I am using -spregress- to perform spatial regression with spatially lagged errors and independent variables. However, I would like to incorporate precision weights (population) for my observations (US counties), and -spregress- doesn't support weights. So I am looking for possible work arounds to achieve the same goal.
One idea I had was to somehow scale the errorlag weighting matrix (I'm using inverse distance spatial matrix) to achieve the same objective, but I can't see how to do that directly.
Another idea was to use the user written program -xsmle-, which supports -aweights-, but -xsmle- seems to only work with panel data - I tried -xtset- with a single time value and -xsmle- balked.
Any ideas welcome.
I am using Stata 17 (upgrading shortly, sooner if it solves this problem).
thanks,
Jeph
One idea I had was to somehow scale the errorlag weighting matrix (I'm using inverse distance spatial matrix) to achieve the same objective, but I can't see how to do that directly.
Another idea was to use the user written program -xsmle-, which supports -aweights-, but -xsmle- seems to only work with panel data - I tried -xtset- with a single time value and -xsmle- balked.
Any ideas welcome.
I am using Stata 17 (upgrading shortly, sooner if it solves this problem).
thanks,
Jeph
