iiinova — where your digital self keeps evolving

The plausibility of this could be tested (not with these data) by repeating the analyses with two assessments a few minutes apart.

010

Jonas Dora @jonasdora.bsky.social · 4h

But on the 0–4 scale, only a 2+ point jump counts as reliable unless the threshold drops to 1 or below. For that, 59–73% of the rating changes between those prompts would have to be "real" shifts in the feeling. Is that plausible? I think right now it's hard to say.

110

Jonas Dora @jonasdora.bsky.social · 4h

Feelings change for many reasons besides stressors, and stressors were only asked about for the 30 min before each prompt. So it's an important limitation: any "real" change hiding in those pairs raises the noise threshold, which widens the gap between felt and detectable change.

110

Jonas Dora @jonasdora.bsky.social · 4h

Good questions! I think @aleksakaurin.bsky.social can respond better, but my attempt: yes, the noise estimate treats all change between two scheduled prompts where no stressor was reported as noise. (That sets the threshold; a big jump in such a pair still counts as "real".)

120

Jonas Dora @jonasdora.bsky.social · 6h

And people's own "I feel different" reports aren't the solution: they catch real changes but often report change the ratings don't show. What could help: finer scales, multiple items per state, and acting on a single change only if it's bigger than the noise (here, 2+ points).

020

Jonas Dora @jonasdora.bsky.social · 6h

The average effect can still hold across many reports, and since noise shrinks it, that could mean the true effect is even larger. But it can't tell you that this person's stress actually rose right now. That matters as soon as you act on single moments, like JIT interventions.

120

Jonas Dora @jonasdora.bsky.social · 6h

My takeaway from this study: on a 0–4 scale, a 1-point change may be just noise, even though people do feel changes that small. So if a model finds that a 1-point increase in stress predicts, say, some behavior, that effect is about shifts we can't reliably detect in any single moment.

120

Jonas Dora @jonasdora.bsky.social · 6h

Big changes were rarely missed (sensitivity .77–.87), and people usually got the direction right. But they often reported change when none was reliable. Specificity was about .5, and only about 1 in 4 "I feel different" reports matched a reliable change.

120

Jonas Dora @jonasdora.bsky.social · 6h

The MIC was 1 point, the smallest change the scale can show. Even when the rating didn't move at all, people said they felt different 35–45% of the time. So the bar for feeling a change (1 point) sits below the bar for measuring one (2 points): Felt, but not detected.

120

Jonas Dora @jonasdora.bsky.social · 6h

The SDC was 1.56–1.92 points on a 0–4 scale. Since ratings move in whole points, that means only a jump of 2+ points counts as reliable. About three-quarters of all consecutive changes fell within the noise.

120

Jonas Dora @jonasdora.bsky.social · 6h

That gives the Smallest Detectable Change (SDC): how big a change must be before you can trust it isn't noise. And we calculated the Minimal Important Change (MIC): at what change size do people start saying "yes, I feel different"?

120

Jonas Dora @jonasdora.bsky.social · 6h

First, we asked how much do ratings vary when nothing much happened? We defined "stable pairs" as two back-to-back scheduled prompts with no stressors reported. It's like stepping on a bathroom scale twice back-to-back: the difference is mostly noise.

130

Jonas Dora @jonasdora.bsky.social · 6h

In this study, 250 adults completed 2 scheduled prompts a day plus self-initiated prompts after stressful events for 14 days. Each time, people rated current stress, positive and negative affect (0–4). They also said whether each had changed since the last prompt (−2 to +2).

130

Jonas Dora @jonasdora.bsky.social · 6h

Very cool new paper by @aleksakaurin.bsky.social, Mario Wenzel, @paulaphilippi.bsky.social using some of my K99 data: osf.io/preprints/ps... EMA studies often ask people how they feel several times a day. But when is a change bigger than measurement noise? And does the person actually feel change?

3125

Reposted by Jonas Dora

Danielle Navarro @djnavarro.net · 05/10/2026

In which the author disagrees with Mr Wolfers and discusses what it *really* looks like when you sit down and have a conversation with your ideological opponents. When you do not have the protections of the ivory tower it turns out to be a much darker experience than he imagines it to be.

blog.djnavarro.net

Dining with wolves – Notes from a data witch

Polite discussion over lunch is not a reasonable option when you are on the menu

37535

Reposted by Jonas Dora

Kevin M. King @kevinmking.bsky.social · 29/09/2026

This is ridiculous. The only avenue for fellowships for international scholars has been the K99. It was already so hard to get, and now there is no way for them to come and train and build a career in the US. It will decimate our ability to recruit the best scholars around the world.

01312

Reposted by Jonas Dora

ProPublica @propublica.org · 26/09/2026

Hey, ProPublica reporter Jake Pearson here. I’m taking over the account to tell you about my new story. For months, I posed as a problem gambler on DraftKings to test the company's "responsible gaming" safeguards. Spoiler alert: They made me a VIP — and I lost $10K. Here's how it all went down. 🧵

Selfie taken by ProPublica reporter Jake Pearson, a man wearing a backwards baseball cap and an orange T-shirt.

19596023538

Reposted by Jonas Dora

Andrea Howard @drandreahoward.bsky.social · 18/09/2026

New paper from me, out soon in Behavior Research Methods: I'm a lil bit obsessed with good online recruitment and the result is this side-quest paper that made use of data we had collected that recruited people online in three ways (met them on Zoom, validated their email, and Prolific). 🧵>>

35618

Reposted by Jonas Dora

Tom Rodebaugh @drrodebaugh.bsky.social · 15/09/2026

I'll be reviewing applications for a new student or two for UNC-CH's clinical psychology PhD program! Social anxiety, pain, intensive longitudinal data, all that fun stuff. What prospective applicants get if they email me (so no need): docs.google.com/document/d/1... Lab: tarheels.live/rodelab/

tarheels.live

Welcome to the Rodebaugh Lab Website - Rodebaugh Laboratory

Welcome to the Rodebaugh lab website! Here you can learn about what Dr. Rodebaugh & colleagues are working on and what research is being conducted.

12318

Jonas Dora @jonasdora.bsky.social · 10/09/2026

I’m reading that as well right now! But I’m only on book 6

110

Reposted by Jonas Dora

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I'm super excited to share a new preprint in which @kevinmking.bsky.social and I propose an update to the affect regulation theory of alcohol use: "A mathematical and computational formalization of the affect regulation hypothesis of alcohol use" doi.org/10.31234/osf...

65325

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I’d agree, but who knows what the kids are up to these days

110

Reposted by Jonas Dora

Kevin M. King @kevinmking.bsky.social · 02/09/2026

This a very exciting piece of work from @jonasdora.bsky.social . Jonas presents a mathematical and computational theory of affect regulation and alcohol use. This is a major advance in the field.

2112

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I'm SO curious what you think of this, Andrea!

110

Reposted by Jonas Dora

PsyArXivBot @psyarxivbot.bsky.social · 02/09/2026

A mathematical and computational formalization of the affect regulation hypothesis of alcohol use: osf.io/znsvf

021

Jonas Dora @jonasdora.bsky.social · 02/09/2026

Multiple of our data collections over the past years were run to inform/prepare this paper, e.g. doi.org/10.1111/add.... (lab) and doi.org/10.31234/osf... (daily life). The code for our model is here: osf.io/8hkvr/. I have been working on this for a long time, so it's so cool to finally share it :)

050

Jonas Dora @jonasdora.bsky.social · 02/09/2026

Huge thanks to @kevinmking.bsky.social; this (to me at least) is the most important step in a line of research we've been working on and towards since I started my postdoc with Kevin all the way back in 2020.

150

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I've left out a LOT here. The paper builds the model piece by piece, with every equation, simulation, and prediction laid out in full, plus much I couldn't fit into this thread. We'd be very happy if you read it and told us what you think. All feedback is very welcome :)

150

Jonas Dora @jonasdora.bsky.social · 02/09/2026

To be clear, should the model be thoroughly refuted within the next five or ten years, we would be thrilled, and would count it a resounding success; for that is what distinguishes a theory that participates in science from one that merely persists.

180

Jonas Dora @jonasdora.bsky.social · 02/09/2026

For the core predictions we didn't just state them. We simulated the proposed study AND a rival account beside it, so both the expected effect and the pattern that would count as support are derived directly from the model.

160

Jonas Dora @jonasdora.bsky.social · 02/09/2026

It also makes novel predictions, and for each we lay out the study design that would test or refute it. We tried to say, in advance and in detail, what would prove our model wrong.

160

Jonas Dora @jonasdora.bsky.social · 02/09/2026

This lets us show that the laboratory–EMA paradox, the pattern verbal theory can't explain, can be an expected consequence of our model rather than an anomaly. And it doesn't only reproduce this paradox; the same architecture recreates several other established lab and EMA findings at once.

160

Jonas Dora @jonasdora.bsky.social · 02/09/2026

So what does the model actually claim? Most important are 3 commitments: affect reaches behavior through the urge to drink; stress biases choice by raising the value of alcohol; and context gates the path from urge to drink.

182

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I cannot overstate how useful that back-and-forth between words, math, and code was for my understanding. And it doesn't stop now: the moment we observe a finding the model can't accommodate, we can point to the exact equation where it goes wrong, and we cannot avoid revising the model.

180

Jonas Dora @jonasdora.bsky.social · 02/09/2026

I cycled between verbal idea, math, and simulation many, many times over the past 5 years, and learned a ton doing it. A claim that sounded sensible in words couldn't be written as an equation; a simulation would refuse to align with the literature it was meant to reproduce, etc.

140

Jonas Dora @jonasdora.bsky.social · 02/09/2026

And words, we think, are simply the wrong language for that. So instead, we specified the affect regulation hypothesis as a system of equations, and implemented it as a computational model we can simulate from.

150

Jonas Dora @jonasdora.bsky.social · 02/09/2026

We argue this is not an issue of data, but of theory. The verbal affect regulation accounts are simply too vague; they can accommodate almost any result, so nothing ever contradicts them. What we need is theory specific enough that it CAN be wrong.

171

Jonas Dora @jonasdora.bsky.social · 02/09/2026

The dominant affect regulation accounts cannot explain this, but also seem oddly untroubled by it; the field's collective belief in affect regulation appears pretty much unchanged since these daily life findings have been established.

151

Jonas Dora @jonasdora.bsky.social · 02/09/2026

While experimentally inducing stress in the lab reliably raises both the urge to drink AND drinking, in daily life negative affect predicts the urge but NOT drinking, and positive affect predicts drinking instead.

160

Jonas Dora @jonasdora.bsky.social · 02/09/2026

Affect regulation, the idea that people drink to relieve stress and negative affect, is one of the central organizing frameworks in addiction science. Yet after decades of research, the field still can't say when, for whom, or why it happens. And it faces a robust empirical paradox.

150

Jonas Dora @jonasdora.bsky.social · 02/09/2026

65325

Jonas Dora @jonasdora.bsky.social · 07/08/2026

As an international postdoc on a temporary visa, this was the only NIH grant I was able to apply for at the time (and I was lucky enough to get it on my second attempt). I'd like to think this relatively small investment (relative to their budget) in me had a pretty decent ROI for NIH so far :)

051

Jonas Dora @jonasdora.bsky.social · 04/08/2026

The short answer: it depends a lot on the research design and methods used. We can talk about this and EMA in more detail in your lab meeting next week, where I believe I'm scheduled to make an appearance ;) (this is a great study btw, I need to read it in more detail soon)

110

Jonas Dora @jonasdora.bsky.social · 23/07/2026

Yep, definitely that’s a big factor in the choice, how much would the 50k impact your life. For somebody for whom it doesn’t change much it’s much easier to choose the option with the higher expected value

150

Jonas Dora @jonasdora.bsky.social · 04/07/2026

It's insane to me how many papers I review in which a slope is plotted with no SE or CI around it. Uncertainty quantification is what statistics is all about!

1110

Reposted by Jonas Dora

Justin Taraska @taraska.bsky.social · 23/06/2026

Looking for your next step after a BS/BA or MS in biology, cell biology, biochemistry, or a related field? My NIH lab in Bethesda, MD is looking for a Lab Manager/Technical Specialist. Interested in basic research? Please email me directly or learn about our lab here: irp.nih.gov/pi/justin-ta...

irp.nih.gov

Justin Taraska, Ph.D. | Principal Investigators | NIH Intramural Research Program

12324

Reposted by Jonas Dora

Aleksa Kaurin @aleksakaurin.bsky.social · 21/06/2026

Thrilled (really) to share recent evidence from our READY study led by @paulaphilippi.bsky.social w/ @aidangcw.bsky.social: "Minutes that matter: A microburst approach to adolescents' resolution of affect in response to daily stressors" 🧵1/9

24918