[R-sig-ME] Mixed Model Specification

Ben Bolker bbolker at gmail.com
Fri Jun 27 01:24:10 CEST 2014

On 14-06-26 05:26 PM, ONKELINX, Thierry wrote:
> Dear Tom,
> Your dataset is very small. Consider yourself lucky when a simple glm
> gives reasonable estimates. A rule of thumb is that your need 10
> effective observations per parameter. The number of effective
> observations in the binomial case is equal to the number of presences
> or absences (the smallest of the two). If you are very lucky: 25/2 =
> 12. So at best you can fit a model with 1 (one) parameter. e.g.
> glm(Presence ~ Substrate) when Substrate has only 2 (two) levels.
> Best regards,
> Thierry

  I should probably re-read this advice on the R-help posting guide:

When responding to a very simple question, use the following algorithm:

    compose your response
    type 4*runif(1) at the R prompt, and wait this many hours
    check for new posts to R-help; if no similar suggestion, post your

(This is partly in jest, but if you know immediately why it is
suggested, you probably should use it! Also, it's a nice idea to replace
4 by the number of years you have been using R or S-plus.)

  except that I think I shouldn't compose my response first!

> [r-sig-mixed-models-bounces at r-project.org] namens Worthington, Thomas
> A [thomas.worthington at okstate.edu] Verzonden: donderdag 26 juni 2014
> 23:00 Aan: r-sig-mixed-models at r-project.org Onderwerp: [R-sig-ME]
> Mixed Model Specification
> Dear All
> I have a question about the use of a mixed effects model. I have
> presence/absence data for a mussel species collected at 25 sites. I
> wish to relate the presence/absence to a number of environmental
> variables and also want to take into account site. Is it feasible to
> use site as a random effect as I have only one replicate per site
> e.g.
> M1<-glmer(Presence ~ Substrate, (1 | Site), family = binomial, data =
> data)
> Best wishes
> Tom

More information about the R-sig-mixed-models mailing list