Thanks to a funding from @fnr.lu 🙏

He will be coming next February. If you are around, please, come to the lecture.

I am happy for you to present it :)

What do you suggest when it is a common practice in some places ?

No. In the paper I read, they use multiple multivariable logistic regressions on 1000s potential exposures of an outcome, to identify those which remain significant at a very low p-value threshold. They do another multivariable analysis, and after something about Mendelian randomisation.

Enjoy and say hello to @pwgtennant.bsky.social :)

The absence of ego 😁

Not sure how to describe me if you were stupid 😬 You are very far from being stupid.

Not here 🤔 It seems that it is more a technical issue than a political shut down.

Not fully there yet ! But this was temporary ! This is a relief 🥲

@fnr.lu @siog-nahs.bsky.social @siogorg.bsky.social

I know about the residual confounding at T0. It was more about the Daggitty comment I was worried about. Thank you for your help. Hope you are doing well :)

Actually, a bit more complicated than that. But basically observed only at T1. I tried with observed/unobserved etc still incorrectly adjusted.

Observed at T1 but not observed at T0.