Does anyone know how to do a power analysis for conditional logistic regression on matched case-control data that uses variable ratio matching? My data has up to 4 controls per case, therefore each case may have 1, 2, 3, or 4 matched controls. I want to calculate the minimum estimated effect of a dataset with 191 cases and up to 4 matched controls. The actual ratio is 1:2.5.
I am using Stata 16.0. I have tried using power mcc (code below), but this assumes fixed ratio matching, which would lead to an overestimation of power:
m() must contain positive integers so replacing m(4) with m(2.5) does not work.
Any advice is much appreciated!
Best,
Anna
I am using Stata 16.0. I have tried using power mcc (code below), but this assumes fixed ratio matching, which would lead to an overestimation of power:
Code:
power mcc .9, power(.8) n(191) m(1) direction(lower) power mcc .9, power(.8) n(191) m(4) direction(lower)
Any advice is much appreciated!
Best,
Anna
