[Openspace] Moran's I with EB Rates and Zero Base Values

[David Stinchcomb]

I am trying to use the "Moran's I with EB Rates" function in GeoDa.  The
denominator for the rate calculation is the number of births (county
level).  Some Texas counties have very few people - hence my interest in
the EB adjustment.  Naturally, there are a handful of counties with zero
births and the function objects, "Error: base values contain zero, no
rate computed."

What is the best way to handle this situation?  I could create a new
shapefile without the zero birth counties.  But then there are
implications for the weights matrix.  I have quite a large number of
different cases to test (different covariates, different birth outcomes,
different time periods) and each will likely have a different set of
counties with zero births for that test case.  I guess I am hoping that
there is a clever way to avoid generating new shapefiles and weights
matricies for each test case.  I the absence of a cool workaround, is it
feasible for a future version of GeoDa (or some relatively simple
customization on my part) to have the Moran's I with EB rates algorithm
handle the zero base values automatically?

Any help will be appreciated!

-- Dave Stinchcomb
    Texas Department of Health