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 lines are drawn;
  • mean squared error (often from a reduced model, see update.);
  • pseudo standard error estimate;
  • 60% trimmed standard error of the effects;
  • adaptive standard error estimate;
  • 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:
fac.aov , qqnorm.fac.aov , update.fac.aov , tfiplot.aov, design.digest .
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)