[R-sig-ME] Is it possible to use newey-west (heteroscedasticity and autocorrelation consistent) standard errors using a generalized linear mixed-model or generalized estimating equation?

[Sun, John]

Dear All,

I am writing to ask whether it is possible to obtain robust standard error corrections when using a generalized linear mixed model or generalized estimating equation via the Econometrics Newey-West method.

I ask this because I think it is safer to assume that the generalized linear model in glmer or glmmTMB does not perform well when the errors have autocorrelation and heteroscedasticity and/or the random-effects have a non diagonal correlation structure.

I am also concerned with generalized estimating equations in that the working-correlation which one uses to obtain standard errors for regression coefficients is a guess based on what I've read and there's no proof according to what I know that generalized estimating equation is robust to model errors having autocorrelation.

Best regards,
John


	[[alternative HTML version deleted]]