R/plot.discrepancy.R

In PoweR: Computation of Power and Level Tables for Hypothesis Tests

plot.discrepancy <- function(x, legend.pos = NULL, ...) {
    
    Fx <- x
	
    if(!inherits(Fx, "Fx")) stop("Method is only for Fx objects!")

	## Retrieve informations from Fx computed by calcFx()
    Fxi <- Fx$Fx.mat
    xi <- Fx$x
    law <- Fx$law
    statnames <- Fx$statnames
    N <- Fx$N

	## Now we plot these (xi,Fxi-xi) on the same graph

	# we trace the 45 degree line
    plot(c(0, 1), c(min(Fxi - xi), max(Fxi - xi)), type = "n", main = paste("P value discrepancy plots", "\n N = ", N, " & m = ", length(xi), sep = ""),
         sub = paste("Data: ", law, sep = ""), xlab = "Nominal Size", ylab = "Size Discrepancy", ...)
    lines(x = c(0, 1), y = c(0, 0), lty = "dotted", col = 1)
	
	# then add theses P value discrepancy plots
    for (i in 1:length(statnames)) {
        points(xi, Fxi[, i] - xi, type = "l", col = i + 1, ...)
    }
	
	# legend
    if (is.character(legend.pos)) legend(x = legend.pos, legend = statnames, col = 2:(length(statnames) + 1), lty = 1)
    if (is.null(legend.pos)) {
        legend.x <- 0.7
        legend.y <- 0.9 * max(Fxi - xi)
        legend(x = legend.x, y = legend.y, legend = statnames, col = 2:(length(statnames) + 1), lty = 1)
    } else {
        legend(x = legend.pos[1], y = legend.pos[2], legend = statnames, col = 2:(length(statnames) + 1), lty = 1)
    }
    
}