influenceFun: Identify highly influential data points.

In p-schaefer/psUtilities: Utility functions for ecological analysis

Description

Use influence.mer from the influence.ME package. Influential observations and/or groups are identified as being above the critical threshold of 4/N, where N is the number of unique observations or groups.

Usage

influenceFun(model, model.variable, data, group.variable,
  inf.type = c("observations", "groups"), plot = T, ...)

Arguments

model	A merMod object from lme4.
model.variable	Character. The name of the random effect variable to be computed.
data	An optional data frame used in model
group.variable	Character. Grouping variable for plots
inf.type	Character vector. Specifying which type of influential variables are returned
plot	Logical. Should a plot be produced?
...	Other arguments passed to ciFun

Examples

require(lme4)
m1 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)
inf1<-influenceFun(model=m1, model.variable= "Days", data=sleepstudy, group.variable="Subject")
inf1

d1<-cbpp
d1$period<-as.numeric(as.character(cbpp$period))
d1$response<-d1$incidence/d1$size
gm1 <- glmer(cbind(incidence, size - incidence) ~ 1 + (period | herd), data = d1, family = binomial)

inf1<-influenceFun(model=gm1, model.variable= "period", data=d1, group.variable="herd")
inf1

gm2 <- glmer(response ~ 1 + (period | herd), data = d1, weights=d1$size, family = binomial)
inf2<-influenceFun(model=gm2, model.variable= "period", data=d1, group.variable="herd")
inf2

p-schaefer/psUtilities documentation built on May 24, 2019, 5:56 p.m.