I hope this is the the right video for calculating minimum detectable effect size if I see an observational study published in a paper and I am reviewing it for discussing in journal club? My main concern is not to jump to an erroneous conclusion of equivalence based on an underpowered observational study which did not even mention any power analysis. This misunderstanding has a potential for negatively impacting patient care. Is there an article and is there a calculator?
thanks for the video, it is very didactic I have some questions: - At 2:37, shouldn't it be ATE instead of ATT? - 22:48, when calculating the variance of the outcome variable, should this be for the treatment, control or whole population? Shouldn't this be sqrt(VARtest/ntest + VARcontrol/ncontrol)? - At 31:10, I can't understand the clustered formula. Imagine that intra correlation is equal to 0, then the MDE would be multiplied by sqrt(1/n), which is a very low number depending on sample size. For me, this segment needs to return 1 when intra correlation is 0. Something like sqrt(1+rho(n-1)) Eager to hear your thoughts (or anyone else)!
awesome video. super clear and really shows formula for calculating MDE and all its nuances. thank you so much for posting this
thank you for sharing your knowledge. I am really indebted to you. I cant tell how much I appreciate what you do
A great video with very clear explanation.
Thank you. Interesting video. What is a reasonable MDE (in % respect to the mean) to consider an experiment good enough?
I hope this is the the right video for calculating minimum detectable effect size if I see an observational study published in a paper and I am reviewing it for discussing in journal club? My main concern is not to jump to an erroneous conclusion of equivalence based on an underpowered observational study which did not even mention any power analysis. This misunderstanding has a potential for negatively impacting patient care. Is there an article and is there a calculator?
Last lonely question here, the t statistic your considering with how many degrees of freedom?
thanks for the video, it is very didactic
I have some questions:
- At 2:37, shouldn't it be ATE instead of ATT?
- 22:48, when calculating the variance of the outcome variable, should this be for the treatment, control or whole population? Shouldn't this be sqrt(VARtest/ntest + VARcontrol/ncontrol)?
- At 31:10, I can't understand the clustered formula. Imagine that intra correlation is equal to 0, then the MDE would be multiplied by sqrt(1/n), which is a very low number depending on sample size. For me, this segment needs to return 1 when intra correlation is 0. Something like sqrt(1+rho(n-1))
Eager to hear your thoughts (or anyone else)!
thanks! very clear overview
Great explanation!
Also, it would be great if you can share the source of your MDE formula in 21:31
Thank you!