LLM just said "I appreciate you being direct".

Are they getting passive aggressive now? It was bad enough when my printer started talking, not sure I can handle sass from my computer.

A fun little update of the data from this book - this year possibly the highest number of floor-crossings ever (at least since 1990s)?

For non-Canadians: multiple floor crossings to the Liberal Party recently, which is on the edge of a majority.

Literally just added a slide to my exam review saying WRITE NEATLY. But its hard for them, even if they try.

Can you say more about that? Is it just if the person exists, or also if the article is pure AI?

I've noticed more than one recent op-ed (in other publications) by well known people that are just obviously AI written, and really downgrades the percieved quality of the publication.

There is a partial second gen effect for sure. Basically everyone is more religious than White non-immigrants (CES data, zero is White non-immigrants).

Also very possible there is a "religion vibes" change among young men though, would be really interesting to see.

👋

The last few years have really made it clear that people (including academics) basically only have substantive preferences, almost no procedural principles at all.

Majoritarianism? Sovereignty? Plain text of constitutions?

All completely flipped ideologically. Even among experts!

Congrats, glad to see this out!

What do you mean by more "consistent" changes? Just less variation in effect size?

A few days ago a student was talking to me while standing up, holding his laptop and balancing a glass of water on the keyboard.

I had to stop him and hold the glass myself, it was stressing me out too much.

This is very good👍. The reason its annoying is that it seems like it was written/said by an author/person, which turns out to be false. If it was actually written by some RA, people would also react badly.

There are objections about abilities, theft or whatever, but the authorship thing is easy.

The fact that it affects correlations and not just levels was news to me!

Getting people to use more reversed questions is very doable. SEM is a much bigger lift though.

This seems like a key point in terms of actually convincing new people: that selection on sig is much higher than it "should" be.

a close up of a man 's face with a shaved line on his head . ALT: a close up of a man 's face with a shaved line on his head .

Can't be over the word count if there isn't a right way to count the words

Revising lecture slides, and I updated an example with a new new cross-tab. The next slide has some step-by-step calculations to show how it works.

So, can AI update the slide? Powerpoint has built in Copilot!

No. AI can't update numbers in a slide. But it will consistant lie that it has. WTF.

Interestingly, due the 2025 citizen reform, I think decendants would keep citizenship forever. Don't imagine they were thinking about separatism when drafting it.

Great idea! We should train an LLM on Aristotle and Kant and let it make IRB decisions. Pretty sure it actually doesn't say anywhere in the regulations that IRB members have to be humans...

Oh yes I agree! I just meant we shouldn't look at the their percentage of nulls as correct/natural/unbiased or something, because its influenced by the behavior of other journals.

They really do deserve a credit for a great journal.

Although, probably some reverse selection on significance for submissions happening there too.

You're too busy and don't have time to do all the reviews?

Easy step one, stop reviewing for Elsevier journals.

Later you can also stop editing and submitting to them. (sorry to Electoral Studies, but this calls for market discipline, time to devalue those assets)

So LLMs have guidelines about discussing presidents?

Trying out ChatGTP5.2, and it reasons:

"OK, I’m thinking through how to respect guidelines on discussing influential politicians. Redirecting to general topics like race and elections research instead of focusing on Obama as an individual."

Reviewing PhD applications for next year, and I can't believe the number and quality of them.

We always have good applicants, but this year we could easily admit three full cohorts. And I suspect the acceptance rate will be really high too.

Thanks for having me!

Sure did! Its possible I have eaten more than one already.

I was an external examiner for a PhD defense for the first time today, and it was a great experience. I went in person rather than zoom. That has costs, but I think its worth it. Partly because its an important event, but also somehow being in person makes it important.

10/10 would examine again.

Thanks! And when the new changes happened, I was so glad we had your work from before! Won't be the last time I'm sure.

HA I believe it, so widespread. Alberta is also not particularly conservative!

Thanks!

Linking of immigration to the housing crisis seems much more likely, given timing and age changes. Polarization has been going on for a while, but this recent shift was among all parties.

Some media analysis with @jeremiedrouin.bsky.social is in the pipeline, maybe that will shed more light.

Thanks!

Thanks to @cbreton.bsky.social and @irpp.org for publishing, @parkinac.bsky.social for the invaluable data series, and to @catherineouellet.bsky.social and @skelalex.bsky.social for helping with the early data work!