This is a multivariable logistic regression sample size issue for mostly categorical variables except 'age' (continuous) = total variables = 20. Research Question (RQ) = Is dependent variable (true or false) for unblocking central venous catheter clog within two hours? Please advise which sample size is the best to answer the above RQ. I have 4 ways of some sizes to approach.
- option 1 = 30 for each predictor = 30 * 20. option 2 = 15 for each
predictor = 15 * 20. option 3 = n = (20 - 1) * 20 = 380.
option 4 = Green's rule of thumb = n >=50 + 8p - 50 + (8 * 20) = 210
Missing data = how could I decide for at what percent of missing data should be replaced with the mean of the relevant independent variable?
I hope you will consider my request favorably. Thank you in advance.