Post

I’m not sure how directly comparable, but Nate Silver basically is doing 538 on his Substack these days

how are you such a legend - this is amazing!

Great, thanks!

Thanks for making this! I used to use their p-hack thing too

Thank you!

🏆🥲

📌

📌

wow this is awesome

📌

Omg! Thank you!!

Next make a version with reasonable informative priors and a hierarchial model?

I love it. I got great publishing results showing that Democrats have a negative impact on the economy (p<0.01) but also a second result showing Democrats are great for the economy (p<0.01). As long as I don't submit to the same journal I should be fine!

A true public service!

📌

So cool, thank you for sharing!

📌 Thank goodness! I use this in my applied stats class each year.

THANKS! And great timing. I teach my class on this next week. I was able to update my link to your website thanks to you.

Well done!

📌

Have you used this before @mcintold.bsky.social ?

This is great! Such a helpful resource for instructors of statistics too.

📌

Thank you!!!

📌

One of my favorite illustrations of p-hacking xkcd.com/882/

Fantastic! Thanks so much

Wow what a public service, use this in intro stats. 🎖️

In a similar vein, I use this game by Felix Schonbrodt to teach... shinyapps.org/apps/p-hacker/

📌

Haha this is awesome!

Thanks so much for doing this!!! I also used the 538 demo in my research methods class

📌

I was just going to test my in-class activity for tomorrow and found that this was gone! Glad I looked at Bluesky (read: procrastinated) rather than prepping a new activity. Thanks so much for saving this resource!

It’s so sad what Silver did to that brand. Blame him far more than ABC.

Oh gosh I just used this in my stats class this semester and it’s always been such a helpful demonstration! Thanks so much for preserving this resource!

Omg thank you!! I used that a lot too and was scheduled to next week and now I'll use yours and of course give you credit!!

thank you, switching out in my syllabus!

Thank you!!! This is great

🙏

THANK YOU!! I have a whole assignment built around the article and widget! So glad to have it saved!

Thanks - that's amazing! 👍🏼 I use this in my teaching every year - so glad it has been preserved

And if a scientist were to track every single z-score/p-value they ever generate, it still shouldn't be a null distribution because then they are *definitely* not enriching over the null distribution in their investigations.

Because they do not publish a non significant finding. Why ? Or, maybe some of this non significant finding are reworked (Data can be filtered, truncated or reshaped, like in the example above) in order to reach the threshold.

@cragcrest.bsky.social bet you ♥️ to see it!

Nice work!

Very cool, and a great example.

📌

Thanks Andrew. Very helpful!

Legend

You can also get R/Python chunks to talk to OJS: quarto.org/docs/interac... Quarto automatically includes Plot and Inputs too, so you can make adjustable viewof things super easily quarto.org/docs/interac...

Are the p-values pre-calculated or calculated somehow on the fly? If on the fly, client or server side?

📌

My research methods lab used the 528 thing for years. Thanks for saving it by putting this together!