I am running a logistic regression analysis with binary dependent variable and a mix of categorical and continuos Indep. Variables with a complex sample design in Spss 20 and windows XP. Want to run an interaction model that tests the interaction between a categorical interaction term and main effect significant predictors (about 6 predictors) but under the complex sample design it forces me to run interactions with all covariates and predictors in model. When I try to run with just interactions terms I am interested in, it kicks the excluded variables out of model analysis thus I am analyzing just interaction terms outside of the main effects model. Are my interaction coefficients still valid? There are about 20 independent variables in model with a sample size of 2600.