Multiple comparison plot.

For grouped data. Draws boxplots for each group and overlays with confidence intervals for pairwise comparison of means.

Usage

mc_plot(data, fit, path = eulerian, col = rainbow(length(data), s = 0.4), 
levels = c(0.9, 0.95, 0.99), varwidth = TRUE, frame.plot = FALSE, 
boxwex = 0.3, cex=0.75, zoom=NULL,ci.yusr=NULL,ci.pos=FALSE,...)

Arguments

data

A list of vectors, such as that returned by split.

fit

Either an aov fit, or else a matrix with columns estimate, followed by confidence intervals. If fit is not an aov fit, the path argument should be a vector.

path

an index vector or a function. If a vector, groups are plotted in order data[path]. By default, it is the function eulerian, and produces an ordering where each pair of groups appears adjacently, with p-values roughly increasing as the sequence progresses.

col

A vector of colours, one per group.

levels

Vector of increasing confidence levels.

varwidth

Passed to boxplot.

frame.plot

Passed to boxplot.

boxwex

Passed to boxplot.

cex

Passed to boxplot.

zoom

If provided, a numeric vector specifying a subsequence of path to display.

ci.yusr

Specifies the vertical par(usr) for the confidence intervals. Defaults to max and min.

ci.pos

If TRUE, all CIs are mu(max) - mu(min), otherwise mu(right) - mu(left).

...

Optional arguments, passed to boxplot and overlayCI.

References

see overview

Author

C.B. Hurley and R.W. Oldford

See also

See also overlayCI

Examples

         
require(PairViz)       

data(cancer)
bx <- with(cancer, split(sqrt(Survival),Organ))



a <-  aov(sqrt(Survival) ~ Organ,data=cancer)
if (FALSE) {
dev.new(height=4.5, width=9.5)
op <- par(no.readonly = TRUE)

par(cex.axis=.75, cex.main = 1.0, cex.lab=1)
par(mar=c(3,5,3,5))

mc_plot(bx,a,main="Pairwise comparisons of cancer types", ylab="Sqrt Survival")

par(op)
}