interaction.plot
Two-way Interaction Plot
Plots the mean (or other summary) of the response for two-way combinations of factors, thereby illustrating possible interactions.
- Keywords
- hplot
Usage
interaction.plot(x.factor, trace.factor, response, fun = mean, type = c("l", "p", "b", "o", "c"), legend = TRUE, trace.label = deparse(substitute(trace.factor)), fixed = FALSE, xlab = deparse(substitute(x.factor)), ylab = ylabel, ylim = range(cells, na.rm = TRUE), lty = nc:1, col = 1, pch = c(1:9, 0, letters), xpd = NULL, leg.bg = par("bg"), leg.bty = "n", xtick = FALSE, xaxt = par("xaxt"), axes = TRUE, ...)Arguments
- x.factor
- a factor whose levels will form the x axis.
- trace.factor
- another factor whose levels will form the traces.
- response
- a numeric variable giving the response
- fun
- the function to compute the summary. Should return a single real value.
- type
- the type of plot (see
plot.default): lines or points or both. - legend
- logical. Should a legend be included?
- trace.label
- overall label for the legend.
- fixed
- logical. Should the legend be in the order of the levels
of
trace.factoror in the order of the traces at their right-hand ends? - xlab,ylab
- the x and y label of the plot each with a sensible default.
- ylim
- numeric of length 2 giving the y limits for the plot.
- lty
- line type for the lines drawn, with sensible default.
- col
- the color to be used for plotting.
- pch
- a vector of plotting symbols or characters, with sensible default.
- xpd
- determines clipping behaviour for the
legendused, seepar(xpd). Per default, the legend is not clipped at the figure border. - leg.bg, leg.bty
- arguments passed to
legend(). - xtick
- logical. Should tick marks be used on the x axis?
- xaxt, axes, ...
- graphics parameters to be passed to the plotting routines.
Details
By default the levels of x.factor are plotted on the x axis in
their given order, with extra space left at the right for the legend
(if specified). If x.factor is an ordered factor and the levels
are numeric, these numeric values are used for the x axis.
The response and hence its summary can contain missing values. If so, the missing values and the line segments joining them are omitted from the plot (and this can be somewhat disconcerting).
The graphics parameters xlab, ylab, ylim,
lty, col and pch are given suitable defaults
(and xlim and xaxs are set and cannot be overridden).
The defaults are to cycle through the line types, use the foreground
colour, and to use the symbols 1:9, 0, and the capital letters to plot
the traces.
Note
Some of the argument names and the precise behaviour are chosen for S-compatibility.
References
Chambers, J. M., Freeny, A and Heiberger, R. M. (1992) Analysis of variance; designed experiments. Chapter 5 of Statistical Models in S eds J. M. Chambers and T. J. Hastie, Wadsworth & Brooks/Cole.
Examples
library(stats)
require(graphics)
with(ToothGrowth, {
interaction.plot(dose, supp, len, fixed = TRUE)
dose <- ordered(dose)
interaction.plot(dose, supp, len, fixed = TRUE, col = 2:3, leg.bty = "o")
interaction.plot(dose, supp, len, fixed = TRUE, col = 2:3, type = "p")
})
with(OrchardSprays, {
interaction.plot(treatment, rowpos, decrease)
interaction.plot(rowpos, treatment, decrease, cex.axis = 0.8)
## order the rows by their mean effect
rowpos <- factor(rowpos,
levels = sort.list(tapply(decrease, rowpos, mean)))
interaction.plot(rowpos, treatment, decrease, col = 2:9, lty = 1)
})
with(esoph, {
interaction.plot(agegp, alcgp, ncases/ncontrols, main = "'esoph' Data")
interaction.plot(agegp, tobgp, ncases/ncontrols, trace.label = "tobacco",
fixed = TRUE, xaxt = "n")
})
## deal with NAs:
esoph[66,] # second to last age group: 65-74
esophNA <- esoph; esophNA$ncases[66] <- NA
with(esophNA, {
interaction.plot(agegp, alcgp, ncases/ncontrols, col = 2:5)
# doesn't show *last* group either
interaction.plot(agegp, alcgp, ncases/ncontrols, col = 2:5, type = "b")
## alternative take non-NA's {"cheating"}
interaction.plot(agegp, alcgp, ncases/ncontrols, col = 2:5,
fun = function(x) mean(x, na.rm = TRUE),
sub = "function(x) mean(x, na.rm=TRUE)")
})
rm(esophNA) # to clear up