Steven Mesquiti

🚨🚨 My lab is hiring a Postdoc! 🚨🚨

Postdoc opening in my lab at Harvard Psych (start: Summer/Fall 2026).

We study identity, group dynamics, language, & politics.

Rolling review begins 3/14
Apply: academicpositions.harvard.edu/postings/15806

Please repost + tag folks who might be a good fit! 👇

Very cool paper, congrats!

Huge congrats to @logicemotionlab.bsky.social PhD student Steven on his excellent work!

Are you a junior faculty member interested in spending 2-4 weeks at Princeton Psych? Consider applying for our Microsabbatical program! It’s a fully funded visit for professional development and creating long-term collaborations.
psych.princeton.edu/diversity/mi...

#AffectSci

Thank you! Hope you're doing well too.

Lastly, huge thanks to my co-authors @dcosme.bsky.social @eriknook.bsky.social @falklab.bsky.social & @shannon47burns.bsky.social for helping me to develop this project as a lab manager and see it through in the first-year of grad school.

Language-based assessments can predict psychological and subjective well-being - Communications Psychology Across three studies, the ability of language-based assessments to capture subjective and psychological well-being was assessed. They performed well for life satisfaction but were less accurate for au...

Paper: doi.org/10.1038/s442...
Code: github.com/cnlab/LP2-we... 7/7

This matters for the field as language-based tools show real promise for assessing hedonic well-being, but we shouldbe cautious about assuming they'll work equally well for more abstract, eudaimonic constructs like autonomy. 6/

Our language visualizations suggest satisfaction with life responses cluster around clear positive emotion and social themes (love, family, grateful). Autonomy responses are more individualized, abstract, and potentially harder for models to pick up on. 5/

When we benchmarked BERT against GPT-3.5 and GPT-4, the newer models significantly improved satisfaction with life predictions (GPT-4: r = .75) but only modestly improved autonomy predictions (GPT-4: r = .49). 4/

Key finding: Language-based models predicted satisfaction with life reasonably well (rs = .43–.75 depending on model & study), replicating prior work by @oscarkjell.bsky.social l et al. But, for us, autonomy was consistently harder to predict across all studies and models. 3/

We asked participants to reflect on their satisfaction with life and autonomy, then used transformer-based models (BERT, GPT-3.5, GPT-4) to predict their questionnaire scores from their language alone. 2/

🧵New paper out in @commspsychol.nature.com We tested whether language-based assessments (LBAs) can predict not just subjective well-being (life satisfaction) but also psychological well-being (autonomy). Across 3 studies, we found that LBAs work well for some constructs—but not all! 1/

New preprint! We investigate the risk of algorithmic bias (across race/ethnicity, gender, and their intersection) in machine learning models predicting suicide attempts across 3 clinical settings in over 1.2 million patients.

osf.io/preprints/ps...

Citations are evidence. Fabricating citations is fabricating evidence, which is just as bad as fabricating data (because data, like citations, is evidence).

🚨HOLY CRAP. An ICE whistleblower just revealed a secret memo authorizing ICE officers to break into homes without a judicial warrant, which DHS's own legal training materials say is unconstitutional!

ICE then hid the memo from the public, passing it along by word of mouth and private conversation.

I submitted five grant applications, and today I got the final rejection (0/5). Trying to get results from a pre-registered report published has also been an uphill battle, including reviewer comments that my work is not science but "activism," and rejections despite glowing reviews.

graphic of the words work and play arrayed in a 2d semantic space constrained by valence and dominance

Teaching semantic distance in my NLP class. Here's a super simple graphic on how to calculate Euclidean distance between work and play in a 2D semantic space constrained by valence and dominance.

We're hiring! The Computational Clinical Science Lab @ Yale is seeking a full-time lab manager/research coordinator to start in early summer 2026.

For more information about the position and to apply: forms.gle/LtQwVgPUfaGk...

Please share widely & consider applying!

Large Language Models Can Forecast Suicidal Ideation From Future Self-Narratives: https://osf.io/fhzum

Seven CEOs and one scientist (half cut out of the picture)

Congratulations to everyone who wrote an op-ed about how he just wants americans to eat better

Trump moves to dismantle major US climate research center in Colorado The Trump administration is breaking up the National Center for Atmospheric Research, taking aim at one of the world's leading climate research labs.

Exclusive: The Trump administration is moving to dismantle the National Center for Atmospheric Research in Colorado, according to a senior White House official, taking aim at one of the world's leading climate research labs.

🚨New paper forthcoming in the Journal of Affective Disorders!

Regulatory flexibility—using strategies based on situational demands—has been a central topic in affective science in recent years. It’s often assumed that more flexibility is better. But is that really the case?

osf.io/preprints/ps...

Just want to share with my American friends how the Australians respond to a shooting tragedy. Action, rather than thoughts and prayers.

Screenshot of title page of an article published in "American Psychologist" titled "Disentangling Autonomy-Supportive and Psychologically Controlling Parenting: A Meta-Analysis of Self-Determination Theory’s Dual Process Model Across Cultures."

Meta-analysis shows autonomy-supportive parenting is positively related to child well-being and psychologically controlling parenting is predicts child ill-being. Effects hold across regions of the world & various measures of cultural ideas. #PsychSciSky #AcademicSky #EduSky
doi.org/10.1037/amp0...

Awesome, can't wait to see where y'all take this work!

Representation in science and trust in scientists in the USA - Nature Human Behaviour Druckman et al. document gaps in trust in scientists in the USA. People from groups less represented among scientists (for example, women and those with lower economic status) are less trusting. Incre...

Paper out today in @nathumbehav.nature.com:

1) those groups (women, African Americans, lower SES, rural) that are underrepresented in science have been less trusting of science.

2) If you improve representation in science, you improve trust among those groups.

www.nature.com/articles/s41...

Intro stats courses need to provide lots of tangible examples of effect sizes. I remember reviewing a paper about the effect of Tylenol on social exchanges once. It had an ES of d~1. So, as strong as effect of opioids on cancer pain. Just, no way. If it was true we’d need warning labels.

1/2