[R] Robust standard error

[Simone Mascia]

Is there a way to estimate Robust standard errors when using a nls()
function? I'm trying to fit some data to a complicated model and everything
works fine with nls() but I also wanted to obtain a robust estimate of my
errors.

I tried "coeftest(m, vcov=sandwich)" and it seems to work, but so does
"coeftest(m, vcov = NeweyWest(m, lag = 4))" or "coeftest(m, vcov =
kernHAC(m, kernel = "Bartlett", bw = 5, prewhite = FALSE, adjust =
FALSE))". They return different error estimates so I wanted you to help me
understand what I should do, if I'm doing something wrong and other stuff.

Thank you

	[[alternative HTML version deleted]]