powerGlmer2: Power analysis for non-normal data 2

In fullfact: Full Factorial Breeding Analysis

Power analysis for non-normal data 2

Description

Extracts the power values of dam, sire, and dam by sire variance components from a generalized linear mixed-effect model using the glmer function of the lme4 package. Options to include one random position and/or one random block effect(s).

Usage

powerGlmer2(varcomp, nval, fam_link, alpha = 0.05, nsim = 100, position = NULL,
block = NULL, poisLog = NULL)

Arguments

varcomp	Vector of known dam, sire, dam by sire, and position and/or block variance components, i.e. c(dam, sire, dam x sire, position/block). If there is a position and a block c(..., dam x sire, position, block).
nval	Vector of known dam, sire, offspring per family, and offspring per position or number of blocks sample sizes, i.e. c(dam, sire, offspring, position/block). If there is a position and a block c(..., offspring, position, block).
fam_link	The family and link in family(link) format. Supported options are binomial(link="logit"), binomial(link="probit"), poisson(link="log"), and poisson(link="sqrt").
alpha	Statistical significance value. Default is 0.05.
nsim	Number of simulations. Default is 100.
position	Optional number of positions.
block	Optional vector of dams and sires per block, e.g. c(2,2).
poisLog	The residual variance component value if using poisson(link="log").

Details

Extracts the dam, sire, dam, dam by sire, and position and/or block power values. The residual variance component for the fam_links are described by Nakagawa and Schielzeth (2010, 2013). Power values are calculated by stochastically simulation data and then calculating the proportion of significance values less than alpha for each component (Bolker 2008). Significance values for the random effects are determined using likelihood ratio tests (Bolker et al. 2009).

Value

Prints a data frame with the sample sizes, variance component inputs, variance component outputs, and power values.

Note

The Laplace approximation is used because there were fewer disadvantages relative to penalized quasi-likelihood and Gauss-Hermite quadrature parameter estimation (Bolker et al. 2009). That is, penalized quasi-likelihood is not recommended for count responses with means less than 5 and binary responses with less than 5 successes per group. Gauss-Hermite quadrature is not recommended for more than two or three random effects because of the rapidly declining analytical speed with the increasing number of random effects.

References

Bolker BM. 2008. Ecological models and data in R. Princeton University Press, New Jersey.

Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, White J-SS. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology and Evolution 24(3): 127-135. DOI: 10.1016/j.tree.2008.10.008

Lynch M, Walsh B. 1998. Genetics and Analysis of Quantitative Traits. Sinauer Associates, Massachusetts.

Nakagawa S, Schielzeth H. 2010. Repeatability for Gaussian and non-Gaussian data: a practical guide for biologists. Biological Reviews 85(4): 935-956. DOI: 10.1111/j.1469-185X.2010.00141.x

Nakagawa S, Schielzeth H. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4(2): 133-142. DOI: 10.1111/j.2041-210x.2012.00261.x

See Also

powerGlmer, powerGlmer3

Examples

#100 simulations
## Not run: powerGlmer2(varcomp=c(0.7880,0.1667,0.1671,0.0037),nval=c(11,11,300,3300),
position=11,fam_link=binomial(link="logit")) 
## End(Not run)

fullfact documentation built on May 29, 2024, 1:21 a.m.