easystats

Official channel of {easystats}, a collection of #rstats 📦s with a unifying and consistent framework for statistical modeling, visualization, and reporting

[easystats @easystats4u]

If you're on bluesky - we're too!
https://t.co/uB8Qw8LNS9

[easystats @easystats4u]

We're on bluesky now! Follow us under @easystats.bsky.social #easystats #rstats

[Dominique Makowski 🧙 @Dom_Makowski]

🤯📈 TAKE YOUR STATS SKILLZ TO THE NEXT LEVEL WITH THIS #R PACKAGE

🔥The @easystats4u {modelbased} package (the successor of #ggeffects) is now published in @JOSS_TheOJ

https://t.co/A897NTJMzZ

Check it out for a demystification of marginal means, contrasts and effects #rlang

[easystats @easystats4u]

@viv_analytics @SophieStats10 Yes, you may look at the {parameters} pkg: https://t.co/xdodfdHKZb
Tidies output, if desired, but also prints nicely and offers a lot more features (standardizing, robust standard errors, effect sizes, ...)

[easystats @easystats4u]

We're working on revisiting and homogenizing outputs from our #easystats packages. This includes consistent coloring of information/warnings/messages, but also: which information is useful in the output, which information should just go into the docs? WDYT?

[Daniel 🕹️ @strengejacke]

Which of the following information below model output (last paragraph, not that one about uncertainty intervals) do you find useful/helpful and think it's worth printing? It's printed once per session. Should some/all information moved into the docs, or kept in output? #easystats

[easystats @easystats4u]

{bayestestR} makes it now much easier to process inputs from packages {marginaleffects}, {emmeans}, or random variable types from posterior draws!
#easystats #rstats

[Mattan S. Ben-Shachar @mattansb]

New update to {bayestestR} expands support for a tidy workflow -
working better with tidy inputs, `rvar`s, and post-modeling estimates, and generating tidy outputs!

@easystats4u #rstats

🧵

https://t.co/i95bgVxfKI

[easystats @easystats4u]

We're on bluesky now! Follow us under @easystats.bsky.social #easystats #rstats

[easystats @easystats4u]

@schneidysenses @dp_moriarity When you install the easystats-package from CRAN, and then run `easystats::install_latest()`, does that resolve the issue for you?

[Mattan S. Ben-Shachar @mattansb]

NEW to {bayestestR} dev version - support for #marginaleffects!

Take it for a test drive: remotes::install_github("easystats/bayestestR")

@easystats4u #rstats

[easystats @easystats4u]

A short update on this feature:
- Improved documentation (https://t.co/Oo9gRxj9KK)
- Streamlined text output
- Improved plots
#rstats #easystats #DAG

[easystats @easystats4u]

A new feature that *might* be added to our #rstats #easystats packages soon: checking models for correct adjustment by using DAGs! `check_dag()` (working title) makes it so easy to check the causal paths of your model and tells you how to address misspecifications!

[easystats @easystats4u]

@mzloteanu But in terms of DAGs, the goals are clearer, or not?

[easystats @easystats4u]

A new feature that *might* be added to our #rstats #easystats packages soon: checking models for correct adjustment by using DAGs! `check_dag()` (working title) makes it so easy to check the causal paths of your model and tells you how to address misspecifications!

[easystats @easystats4u]

@jpmarocox Yes, random slopes or more than one higher level should work.

[easystats @easystats4u]

Are you working with mixed (multilevel) models in #rstats and wondering how to calculate R2? Grab the latest updates of our #easystats {performance} and {insight} packages from CRAN and try out "r2_nakagawa()" (or simply "r2()" for mixed model):
https://t.co/WG374eDB43 /1

[easystats @easystats4u]

We improved the accuracy for many new model families and validated the results against examples from the paper that has proposed this method: https://t.co/b0S99cIWEp
"r2_nakagawa()" is probably one of the most accurate functions to return R2 for mixed models in #rstats.

[Mattan S. Ben-Shachar @mattansb]

Today Tom Geva (from @bengurionu's Statisticas and Data Analysis program) presented the work he's done over the last few months for @easystats4u's correlation package 👏

[ᴅʀ ᴍɪʀᴄᴇᴀ ᴢʟᴏᴛᴇᴀɴᴜ 🌺🌞🍃 @mzloteanu]

#statstab #129 Structural Models (EFA, CFA, SEM, ...) w/ {parameters}

Thoughts: Lots of debate about #EFA vs #CFA; very confusing. Once I figure out what to use, this #R package seems to have lots of functionality.

#rstats #factoranalysis #r #stats

https://t.co/gDY41o2g6h

[easystats @easystats4u]

And another "small but great" piece in our #easystats collection...

[Mattan S. Ben-Shachar @mattansb]

{effectsize} has been updated!

This update includes the new repeated_measures_d() function that can compute not 1, not 2, but 6 (!) types of standardized mean differences for repeated measures.

Try it out:
https://t.co/UuD8QvqPQS #RStats

[easystats @easystats4u]

You may think it's just s small addition to one of our packages, but overall, the many small pieces form a marvelous "R Framework for Easy Statistical Modeling, Visualization, and Reporting" #easystats

[Dominique Makowski 🧙 @Dom_Makowski]

A long overdue little feature just dropped: report() now works with #BayesFactor objects.
Use it to make your stats reports more reproducible and easy to read.
#Rstats @easystats4u

[easystats @easystats4u]

🚨New publication in Behavior Research Methods!🚨

We review outlier detection methods and how to achieve them in R using #easystats' {performance} 📦

📄paper: https://t.co/4RIbfKfCEd
📃preprint: https://t.co/UMD3CEYgAG
💻performance: https://t.co/k8DkbIyG9W

#rstats

[easystats @easystats4u]

read the case study how to arrive at the best model fit (https://t.co/rUdEzG1DJv), using `check_model()` and other tools from the {performance} package. Want to use the latest features? Install the {easystats} package from CRAN, run `easystats::install_latest()` and have fun! /3

[easystats @easystats4u]

Checking model assumptions is important, and doing so requires accurate tests and visuals appropriate for the given model. Hence, we revised our #easystats #rstats packages {see} and {performance}, which now provide methods to simulate residuals for complex models, to perform /1

[easystats @easystats4u]

accurate tests using simulated residuals, and improved diagnostics plots (in particular, Q-Q plots). Check out the two vignettes how to check your model based on simulated residuals (https://t.co/zHsoLiFjCR) and /2