categorizing monthly gross income what is the best way to categorize or even bin monthly gross income? As of now, the variable for income is numeric and continuous. I have seen many examples for annual income. Should I categorize variables into quartiles? The purpose would be to fit the transformed variable into a logistic regression. As of now, I do not feel comfortable with interpreting the continuous variable, as it seems strange to say for every dollar increase? Thanks!
 A: You don't need to bin the data to use as an explanatory variable in a logistic regression.  
If you need to bin it for some other reason than for the logit model (e.g. for use in a contingency table), then you can transform it first and then bin it.  
You could fit a separate slopes model with dummy interaction effects (that designate each income category).  This allows you to treat income as continuous as well as allowing income categories to enter the model.  You can test for the significance of these categories and/or the continuous income variable and see what sort of model fits best.
A: In general, one tries to avoid discretizing variables (why make life harder by throwing out information?).  If you are performing a logistic regression, you can keep the variable as-is if it's an independent variable; only the dependent variable needs to be categorical. 
Furthermore, there's nothing wrong with describing changes on a per-dollar basis. The idea of marginal costs or benefits is a fairly common one in business economics. You just need to be aware of the range of your data and don't assume that a marginal change goes on forever. 
If you must transform your data, net profit/loss could be useful (assuming you have data about expenses)...
