mzloteanu

@mzloteanu@mastodon.social

SL in Investigative Psych @KingstonUni | Deception Detection; Emotions; Decision-Making | Open Science; R; Bayes | @ukrepro ReproTea & StatsTea | 🇷🇴 🇬🇧🌍 https://www.buymeacoffee.com/mz555

This profile is from a federated server and may be incomplete. Browse more on the original instance.

mzloteanu, to random

Using tidybayes with the posterior package by @mjskay

Thoughts: Going means a lot of wrangling the posteriors to get the info you want. A crucial skills I'm still learning, but invaluable.

https://mjskay.github.io/tidybayes/articles/tidy-posterior.html

mzloteanu, to random

#8 Step away from stepwise

Thoughts: Pretty straightforward, don't use stepwise regression. It is atheoretical and can produce causal inference mistakes; it is also a misuse of p-values.

https://journalofbigdata.springeropen.com/articles/10.1186/s40537-018-0143-6

mzloteanu, to github

I'm updating my workflow for 2024, thinking of going full #R + #quarto + #github, but i need something for easy collab with non-R people (revisions, comments, etc.). Suggestions? 🤔
#advice #research #openscience

lakens, to random Dutch
@lakens@mastodon.social avatar

As a scientist, every decision you make should follow from your philosophy of science. If and why you would collect data. How you analyze it. How you write a paper. If you can not justify your actions, you are just imitating others, which might be fine, but it might not be fine.

mzloteanu,

@lakens when using robust statistics, what is the hypothesis that a researcher is actually investigating? what is the phil. of science in those cases?

mzloteanu,

@lakens that is what I mean. Typically, we speak of, say, an average population effect (group 1 =! group 2), which can be seen as suggesting the two populations are different. But what does the robust hypothesis say. "The trimmed population effects differ"? Does it make "epistemological" sense to speak this way, or simply "statistical" sense?

lakens, to random
@lakens@mastodon.social avatar

People are often surprised to learn they can design a study where you look at the data twice (a sequential design) at no cost (because the alpha correction is so small, there is no increase in the sample size). To learn how, see https://lakens.github.io/statistical_inferences/10-sequential.html#sample-size-for-sequential-designs

mzloteanu,

@lakens wait until they find out they can look at the data how many times they want if they go 😉

MarkRubin, to statistics

In this new preprint, I argue that “Type I error rates are not usually inflated.”

https://doi.org/10.31222/osf.io/3kv2b

mzloteanu,

@MarkRubin this is in the category of "technically yes, but realistically no".

mzloteanu, to statistics

You have an ordinal DV, a 4-level RM factor and 2 continuous covariates. Which statistical software has an analysis option for this design (w pairwise followups + plots)? 🤔

*you cannot say R or "just code it"

mzloteanu,

@DataAngler SAS is also basically coding, not point-and-click, right? I'm curious if there is software that would just do the analysis directly.

(for context, my curiosity is that if i didn't know R, I don't know how I would have handled this study, and i wonder what others would do. Since ANCOVA gives completely wrong answers)

andrew, to random

Q about stuff: I'm finding the probability of direction (proportion of posterior that's >0 or <0), but I never know how to report these, since sometimes they're positive and sometimes they're negative. My current solution is to use a column for each direction—is there a better way tho?

mzloteanu,

@andrew @jdonland just to be pedantic, if you use PD as in that ref it will be different than p(x)>0, as PD is 50%-100%, while the latter is 0-100. (my view is ROPE, if you can define it, is best)

Private
mzloteanu,

@scholzmx @rstats sounds great!

  • knows none of those distributions *
MarkRubin, to random

Scrolling, scrolling scrolling….

Our new research finds passive social media use is associated with poorer mental health, especially among younger people.

Open access: https://doi.org/10.3389/fpsyg.2023.1181233

mzloteanu,

@MarkRubin * scrolls, stops to read this post, feels worse, keep scrolling *

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • InstantRegret
  • mdbf
  • ethstaker
  • magazineikmin
  • cisconetworking
  • rosin
  • thenastyranch
  • Youngstown
  • GTA5RPClips
  • slotface
  • khanakhh
  • kavyap
  • DreamBathrooms
  • megavids
  • Durango
  • everett
  • tacticalgear
  • tester
  • osvaldo12
  • modclub
  • normalnudes
  • ngwrru68w68
  • cubers
  • Leos
  • provamag3
  • anitta
  • lostlight
  • All magazines
    •