0
$\begingroup$

Is it possible to add a parameter to lmer model which will be modeling the error structure? Sth similar to TOEP(X) and SP(POW) from SAS???

$\endgroup$
3
  • 2
    $\begingroup$ I don't think so, but Ben Bolker might correct me. However, covariance structures can be specified with lme from the nlme package. See also: stats.stackexchange.com/a/49793/11849 $\endgroup$
    – Roland
    Dec 16, 2014 at 9:56
  • $\begingroup$ For future reference and for the SAS-impaired, it would be nice to describe what TOEP(X) and SP(POW) statements actually do (I'm guessing a Toeplitz structure and some kind of spatial power autocorrelation function)? $\endgroup$
    – Ben Bolker
    Dec 17, 2014 at 3:24
  • $\begingroup$ As you wrote Ben Bolker. The first one is Toeplitz and the second is spatial power $\endgroup$
    – Legionista
    Dec 17, 2014 at 7:11

1 Answer 1

3
$\begingroup$

No, there is no such parameter. However, you could try nlme library, the ancestor of lmer. nlme library enables you to define a correlation structure, e.g. Gaussian, Autoregressive, or ARMA.

For detailed description, tutorial and examples check this book: Pinheiro, J.C., and Bates, D.M. (2000). Mixed-Effects Models in S and S-PLUS. Springer.

You can also find some examples here and here.

$\endgroup$
2
  • $\begingroup$ Why would the ancestor be more powerful than the latest ? Is this a choice or is this pending ? $\endgroup$ Sep 17, 2018 at 10:15
  • $\begingroup$ @DanChaltiel don't ask me about the decisions made by the designers of both packages... $\endgroup$
    – Tim
    Sep 17, 2018 at 10:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.