iiinova — where your digital self keeps evolving

Scott James Perry @sjperry.bsky.social · 09/10/2026

Me: Okay, so a model like this would assume that first language speakers and second language speakers are identically variable, although one group may be faster of slower: Student: But I don't want to assume that! Me (imitating Palpatine): Good, goooood!

110

Scott James Perry @sjperry.bsky.social · 09/10/2026

I taught an Intro to Bayes course at a summer school a few years back and there was a first-year PhD student who was absolutely scandalized when he finally understood that 99% of the mixed models used in papers he had read assumed constant variance for EVERYTHING.

110

Scott James Perry @sjperry.bsky.social · 08/10/2026

Looks to me like the ones who think they are experts just have less variability in the perception score, so the top of the density is higher.

000

Scott James Perry @sjperry.bsky.social · 02/10/2026

I have access through my institution, I've requested it and I'll forward it along when I have it!

110

Scott James Perry @sjperry.bsky.social · 18/08/2026

It's on Fridays, so it's not so bad. Four years of being a dad to a terrible sleeper has me conditioned to handle it. And we use Fantasy Grounds to play DnD. I agree that in person is better but having a game every week is awesome.

022

Scott James Perry @sjperry.bsky.social · 18/08/2026

I've got a great online group that's been going for years. We miss a week here or there, but generally everyone makes it every week (and also everyone pays attention, knows the rules, and is engaged). The only catch is I have to wake up at 3am to play...

110

Scott James Perry @sjperry.bsky.social · 24/06/2026

Idk man, but here in Catalonia my mother in law brought a sweater over to our house to wear while visiting because we have our AC set to 26 degrees. Last month when it was actually 26 degrees outside she was not wearing sweaters.

040

Scott James Perry @sjperry.bsky.social · 28/05/2026

New paper! At ICPhS in 2023 Gemma invited me onto a project looking at duration as an acoustic correlate of the tap-trill contrast in Spanish heritage speakers from California. We look at children and adults, comparing them to age-matched monolinguals. Some fun Bayesian modelling as a bonus!

cambridge.org

Duration as an acoustic cue of the Spanish tap-trill contrast according to age and language background | Journal of the International Phonetic Association | Cambridge Core

Duration as an acoustic cue of the Spanish tap-trill contrast according to age and language background

040

Scott James Perry @sjperry.bsky.social · 24/05/2026

Wow, it sounded really gross to me until this, now I soft of want some!

030

Scott James Perry @sjperry.bsky.social · 22/05/2026

When we Canadians talk about "brown sugar" we confusingly don't mean "sugar that is brown". We take regular old white (granulated) sugar and then add molasses to it. Depending on how much molasses you add you either get "light brown" or "dark brown" sugar. I live in Spain and have to buy it online.

010

Scott James Perry @sjperry.bsky.social · 06/05/2026

It is a truth universally acknowledged that Eduroam always sucks (posted using mobile hotspot from my university campus)

020

Scott James Perry @sjperry.bsky.social · 04/05/2026

For one of them, yeah. The other was actually 5 days, so a big improvement there!

010

Scott James Perry @sjperry.bsky.social · 01/05/2026

What's that? After waiting for a good bit on proofs for two articles they both show up on the same day right before a holiday weekend?

100

Scott James Perry @sjperry.bsky.social · 13/04/2026

Lots of people (mostly older professors) laughed, and no comments were raised about this dubious fishing expedition. But I was super happy to see several MA students from my quantitative methods class look across the room at me with horror in their eyes. I must have done something right.

010

Scott James Perry @sjperry.bsky.social · 13/04/2026

I was at a conference talk and the professor told us they instructed their PhD student to "check all the ways this property has been measured in the literature and run all of them to see if any turn out significant".

110

Reposted by Scott James Perry

Brian Wong @brianwwl.bsky.social · 20/03/2026

📢 New paper in JEP:HPP Speech perception gradiency (sensitivity to fine-grained phonetic detail) has been shown to enhance speech perception flexibility (recovery from misleading information) in L1 English. But does it play the same role in L2? doi.org/10.1037/xhp0... @bcbl.bsky.social

doi.org

APA PsycNet

1125

Scott James Perry @sjperry.bsky.social · 19/02/2026

Not fully randomized, but sort of quasi-experimental. Looking at verb choice during an elicited narration. There were three different possible verbs for the grammatical construct under study. If this is close enough I can probably find the data.

000

Scott James Perry @sjperry.bsky.social · 18/01/2026

Invited talk to a doctoral seminar tomorrow, and we're taking time to talk about the estimand. I feel like senior researchers generally have an implicit understanding of this concept, but I've found it helps people a lot to talk about it explicitly at the beginning of statistical training.

A picture of a slide titled "The estimand" with the subtitle (The most important word you'll never hear).

010

Scott James Perry @sjperry.bsky.social · 13/11/2025

In data cleaning this week, I learned that ePrime often defaults to entering a value of '0' when there is no response. This is pretty bonkers, and it's crazy that a dedicated experimental software that you pay a lot of money for does this. Needless to say I have to re-write some scripts...

000

Scott James Perry @sjperry.bsky.social · 23/09/2025

I was hired to analyze some data a while back and the professors were surprised at the end when I sent a folder with all of the scripts + a README file. Apparently the consultant before me never shared scripts because he said "then the next time you'll just use the script and not have to hire me".

250

Scott James Perry @sjperry.bsky.social · 16/07/2025

The numbers on sharing these things later on "reasonable request" are pretty bleak, so if you don't share it puiblicly in a way that is linked with the paper somehow that information is/will be gone forever. If you can't share the evidence of a claim, the strength claims make by the paper is weaker.

000

Scott James Perry @sjperry.bsky.social · 16/07/2025

I always go through scripts shared during peer-review and make sure they are clear, run properly, and check it for common statistical errors. I have found everything from silly mistakes in coding to serious errors in data cleaning that invalidated the claims of the paper.

100

Scott James Perry @sjperry.bsky.social · 16/07/2025

b) Unless there is a valid reason for the data to NOT be shared (and there are several), it should be the default because it's hard to verify what was done with a script and no data. Or at least it's harder to tell if what was done was reasonable.

100

Scott James Perry @sjperry.bsky.social · 16/07/2025

IMO data cleaning scripts also need to be in the open materials, although I've had spirited discussions about this with co-authors before who think that only the bare minimum code to replicate the numbers in the paper needs to be shared.

200

Scott James Perry @sjperry.bsky.social · 16/07/2025

a) It stands as a record of what you did, including details that will never make it into a paper. As analyses get larger and more unwieldy, less details can be reported in papers that sometimes have strict word limits.

200

Scott James Perry @sjperry.bsky.social · 01/07/2025

An exception: you want cheddar in Spain and all the locally-made stuff is white and the KerryGold is yellow.

010

Scott James Perry @sjperry.bsky.social · 03/06/2025

I honestly think that I've had to cite this paper in more than half of the reviews I've ever done. At least in linguistics, it seems to be a super common misinterpretation.

030

Scott James Perry @sjperry.bsky.social · 25/05/2025

My son tried a bagel for first time recently. My wife taught him to correct my pronunciation. Now, when I say "do you want a [bægəl]?", his response is "No, [beɪgəl]!".

010

Scott James Perry @sjperry.bsky.social · 21/05/2025

Yeah, I taught a class in a new department during the pandemic when masks were mandatory. When the mandate lifted and they all stopped wearing them I realized that I couldn't recognize most students because I filled in the lower half of their faces and was not accurate not even once.

020

Scott James Perry @sjperry.bsky.social · 20/05/2025

To be fair I built and troubleshot the model/priors on simulated data a few months ago before we built the survey, but I always feel more comfortable when something goes wrong and I dig around to fix it. Now I feel like a cartoon character that's walked off a cliff but hasn't looked down yet.

000

Scott James Perry @sjperry.bsky.social · 20/05/2025

Me when my Bayesian SEM model samples properly and quickly the first time around:

120

Scott James Perry @sjperry.bsky.social · 20/05/2025

I'm still very novice in terms of Stan code (simple models I can code up - hierarchical models kick my butt), but if your target audience is brms people looking to learn more Stan then that's me and I'd be happy to work through drafts and give feedback.

050

Reposted by Scott James Perry

Iris van Rooij 💭 @irisvanrooij.bsky.social · 16/08/2024

🚨Our paper `Reclaiming AI as a theoretical tool for cognitive science' is now forthcoming in the journal Computational Brain & Behaviour. (Preprint: osf.io/preprints/ps...) Below a thread summary 🧵1/n #metatheory #AGI #AIhype #cogsci #theoreticalpsych #criticalAIliteracy

The idea that human cognition is, or can be understood as, a form of computation is a useful conceptual tool for cognitive science. It was a foundational assumption during the birth of cognitive science as a multidisciplinary field, with Artificial Intelligence (AI) as one of its contributing fields. One conception of Al in this context is as a provider of computational tools (frameworks, concepts, formalisms, models, proofs, simulations, etc.) that support theory building in cognitive science. The contemporary field of Al, however, has taken the theoretical possibility of explaining human cognition as a form of computation to imply the practical feasibility of realising human(-like or -level) cognition in factual computational systems; and, the field frames this realisation as a short-term inevitability. Yet, as we formally prove herein, creating systems with human(-like or -level) cognition is intrinsically computationally intractable.

22491167

Reposted by Scott James Perry

Emily M. Bender @emilymbender.bsky.social · 13/05/2025

Happy US book release day to me, my amazing co-author @alexhanna.bsky.social and everyone else who can now open up their own copy!! All the details on ordering, events, and news coverage at thecon.ai

thecon.ai

THE AI CON

How to Fight Big Tech's Hype and Create the Future We Want

1627762

Scott James Perry @sjperry.bsky.social · 06/05/2025

I've recently spent lots of time on the Discourse pages of different academic softwares, and it made me appreciate the Stan site so much. I found horrendous amounts of condescension and general unhelpfulness out there, which is not something I see much of over at discourse.mc-stan.org

discourse.mc-stan.org

The Stan Forums

A community to discuss Stan and Bayesian modeling.

020

Scott James Perry @sjperry.bsky.social · 04/05/2025

No, YOU just wasted an hour checking whether or not the list randomization in your experiment was working before realizing you'd copy and pasted the same file into all parts of the ifelse statement...

010

Scott James Perry @sjperry.bsky.social · 11/04/2025

Tagging @dingdingpeng.the100.ci and @vincentab.bsky.social because I told them I'd keep them in the loop. Also, I'll be writing up the code and examples used in this class into a blog post directed towards linguists that want to start to use the package marginaleffects!

240

Scott James Perry @sjperry.bsky.social · 11/04/2025

For the few it didn't work for, I'm trying to keep in mind that the previous version had been honed for 4 years, while this was my first time teaching this. I'm sure I have lots of room for improvement in walking others through this material. Hopefully this can mitigate the confusion in future years

110

Scott James Perry @sjperry.bsky.social · 11/04/2025

Overall, I think I will pivot to teaching marginaleffects for model interpretation. It seems like it was positive for many students, neutral for quite a few, and potentially discouraging for a small minority.

110

Scott James Perry @sjperry.bsky.social · 11/04/2025

A couple students said they understand LMMs and GLMs worse than before the course. This was hard to read, because I want to motivate students to learn stats, not discourage them. I'm hoping this is a "I'm confused because I'm paying attention" thing, but maybe this change didn't work for everyone.

110

Scott James Perry @sjperry.bsky.social · 11/04/2025

We also used non-aggregated predictions and comparisons to help us understand what is actually happening in a mixed-effects model (i.e., varying intercepts and slopes). Of those who already had experience with these, more than half said that doing this made them understand the models better.

110

Scott James Perry @sjperry.bsky.social · 11/04/2025

For predictions/comparisons, I asked how comfortable people felt from "1" (I don't feel comfortable using this at all) to "5" (I could use this in my own research). Most answered between 2-4. Those with previous stats knowledge answered higher. predictions() were rated higher than comparisons().

120

Scott James Perry @sjperry.bsky.social · 11/04/2025

25% of students took this course previously. Of that 25%, half thought that model interpretation was more straightforward with marginaleffects compared to summary tables. The other half said both were equally hard to understand. No one thought summary tables were easier, or that both were easy.

130

Scott James Perry @sjperry.bsky.social · 11/04/2025

Some context: 1/3 of my students were learning about regression for the first time. Another third took an MA course with me where we learned to fit models and interpret summary tables, along with visualizing model predictions. Another third were PhD students with previous knowledge about regression

120

Scott James Perry @sjperry.bsky.social · 11/04/2025

The past 4 weeks I've taught a regression crash course to graduate students in linguistics. This is my fifth year teaching the course, and I decided to switch to teaching model interpretation using the marginaleffects package. Here are my thoughts along with info from an anonymous student survey 🧵

1113

Scott James Perry @sjperry.bsky.social · 23/03/2025

I'm hoping it ends up being positive for everyone, and not just because overhauling a course instead of making minor tweaks means me investing a good amount more time. The silver lining either way is that I'm gaining a deeper understanding of all of this when I teach it!

130

Scott James Perry @sjperry.bsky.social · 23/03/2025

I'm one week into short stats course for linguistics grads students where half my students have no experience with stats or R. This year I'm teaching marginaleffects for interpretation, and will be asking for feedback from students after.

120

Scott James Perry @sjperry.bsky.social · 23/03/2025

The main reason is to avoid student confusion, as all of the different models, link functions, transformations, etc. definitely take a toll on my students. I'm hoping that teaching a unified approach to interpretation that works for all models will work better. Will report back in three weeks!

140

Scott James Perry @sjperry.bsky.social · 23/03/2025

I have taught a statistics and R crash course to graduate students in linguistics for four years. We cover an intro to R and then lm(er) and glm(er). This year, I'm overhauling the course and teaching everyone to use marginaleffects from @vincentab.bsky.social for model interpretation.

130