Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦 @mattansb Twitter profile

Last Seen Profiles

@SmithsonianDPO

@retamboria

@JimBrowning11

@StephaneWinkel

@BobbyBrouno

@sprooggy

@BinorRaja

@Tamia_Nico_

@clematis_taiga

@manunkama

@brookielove44

@NielsPeek

@BoWs1rJr

@Tranace

@Deox6_SSB

@stern

@BinorRaja

@s_otani_brass

@zoewhittall

@sebastiankehl

@La1_tve

@c_xair

@squonchypossums

@xcmag

@DZ4UOUTDOOR

@anarkopollas

@pengen_stw

@AUTFakultesi

@AwesomeNintend0

@Raincolt3

@f_sin_cassette

@CoullLochguy

@JuniorschoolCH

@Vargore

@mei_asmr_0120

@Dalzell60

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Did I just discover a huge bug in my EEG analysis code? Yes. Did I only check because the results were too good to be true? Also yes. What's the opposite of p-hacking?

50

18

528

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

Today I finish teaching SEM in #rstats in #lavaan . If anyone wants a peek at my materials, you can find them here:

GitHub - mattansb/Structural-Equation-Modeling-foR-Psychologists: Lesson files used in the Struct...

Lesson files used in the Structural Equation Modeling for Psychologists. - mattansb/Structural-Equation-Modeling-foR-Psychologists

github.com

9

138

486

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Did I waste 3 hours trying to figure out how to rotate a #ggplot but not the legend? Of course not! It wasn't a waste - it was an investment!

8

23

448

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 months

Reading a paper. Title: Using AI to predict {behavior} from {physiological measure} Methods: Using a linear regression. Results: Model validated on training set. What a time to be alive! We truly 𝘢𝘳𝘦 living in the future!

12

40

421

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

library(MASS, exclude = "select")

12

26

406

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

@christapeterso They're hilarious, but I hate seeing them in real papers, so I tend to discourage my students from using them (I think they're leftovers from essay writing in HS). I often find myself marking the 1st sentence of the 2nd paragraph with "you really could have started from here..."

4

5

377

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

{effectsize} 📦 version 0.8.0 is out on CRAN! We have new effect sizes, we have better vignettes, we even dabble with some Unicode outputs - a 🧵 1/ @easystats4u #rstats

4

53

352

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

{effectsize} 0.3.0 is now out! #rstats Among other changes - all effect sizes now support confidence intervals! Check it out! And the best documentation: @easystats4u @Dom_Makowski @strengejacke and thanks to @patilindrajeets !

2

51

254

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

(Finally) finished teaching my intro+ #Rstats course for psych MA! This year I've leaned much more on #tidyverse & @easystats4u , which I think has really helped the students grasp many data and statistical concepts. As usual, my martials are online:

GitHub - mattansb/Practical-Applications-in-R-for-Psychologists: Lesson files for Practical...

Lesson files for Practical Applications in R for Psychologists. - mattansb/Practical-Applications-in-R-for-Psychologists

github.com

6

47

241

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Collaborating with others be like #rstats

10

32

231

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

#statstwitter I've often seen the following argument: 1. When groups are of unequal size there is reduced power. 2. Better throw out observations from the larger group to make them equal! W... Why? Who is teaching this? Can they stop?

32

10

211

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 months

{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: #RStats

3

39

198

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

This week I finished teaching analysis of factorial designs (ANOVA / (G)LMM) in #rstats w/ ( @HenrikSingmann 's amazing) {afex} and {emmeans}. If anyone wants a peek at my materials, you can find them here:

GitHub - mattansb/Analysis-of-Factorial-Designs-foR-Psychologists: Lesson files used in the...

Lesson files used in the Analysis of Factorial Designs for Psychologists. - mattansb/Analysis-of-Factorial-Designs-foR-Psychologists

github.com

3

42

183

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

This week I finished teaching *Intro to #rstats * for psych-grad students (data wrangling + principals of statistical modeling)! The course was built with no-previous-coding in mind, and I think it worked quite well! Full materials can be found here:

4

41

184

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

It's official!! 🎉🥳🎓 Time to get out your physical address book, flip to the B tab, and update my entry to Mattan S. Ben-Shachar, 𝙋𝙝𝘿 #PHinisheD

25

1

169

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

#rstats

1

12

162

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

This week I finished teaching graduate level #rstats for psych! Course covered data wrangling and plotting and the basics of building and evaluating regression models. As always, if anyone wants a peek at my materials, you can find them here:

GitHub - mattansb/Practical-Applications-in-R-for-Psychologists: Lesson files for Practical...

Lesson files for Practical Applications in R for Psychologists. - mattansb/Practical-Applications-in-R-for-Psychologists

github.com

4

46

162

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Hey, you know how ANOVA is just a special case of regression? So no. This post is my attempt at giving the broadest overview of what ANOVAs are (and are not) and explain the differences between their various types (with #rstats )

4

32

158

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Don't update to #rstats 4.2 before the semester is over... Don't update to #rstats 4.2 before the semester is over... Don't update to #rstats 4.2 before the semester is over... Don't update to #rstats 4.2 before the semester is over... Don't update to #rstats 4.2 before the semes

11

7

154

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Glad to have contributed to the increase the #emmeans support in #brms ( @paulbuerkner )! You can now use #emmeans w/ your Bayesian multilevel, multivariate, smooth or monotonic, generalized models with modeled distributional parameters! #rstats

1

16

150

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Just finished my workshop on Evaluating Evidence & Making Decisions using Bayesian Statistics w/ #brms and #bayestestR ( @easystats4u ) - opening the online #ISCoP2021 ! Love talking about Bayes! Here's a link to my slidedeck (will post vid when it's up!):

6

34

148

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

🌀🚨New Paper!🚨🌀 Phi, Fei, Fo, Fum! We present a new (!) effect size פ (Fei) for χ² test of goodness-of-fit, and how it fits into the family of related effect sizes! Now out @MDPIOpenAccess ! w/ @patilindrajeets @RemPsyc @bmwiernik @strengejacke ( @easystats4u ) 1/

5

32

143

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

🚨🚨New Blog Post🚨🚨 "Order Constraints in Bayes Models (with #brms )" In which I introduce order constraints and show how they can be used / applied in #rstats #brms and @mcmc_stan (but also generally with any MCMC sampler). Dig in!

Stat’s What It’s All About - Order Constraints in Bayes Models (with brms)

blog.msbstats.info

5

43

139

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Inspired by @bmwiernik

2

17

141

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 months

PSA: if you have an interaction in your model Y~X+G+X×G Unless you did something special*, the coefficient of X is the simple slope of X when G=0, and not the average slope of X. *Centering G (if it is numeric) or using some form of effects coding (if categorical).

5

14

139

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

6 years

@FraudSyndrome @CoryjTaylor @FilmCritHULK Hagrid is Animal.

11

0

130

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 months

⚠️⚠️New Blog Post! ⚠️⚠️ Centering in Moderation Analysis: A Guide We talk about: how to center variables (including categorical variables!), when you need to use it (moderation analysis, ANOVA) and how to do it ( @easystats4u 's datawizard)! #rstats

Stat’s What It’s All About - Centering in Moderation Analysis: A Guide

blog.msbstats.info

2

34

132

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

I'm P-h-Done! 7 years in the making, 36,892 words over 109 pages describing 4 experiments (6 pilot studies) on a total of nearly 200 humans between the ages of 6 months and 35 years... #PHinisheD

16

1

126

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Does anyone know what's the source of the N > 30 "rule of thumb" for the CLT to kick in?

23

17

125

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Hello confidence intervals for random effects' variances! @easystats4u #RStats

Daniel 🕹️

@strengejacke

3 years

The {parameters} 📦 updated on CRAN. thanks to @bmwiernik we now have more clarity and consistency regarding different methods to compute confidence intervals and p-values for so many different models. Read more about "CI and approximation of df" () #rstats

3

12

67

3

20

123

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 years

I've often noticed that researchers / students who approach me for help with power calculation are unhappy with the N I provide ("no, that is way too many!"). For them, I have made the following, ultimate #rstats #shiny app:

7

25

117

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Another day, another paper reporting CIs and p-values after stepwise variable selection...

5

4

109

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

@jacasiegel "bUt woMAn twEEt = iNVitE to DM"

1

107

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Nobody: Economists using LPMs:

8

6

108

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

11 months

Any day in which I can convince someone *not* to transform their variables just because they aren't normally distributed is a good day. Today was a good day.

3

8

104

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

#Bayes folks, is it true that observing data can only ever decrease our uncertainty? That is, the posterior will be more specific than our prior? This feels like it should be right, but also like there's no way it is...

19

5

102

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

6 years

@FraudSyndrome @CoryjTaylor @FilmCritHULK Hermione is Piggy, Ron is Gonzo, Harry is Kermit.

7

1

94

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 months

Here is something I suspect happens *a lot*: Researchers have nested data. Run rmANOVA. Don't get significant results. "Hey, how about multilevel models?" Run MLM with only random intercept. Everything is significant. "I guess MLMs have more power!" Publish.

6

12

90

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

{effectsize 3.3.0} now on CRAN! The pkg for all you effect-size needs, now with Pseudo-standardized coefficients for mixed models, and more! 🥳 #rstats @easystats4u

0

33

85

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

How to compare coefficient of different predictors within a single model, in #rstats ? Here are 4 methods!

Stat’s What It’s All About - Testing The Equality of Regression Coefficients

shouldbewriting.netlify.app

8

20

82

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

Ten Common Statistical Mistakes to Watch Out for When Writing or Reviewing a Manuscript (by Tamar Makin & @jjodx ). Should be required reading for any undergrad (and grad!) methods course!

2

26

81

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

#mixedmodels folks, I've often seen random effects like: y ~ Time + (1 | ID / Time) Where I would expect: y ~ Time + (Time | ID) Are they equivalent? When would one prefer which of the parameterizations? (Assume time is categorical)

13

11

81

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

effectsize 0.3.1 is on CRAN! #RStats Together w/ some other @easystats4u goodies, it now powers this any-effect-size-from-any-test-statistic #shiny calculator (with confidence intervals!) Give it a spin!

2

26

78

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

1st preprint from my PhD! Individuals’ RT slopes in mental rotation tasks are related to visual-spatial abilities - often interpreted as reflecting greater SPEED of mental rotation. But, does it? In this paper I use EEG and computational models to show that, well, no.🧵

PsyArXiv-bot

@PsyArXivBot

1 year

Are Visual-Spatial Abilities Related to the Speed of Mental Rotation?

0

1

2

17

82

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

7 months

Tired of the war. Exhausted by the horror stories that are still coming out nearly two months later. Can't deal with the images eched onto my mind forever. The news. The sirens. The worries. Wish I could just post about stats, about R, about literally anything else. ❤️‍🩹

5

1

75

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

New blog post after almost a year! Modeling Signal Detection with Ordinal Regression #rstats #brms

Stat’s What It’s All About - Ordinal Regression as a Model for Signal Detection

blog.msbstats.info

6

20

75

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

My snippet that has saved me thousands of mistakes! #RStats

Amit Grinson

@Amit_Levinson

2 years

Snippets are fantastic time savers, especially when you custom make them. You can also set the cursor to appear at a location you want (or several!) My frequent one is for setting up an sql connection, but I imagine I can make a few more if I dig deeper. #rstats

1

4

23

5

12

74

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

From LinkedIn: the mean is only meaningful when the distribution is normal - look at the median and Mode instead 34k 👏 44k 👍 13k 🔁 🗨️ Wow! 🗨️ Brilliant! 🗨️ Never thought about that! --- Top quality content 🤦‍♂️🤦‍♂️🤦‍♂️

6

0

71

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

They said it couldn't be done. They said I was crazy. But I did it. I taught 7 (!) courses this semester that ended today (two of them completely new!) all while also continuing to grow my consultancy services! 1/

5

2

72

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

This week I finished teaching SEM in #rstats with #lavaan and @cjvanlissa 's awesome tidySEM (for plotting)! You can find all the (shoddy) materials here:

GitHub - mattansb/Structural-Equation-Modeling-foR-Psychologists: Lesson files used in the Struct...

Lesson files used in the Structural Equation Modeling for Psychologists. - mattansb/Structural-Equation-Modeling-foR-Psychologists

github.com

3

18

67

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Modern pipes require modern hex stickers #RStats

4

8

68

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

{effectsize📦} 0.4.0 in now on CRAN! @easystats4u #rstats This update comes packed with many new features - so here's a short thread of some of the highlights.

Indices of Effect Size

Provide utilities to work with indices of effect size for a wide variety of models and hypothesis tests (see list of supported models using the function insight::supported_models()), allowing...

easystats.github.io

2

23

68

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Bayesians: Bayesian stats allow you to incorporate prior information into your conclusions! Also Bayesians: use flat priors / weakly informed priors / default priors. (Who you calling a subtweet???)

8

5

67

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

9 months

Today my son is one month old. For the last 10 days (so a third of his life) I've been staring at him trying not to think what might have been had we lived somewhere else last Saturday. My almost 3yo is tired of me embracing her. #HamasisISIS

2

1

63

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Good model fit ≠ theory is correct

Eiko Fried

@EikoFried

1 year

"Using structural equation models, we discovered underlying triangles in the data." 🙃 #StatisticalEquivalence

8

39

304

2

9

62

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

11 months

How to become "the stats person": When asked if I know method X, 1st time: no, what's that? 2nd time: think I've heard about it 3rd-6th time: I'm not sure, but I can take a look & think with you. 7th: yeah, what do you need? After 10th time: .. Adding to my CV.

4

3

60

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

9 months

We're safe. But devastated. Can't think of what else to post. #IsraelUnderAttack

0

60

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 months

@MatthewBJane People are walking around thinking all effect sizes are R²≥0.8 when really a hopeful est. would be R²≈0.08. The sex difference in height is only about R²=0.25...

6

4

59

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

Abandon p values (and CIs), AND effect sizes... Let's not forget that anything Bayesian is LITERALLY data fabrication... Do we even need to bother collecting data? Guess I'll hand in a packet of blank pages for my PhD dissertation 🤷‍♂️

PsyArXiv-bot

@PsyArXivBot

4 years

Do Effect Sizes in Psychology Laboratory Experiments Mean Anything in Reality?

2

32

136

2

5

59

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Feeling pretty proud of myself for writing this semi-arbitrary distribution in @mcmc_stan (to demo in class). 1. R code for weighted mixture "prior" 2. ggplot of (normalized) arbitrary function 3. Stan code for 1 4. ggplot with MCMC samples on-top of target dist.

4

56

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Close enough! #rstats

1

6

57

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 months

@ajordannafa There, fixed it for ya:

4

3

56

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

"Just add the data" How do you do this for SEM? Difficult to do with adjusted confounders, or small effect sizes, or hierarchical data with high heterogeneity, or with.... If it was as easy as showing the data, we wouldn't need stats. But sure, if you have a t-test, go for it.

Ethan Mollick

@emollick

1 year

We are visualizing our data all wrong. Typically, academic graphs show statistical confidence intervals or standard errors. This paper shows that everyone from doctors to tenured faculty misread these graphs, assuming effects are larger. Show variability!

25

260

1K

4

3

55

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

{effectsize} 📦 0.6.0 is out! #rstats @easystats4u This update brings some bug fixes, as well as: 1) Common-language effect sizes (CLES) for independent group differences (both parametric and non-parametric). You got your p(sup), U3, overlap... 1/2

1

12

56

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Today I presented my dissertation in the department seminar... Now I just need to write the thing! #phdlife #relief

7

0

55

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

8 months

Never in a million years would I have guessed that I would be unfollowing, muting, and blocking people on this site because they are somewhere between indifferent to the slaughtering of my people and blatantly antisemitic. What a time to be alive (a shitty time).

4

2

55

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

You'd think a HUGE change to the Bayesian rmANOVA (that essentially invalidates all previous uses of the analysis) would perhaps be emphasized and not buried at the end of a version update announcement? Nah.

JASP Statistics

@JASPStats

2 years

Introducing JASP 0.16.3: Quality Control, GLMs, Bayesian State Space Models, and More! #OpenSource

1

20

55

13

16

52

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Client: <tells me about their data and research> Me: that sounds doable! Me: ... Me: Wait... do you have any missing data? Client: boy, do I! This happens way too often...

4

0

52

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Of all the "canonical" link functions, logit* is by far the worst if you actually want to understand your results (infinitely so if you have an interaction term in your model)... #statstwitter (* Yes, probit is just as bad)

13

4

49

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

#ChatGPT probably should not be used as a tool for knowledge acquirement. BUT it can be used to articulate an idea we a-prior know to be true. E.g., I was having trouble explaining the difference between Binomial and Poisson distributions. This was pretty useful: #statstwitter

4

0

45

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

11 months

Okay, once an for all, I will give the definitive definition of a confidence interval, and how one should interpret a single realized confidence interval. 🧵1/312

Stephen Wild

@stephenjwild

11 months

@mjskay @Corey_Yanofsky @sharoz I get the confusion around CIs. A reasonable question is how do you interpret any *one* CI? I have yet to see a good, short, intuitive answer for confidence intervals that I like.

4

0

9

2

1

46

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Anyway - update JASP and re-run your analyses - see how drastically your conclusions change. This is a reminder NOT to run analyses you don't fully understand only because they come in pretty packaging...

Bayesian Repeated-Measures ANOVA: An Updated Methodology Implemented in JASP

Analysis of variance (ANOVA) is widely used to assess the influence of one or more (quasi-)experimental manipulations on a continuous outcome. Traditionally, ANOVA is carried out in a frequentist...

osf.io

3

8

45

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

Come see me at #ISCoP2023 !

4

3

45

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Is there any way to just get a portion of the polar plot? Say, just the right side? #rstats #ggplot

6

7

43

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 years

New blog post: Signal Detection Theory vs. Logistic Regression - are they one and the same??? (yes) #rstats

4

12

44

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

So here I was, about to finish writing a blog post about modeling signal detection with multiple responses using #brms (that I started writing a year ago), when I came across @vuorre 's amazingly detailed blog post - go read it! All of it!

Matti’s homepage - Bayesian Estimation of Signal Detection Models

Signal Detection Theory (SDT) is a popular theoretical framework for modeling memory and perception. Calculating point estimates of equal variance Gaussian SDT parameters is easy using widely known...

mvuorre.github.io

5

3

42

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

And once again, I find myself recommending @_MiguelHernan 's "The C Word" to someone!

The C-Word: Scientific Euphemisms Do Not Improve Causal Inference From Observational Data | AJPH |...

Causal inference is a core task of science. However, authors and editors often refrain from explicitly acknowledging the causal goal of research projects; they refer to causal effect estimates as...

ajph.aphapublications.org

4

10

41

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 months

Ah yes, peer review, what a system. Splendid.

Eiko Fried

@EikoFried

3 months

Another Elsevier article clearly written by hooomans: "I'm very sorry, I don't have access to real-time informatino as I am an AI language model" 🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖🤖 ()

34

192

813

4

2

42

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Bayes factors are sensitive to parameterization, priors and more. If you don't have a good grasp of these, or a good idea of what they should be, you probably shouldn't use any "defaults". ... and if you do, you probably don't need the defaults.

Fabian Dablander

@fdabl

3 years

In a now published paper, we compare two Bayesian tests, find puzzling differences, and argue for a new default. But the key takeaway is this: even behind seemingly simple statistical tasks there may be dragons. Beware. @khuth6 @AlxEtz @EJWagenmakers

1

28

118

2

39

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 years

New blog posts: How to model individual-level and group-level processes using linear-mixed models (and when you don't need to!) This is a follow-up to @SolomonKurz 's excellent post (and the following twitter discussion that sent me back to the books).

Solomon Kurz

@SolomonKurz

5 years

New blog up! It’s #2 of the idiographic and small-N statistics series. This time we continue motivating the topic by covering the ecological fallacy.

3

34

88

1

8

40

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

ggplot() + geom_T()

Josemari Feliciano

@jmtfeliciano

2 years

Wait a second, these are not real error bars … the author literally just put the letter “T” above the bar graphs 😭

531

4K

53K

1

37

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

I propose we stop calling it "linear regression" and start calling it "gaussian regression".* Linear = identity link function (which can be used with any likelihood function)† But more often, people mean "yeah yeah, the one with the normal errors"‡

11

2

38

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

I tend to keep my presence here free of politics, but waking up to the smell of smoke.. I cannot remain silent. Our democracy is in danger. Shards are now missing from where there were once only cracks. This Passover we will be fighting to not be shackled once more. #Israel 💙

5

1

38

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

@cameron_pfiffer <Me, explaining to students log odds> Them: and why are we doing this? Me: we really like linear spaces and really hate being able to interpret coefficients....

1

2

36

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Confidence intervals on p-values.

4

6

37

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Shoutout to the #rstats @mcmc_stan team for the rvar() data type (from {posterior}) which makes working with posterior draws sooooooo easy. Extra shoutout to @mjskay for streamlining #ggdist 's support for rvars a thing of beauty!

1

4

36

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

10 months

Oh, we ( @easystats4u ) did a thing and you liked it! Many thanks to whoever nominated us and to @improvingpsych for choosing easystats to appear alongside all these other great projects!

SIPS (Account Archived - See Bio)

@improvingpsych

10 months

easystats ( @easystats4u ) is a collection of #rstats packages aimed at providing a consistent framework to explore, understand and report statistical models. @strengejacke , @Dom_Makowski , @mattansb , @patilindrajeets , @bmwiernik , @EtienneBacher , @RemPsyc

1

10

38

2

7

37

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

5 months

This one is for all you "extrapolation bad, but interpolation okay" people.

Randall Munroe

@xkcd

5 months

Sphere Tastiness

231

2K

23K

1

6

36

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

The #mcmc life comes with built-in coffee breaks! (I feel like someone posted this, but couldn't find it, so had to recreate it... for pedagogical reasons)

1

2

36

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

Multicollinearity is messy - reducing power and making estimates unstable and interpretation difficult. In this post I demonstrate how ignoring co-variation between predictors can lead to odd results... and how to deal w/ this using #rstats 's { #emmeans }.

Stat’s What It’s All About - (What to do) When Predictors Co-Vary

shouldbewriting.netlify.app

1

7

35

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

{cheatR} is finally on #rstats CRAN - developed with @almogsi ! cheatR provides a set of functions to compare text files for similarities - originally developed for detection of "overlaps" (🤨) in course hand-ins! The shiny app:

1

13

33

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

Are the days of emmeans soon behind us? It feels almost like heresy... I have got to get on this train!

3

4

33

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

@datingdecisions @bmwiernik sandwich %>% melt()

1

0

34

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

4 years

Just in time to close off 2020: Our ( @Dom_Makowski @strengejacke ) {effectsize}📦 got accepted at @JOSS_TheOJ ! All your effect size needs (with confidence intervals and interpretation!) in one place - powered by @easystats4u ! Check it out:

JOSS

@JOSS_TheOJ

4 years

Just published in #JOSS_theOJ : 'effectsize: Estimation of Effect Size Indices and Standardized Parameters'

0

3

10

1

13

35

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

1 year

I taught a Mixed Models course for MA psych students at both @BenGurionUni and at @TelAvivUni in R. All of the R materials, built with my amazing TA - Yael Bar Shachar, can be found here: 2/

GitHub - mattansb/Hierarchical-Linear-Models-foR-Psychologists: Lesson files for Hierarchical...

Lesson files for Hierarchical Linear Models for Psychologists. - mattansb/Hierarchical-Linear-Models-foR-Psychologists

github.com

1

5

34

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

GAM people, is there a way to separate a smooth term from it's "by"? In other words, is y~f+s(x,by=f) Equivalent to? y~f+s(x)+t2(x,by=f) Is yes / no, is there a way (other than model comp) to get the "interactive term" (statistic, effect plot...)? #rstata #mgcv

4

5

34

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

2 years

I have opinions.

2

3

34

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

11 months

Thinking today about that time someone asked in a FB stats group how to select confounders, and this guy said "stepwise regression", and he got so furious when I and others explained (politely and patiently) that 'no actually', that he called us all babies and left the group.

4

3

34

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

I have on my CV, under "preprints", manuscripts I have no intention of ever submitting to a journal (for reasons). Is this fraud? Do we have a better name than "preprint"? #AcademicTwitter

13

7

32

Mattan S. Ben-Shachar 🎗️🇮🇱🇺🇦

@mattansb

3 years

Check out the new vignette about assumption checks for ANOVAs, with {performance} 📦 @easystats4u

Henrik Singmann

@HenrikSingmann

3 years

afex version 1.0 is now on CRAN with breaking changes to defaults: (1) For RM-ANOVAs, follow-up tests use the multivariate model per default. (2) For mixed(), p-values are based on Satterthwaite method ("S"). Also, @mattansb has contributed a new vignette.

0

14

51

1

11

34