Plotting mean and error (2sd) as multiple normal curves in R

Question

I have a series of values with a mean and a 2sd error:

structure(list(Site = 1:5, Value = c(0.54, 0.36, 0.13, 0.25, 
0.05), Error = c(0.26, 0.27, 0.25, 0.4, 0.24)), .Names = c("Site", 
"Value", "Error"), class = "data.frame", row.names = c(NA, -5L
))

I am trying to represent this a series of normal curves on one graph where the mid point of the curve is the mean and the range of the base of the curve is the mean+error/mean-error. The height of the curves can all be the same as we give each mean value the same weight.

I've had a search and I am really stuck. Sorry if I am missing somewhere where this may have been answered.

Answer 1

First you need to set up the plot but give 'plot' an NA to suppress any plotting. When you do that, plot requires ranges for X and Y

plot(NA, xlim=c( min(dat$Value)-max(dat$Error), 
                 max(dat$Value)+max(dat$Error) ),
         ylim=c(0,1) )
apply(dat, 1, function(x){ xx <-seq( x['Value']-x['Error'], 
                                    x['Value']+x['Error'], length=20);
                           yy=dnorm(xx, x['Value'], x['Error']/2);  sd is 1/2 'Error'
                           lines(xx,  yy/max(yy)) })  # normalize to peak == 1

If you want a smoother plot near the means, you can always increase the length of the 'xx' sequence.