I am working with a subset of data from a national data set (BRFSS) that uses complex sampling. I will be building a multiple logistic model to predict physical distress. My unweighted sample size is N=287, with n=47 cases of physical distress and n=240 cases of no physical distress. I have several covariates that I am interested in but I am concerned about over-fitting the model. When considering the number of covariates, should I focus on the unweighted sample size, or the weighted sample size of N=139,880 (n= 27414 cases of physical distress). I would appreciate any help!