Does a multiple logistic regression imply that you are controlling for the other independent variables in the model when looking at an odds ratio for any 1 particular independent variable? In other words, can you assume the odds ratios for each categorical variable in this logistic regression are more accurate then the odds ratios gathered from their respective chi square analyses? Also, when is it appropriate to include interaction variables in the MLR model? If there is no association between dependent variables in a chi square, does that mean you do NOT need to include interaction variables?
