Geographically weighted regression (GWR)


I am looking for some info on possibility of making a GW regression with a dependent variable as floating point numbers and independent variable(s) as binary data (1/0). I experimented many different software, like arcgis 9.3, STIS, SAM..

Since I do not have in-depth knowledge on statistics, I appreciate If someone inform me on applicability of this situation and which model can handle this "floating point--binary" combination ; linear, logistic or poisson?