This massive new social media deactivion study is bring framed as a nothingburger. But look closely at the data:

Facebook and Instagram deactivation:
-reduced affective polarization by .03 SD (p=.049)
-increased Trump vote share by 2.6% (p=.015)

pnas.org/doi/10.1073/pn… These effects are quite small, but potentially consequential if they actually scale.

Affective polarization has grown by an average of .021 SD per year since 1978.

A 2.6% vote swing to Trump could have won him the 2020 election (and might swing the Senate and House).

The legal document says that Harvard "found her guilty of all charges" and recommended termination

But she alleges that her Research Assistants altered the data files and therefore it's not her fault (but she can't produce any of the original data). https://t.co/F68FP0RxIUstorage.courtlistener.com/recap/gov.usco…

https://twitter.com/danengber/status/1686851180771557376

Here is Gino's argument that her RAs someone prepared the data for analysis and conducted preliminary analysis and she did not run the studies nor handle the data.

Elite private #college prioritize wealth:

For college applicants with the same SA/ACT score, children from rich families were 34% more likely to be admitted than the average applicant, and those from the top 0.1% were more than twice as likely to get in. nytimes.com/interactive/20… The advantage given to extremely rich kids is from a combination of legacy points, reference letters from private high schools, and athletics (which disproportionally benefit wealthy families).

Prestigious public universities do not privilege rich kids in the same way… https://t.co/4LTr99FBHXtwitter.com/i/web/status/1…

Do growth mindset interventions impact students’ academic achievement?

A meta-analysis (63 studies, N=97,672), found major flaws in study design, analysis & reporting

The overall effect (d=.05) was nonsignificant after correcting for publication bias! https://t.co/qdElBfHFRcpsycnet.apa.org/record/2023-14…
This is a pretty shocking finding from the #GrowthMindset meta-analysis:

"Authors with a financial incentive to report positive findings published significantly larger effects than authors without this incentive." https://t.co/HTChj9aCImpsycnet.apa.org/record/2023-14…

43% of White students at Harvard are legacy, athletes, related to donors or staff.

All the universities making social media statements today should end legacy admissions tomorrow. nbcnews.com/news/us-news/s… The use of legacy admissions dates back to the 1920s, when elite colleges, traditionally the domain of wealthy Protestants, became concerned that spots were being taken by Jews and Catholics. nytimes.com/2022/07/13/us/…

In a new paper we motivated people (n = 3,364), to be more accurate by providing financial incentives for correct responses about the veracity of true + false political news.

We improved accuracy + reduced partisan bias in judgements of headlines by 30% nature.com/articles/s4156… @steverathje2 @Sander_vdLinden @roozenbot @nyuniversity @vanbavellab @Cambridge_Uni Across 4 experiments, we found that:

-accuracy incentives improved truth discernment and decreased partisan bias in accuracy judgements

-whereas incentives to find articles that would be liked by one's political in-group if shared on social media *decreased* accuracy judgments!

We recently studied social media & COVID vaccine hesitancy.

Joe Rogan's Twitter followers were extremely distrustful of vaccines: only 27% of his followers planned to get vaccinated.

Amplifying conspiracies on podcasts & social media has consequences. academic.oup.com/pnasnexus/arti… This was the list of social media influences with the lowest rates of vaccine confidence and vaccination rates in the entire sample:

Joe Rogan ranks #2 in the US, right behind Candace Owens.

There was not a political figure or celebrity with followers this low in the entire UK.

Our new paper suggests that #GPT is a cheap, easy & effective tool for text analysis.

We find that GPT (3.5 & 4) is able to detect psychological constructs (eg emotions) WAY better than dictionary methods across 12 languages!

See paper for details: psyarxiv.com/sekf5/
Our paper suggests AI (GPT) will might make automated psychological text analysis more accurate & efficient.

IMHO it will revolutionize text analysis in the social & behavioral sciences.

This paper was led by @steverathje2 @DanMirea4 @sucholutsky @RajaMarjieh @CRobertson500

The @nytimes analyzed partisan language over the past 10 years from 3.7 million tweets, Facebook ads, newsletters and congressional speeches.

Republicans who voted to reject the Electoral College results used far more divisive language than anyone else. nytimes.com/2022/10/22/us/… @nytimes The @nytimes tallied divisive political words from our online polarization dictionary (created by @almogsi & @william__brady) academic.oup.com/pnasnexus/arti…

Republicans used divisive words more than 2X as often as Democrats in tweets, and 6X as often in emails to constituents.

Our new paper provides a rigorous assessment of 742 scientific articles on human behavior during #COVID19

Two independent teams evaluated 19 predictions from our prior paper & found evidence in support of 16 claims, 2 null effects. and 1 with no studies: psyarxiv.com/58udn This was a comprehensive analysis of the claims from our 2020 @NatureHumBehav paper "Using social and behavioural science to support COVID-19 pandemic response" nature.com/articles/s4156…

Our paper made several claims about how behavioral science would be relevant to the pandemic.

A megastudy on a national sample of partisans (n=32,059) tested 25 interventions designed to reduce anti-democratic attitudes & partisan animosity.

The top 3 were:
-intergroup contact
-addressing mistrust in the media
-creating a common ingroup identity
strengtheningdemocracychallenge.org/press-release In my view, this is one of the most impressive studies in the history of the field.

The authors compared 25 interventions on an urgent and critically important issue-reducing intergroup animosity and supporting democracy.

Impressively, 23 of the interventions worked!

Our paper on national identity and public health support during the #COVID19 pandemic is out today @NatureComms

We find that national identification predicts public health attitudes in 67 countries (N = 49,968) and google mobility data in 42 countries nature.com/articles/s4146… To learn more about the key findings from the project and our team of over 250 authors from around the globe, please read this thread for a brief summary:

https://twitter.com/jayvanbavel/status/1448741986782089216

If you want to see how people respond to a fact check over something that is seemingly trivial (eg about the pseudoscientific term "mass formation psychosis") read the thousands of comments on this AP Fact Check.

There is a massive collective effort to bolster false beliefs.

https://twitter.com/APFactCheck/status/1479886452641120261

This video of the Asch Experiment is being shared as evidence of Mass Formation Psychosis.

But people sharing it don't appear to understand what's happening psychologically.

All it takes is a single dissenter and about 95% of people will no longer conform with the majority.

Dr. Robert Malone was on the Joe Rogan Experience and claimed that people advocating for vaccinations + public health are suffering from "mass formation psychosis".

I was asked if this exists as a concept.

I just co-wrote a book on collective behavior and the answer is NO. I have never even seen the term "mass formation psychosis" in a scientific paper.

I did a search on Google Scholar and there wasn't a single published scientific paper using this term.

There was only one website I could find on Google Scholar and it was citing Robert Malone!