PlotGLMERFactor: Make an error-bar plot showing model coefficients

In timnewbold/StatisticalModels: What the package does (short line)

Make an error-bar plot showing model coefficients

Description

Plots modelled effects of specified factors on a specified response variable.

Usage

PlotGLMERFactor(model,data,responseVar,seMultiplier=1.96,
                     logLink="n",catEffects=NULL,
                     xtext.srt=0,ylim=NA,yaxp=NULL,order=NULL,rescale=NULL,
                     errbar.cols=NULL,pt.pch=NULL,
                     errbar.lty=1,
                     params=list(),add=FALSE,offset=0,
                     plotLabels=TRUE,cex.txt=NULL,pt.cex=1,
                     pt.bg="white",main=NULL,type="percent")

Arguments

model	The model to derive parameter estimates from
data	A data frame containing all variables in the model
responseVar	The name of the response variable fitted, as text, to display on the y-axis of the graph
seMultiplier	The number of standard errors to use for the error bars. Default is 1.96, which shows 95 percent confidence intervals
logLink	The transformation applied to the response variable before or during modelling: one of "n" (identity), "e" (log, base e), "10" (log, base 10) or "b" (logit - used in binomial models)
catEffects	The categorical effects (factors) to plot, as a vector
xtext.srt	The number of degrees by which to rotate the x-axis labels
ylim	Fixed y-axis limits. If not specified, then resolved automatically
yaxp	Specified values to include on the y axis, as used in general R graphical parameters
order	The order to plot the categorical factor levels in. If not specified, then as in the original data frame
rescale	The amount by which to rescale y-axis values for the categorical effects by. If not specified, then no rescaling is performed
errbar.cols	Colours to use for the error bars
pt.pch	The point types (using R's 'pch' values), with which to overwrite the default
errbar.lty	The line types to use for the error bars; values as in the general R graphical parameter 'lty'
params	Any R graphical parameters, which will overwrite the defaults
add	Whether to add the plot to an existing plot; default is FALSE
offset	The amount by which to offset points on the x axis; default is zero
plotLabels	Whether to plot individual labels for each factor level
cex.txt	Text size for additional labels (i.e. text not on the axes or in the axis labels, which should be specified using the params argument)
pt.cex	Point size to use. Default is 1.
pt.bg	The background colour to use for the points (as long as this is compatible with the point type used)
main	The title for the plot
type	Whether to plot as a percentage change ('percent' - the default) or directly as the response variable ('response')

Details

Code for calculating predicted values and confidence intervals for continuous effects was taken from the GLMM wiki (see references).

Author(s)

Tim Newbold <[e-mail]>

References

http://glmm.wikidot.com/faq

Examples

# Load example data (site-level effects of land use on biodiversity from the PREDICTS database)
data(PREDICTSSiteData)

PREDICTSSites$LandUse <- factor(PREDICTSSites$LandUse,levels=c("Primary Vegetation","Secondary Vegetation",
"Plantation forest","Cropland","Pasture","Urban"))

# Run a model of log-transformed total abundance as a function of land use, human population density
# and distance to nearest road (with an interaction between human population density
# and road distance)
m1 <- GLMER(modelData = PREDICTSSites,responseVar = "LogAbund",fitFamily = "gaussian",fixedStruct = "LandUse+poly(logHPD.rs,2)+poly(logDistRd.rs,2)+poly(logHPD.rs,2):poly(logDistRd.rs,2)",randomStruct = "(1|SS)+(1|SSB)",REML = TRUE)

# Plot the effect of land use as an error bar, showing +/- 1 standard error around the mean coefficient estimates
PlotGLMERFactor(model = m1$model,data = m1$data,responseVar = "Total abundance",seMultiplier = 1,
logLink = "e",catEffects = "LandUse")

timnewbold/StatisticalModels documentation built on Aug. 25, 2023, 4:58 p.m.