty:
christophertkenny.com/geomander/re...
Suggest Connections for Disconnected Groups — suggest_component_connection
Suggests nearest neighbors for connecting a disconnected group.
christophertkenny.com
Reposted by Cory McCartan
upshot.nytimes.com
If Redistricting Goes as Expected, Which Party Will Come Out Ahead?

Democrats would probably need to win the national popular vote by two or three percentage points to retake the House next year.

www.nytimes.com/2025/08/31/u...
Reposted by Cory McCartan
chriskenny.bsky.social
Updated working paper for "Individual and Differential Harm in Redistricting" with @corymccartan.com

Provides an better method to evaluate electoral fairness, focused on individuals and explicit counterfactuals. One framework for party, race, ideology, religion, etc.

Updated draft: osf.io/nc2x7_v2
OSF
osf.io
corymccartan.com
DJT in probability class: "thank you for your attention to this measure"
Reposted by Cory McCartan
chriskenny.bsky.social
Take a look at our new working paper, which separates out changes in gerrymandering and geography to better understand bias in congressional redistricting.

US geography is becoming slightly more balanced in terms of House seats, but competition is decreasing, largely due to geographic polarization.
corymccartan.com
Timely WP from our ALARM project team: we generated over 200k simulated maps for the 2010 cycle, to compare against our existing 2020 simulations. This lets us disentangle changes in gerrymandering from changes in political geography!

alarm-redist.org/papers/ggpre...
corymccartan.com
Overall, geographic bias is a much bigger part of why the House map favors the GOP than gerrymandering is. But gerrymandering makes the House even less responsive to shifts in public opinion.

Time for alternative electoral systems that mitigate these biases and foster competition: MMDs, PR, etc.!
corymccartan.com
Dems gained about 4 seats, on average, from political geography shifts (which were much more extreme on the state level, but canceled nationally)

Plus another 2 from gerrymandering in blue states

But the price was fewer competitive seats, bc of polarization + gerrymandering
corymccartan.com
Timely WP from our ALARM project team: we generated over 200k simulated maps for the 2010 cycle, to compare against our existing 2020 simulations. This lets us disentangle changes in gerrymandering from changes in political geography!

alarm-redist.org/papers/ggpre...
Reposted by Cory McCartan
simko.bsky.social
Great article, and a good example of the continuing value of public, open source research tools (we published these simulations in 2022). When new claims and arguments arise, anyone can use existing resources to examine them.
corymccartan.com
NYT used our ALARM project redistricting simulations to talk about geographic bias against the parties and what counts as "fair" in redistricting!

www.nytimes.com/interactive/...
Is Massachusetts a Gerrymandered State?
Can you recognize a gerrymander?
www.nytimes.com