Peak 2026 is when you choose your preferred burrito taxi according to who has the less terrible customer service chatbot.

K10 ftw!

Charlie lying on my desk looking forlorn.

Mondays, amirite?

Is that… good?

butlerian community planning hearing

Two new posts from us at @pewresearch.org today (links in replies).

First, it is generally the case that higher income countries are less likely to say their political system needs big changes. But the U.S. is an exception.

This post shows how the U.S. both does and does not stand out on this.

I just spent a bunch of time talking with colleagues about interpreting statistical significance or a lack thereof, and now my head hurts. Their heads hurt too.

Yeah, I am not optimistic...

Hey, accurate frame information is critical!

@jonrobinson2.bsky.social I know you're out of the biz but maybe you know?

Campaign people. If one (or perhaps several) of the national voter files incorrectly have my phone number associated with someone else's record, is there a way to get that fixed?

Asking for a friend who's really tired of getting all these texts...

Is a 55% Replication Rate Too Low, Too High, or Just Right? Tyner et al.

Is a 55% replication rate too low, too high, or just right? Some thoughts on Tyner et al.’s (2026) recent study.

#MetaSci #PhilSci

I had been planning to skip this one but now maybe not.

Like I've said before, it's not fraud if you do it to yourself!

I think that's just called lying

Plus, suppose you ran a clinical trial in Georgia. You'd probably have good reason to believe that the results would hold up in Montana.

I don't see this as simulating anything. I just see this as making some extremely heroic assumptions that are unlikely to hold up.

I disagree. With those kinds of samples, your modeling assumptions may be harder to validate, but the fundamental task is the same. Randomization lets you rule out (or at least put bounds on) certain kinds of problems, which is *incredibly valuable*, but it's still the same fundamental task.

There's a sense in which you're using your sample to extrapolate to people that are not in the sample, but if that's simulation then it seems like all of inferential statistics would count as simulation under that definition.

With IPW, you're trying to model the odds of inclusion for the people in your sample vs. the rest of the population. For model based/assisted weighting, you're trying to predict the unobserved values of Y for all of the non-sampled units in the population.

I tend to think of simulation as something you do to understand a complex process, where you can manipulate various knobs and see how the results play out. In survey weighting, we're not usually (ever?) trying to do that.

I had literally exactly this thought yesterday: "No matter what kind of mood I'm in, by the time I'm done listening to Ante Up I'm ready to go rob a bank."

#30daychartchallenge

Day 11: Physical

Based on Royal Kennel Club registration data from 2016 to 2025, I give you 10 years of the most popular dog breeds in the UK, a crocheted slope chart.

We have Year on the X- and popularity ranking on the Y-axis, as well as a handmade legend.

Mos Def - Do It Now (feat. Busta Rhymes) YouTube video by Eclectic Avenue

m.youtube.com/watch?v=mb2h...

Tribe Called Quest - Scenario (ft. Busta Rhymes) YouTube video by m.usic

Exhibit B.

m.youtube.com/watch?v=Xfgj...

Ante Up Remix (featuring Busta Rhymes, Teflon, and Remy Martin) (Official Video) YouTube video by MOPVEVO

Case in point.

m.youtube.com/watch?v=PLT6...

There a no better sign that a track is an absolute banger than “Feat. Busta Rhymes”.

Impeccable timing

Danger Mouse ‎– The Grey Album This album was made by using two albums. The Black Album by Jay-Z and The White Album by The Beatles. That's all.. every kick, snare, bassline, hi-hat, guitar etc. was taken from the White Album, and

It slaps.

on.soundcloud.com/HPRTzmM9Fzl8...