A graphic showing the growing partisan divide in the U.S. -- a pattern that is specific to partisanship. In 1994, the average difference across 10 political values items was 15 percentage points. That gap more than doubled by 2017. Gaps by other demographic groupings remained more or less constant during the same period.
I recently had the opportunity to dig up this chart from an old Pew report:
www.pewresearch.org/politics/201...
In my years at Pew, this was the most striking image that I had a part in producing. The graphic takes a little explaining, but the payoff is worth it. (1/
06.03.2026 15:32
👍 7
🔁 3
💬 2
📌 1
The same is seen for other demographic groupings (age, education, religiosity, gender) ...
The point is, most of us have spent our formative years during a period when partisanship has come to dominate politics in a way that is actually a pretty recent phenomenon. (5/5
06.03.2026 15:32
👍 2
🔁 0
💬 0
📌 0
A graphic showing the growing partisan divide in the U.S. -- a pattern that is specific to partisanship. In 1994, the average difference across 10 political values items was 15 percentage points. That gap more than doubled by 2017. Gaps by other demographic groupings remained more or less constant during the same period.
The increase is specific to partisanship.
In 1994, the difference between Dems and Reps was roughly comparable to what the difference was between White people and Black people.
The Black/White gap stayed roughly the same over this period, but the partisan gap more than doubled (4/
06.03.2026 15:32
👍 1
🔁 0
💬 1
📌 0
A graphic showing the growing partisan divide in the U.S. across 10 political values items.
So, if we focus in on the partisan difference, we can see a dramatic rise during this period. In 1994, the average difference between Republicans and Democrats across the 10 items was 15 percentage points.
In 2017, that difference had increased to 36 points. (3/
06.03.2026 15:32
👍 1
🔁 0
💬 1
📌 0
A series of small multiples showing the 10 different political values questions included in the scale.
Each line in the plot shows the average difference between by a certain demographic grouping across 10 political values questions that were the favorite children on Andy Kohut from before my time at Pew and so we kept asking them.
The graphic below shows the 10 items in the set. (2/
06.03.2026 15:32
👍 0
🔁 0
💬 1
📌 0
A graphic showing the growing partisan divide in the U.S. -- a pattern that is specific to partisanship. In 1994, the average difference across 10 political values items was 15 percentage points. That gap more than doubled by 2017. Gaps by other demographic groupings remained more or less constant during the same period.
I recently had the opportunity to dig up this chart from an old Pew report:
www.pewresearch.org/politics/201...
In my years at Pew, this was the most striking image that I had a part in producing. The graphic takes a little explaining, but the payoff is worth it. (1/
06.03.2026 15:32
👍 7
🔁 3
💬 2
📌 1
For the Old Heads
28.02.2026 15:32
👍 431
🔁 71
💬 13
📌 1
The @ifbookspod.bsky.social series on this book was a masterpiece open.spotify.com/episode/1YsX...
21.02.2026 21:09
👍 0
🔁 0
💬 0
📌 0
An image of The Score by C. Thi Nguyen
I just finished @add-hawk.bsky.social 's very insightful book on games and life and yo-yoing(?). It defies easy categorization... but it nicely captures some things I've been thinking about for years. Highly recommend!
21.02.2026 20:22
👍 1
🔁 0
💬 0
📌 0
big "AI is so powerful it will be the death of humanity" energy
18.02.2026 15:21
👍 0
🔁 0
💬 0
📌 0
So, about the GOP’s argument that you need ID to prove you’re old enough to drink, so why not in order to prevent “voter fraud”? www.ms.now/opinion/save...
17.02.2026 20:18
👍 232
🔁 73
💬 11
📌 5
Network structure of Colorado Democratic candidates by donor base similarity
Notes: Major Democratic candidates for office in 2026 (Harris, Polis and the DNC have been added for comparison). The positioning of each node of the network shows the relative similarity between the donor bases of each candidate, and for each node in the network the top five connections (in terms of mutual similarity of donor networks) are drawn. Sources: TRACER (2017-2026) for contributions to state office, DIME 4.0 (2017-2024) for contributions to federal office prior to 2025 and FEC for contributions in 2025 and 2026. Record linkage by the author.
New analysis up today on what networks of donors can tell us about the state of the race in 2026:
www.co-political-landscape.org/donor-networ...
17.02.2026 14:36
👍 0
🔁 1
💬 0
📌 0
Network structure of Colorado Democratic candidates by donor base similarity
Notes: Major Democratic candidates for office in 2026 (Harris, Polis and the DNC have been added for comparison). The positioning of each node of the network shows the relative similarity between the donor bases of each candidate, and for each node in the network the top five connections (in terms of mutual similarity of donor networks) are drawn. Sources: TRACER (2017-2026) for contributions to state office, DIME 4.0 (2017-2024) for contributions to federal office prior to 2025 and FEC for contributions in 2025 and 2026. Record linkage by the author.
New analysis up today on what networks of donors can tell us about the state of the race in 2026:
www.co-political-landscape.org/donor-networ...
17.02.2026 14:36
👍 0
🔁 1
💬 0
📌 0
The box cover of Neil Patrick Harris's Box One
My wife and I just finished @nphstuff.bsky.social 's excellent Box One. A thing that was obviously designed and produced with love. If you like puzzles, get yourself a copy of this lovely box.
15.02.2026 03:48
👍 1
🔁 0
💬 0
📌 0
The political science research on VID laws is clear
1.) It doesn't stop fraud, because fraud is non-existent.
2.) It does decrease turnout by 1-2%, but not in a way that advantages Rs.
3.) Much of research on VID was done before Trump, who wins low propensity voters.
13.02.2026 22:57
👍 137
🔁 53
💬 11
📌 3
Of all the small human joys that are spoiled by AI, casting doubt on cute animal videos ranks high on my list
13.02.2026 18:07
👍 3
🔁 0
💬 0
📌 0
Post a tree you've photographed
13.02.2026 01:46
👍 3
🔁 0
💬 1
📌 0
In-person voter fraud is vanishingly rare (and in almost all cases involves good-faith misunderstandings of the law). If you are going to rig an election, you don't do it by trying to orchestrate a massive, coordinated conspiracy involving tens of thousands of individuals.
12.02.2026 15:41
👍 1
🔁 0
💬 0
📌 0
the naive kind of hypothesis testing that says something like:
H0: The effect of education on voting is zero
H1: The effect of education on voting is not zero
Given a sufficient sample size, H0 will always be rejected, but it doesn't really say anything.
11.02.2026 19:02
👍 8
🔁 1
💬 1
📌 0
I might not be posing it well, but it comes out of things I've read from Andrew Gelman on the topic (e.g. this sites.stat.columbia.edu/gelman/resea...).
The difference is between a rigorously derived hypothesis based on theory (think the kinds of theorizing that happens in physics) and ...
11.02.2026 19:02
👍 5
🔁 0
💬 1
📌 0
Surely some of this is the prevalence of "straw man" null hypotheses in the literature, no?
If I posit *some* non-zero effect of education on voting , I will always find a significant result, but the problem is not selection bias in the publication process, it is a poorly specified hypothesis.
11.02.2026 18:30
👍 8
🔁 0
💬 1
📌 0
