rpt.poisGLMM.multi: GLMM-based Repeatability Using PQL Estimation for Count Data

In rptR: Repeatability for Gaussian and non-Gaussian data

Description

Calculates repeatability from a generalised linear mixed-effects models fitted by PQL (penalized-quasi likelihood) estimation for count data.

Usage

	rpt.poisGLMM.multi(y, groups, link=c("log", "sqrt"), CI=0.95, nboot=1000, npermut=1000)

Arguments

y	Vector of a response values.
groups	Vector of group identitites.
link	Link function. log and sqrt are allowed, defaults to log.
CI	Width of the confidence interval (defaults to 0.95).
nboot	Number of parametric bootstraps for interval estimation. Defaults to 1000. Larger numbers of permutations give better asymtotic CI, but may be very time-consuming.
npermut	Number of permutations for a significance testing. Defaults to 1000. Larger numbers of permutations give better asymptotic P values, but may be very time-consuming.

Details

Models are fitted using the glmmPQL function in MASS with quasipoisson family.

Value

Returns an object of class rpt that is a a list with the following elements:

datatype	Type of response (here: "count").
method	Method used to calculate repeatability (here: "PQL").
link	Link function used (here: "log" or "sqrt").
CI	Width of the confidence interval.
R.link	Point estimate for repeatability (ICC) R on the link scale, i.e. the mode of the posterior distribution
se.link	Standard error (se) for repeatability (ICC) on the link scale, i.e. the standard deviation of the posterior distribution. Note that the distribution might not be symmetrical, in which case the se is less informative.
CI.link	Vonfidence interval for repeatability (ICC) on the link scale based on the posterior distribution of R
P.link	Approximate P value from a significance test for the link scale repeatability
R.org	Point estimate for repeatability (ICC) R on the original scale, i.e. the mode of the posterior distribution
se.org	Standard error (se) for repeatability (ICC) on the original scale, i.e. the standard deviation of the posterior distribution. Note that the distribution might not be symmetrical, in which case se is less informative.
CI.org	Confidence interval for repeatability (ICC) on the original scale based on the posterior distribution of R
P.org	Approximate P value from a a significance test for the original scale repeatability
omega	Multiplicative overdispersion parameter.
R.boot	Named list of parametric bootstap samples for R. R.link gives the samples for the link scale repeatability, R.org gives the samples for the original scale repeatability.
R.permut	Named list of permutation samples for R. R.link gives the samples for the link scale repeatability, R.org gives the samples for the original scale repeatability.

Note

Confidence intervals and standard errors are inappropriate at high repeatabilities (omega < 1), because parametric bootstrapping allows only omega greater than or equal to 1.

Author(s)

Holger Schielzeth ([e-mail]) & Shinichi Nakagawa ([e-mail])

References

Carrasco, J. L. (2010). A generalized concordance correlation coefficient based on the variance components generalized linear mixed models with application to overdispersed count data. Biometrics 66: 897-904.

Carrasco, J. L. and Jover, L. (2005). Concordance correlation coefficient applied to discrete data. Statistics in Medicine 24: 4021-4034.

Nakagawa, S. and Schielzeth, H. (2010) Repeatability for Gaussian and non-Gaussian data: a practical guide for biologists. Biological Reviews 85: 935-956

See Also

rpt.poisGLMM.add, rpt, print.rpt

Examples

## Not run: 
# repeatability for female clutch size over two years.
data(BroodParasitism)
attach(BroodParasitism)
(rpt.Host <- rpt.poisGLMM.multi(OwnClutches, FemaleID, nboot=10, npermut=10))  
	# reduced number of nboot and npermut iterations
detach(BroodParasitism)

# repeatability for male fledgling success
data(Fledglings)
attach(Fledglings)
(rpt.Fledge <- rpt.poisGLMM.multi(Fledge, MaleID, nboot=10, npermut=10))  
	# reduced number of nboot and npermut iterations
detach(Fledglings)

## End(Not run)

rptR documentation built on May 2, 2019, 5:01 p.m.