Jordi Real

Volatil

[Pedro Vallín @pvallin]

Dice el Veterano Estadista Socialista™ que la democracia española, que se fundó amnistiando a quienes violaron los derechos humanos durante 40 años, no se puede permitir amnistiar al dire de un cole que abrió el centro para que la gente votara.

[Jordi Real @JordirealReal]

@lauraferni12 @ManuelBartual @Premios_Ondas Enhorabuena por la interacción de @ManuelBartual

[Matt Dancho (Business Science) @mdancho84]

I think the message in #DataScience needs to be: Don't believe everything you read. 🧵

[Jordi Real @JordirealReal]

@AlbertRoso @IDIAPJGol @ARC__KI @UABBarcelona Felicitats superdoctor

[Jordi Real @JordirealReal]

@AnaBayes R base mínimo imprescindible previo a cualquier otro paquete que requieran código. De acuerdo en que será más sencillo tidyverse después. Ademàs con R base podrias hacerlo todo, con tidyverse no.

[Frank Harrell @f2harrell]

@karlahemming @ADAlthousePhD Neither. RCTs do not sample from populations. Report uncertainty intervals for differences in outcomes between treatments. https://t.co/RZgl93Mngj

[Chelsea Parlett @ChelseaParlett]

sometimes a statistical consultant isn't there to help you build fancy models, but to help you to figure out that you don't need one.

[Frank Harrell @f2harrell]

Interesting study, incompetent analysis: “We found that men in the highest quintile of ultraprocessed food consumption, compared those in the lowest quintile, had a 29% higher risk of developing colorectal cancer,” - https://t.co/m9RzRtJSQE

[Maarten van Smeden @MaartenvSmeden]

Should a prediction model be developed?

[Maarten van Smeden @MaartenvSmeden]

Stop calling everything AI is a good suggestion
https://t.co/0kQdcYOGhy

[Matt Dancho (Business Science) @mdancho84]

This used to be me when I was first learning linear regression. 😂

But here's what's changed for me. 🧵

#rstats

[Gary Collins @GSCollins]

The next in a very long line of studies concluding

All [prediction model] studies were at high risk of bias due to inappropriate analysis methods, omission of important statistical considerations, suboptimal validation

-> https://t.co/8P9fJbtdif

#statstwitter #researchwaste

[Valentin Amrhein @vamrhein]

A single p-value for the null hypothesis of no effect is not enough.

https://t.co/LtskirFZQ8

[Maarten van Smeden @MaartenvSmeden]

This is a MUST read for anyone involved in prediction modeling. Not only has it the best title, it also has the best content
https://t.co/iuSxProlPX

[Maarten van Smeden @MaartenvSmeden]

Doing descriptive statistics can be helpful for generating new hypotheses. My new hypothesis is that my data are wrong

[Maarten van Smeden @MaartenvSmeden]

The best thing about being a statistician is that you get to play in everyone's backyard, and then get to tell them their yard would have been so much nicer if they had consulted you earlier

[Oscar Arias @OACerebro]

Para todos los amantes de la ciencia 🔬

¡Una lectura obligatoria!

Archivo descargable 📂 https://t.co/6QtYMVV4gI

[Christoph Molnar 🦋 christophmolnar.bsky.social @ChristophMolnar]

Why I hate cluster analysis

• What's longer, 10 cm 📏, 200$ 💵or a change from red to blue 🎨???
• What's the difference between banana 🍌 and apple 🍎? What about lemon 🍋?
• How many clusters? 2? 10? 100? I can do whatever you want, no wrong answers!

[Christoph Molnar 🦋 christophmolnar.bsky.social @ChristophMolnar]

I don't *hate* clustering, but the ambiguity makes me uncomfortable.

[Gary Collins @GSCollins]

REMINDER: Professional biostatisticians go through 3-4 years of undergraduate training, then maybe a 1 year Masters degree, then maybe 3-4 years of a PhD.

Biostatistics is a profession, not a hobby.

#subtweet #statstwitter #medtwitter