iiinova — where your digital self keeps evolving

Reposted by Tim Hayes

Andreas Brandmaier @brandmaier.bsky.social · 01/06/2026

Cohen’s d is one of the most widely used ways to communicate mean differences across studies. But what, exactly, is it telling us? A perspective on detectability of effects -- just out in PARE: openpublishing.library.umass.edu/pare/article... #openaccess

1115

Reposted by Tim Hayes

Amanda Kay Montoya @akmontoya.bsky.social · 28/05/2026

If you’re at #aps26bcn join me at 2pm for a workshop on causal mediation analysis with the #PROCESSmacro! Learn all about potential outcomes, sensitivity analysis, and more!

082

Tim Hayes @letsnotbehayest.bsky.social · 03/04/2026

Varimax, Promax, Equimax, Parsimax … why do orthogonal factor rotation methods all sound like cheap rental car places?

010

Tim Hayes @letsnotbehayest.bsky.social · 27/03/2026

I pass this out as supplemental content every year in my own regression class. Happy to download the most recent version!

110

Tim Hayes @letsnotbehayest.bsky.social · 13/03/2026

At home with a 101 degree fever (thanks, my daughter’s daycare) prepping for a lecture on exploratory factor analysis. Normally I do all of my own work as a nonnegotiable point of pride, but I’m so tired and out of it, I’ll admit it: I decided to lean on ChatGPT for this one …

020

Tim Hayes @letsnotbehayest.bsky.social · 14/02/2026

If you cluster mean center your predictors, they’re really not that bad … =P

010

Tim Hayes @letsnotbehayest.bsky.social · 24/01/2026

Very interesting stuff! Reposting so students (eg, in my SEM class this semester) and colleagues can learn more.

010

Tim Hayes @letsnotbehayest.bsky.social · 31/12/2025

“Human psychology for people who can’t stand people”. To be fair, if someone wrote a book with that name, I would buy it and read it. 😂😂😂

130

Tim Hayes @letsnotbehayest.bsky.social · 19/12/2025

Wes is the bes! Catch him and Sonja (also the best) on the Wes coast!

020

Tim Hayes @letsnotbehayest.bsky.social · 22/11/2025

Fun fact: it was actually on the basis of responses to this exact behavior that he originally came up with the idea for the “strongly disagree” response option.

010

Tim Hayes @letsnotbehayest.bsky.social · 22/11/2025

Yeah but after you pronounce it correctly you have to lick your lips and make hard, aggressive eye contact for an uncomfortably long time (as Likert himself did, I can only assume).

110

Tim Hayes @letsnotbehayest.bsky.social · 10/10/2025

Happy Friday! I am clearly in a mood.

020

Tim Hayes @letsnotbehayest.bsky.social · 02/10/2025

Extremely proud of this paper, which finally has an issue number! psycnet.apa.org/record/2024-...

030

Tim Hayes @letsnotbehayest.bsky.social · 17/09/2025

“Give him .. help. Be a beacon in his sad and lonely life …”

010

Tim Hayes @letsnotbehayest.bsky.social · 17/09/2025

No more secrets, Marty … er, Andrea.

110

Tim Hayes @letsnotbehayest.bsky.social · 17/09/2025

I knew that you were a quantitative methods expert, but I didn’t know that you were also versed in SETEC ASTRONOMY. Respect. :)

120

Tim Hayes @letsnotbehayest.bsky.social · 28/07/2025

Who needs ordinal regression. I mean, it’s not like nearly *all* of our variables are ordinal or anything … 🤣😂😢

030

Tim Hayes @letsnotbehayest.bsky.social · 24/07/2025

University requirements vary, but I’m a fan of chapters that are written like journal papers with focused, driving arguments rather than long, meandering lit reviews just for the sake of “comprehensiveness.” As a frequent reader of these documents, focused, clear, logical writing is so appreciated.

030

Tim Hayes @letsnotbehayest.bsky.social · 25/06/2025

FIU is disabling email forwarding as of tomorrow, which is understandable for cybersecurity but means I’ll have to send all emails using their garbage Microsoft Outlook interface. On the bright side, I guess this’ll help my compulsive email checking, since I refuse to get outlook on my phone.

040

Tim Hayes @letsnotbehayest.bsky.social · 21/06/2025

Great thread in which @kevinmking.bsky.social says everything I think and feel about latent classes. From now on, any time people ask me about these analyses, I’m going to just send them a link to this thread.

041

Tim Hayes @letsnotbehayest.bsky.social · 08/06/2025

“Most youth own smartphones.”

130

Tim Hayes @letsnotbehayest.bsky.social · 23/05/2025

Had a blast teaching SEM at #APS2025! Great group of workshop attendees asking thoughtful, engaging questions. Hopefully folks learned something and had fun in the process!

030

Reposted by Tim Hayes

Amanda Kay Montoya @akmontoya.bsky.social · 18/05/2025

Last chance to sign up for Intro (June 9-13) and Advanced (June 16-20) Mediation & Moderation at GSERM either in-person in beautiful St. Gallen, Switzerland or online from the comfort of your own home/office. gserm.org/programme/co...

gserm.org

Course table - University of St.Gallen – GSERM

The highlight of GSERM summer school regularly takes place in June: more than 30 courses at the campus of the University of St.Gallen.

032

Tim Hayes @letsnotbehayest.bsky.social · 14/05/2025

8 mo home sick from daycare. No big deal but marathon kinda day. Circa hour 11 of the waking day we’re watching Ms Rachel-throwing that at the wall in case it sticks-and all of a sudden, I swear, she goes, “Can you say … help? … HELP … heeeeeelp …” And my brain starts to break, just a little. 😅😂🤣

000

Tim Hayes @letsnotbehayest.bsky.social · 14/05/2025

Definitely take the free SEM workshop with @centerstat.org! And while we’re also shamelessly promoting our own things, I know of an SEM workshop at APS next week that I can recommend as well … 😉 www.psychologicalscience.org/conventions/...

Structural Equation Modeling in the Psychological Sciences workshop information (see hyperlink in post for more)

020

Reposted by Tim Hayes

Wes Bonifay @wesbonifay.bsky.social · 14/05/2025

If you are planning to attend #APS2025 next week, please come by our (@winterstat.bsky.social & @ashleylwatts.bsky.social) invited symposium on the myriad problems of widespread statistical model evaluation practices (e.g., goodness-of-fit testing), along with some insightful solutions

“It is impossible to escape the impression that people commonly use false standards of measurement — that they seek power, success and wealth for themselves and admire them in others, and that they underestimate what is of true value in life.”
― Sigmund Freud, Civilization and Its Discontents

Model Evaluation and Its Discontents
Invited Symposium
Saturday May 24, 2025 10:30 AM - 11:50 AM ET

Subject: Methodology

Wes Bonifay (Chair)
Talks:
How to Lie With Fit Statistics
Wes Bonifay (Presenting Author)
The Shape of Good Model Fit
Sonja Winter (Presenting Author)
Consideration of Indicator Interchangeability Can Strengthen Construct Validation Efforts
Ashley Watts (Presenting Author)
Statistical models enable scientists to describe data, explain phenomena, form testable predictions, and (ideally) understand the world. To achieve such aims, a model and its components must undergo thorough evaluation. In this symposium, we confront the widely accepted methods of model evaluation, expose their limitations, and explore more insightful alternatives.

083

Tim Hayes @letsnotbehayest.bsky.social · 29/04/2025

I have neither beneficence nor respect for the persons who created the extremely long CITI Training Good Clinical Practice modules and mandatory 100% quiz question pass rate requirement. 😒

000

Tim Hayes @letsnotbehayest.bsky.social · 12/04/2025

Matrix overview slides for my regression class are coming along.

010

Tim Hayes @letsnotbehayest.bsky.social · 09/04/2025

I’m amazed that I accomplished it at all. At the end of a long day, I feel like I’ve had a LOBOTOMY.

010

Tim Hayes @letsnotbehayest.bsky.social · 09/04/2025

Hey @wesbonifay.bsky.social, I finally got a NYT Letterboxed in 2 words, so now I’m preparing to ascend to heaven as a being made of pure light. 🙌🏻🌅 Just figured you’d want to know.

110

Tim Hayes @letsnotbehayest.bsky.social · 08/04/2025

The negative binomial is just like the regular binomial, except everything is terrible and nothing matters and what’s even the point?

150

Tim Hayes @letsnotbehayest.bsky.social · 04/04/2025

It is only our hubris that ever led us to name it “the negative binomial distribution” instead of “the Poisson distribution with wiggly gamma means.”

020

Tim Hayes @letsnotbehayest.bsky.social · 02/04/2025

For Q+A (at real talk): treat every question as if it is from a friend in lab meeting who is genuinely interested in understanding + improving the study (even when it is obvious the questioner’s intentions aren’t benign IRL). Foster an interested, curious dialogue + don’t be afraid to ask Qs back.

010

Reposted by Tim Hayes

Dan McNeish @dmcneish.bsky.social · 02/04/2025

New paper on dealing with MNAR intensive longitudinal data. Ran into this problem in an empirical study and didn't find too much in the methods literature on MNAR ILD, so this was the best I could come up with. Lots of opportunity to improve methods in this area! psycnet.apa.org/record/2025-...

2205

Tim Hayes @letsnotbehayest.bsky.social · 01/04/2025

I know scientific funding is precarious right now, but maybe you could volunteer your time and expertise to offer training to some aspiring shitpost-docs… 😂😂😂

130

Tim Hayes @letsnotbehayest.bsky.social · 15/03/2025

The mitml package in R is also excellent, as is mdmb. Many, many phenomenal alternatives to mice out there that leave it in the dust.

100

Tim Hayes @letsnotbehayest.bsky.social · 15/03/2025

If I was going to impute using R, I’d either learn the forthcoming rblimp package or use jomo by Matteo Quartagno, which is also excellent and has features and defaults lightyears better than mice.

000

Tim Hayes @letsnotbehayest.bsky.social · 15/03/2025

Meanwhile, mice defaults to PMM w 5 iterations when most models require thousands to converge (Because the algorithm supposedly possesses “ a touch of magic” :: cough:: bullshit) and makes diagnosing convergence all but inaccessible, except the experts like Kevin, who know the workarounds.

020

Tim Hayes @letsnotbehayest.bsky.social · 15/03/2025

Blimp also has FCS available but I find the defaults much better and it also has more easily accessible convergence diagnostics. Plus state of the art categorical handling, SMC/Fully Bayesian imputation for interactions/random slopes/etc and a variety of other bells and whistles (eg MNAR models).

110

Tim Hayes @letsnotbehayest.bsky.social · 15/03/2025

I didn’t have the chance to comment on the original thread but Blimp is the best! It’s the only program I ever use for imputation these days. Glad it worked for you!!

140