A question for #StatsTwitter

We are using propensity score matching to estimate a treatment effect and 95% CI. Because of small N, we are using bootstrapping. Why is my treatment effect always bigger when I use bootstrapping versus not?

Demetri A. Jordan Nafa Chelsea Parlett-Pelleriti

The Rdatasets website now hosts over 2000 datasets in CSV format. All data are free and documented. Perfect for teaching or if you're just looking for some data to play with. vincentarelbundock.github.io/Rdatasets/inde… #RStats #StatsTwitter #EpiTwitter #EconTwitter #PolisciTwitter

Me reading the #statsTwitter discourse

Delighted my book “Introduction to Clinical Trials” is released next week!

Available @ Amazon
amazon.co.uk/Introduction-C…
or via OUP
global.oup.com/academic/produ…
and soon all good bookshops 😇

#statstwitter #epitwitter #ClinicalTrials #clinicalresearch

Okay which one of you nerds did this? #epitwitter #statstwitter

I'm getting really tired of significance testing... #statstwitter

Have you seen a really great application of distance metric learning? I wanna see! #statstwitter

Delighted to have picked up Boen & Zahn 1982 but note the previous reader circled this section on 'anger' 😟

#statstwitter

📣 JMbayes2 R package now offers time-varying associations between the biomarkers and the events of interest. Curious? Explore our latest vignette! 🚀

#rstats #statstwitter #epitwitter

What does #statstwitter think about removal of outliers. Are there any papers with the authoritative answer?

A. Jordan Nafa Demetri Cameron Patrick

🙏Please share your best resources, tools, approaches, etc., for conducting power calculations of INTERACTION EFFECTS 😱

Retweets are also greatly appreciated! 💪💪💪

#EconTwitter #PopTwitter #SocialScienceTwitter #PolicyTwitter #StatsTwitter #AcademicChatter #AcademicTwitter

Hi #rstats #statstwitter ,
With a multilevel GLM with a binomial link function, e.g., glmer(y~treatment * time + (1 | id), what is the best way to report the probability differences between the groups in the text of an article? (1/3)

#StatsTwitter what go-to books and/or resources do you find yourself revisiting often?

Hey #statstwitter , can you help me figure out what's up with this weird smoothing procedure?

Is this LOESS?

My statistical test assumes that I randomize.

I indeed randomize.

Do I need to do a balance test to check whether the randomization “worked”?

#rstats #statstwitter #econtwitter

📣I'll deliver a course on #JointModels at the upcoming congress of Sociedade Portuguesa de Estatística! I’ll present our #RStats 📦 JMbayes2, developed with Dimitris Rizopoulos and Greg Papageorgiou.

Join me in the charming city of Guimarães. Abstract submission deadline: 18 June. #statstwitter #epitwitter

We're pleased to share our latest guidance for reporting clinical prediction model studies that have accounted for clustering (eg, multiple hospitals/centres or multiple datasets)

checklist tinyurl.com/ytjp4bmj
explanation tinyurl.com/3sykc65e

#statstwitter #epitwitter

I find it really awkward to display binary variables in summary statistics tables like they’re continuous variables (min = 0, max = 1, etc). Does anyone have a better approach for summarizing/visualizing your data’s binary variables?

#RTwitter #StatsTwitter #EconTwitter

I am teaching model building/ variable selection using DAGs in my regression course.

I'd love to provide a few different real-world examples of the consequence of conditioning on a collider. Does anyone have a favorite teaching example? #epitwitter #statstwitter

When I was an undergrad at UChicago, I worked for the stat department as a consultant, mostly helping masters students with limited quant knowledge to finish their theses. During the first consult the student asked me, “Can you help me make my p-values <0.05?”

#StatsTwitter