iiinova — where your digital self keeps evolving

Reposted by Robert (Bob) Kubinec

Matti Vuorre @matti.vuorre.com · 09/10/2026

Send your best work to PCI Psychology (psych.peercommunityin.org) where it will receive high quality & open peer review free of commercial interference. We review a variety of article types from a wide range of psychological topics (psych.peercommunityin.org/help/help_ge...).

23320

Robert (Bob) Kubinec @rmkubinec.bsky.social · 15h

or some kind of Freudian slip?

000

Reposted by Robert (Bob) Kubinec

Xela Hart @xelahart.bsky.social · 20h

🎯🎯🎯

253143763615

Robert (Bob) Kubinec @rmkubinec.bsky.social · 17h

cool idea! we might need an acronym revision, though. what about PANS?

000

Reposted by Robert (Bob) Kubinec

Lukas Stoetzer @lstoetze.bsky.social · 22h

Ever wondered what attributes are relevant in your conjoint experiment? And what assumptions you need to estimate it? Asya and I got you covered. In a new working paper we propose the APNS: the probability that an attribute's level change is necessary and sufficient for the selection of a profile.

34618

Reposted by Robert (Bob) Kubinec

Omar Wasow @owasow.bsky.social · 09/10/2026

“Our research suggests that Trump was not necessarily attracting male supporters who are as confidently masculine as the former president presented himself to be. Rather, our findings suggest that Trump appeared to appeal more to men who may be insecure about their manhood.” — Knowles & DiMuccio

psychologytoday.com

Why Do Some Men Support Aggressive Policies and Politicians?

Does the idea of "precarious manhood" contribute to aggressive politics?

1715939

Robert (Bob) Kubinec @rmkubinec.bsky.social · 09/10/2026

this election season is a comedy of GOP errors. Paxton and Darline Graham are both absolute unicorn failures.

000

Robert (Bob) Kubinec @rmkubinec.bsky.social · 08/10/2026

There's a lot of writing about the shifting loyalties of working class voters but very little on libertarians. Probably enough of them to swing a close election.

021

Robert (Bob) Kubinec @rmkubinec.bsky.social · 08/10/2026

All these white supremacists care about is SI (supplementary information). Mainly because they are unable to write anything coherent in the main text.

040

Robert (Bob) Kubinec @rmkubinec.bsky.social · 08/10/2026

There is secular extremism just as there is religious extremism.

000

Robert (Bob) Kubinec @rmkubinec.bsky.social · 08/10/2026

New York Times needs to edit better (are they using AI? 🤔). Currently their home page is displaying an incorrect headline (for their own article!). The professor who made anti-semitic posts is at "Haverford", not "Harvard."

160

Reposted by Robert (Bob) Kubinec

Alan Elrod @aselrod.bsky.social · 07/10/2026

This is just a relentlessly horrific story. It’s hard to adequately contextualize the moral desolation of this unfolding in the midst of a midterms against a ruling party helmed by inveterate rapists and rape apologists. www.phoenixnewtimes.com/news/congres...

phoenixnewtimes.com

Congressional candidate is raped while repairing her campaign signs

Embarrassed, she initially didn’t want to report attack. Maricopa County Sheriff’s Office's investigation is ongoing. A Phoenix New Times exclusive.

331790546

Robert (Bob) Kubinec @rmkubinec.bsky.social · 07/10/2026

Agnostics are smarter than atheists, finally we have proof

061

Reposted by Robert (Bob) Kubinec

shira mitchell @shiraamitchell.bsky.social · 07/10/2026

blog post: 2 different uses of simulated data 1. Simulated data to learn about the external world. 2. Simulated data to understand how a model works

031

Reposted by Robert (Bob) Kubinec

Stephen Vaisey 🇺🇦 @stephenvaisey.com · 07/10/2026

BTW this means no funding for GSS, ANES, or PSID. GSS 2026 is in the field now, but then it looks like that's it. Devastating for social science.

09797

Reposted by Robert (Bob) Kubinec

Stella Rouse @stellarouse.bsky.social · 07/10/2026

One surprising thing from the Trump era is the level of cowardness and capitulation exhibited by institutions of power. Even in the face of evidence that pushback works and that the law is on their side, they continue to fall in line. It's a stain that will be hard to erase from history.

714748

Robert (Bob) Kubinec @rmkubinec.bsky.social · 07/10/2026

If we can infect chatGPT with the woke mind virus, perhaps the world can still be saved 🦸

081

Robert (Bob) Kubinec @rmkubinec.bsky.social · 06/10/2026

People are allowed to think, but not about their social, economic, or political context 🤔

020

Reposted by Robert (Bob) Kubinec

Dan Garisto @dangaristo.bsky.social · 06/10/2026

NEW: In March the White House released an FY27 budget request that zeroed out the National Science Foundation's directorate of Social, Behavioral and Economic Sciences (SBE). For months it was unclear what would happen. Now, with FY26 over, the damage to SBE is coming to light. My reporting:

science.org

‘Complete devastation’: How NSF gutted its social sciences research

Agency body that funds sociology, linguistics, and archaeology studies in U.S. gave 90% fewer new grants this year

9661515

Reposted by Robert (Bob) Kubinec

Dan Garisto @dangaristo.bsky.social · 06/10/2026

How did this happen? Officially, proposals went through the same process as before: peer review, program officer recommendation, section head approval. Unofficially, the choice was taken out of SBE's hands, going above the directorate head, Christina Freyman and to NSF leadership.

In the past, panels of independent scientists vetted proposals; program officers recommended those found worthy, and section heads approved them. But a new, unofficial process has removed decision-making power from all directorate staff and leadership. Only a few proposals in administration priority areas, such as cognitive science and AI, were allowed to move ahead, according to the former staffers. Program officers sent them to Freyman, who met with Simon Malcomber, NSF’s chief science officer. “Then he would come back with some decision,” a former staffer explains. “So it wasn’t his decision-making. He was in consult with others.” If Malcomber said a proposal was acceptable, a program officer was given the green light to officially recommend it. “They were very, very careful in terms of never putting anything in writing about what the process was,” the former staffer says. NSF did not respond to questions about the process.

213843

Robert (Bob) Kubinec @rmkubinec.bsky.social · 06/10/2026

I am looking forward to seeing studies of the effect of campaign finance from this election. My impression is that big spending has been wildly ineffective so far (though of course we don't observe the counterfactual, etc).

120

Robert (Bob) Kubinec @rmkubinec.bsky.social · 06/10/2026

Generally speaking, proportions are best when you want to standardize, i.e., you want to compare across scales. The binomial works best when you are dealing with aggregated data like surveys. Most poll-tracking models are beta-binomial. For proportions as such beta seems a better fit than binomial.

000

Reposted by Robert (Bob) Kubinec

Aaron Reichlin-Melnick @reichlinmelnick.bsky.social · 06/10/2026

A few weeks ago, US Citizenship and Immigration Services released quarterly data on the legal immigration system through June. The data presents a stark picture of an agency going through a period of total dysfunction. As of June, net application backlogs have risen by 80% since Trump took office.

Chart showing total net USCIS immigration application backlog, from Q4 Fiscal Year 2023 through Q3 Fiscal Year 2026. The data shows backlogs were slowly dropping from net 4.3 million applications in Q4 FY 2023 to 3.8 million in Q3 FY 2024, after which applications began to rise again, increasingly every single month until Q3 FY 2026, where they stand at 7.7 million.

19364141

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

That's where you have Claude write you a web scraping script...

000

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

The next big indie folk band should be named Trending Toward Significance.

030

Reposted by Robert (Bob) Kubinec

🎃Darin S Pumpkins🎃 @darinself.com · 05/10/2026

Yeah, this is a big time impeachment abuse of power.

22823232

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

The race between Hinojosa and Abbott has become sooo close. Two recent polls come to divergent numbers based on "likely voters". This means the two polls made different assumptions about who will turn out to vote. Personally, I'd go with NYT over Fox News for sophistication. In which case, ... 😱

overall Hinojosa abbott polling average (Elliott morris)

010

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

I mean we already let them get away with violating an astonishing amount of copyright, why is anyone surprised that they want to break other laws?

050

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

No one is talking about South Carolina... The two most recent polls have Lindsey Graham's sister up by only 1 point over the Democratic challenger--including a poll by a GOP polling house (!!). Not a Paxton situation per se but giving the seat to Graham's sister did put the seat back in play.👀

051

Robert (Bob) Kubinec @rmkubinec.bsky.social · 05/10/2026

This👇 is an interesting feature of campaign ad spending that I did not know and is apparently boosting Democratic fortunes despite GOP outside group spending. More grassroots fundraising = cheaper ads From David Nir's Downballot substack: open.substack.com/pub/thedownb...

020

Robert (Bob) Kubinec @rmkubinec.bsky.social · 04/10/2026

You could always try fitting an ideal point model (assume ads exist along a polarizing latent dimension) but I'm not sure the clients would enjoy the statistical results.

110

Robert (Bob) Kubinec @rmkubinec.bsky.social · 03/10/2026

Academics who constantly rant about their institution but then start promoting a job opening in their dept

0181

Robert (Bob) Kubinec @rmkubinec.bsky.social · 03/10/2026

Take the feeling thermometer scale. You could think of it as a sum of independent trials if the respondent was asked 100 times, "do you like x?", and the slider then was set at the number of Yes's. But that's probably not the likely way someone would interpret the scale without being told to do so.

010

Robert (Bob) Kubinec @rmkubinec.bsky.social · 03/10/2026

trials in the beta-binomial are independent trials, like flipping a coin again and again. I would not, for example, say that slider data is easily represented as a set of independent trials (unless the respondent was told to think of it that way). so depends as always on what the data represent.

200

Robert (Bob) Kubinec @rmkubinec.bsky.social · 03/10/2026

Kalshi and other prediction markets have plenty of issues but one valuable trait is that they are bit harder to bluff given the incentive to make money. Kalshi now has Hinojosa with a greater than 25% chance of winning the TX governor. That's *way* higher than many pundits.

020

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

I enjoyed working on this project with Jason. It contains a wealth of statistical knowledge of bounded outcomes *but also* computational tools & R packages for fitting and understanding models. I also snuck in an argument about beta distributions & entropy that might become its own paper... 👀

090

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

that's enough, Jarrett, time to log off

030

Reposted by Robert (Bob) Kubinec

JsonGeller @jgeller1phd.bsky.social · 02/10/2026

New tutorial finally in press with @rmkubinec.bsky.social @chelseaparlett.bsky.social and @matti.vuorre.com. It has everything you could possible want: Quippy headings, a whole lot of beta, bayesian analyses and frequentist analyses side by side in harmony. link.springer.com/article/10.3...

rdcu.be

27825

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

In the surveys I've run since LLMs came on the scene--using my own recruitment techniques via social media--I have only seen LLM responses for surveys in which a respondent can get a specific payment. Even then, most of the time respondents don't seem to bother with LLMs; they just make up data.

100

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

I buy that an LLM agent can fill out a survey. At the same time, whether or not someone will do this largely depends on survey incentives. It seems to be mainly applicable to survey panels in which people can fill out many surveys. Otherwise, why would anyone care to do this? It's time & money.

140

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

A lot of attention on manufacturing but a lot of professional occupations have seen minimal job growth, especially IT. Most job growth is in health care which is probably why women have seen such a large increase.

010

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

It would be an interesting move for the Democrats to try to pick up votes from the predominantly male occupations that got torched by AI (comp sci, management consulting, etc)

130

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

A related problem is the rate of false positives: how easy is it to make an argument that a given rock looks like the Arc? If you can see the Arc in any given rock formation... well good luck finding the real thing 😅. Even so, just the prior probability alone should make you very skeptical.

020

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

The probability that a rock would be boat-shaped if it were Noah's arc is pretty high--we should expect that to be true (99% probability). BUT, the probability that any given rock is a long-lost Biblical boat is extremely low--maybe 0.001% probability). Combine them: 0.001*99 = 0.099% -- very low

110

Robert (Bob) Kubinec @rmkubinec.bsky.social · 02/10/2026

Example of bad Bayesian thinking: What they want to know (posterior): probability that the rock is Noah's ark given that it is boat-shaped. What they know (likelihood): if the rock were Noah's arc, it would be boat-shaped. What they forgot (prior): probability that any given rock is the Arc

230

Robert (Bob) Kubinec @rmkubinec.bsky.social · 01/10/2026

the problem is that people don't pay attention to statistics, they pay attention to unusual stories. so the one person killed in Central Park stays in people's minds, while the many killed by cars are just faceless numbers.

110