Closed darunabas closed 1 year ago
The underlying reason is that there was no need to estimate the full variance covariance matrix when the function was written almost 20 years ago to implement Waller and Gotway, as the fixed part (covariates) is given once the spatial coefficient is estimated. Since then, a route via Hessian was added, but the public health ML approach doesn't need the vcov tools, as the spatial coefficient is just a nuisance parameter. If you'd like to contribute a well-tested PR for all of the SAR, CAR and SMA cases, I'll look at it. If you only need SAR, use errorsarlm()
not spautolm()
.
Is there a way to obtain confidence interval for a
spautolm
object?While it is straightforward to get confidence intervals for objects from functions like
lagsarlm
by applyingconfint(mod)
to, it didn’t work forspautolm
object. Could anyone please provide some guidance on how to get confidence interval from aspautolm
object?Instead, I get an error: