Every time you write "there is no effect/difference (p>0.05)", when you mean "we found no evidence for an effect/difference (p>0.05)" a baby kitten dies - ThreadSky

maartenvsmeden.bsky.social • 96 days ago

Every time you write "there is no effect/difference (p>0.05)", when you mean "we found no evidence for an effect/difference (p>0.05)" a baby kitten dies

Comments

havhax.bsky.social•95 days ago

denismongin.bsky.social•96 days ago

I will put your post on my beginner stat lecture next week

stingray68.bsky.social•95 days ago

My God! What have I done? So many kittens.

eoghan01.bsky.social•95 days ago

I try SO HARD to stress this to my students, but then all those professional researchers go and set the worst examples imaginable.

kenquarrie.bsky.social•95 days ago

But if it happens to be Schrödinger's kitten it might not be dead. Maybe we should replicate the study and check?

lordp666.bsky.social•96 days ago

Stop that. No killing kittens or you'll go to bed without having supper.

maartenvsmeden.bsky.social•96 days ago

I have found no evidence of killing kittens

houfek.bsky.social•95 days ago

We need at least a p>0.10 of a baby kitten dying.

filiphenome.bsky.social•96 days ago

How about "we found a trend in the association (p = 0.06)"?

maartenvsmeden.bsky.social•96 days ago

Hilarious to claim a trend based on 1 effect estimate. And reminds me of this list:
https://mchankins.wordpress.com/2013/04/21/still-not-significant-2/

apsmunro.bsky.social•96 days ago

Don't forget to reference this bad boy

https://www.bmj.com/content/348/bmj.g2215

uphillwheezer.bsky.social•95 days ago

The kitten situation seems to involve correlation rather causation, if we're being accurate.

gaelvaroquaux.bsky.social•96 days ago

There's a typo in your post:

I don't think that you meant to put in the word "when"

😀

stephensenn.bsky.social•96 days ago

Especially if categorical data are involved.

conjugateprior.org•96 days ago

Ah, the famous diecatomania.

andrewtanyongyi.bsky.social•96 days ago

There is no difference between "there is no difference (p>0.05)" and "we found no difference (p>0.05)" (p>0.05).

rankedchoice.bsky.social•95 days ago

I took an introductory statistics course recently, and this detail was emphasized more than anything else in the course.

expatrob462.bsky.social•96 days ago

Now I don't feel so guilty about masterbating.

tq.tqwhite.com•93 days ago

I get so tired from all the language imprecision. Nobody uses language carefully even in situations where they are pretending to. My kittens are all dead and mouldering.

jeem-reeve.bsky.social•96 days ago

https://bsky.app/profile/jeem-reeve.bsky.social/post/3lbyz6ighw42g

ppaxisa.bsky.social•95 days ago

opherdonchin.bsky.social•95 days ago

Isn't it more accurate to say: "our evidence for an effect failed to reach significance"?

jjodx.bsky.social•96 days ago

Funny, b/c I write it always as an absence of evidence and some reviewers do not like it... I then have to write an intro to p-value in my response to their comments...

maartenvsmeden.bsky.social•96 days ago

just refer to this tweet from now on, you are welcome