?is there a test for this?

I have one nominal (disease, no disease) dependent variable and two independent variables I wish to look at regarding whether or not they are significantly influencing the dependent variable. One independent variable is also nominal (gene, no gene) and the other is a continuous variable (dose of medicine). Is it possible to run some sort of analysis to examine this? If so, is there some way of ascertaining how much of an influence each independent variable is having on the dependent variable?

Logistic regression would allow you to perform this.

Logistic will give you a probability of having the disease or not given a set of covariates.

The regression parameters that you get out of the model can give you a log odds that will let you make statements about the effect of each covariate.


johank's suggestion is better than the one I am about to give but more complicated. You could dummy code both of the binary variables and run multiple linear regression. The standard practice with binary codes is to run log regression for a few reasons: mainly because, if you are looking at the predicted valaues, and you move far enough on the x axis (either - or +) you would get a predicted value greater than the maximum value of your binary code (which is not possible). That caveat being said, linear regression can provide analysis for your design.