PMAP 8521 • Example: Matching and IPW with R: 5: Inverse probability weighting

[@amanielkefleyesus3236]

Kudos+++ for this clear and well-demonstrated concept of matching and IPW ! Thanks

[@TheYepyep3]

Your videos are extremely helpful, thank you so much for making these!

[@muhammedhadedy4570]

Wow. That's amazing. This is propaply the best tutorial in explaining propensity score analysis on KZbin. Thank you so much for your work. You literally saved my life. :D :D

[@paolatello8083]

I'm veryyyyyy happy with your explanation! Keep doing this videos!

[@kjevlevigen]

Fantastic procrastination before my Epidemiologi exam! Thank you :-)

[@gordiballester]

Really helpful. Greetings from Spain, Viva España.

[@noahcat9064]

This is super helpful! Thanks for the video!

[@austinpower14]

Do anyone know what to do if you have a multinomial distribution? I have 4 potential outcomes.

[@kenkoonwong2166]

after calculating IPW, do we need to add the confounding variables back to lm to control after the weight has been provided? Thank you for the explanation. the video is very helpful.

[@azulchincarini8924]

Is there the possibility of combining differences in differences with matching propensity score?

[@dataanalytics3001]

If there are various estimates from adjusted regression, PS, and IPW all of them are significant and in the same direction. Is there any statistical tests to decide which one is correct or closer to the truth?

[@priyankaroy3686]

extremely useful video for me. Thank you. However, I would like to ask how did you convert net into a new numeric column called net_num? I have a variable with "yes" and "no" which i had converted into factors with "as.factor" function. But that is not working for generating IPWs. please help

[@munthirmansour]

what is the package used for ipw

[@yeobin_mingyu]

Thank you SO much sincerely ! 감사합니다 !

[@TheBjjninja]

Thanks you. Around 9 minutes in, I would probably use mutate(ps= predict(model, type='reponse'). Much easier and less typing to just use the model and run predict()

[@trueperson22]

Can we just add family=binomial without link="logit" at 3:27? Thanks a lot for this great lecture

[@phamnguyenductin]

Yes, the default link function for family = binomial is the logit function. Only if we intend to use another link function do we need to specify otherwise.

[@florencenyaguthii5022]

Thanks for this video very helpful

[@dr.kingschultz]

Very good video

[@taiwankyh]

Thanks so much

[@caviper1]

Gracias