Can I do inferential statistics with regression if there are zero cells in cross tabulation results of predictors and dichotomous dependent variable in a certain research?
Hi, using simple regression based on crosstab data is not recommended but using generalized linear model is recommended. Would you write us more about your data?
I agree with Jochen Wilhelm's response, especially if you're contemplating a nominal x nominal type of analysis. However, do remember that, if you have a nominal variable with only two levels, and one has no cases, then that characteristic is a constant for your data set, and not a variable (and therefore should be reported as such, but omitted from model-building exercises).