Kevin Cummins

Public health researcher, biologist, research methodologist, educator, and open gov advocate

[Kevin Cummins @Kcummins760]

One good experiment is worth 10 meta-analyses. Wisdom does not emerge from 100 poorly designed short-term studies ~ Charles Krebs #science

[Kevin Cummins @Kcummins760]

@DasOptimist @SwipeWright @mitzi_turpin No

[Dale Chu @Dale_Chu]

More proof that the battle for student success is won or lost in the early years.

New data from 400K students shows that if a child is behind at the end of 1st grade, their chances of hitting 3rd grade proficiency drop to a brutal 2%.

[Eric Topol @EricTopol]

We stress tested many frontier AI models for multimodal medical reasoning (including GPT-5, Claude 3.5, Gemini 2.5 Pro). They’re not ready. Faulty reasoning, use of inappropriate shortcuts, hallucinations. Published today @NatureMedicine https://t.co/P6eHZEmfbW

[Frank Harrell @f2harrell]

Wow this is a truly great paper. I'm a big fan of David Hand and own his Measurement book, so I don't know how I missed this for years.

[Steve Hare @sharemath]

In 𝘏𝘰𝘸 𝘞𝘦 𝘓𝘦𝘢𝘳𝘯, Stanislas Dehaene—one of the world’s leading cognitive neuroscientists and winner of the Nobel-equivalent Brain Prize—identifies the 4 Biological Pillars of Learning. Without all 4 of these pillars in place, learning is fragile and will not last: 🧵

[Olivier George @brainaddiction]

@BruceHope4 @yavinshaham @NIDAnews And a cue that signaled "drug not available" suppressed seeking just as long, even when the rat received an injection of heroin.

This is actually the most important finding here when you think about rehab/treatment. We should build stronger learned inhibitors of drug seeking.

[Florian Ederer @florianederer]

Access to broadband mobile phone networks radically decreased in-person socializing among teens and led to a decrease in teen fertility and an increase in teen suicide.
https://t.co/plImh8n4Zy

[Kevin Cummins @Kcummins760]

@jayvanbavel Maybe it’s not the choice but the chooser.

[Kevin Cummins @Kcummins760]

@cremieuxrecueil @Nullsci1 Is there range restriction? I think Ppl who get a low score on CBest don’t get to teach (until recently that is).

[Kevin Cummins @Kcummins760]

@drantbradley Seems like some state schools near me plan to address this by lowering entrance standards and setting up automatic acceptance/matriculation for local high school students. This is not good.

[Kevin Cummins @Kcummins760]

@Noah_van_Dongen @lakens Do you need informed consent?

[Jonathan Haidt @JonHaidt]

As we all comb through Meta's internal docs, released via the lawsuits, it's fun to discover that @jean_twenge and I were getting under Meta's skin by 2019. See the screenshot below, from:
https://t.co/ARcQn8uh35

They then ran their own RCT to show that Jean and I were wrong, but oops, they found that when people were randomly assigned to quit Insta or FB for a week, their mental health improved.

They called it "Project Mercury."
It is study 6.2 at
https://t.co/Nv14Y2n2kH

They found that those who stopped using Facebook for a week "reported lower feelings of depression, anxiety, loneliness, and social comparison"

Their own researchers (all of whom had Ph.Ds) concluded that: “the Nielsen study does show causal impact on social comparison.”

So whenever you hear the Meta talking point that the evidence is "just correlational," or that there is "no evidence of causality," know that this is not true, and they have known it's not true since 2019.

Please check out the other 30 internal Meta studies we found and catalogued at
https://t.co/Nv14Y2n2kH

Also, there is SO MUCH additional academic research demonstrating causality since 2019:
https://t.co/e2ifxADjd9

[Frank Harrell @f2harrell]

@5_utr I hate to see @JAMA_current favor any multiplicity adjustment procedure. Multiplicity is MUCH better addressed by making assertions that factor in clinical significance instead of testing null hypotheses: https://t.co/pEtwUVCnvA

[Olivier George @brainaddiction]

Very happy to share a new paper in @npp_journal
that addresses a long-standing question in addiction science:
What actually drives the escalation of cocaine use over time?
Big thanks to Jarryd Ramborger, undergraduate student extraordinaire, who led this work.
I thought I knew what would predict escalation of cocaine use. Well, I was wrong. After tracking behavior daily for weeks, the answer turned out to be the opposite of my prediction, and it changed how I see addiction. 👇🧵

[Kevin Cummins @Kcummins760]

@CoachJenna10 @tomloveless99 Does that body of research distinguish between types and quality of homework?

[Jonathan Haidt @JonHaidt]

Here are the 2 largest projects we've ever done to catalogue the evidence that social media is harming teens at an industrial scale:

1) A review paper, in press
2) A new website that presents 31 internal Meta studies

https://t.co/B8bAgWTfEG

[Steven Pinker @sapinker]

Excellent essay by Peggy Sastre on Napoleon Chagnon's persecution by his anthropological colleagues for his empirical approach & politically incorrect findings about the Yanomamo. (I knew him well from 2 sabbaticals at UCSB, & Sarah Blaffer Hrdy's characterization is right on.) https://t.co/Xza72FS93U

[Anna Stokke @rastokke]

💡 If we want better results in math, we need a better approach.
Better math curricula.
Better teacher preparation.
Instruction aligned with how humans learn.
Reliable checks of student learning.
It's overdue.

[Kevin Cummins @Kcummins760]

@drantbradley expectations of grade inflation plus AI in the mix has me self-exiled from the classroom until I find an education system that wants to take this on. We are failing students.

[Kevin Cummins @Kcummins760]

@eduleadership It does in my classrooms… been doing SBG for 15 plus years. now ask me the consequences when I get a big cohort of students seeking to gather points rather than knowledge and skills who are accustomed to heavy grade inflation.