David Epstein

@davidnotdave.bsky.social

1199 Followers 1136 Following

Literature major, then neuroscience Ph.D., then addiction researcher. Enough-knowledge-to-endanger-myself in social sciences and statistics. He/him. Views my own.

To appear in: Contemporary Clinical Trials Communications Received Date: 2 June 2025 Revised Date: 5 June 2026 Accepted Date: 26 June 2026 Results This study is ongoing and data collection is expected to be concluded in July 2025, with plans to publish the study results by November 2025.
Screencap from John Oliver's May 4 "Last Week Tonight," showing a paper about tianeptine that was written by Kirsten Smith, Jeff Rogers, Justin Strickland, and me.
Screenshot of an 1970s SNL commercial parody called "Placenta Helper," which famously never aired.
A graph of approval/disapproval ratings for 47 as their shape widens over time like a sideways letter V (disapproval up from 41% to 62%, approval down from 47% to 36%). They form a set of shark's teeth, inside the mouth of a shark in pursuit of him. Header, above the shark: "Help him 'tank.'"
Side one of sign: Stephen King's "Misery" book cover Photoshopped to say "No Kings, Less Misery." Side two of sign: Stephen King's "It" book cover Photoshopped to say "Kings Ain't It."
Text excerpts from "Latent Classes of Adverse Childhood Experiences (ACEs) and Adult Substance-use Problems and Psychosocial Outcomes: Complex and Heterogeneous Associations." (International Journal of Mental Health and Addiction)
Micro interventions based on real-time monitoring of affective instability: a proof-of- concept study. Merlijn Olthof et al. Excerpt 1 of 3: Design: We used a quasi-randomized within-person experimental design for testing the micro-intervention effects during increased EWMSD [exponentially weighted moving standard deviation of affect ratings] vs. control periods (with EWMSD below the threshold). At the start of the study, participants were assigned at random to one of two phases: the experimental phase (micro-intervention is applied when there is increased EWMSD) and the control phase (micro-intervention is applied semi-randomly). [David’s comment: This is the appropriate control for Forer effects, and no one else ever uses it!] Excerpt 2 of 3: Interestingly, from the data viewing sessions, we also learned that there was a large heterogeneity in how persons experienced - and reacted to - the intervention and burst measurements. While some persons were really enthusiastic about the intervention, others found it annoying or confronting because they found that they had no time to engage in the intervention. Some persons never actually performed the activity they selected, some nearly always performed it in the next 30 minutes, but others used it as 'a reminder'… [David’s comment: This is the kind of frankness we need more of.] Excerpt 3 of 3: These results can inform future research, which, who knows, may at some point be able test the hypothesis that we were after. [David’s comment: OK, marry me.]
Tweets from June 8, 2020. Payton Jones: “You may want to reconsider your use of trigger warnings. Our new paper, just appearing in Clinical Psychological Science, suggests they may do more harm than good.” Me, in a series of five replies: “This is a nicely conducted experiment, but it's a very limited operationalization of the broad question it means to address— limited both by the stimuli used and the responses assessed….The "markedly distressing" cues (text from Dosteovsky, Ballard, etc.) were tame indeed. I don't see anything in the Results to indicate that they even markedly bothered anyone! So, not a very stringent test of whether trigger warnings help or harm. Re responses assessed: If nothing else, use of TWs conveys courtesy and respect. Presentation of highly disturbing material without TWs can convey an indifference or erasure that might be inherently distressing to people with trauma histories. Your study didn't assess that aspect of people's responses to your stimuli and TWs--a major omission in any consideration of TWs' cost:benefit or risk:benefit ratio. (But, again, you would have needed nastier stimuli to assess this.) In sum: an interesting demonstration and a very worthy endeavor, but I wouldn't use it to conclude that if I plan to show graphic films of sexual assault to a roomful of survivors, the best thing I can do for them is omit a warning. This study won't generalize to that.”
Side one of sign, from an old issue of the Onion's "American Voices" feature: "Phrases like 'massive redistribution of wealth' have historically brought rivers of blood. That said, it's time for a massive redistribution of wealth." Side two of sign: "Austerity? You go first, E.M. I'll join in if it looks good."
Graph of Tesla stock freefalling: "You can't keep a good man down. But THIS schmuck? You can."
Side one of sign: "Let's get a little more juice in that third rail." Side two of sign: "Uncap FICA."
Abstract: Craving is a central phenomenon in cocaine use disorder (CUD), significantly influencing relapse risk and treatment outcomes. The STANDUP® method, an innovative telemedicine-based intervention, integrates structured behavioral modification, cognitive restructuring and digital therapeutics to target reward, relief and obsessive craving subtypes. The six-month structured routine program includes daily behavioral exercises that focus on: • Breaking addictive cycles through structured daily planning (Mappa del Giorno). • Cold showers as a physiological and psychological tool for increasing self-discipline and stress tolerance.
Alcohol use disorder symptoms are kinda detectable in daily life (but it depends on how they are operationalized) Kevin King Dani Kang Megan Schultz Christine Lee Jonas Dora Cassandra Boness Ashley Watts
Screencaps of tweets from 2022, in which someone wrote: "i feel like you could shrink joss whedon's face and no one would notice. like 15-20% shrinkage and you could do it undetected. he might not even notice it in the mirror" So I obliged.
From the ACNP website: "NEW in 2024: Member invited guest invitations will be limited to 400. The College has reached the limit of 400 member guest invitations. Member guest invitations are no longer available for the 2024 Annual Meeting."
Screenshot of output from an AI app called the Bluesky Roaster (blueskyroast.com), based on my Bluesky content. "David Epstein: The Neuroscience Nerd with a Side of Sarcasm." "David's tweets are like an academic paper that took a detour through a comedy club—insightful, often dense, but always with a punchline waiting at the end. Just when you think you're lost in the jargon, he'll throw in a zinger about convenience samples that feels like a punch to the gut (in a good way)."
At the time of preregistration, we liked this idea because it is a more stringent criterion compared to statistical significance. However, applying this criterion in practice made us realize that this is not the most suitable strategy for replication: 1) it is arguably overly stringent, requiring near-perfect alignment between the studies, which is unrealistic given natural sampling variability; 2) it can lead to an apparent lack of replication even if estimates are close if Cls are narrower in the replication study. For that reason, below we discuss our results both in light of this preregistered criterion and the overall strength and direction of effects. Megan E. Schultz, M.S. Jonas Dora, Ph.D. Kevin M. King Ph.D.
Illuminated white pumpkin on our steps, HARRIS-WALZ '24 carved in.
Mood induction: Three 233-second-long videos were shown to participants on a 35-inch color television. The boring video was created for this study and portrayed two men hanging laundry, occasionally asking each other for a peg. We validated this clip in a separate study of 48 participants (mean age 21.5 years; 33 female) to ensure the video reliably induced feelings of boredom
Andrew Wyeth's Christina Photoshopped onto the slope of the grass in Kurt's photo
...the following report, which we originally planned to publish in 2023. We had initially framed it as an update on a rapidly evolving situation of kratom product diversification...but we now offer it for a more enduring purpose: to help ensure that peer-reviewed publications about kratom do not reflect the polarization seen in popular discourse. If specific groups of investigators consistently emphasize only positive or only negative aspects of kratom use, the credibility of their contributions may suffer. Our recent publications about kratom largely illustrated how kratom use appears to benefit many consumers. Most of our participants had found products and dosage patterns that appeared...satisfactory to them and not detrimental to their health or daily functioning. We stand by those findings, but we have always noted that they are not the whole story....Here, we give voice to consumers who believe that their relationship with kratom was not unambiguously positive or beneficial...
Screenshot of paper titles from a Google Scholar search: From lab to life: Evaluating the reliability and validity of psychophysiological data from wearable devices in laboratory and ambulatory settings. From Lab to Life: Bridging Gaps in Motion Capture to Increase Public Usability through Integrated Hardware and Software Solutions. From Lab to Life: Phoneme Recognition Systems Making a Difference in Assistive Technologies. From lab to life: Decoding the predictive power of present bias on adolescent self-control behaviors. From lab to life: how wearable devices can improve health equity. From Lab to Life: Challenges and Perspectives of fNIRS for Hemodynamic Neurofeedback in Real-World Environments. From lab to life: Fine-grained behavior monitoring in the elderly's home. Serial versus concurrent multitasking: From lab to life. Translations in gerontology: From lab to life: Psychophysiology and speed of response.
I bought a cow for $800. I sold it for $1000. I bought it again for $1100.1 sold it again for $1300. How much did I earn?
text excerpts: "The importance of the results from Study I are threefold." "The argument against class criterion-oriented psychometrics is thus two-fold." "the number of pharmacies distributing buprenorphine was 43.1-fold greater than OTPs distributing methadone in 2010. However, this ratio decreased to 33.3-fold in 2021." "MOR agonist that is 100-fold more potent than morphine" "Aims were threefold. In study one..." "The aim of the simulation study was threefold." "two-fold higher than crystal MDMA"
Screenshot from a published protocol called "Personalized Deep Learning for Substance Use in Hawaii: Protocol for a Passive Sensing and EMA Study." Excerpt: "As a precaution, the interface on the app will not be labeled as 'substance consumption' and 'substance use' but rather as 'banana consumption' and 'banana use.' Furthermore, our data will be stored and labeled as 'fruit' rather than 'substance use.'"
John Hoenig and Dennis Heisey: "We feel that researchers often do not need a rigorous understanding of confidence intervals to use them to good advantage. Although we cannot demonstrate it formally, we suspect that imperfectly understood confidence intervals are more useful and less dangerous than imperfectly understood p values and hypothesis tests. For example, it is surely prevalent that researchers interpret confidence intervals as if they were Bayesian credibility regions; to what extent does this lead to serious practical problems? .... If informally motivated confidence intervals lead to better science than rigorously motivated hypothesis testing, then perhaps the rigor normally presented to students destined to be applied researchers can be sacrificed."
Two screenshots from competing papers. Geoff Cumming, 2013: "A value close to the center of the CI is about 7 times as likely to be u as is a value near a limit of the 95% CI. Thus, the black area is the likelihood profile, or beautiful 'shape' of the 95% CI. Our CI defines an interval of plausible values for u, but plausibility varies smoothly across and beyond the interval, as the cat's-eye picture indicates." John Kruschke and Torrin Liddell, 2018: "Notice that a confidence interval has no distributional information. There is no direct sense by which parameter values in the middle of the confidence interval are more probable than values at the ends of the confidence interval. This absence of distributional information is why the confidence interval in Fig. 3 is drawn as a flat line."
If your sample size is 50, then the smallest correlation that will be statistically significant (at .05, two-tailed) is .28. An observed r of .27 in that sample cannot be statistically significant (at .05, two-tailed). And yet… A power analysis would tell you that you have 48% power to detect that correlation with n = 50. The power analysis actually means you have a 48% probability of getting a p value <.05 with n = 50. When that happens, it can only mean that the correlation in your sample is spuriously large: bigger than .27. It's impossible, with n=50, to detect the correlation at its true size (r=.27) and have it be statistically significant. Presumably, this problem is less extreme at conventional power levels (80% or more), but the underlying fact seems to remain.
The Science and Ethics of Measuring and Modeling Individual and Group Behavior. Cheryl Corcoran, Chair. Holly Moore, Co-Chair. Elizabeth Stafford, Moderator. Laura Cabrera, Justin Baker, Tingting Liu, Satrajit Ghosh, and Sara Berger, Participants.
I don’t usually say “reject.” I usually say, “Just present the findings more tentatively.” And I’m sure the authors will want to do so. They’ll say something like, “Let’s publish this as a preliminary finding. It will inspire future research to resolve the ambiguities.” But it’s not a case of “no harm done.” People love genetics and they love to overread it. There will be a press release and a tweet and a TikTok dance, all trumpeting how genes determine...