felipefv.bsky.social
PhD researcher @UGent | https://felipelfv.github.io | I like statistics, psychometrics, and metascience sometimes; pasta and ice cream, always | Donate to lavaan: https://lavaan.ugent.be/
130 posts
1,083 followers
784 following
Regular Contributor
Active Commenter
comment in response to
post
I used to do that during my bachelors and (less but still) masters. That habit got me in touch with my previous internship supervisor. It also allowed me to exchange a few emails with cool people, like Yalom, and it is also how I got in touch with my phd supervisor
comment in response to
post
Nothing. It does precisely the job it is designed to do, but users don't typically realise that sampling from a multivariate normal distribution involves matrix decomposition that is inexact and not invariant across machines, so results are irreproducible even when you use set.seed()
comment in response to
post
Very beautiful moment tbh
comment in response to
post
Not-so-modest shoutout to our own paper on interactions in which we discuss both issues — scale dependence and confounding 😋 journals.sagepub.com/doi/10.1177/...
comment in response to
post
Omg, I recently had a chat with LLM about this and out of nowhere it finished with a quote (and more) from Alfred Korzybski: "The map is not the territory — the scale you choose shapes what you see"
comment in response to
post
Also note that "removable" here is a mathematical term: an interaction is removable if we can nullify it with a monotonic transformation. But just because an interaction is "removable" doesn't mean it isn't substantial!
comment in response to
post
Oh, completely missed that reply. Thanks!
comment in response to
post
Do you know of any such service that is currently available?
comment in response to
post
"It’s an odd mix of weirdly capable, but also sort of dumb and forgetful."
LLMs: they're just like me.
comment in response to
post
Honest question: what do you mean when you say it doesn’t “use” eigenvalues?
comment in response to
post
I can see the problem was solved, but I just wanted to let you know that lavaan.mi is now a separate package in R. If you install "lavaan.mi", and just load semTools, that code should work (without even needing to load lavaan.mi).
comment in response to
post
Not textbooks but this may be interesting (?): arxiv.org/abs/2203.06469 and www.tandfonline.com/doi/full/10....
comment in response to
post
Chapter 2 here might be nice for students to get a summary of the mess: www.repository.cam.ac.uk/items/895d37...
comment in response to
post
Welcome :)
comment in response to
post
This is a quite straightforward and simple document for that: ucsb-meds.github.io/creating-qua...
comment in response to
post
medoutcon is a cool R package that allows you to integrate machine learning
comment in response to
post
📌
comment in response to
post
📌
comment in response to
post
This is exactly it, and the fun thing about complexity is often those inner stat uncertainties wash out quite well when you model something at scale. Even there you can make a lot of progress:
www.pnas.org/doi/10.1073/...
comment in response to
post
This reminds me of a good friend I have. Whenever we had a quiz evening with other friends, he would know the answer to Latin related questions (as an example), but then he wouldn’t know Billie Eilish
comment in response to
post
Fear of climate change is a fear*
comment in response to
post
Independently of my judgement and the entire post, the argument seems to be that climate change is a fear that is generated from real/specific dangers. So, a rational fear; instead of a general fear (“dooming”) about any other disasters (i.e., nuclear war, zombies bs).
comment in response to
post
Oh oh, you are now a communist for many people in Brazil