Pareto Plot of Fractional Factorial Effects

DESCRIPTION:

Creates a Pareto plot (vertical bar plot) of effects from a fractional factorial design.

USAGE:

pareto(x, method=<<see below>>, sig=.05, xmax=<<see below>>, 
          prt.values=T, shade=F, ...) 

REQUIRED ARGUMENTS:

x
a fac.aov object, typically created by fac.aov.

OPTIONAL ARGUMENTS:

method
character string specifying the method used to estimate the standard error of the effects for drawing significance lines on the plot. Possibilities are:
`"all"'

all lines are drawn;

`"mse"'
mean squared error (often from a reduced model, see `update.);
`"pse"'
pseudo standard error estimate;
`"tse"'
60% trimmed standard error of the effects;
`"ase"'
adaptive standard error estimate;
`"none"'
no lines drawn.

The default is "pse" if df.residual = 0, otherwise, if "mse" is available, both "mse" and "pse" are used.
sig
value of significance level for significance lines.
xmax
upper value of graphical parameter xlim.
prt.values
a logical flag for whether values are printed at end of bars.
shade
logical flag for whether bars will be shaded or filled with solid colors.
...
arguments passed to barplot.

SIDE EFFECTS:

A Pareto plot is produced on the current graphics device.

DETAILS:

If fac.aov or update.fac.aov are used to create an unsaturated model, then the Pareto plot will show the seffects (from the saturated model) rather than the feffects. The significance line for the pse method will also be based on the seffects.

REFERENCES:

Haaland, P. D. (1989), Experimental Design in Biotechnology, New York: Marcel Dekker.

Haaland, P. D. and M. A. O'Connell (1994), Inference for effect saturated fractional factorials, to appear in Technometrics.

SEE ALSO:

, , , , .

EXAMPLES:

buffer.fac <- fac.aov(buffer.df) 
pareto(buffer.fac)                # use all defaults 
pareto(buffer.fac, sig=.1)         # use .1 significance level to plot lines 
pareto(buffer.fac, method="none")  # do not plot significance lines 
pareto(buffer.fac, "all")          # plot all possible significance lines 
pareto(buffer.fac, "ase")          # plot only "ase" significance line 
# show what happens with an unsaturated model 
buffer.fac1 <- update(buffer.fac,~pH*thimer+pH*gent) 
pareto(buffer.fac1) 
buffer.fac2 <- fac.aov(rate~pH*thimer+pH*gent,buffer.df) 
pareto(buffer.fac2)