[R-sig-ME] ICC for quasipoisson?

[Jarrod Hadfield]

Hi,

You can find the relevant equations in Table 2 of:

Repeatability for Gaussian and non-Gaussian data: a practical guide  
for biologists. Biological reviews. 2010. Nakagawa1 & Schielzeth

Cheers,

Jarrod

Quoting Elizabeth Oliva <elizabeth.oliva at gmail.com> on Fri, 9 Sep 2011  
08:34:50 -0700:

> Hi all,
>
> I’ve been searching at length for a way to figure out how to calculate the
> ICC for a mixed effects quasipoisson model in R from the output below (i.e.,
> intercept only model) and can't figure out the correct equation to use. For
> example, I've run mixed effects logistic regression models for which I used
> the following equation:
>
>                c<-a*a
>
>                icc<-c/(3.289868+c)
>
> [a being the standard deviation of the intercept]
>
>
>
> I'm not sure what the corollary type of equation is for quasipoisson mixed
> models.
>
>
>
> If you had any suggestions I would greatly appreciate it.
>
>
>
> Best,
>
> Elizabeth
>
>
>
>
>
> womentxengintonly <- glmmPQL(NewSpecOut_SUM ~ 1, random = ~1 |NEPEC3N,
> family =quasipoisson, data = women)
>
>
>
>
>
> summary(womentxengintonly)
>
>
>
> Linear mixed-effects model fit by maximum likelihood
>
> Data: women
>
>   AIC BIC logLik
>
>    NA  NA     NA
>
>
>
> Random effects:
>
> Formula: ~1 | NEPEC3N
>
>         (Intercept) Residual
>
> StdDev:   0.4679934  7.26936
>
>
>
> Variance function:
>
> Structure: fixed weights
>
> Formula: ~invwt
>
> Fixed effects: NewSpecOut_SUM ~ 1
>
>                Value  Std.Error   DF  t-value p-value
>
> (Intercept) 3.046308 0.04805773 4960 63.38852       0
>
>
>
> Standardized Within-Group Residuals:
>
>        Min         Q1        Med         Q3        Max
>
> -1.0333625 -0.5507037 -0.3717204  0.2138915 12.4856954
>
>
>
> Number of Observations: 5099
>
> Number of Groups: 139
>
> 	[[alternative HTML version deleted]]
>
>



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.