Ben Schneider

I have my students watch this now-classic video to get them to come to office hours for intellectual vibing (but so few come 😭)

Why most polls overstate support for political violence Misperceptions about the popularity of violence increase public support for it — but you can help change that.

Terrific piece from @gelliottmorris.com this morning.

Whenever you ask if something “might” happen or “may” be necessary, it’s very easy for respondents to think of scenarios where the answer is yes, no matter how unlikely.

I dare not

Learning about your statistics software’s quantile definitions is like learning about your knee cartilage or your house’s plumbing. You only find yourself learning about it if something’s gone wrong.

Obama Aides Say 'Veep' More Accurate Than 'West Wing,' 'House of Cards' "You guys nail the fragility of the egos, and the, like, day-to-day idiocy of the decision-making"

Even back when it was airing, it was obviously an unrealistic fantasy of how persuasion works in politics and government. There are many reasons why Veep is viewed by politicians and staffers as the more true-to-life series. www.thewrap.com/obama-aides-...

It’s not a physical paper, but The Banner does good local reporting for MoCo and Maryland.

There has been basically no (public) reckoning with the democratic strategists who said that "raising the salience" of immigration/deportations in ~March would help Trump, which was obviously wrong at the time and has been proved wrong over the last 6 months www.gelliottmorris.com/p/kilmar-abr...

Excellent reporting and data visualization in this story

There’s no rule of journalistic neutrality that requires you to say something is “unclear” when that is objectively untrue. It would be illegal. There’s absolutely zero ambiguity about that fact. Saying it’s unclear is making a false statement in your reporting.

House, PhD -

A drama, about a brilliant arrogant medical statistician and his team of graduate students - every week receives a new case: collected clinical data. Their job: to 'diagnose' what the research question is.

Quarto To Tell Your Story with Data GASP 2025

In June, I gave a webinar at @amstatnews.bsky.social GASP 2025, "Quarto--To Tell Your Story with Data"

and the recording is now online!

🕸️ #RStats & #Python materials: ivelasq-gasp2025-quarto.share.connect.posit.cloud
📹 Recording: www.youtube.com/watch?v=P5Yf...
🐙 GitHub: github.com/ivelasq/2025...

Who Goes Nazi?, by Dorothy Thompson

Every few months now I re-read this "Who Goes Nazi?" piece from 1941 and am blown away by how it captures the people we are dealing with 80 years later.

harpers.org/archive/1941...

Imagine saying that you’re mad at a dad for requesting sidewalks to keep his disabled son safe. Just…saying it. Out loud. To a reporter.

www.thebanner.com/community/lo...

We are thrilled to announce that our NEW Large Language Model will be released on 11.18.25.

Government watchdog websites go dark as OMB withholds funds from IG committee The Council of the Inspectors General on Integrity and Efficiency is funded largely through a no-year revolving account, rather than the regular appropriations that lapsed Oct. 1.

NEW: At least 10 watchdog websites are dark after OMB withheld funds from a watchdog council, removing access to thousands of watchdog reports and required whistleblower links.

OMB's move to withhold funds - which has received Republican pushback - is separate from the shutdown.

Essentially, you might have categories “a”, “b”, “c” (unordered) or “low”, “medium”, “high” (ordered). For the former you would set ordered=FALSE, for the latter you might want to set ordered=TRUE, so that certain R functions know that your categorical variable is ordinal.

Factors are meant to represent categorical variables, and ordinal categorical variables are a special case. Section 16.6 in the book “R for Data Science” explains the distinction between unordered and ordered factors:

r4ds.hadley.nz/factors.html...

Excited to introduce Vibes from Meta. Eat your slop, piggies!

The following R code: library(dplyr) library(forcats) # Order the `species` levels by frequency penguins2 <- penguins |> mutate( species = fct_infreq(factor(species)) ) # Check levels before reordering penguins$species |> table() # Check levels after reordering penguins2$species |> table()

Yeah, exactly. That's the right tool for this.

Many of you have asked about gifts to us during a shutdown, mainly doordash types of things. No. Instead, donate to local food banks like Manna in MoCo, Capital Area Food Bank in DC, Find Food in PGC, and Food for Others in Fairfax. These serve everyone. Please repost. #fedstrong #shutdown

Yep. It’s very common and reasonable in social surveys. For instance in NHANES or NHATS where sensitive detailed information is used in weighting but can’t be shared without greatly risking respondent confidentiality.

In general I think it's hard to combat scientific misinformation when some of the best research is locked behind an academic paywall, while lots of nonsense gets published free for everyone to read in predatory journals.

Problem #2 is pretty self-evident. Once you include more predictors in your model, you have a different model. This can be annoying in terms of interpreting the model, but it can be a more serious problem if, say, the added predictors are collider variables.

Problem #1 is elaborated on in this response from Breidt and Opsomer to the famous Gelman paper on weighting: projecteuclid.org/journals/sta...

New blog post on the right and wrong way to rake a survey's data to estimates from another survey. As an example, it shows what can go wrong when we rake data from a poll to benchmark data from Pew's National Public Opinion Reference Survey.

www.practicalsignificance.com/posts/npors-...

There’s good literature on this. But there are two problems with this approach. #1 is the data usually doesn’t include all the variables used in weighting, to protect confidentiality. #2 is that by including all these vars in the model, you get a different model than the one you wanted.

I’ve had very good experiences with using ‘uv’ instead of PyEnv/poetry/etc. and with using Positron instead of VS Code.

A little bit of self-promotion, I @ivelasq3.bsky.social will be giving a talk on Positron for RStudio Users!

Hoping to share some perspectives, extensions, and other tips in case you're interested in exploring Positron.

Sign up here: www.meetup.com/rladies-gabo...

Did an honesty researcher fabricate data? Duke professor and behavioral scientist Dan Ariely has been accused of using falsified data in research into ways to make people more honest. New info makes the case against him look stronger.

I mean, Dan Ariely is also infamous for polluting academic literature and pop psychology with dirty data.

www.npr.org/2023/07/28/1...

Q: how did you come to know him?
A: I know a lot of painters
Q: but he wasn't just a painter
A: True. A lot of people don't realize his main interest was architecture!
Q: and his political career?
A: I don't know much about that.
Q: the war?
A: I don't mind if we go light on the Nazi stuff