[R-sig-ME] glmADMB mcmc output

Dean Castillo dmcastil at umail.iu.edu
Mon Mar 12 17:04:53 CET 2012


Dear list members,

I have run the ML and mcmc for my model. I am wondering if there is a
simple transformation between the the ML coefficients and mcmc
coefficients. Or in general how do I interpret the mcmc coefficients
(I know how to interpret the ml coefficients). For example e^4.03 is
not in the range of the HPD for beta1 of the mcmc

nb1<-glmmadmb(D2_eggs~D2_female*(Bacteria+Enviro)+(1|Pop), data=data,
family="nbinom1")
> summary(nb1)

Call:
glmmadmb(formula = D2_eggs ~ D2_female * (Bacteria + Enviro) +
  (1 | Pop), data = data, family = "nbinom1")


Coefficients:
                  Estimate Std. Error z value Pr(>|z|)
(Intercept)          4.03120    0.26620   15.14  < 2e-16 ***
D2_female            0.05115    0.00532    9.62  < 2e-16 ***
BacteriaE            0.13913    0.25159    0.55     0.58
EnviroEC             1.06900    0.22090    4.84  1.3e-06 ***
D2_female:BacteriaE  0.02622    0.00578    4.54  5.7e-06 ***
D2_female:EnviroEC  -0.03357    0.00471   -7.13  9.7e-13 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Number of observations: total=95, Pop=16
Random effect variance(s):
Group=Pop
          Variance StdDev
(Intercept)  0.04716 0.2172
Negative binomial dispersion parameter: 243.68 (std. err.: 48.422)

Log-likelihood: -572.814

>nb_mcmc<-glmmadmb(D2_eggs~D2_female*(Bacteria+Enviro)+(1|Pop), data=data, family="nbinom1",mcmc=T, mcmc.opts=mcmcControl(mcmc=10000))
> m<-as.mcmc(nb_mcmc$mcmc)
> head(HPDinterval(m))
          lower     upper
beta.1 50.4190302 55.231164
beta.2  6.3920598  9.033473
beta.3  0.5165269  4.145737
beta.4  1.7327276  5.167999
beta.5  1.6289002  3.791544
beta.6 -3.9886068 -2.057463

Thanks

Dean




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