comparing bootstrapped confidence intervals

["Data Analytics Corp." <dataanalytics@earthlink.net>]

Hi,

I have a question about comparing bootstrapped confidence intervals. I 
have a slightly complicated function to calculate a statistic called a 
penalty often used in the food sensory area.  Basically, it's just the 
difference between the middle point on a 5-point liking scale and the 
weighted average of the other points.  This is done for all attributes 
of a product.   Let's say there are five attributes so there are 5 
penalties.  I want to bootstrap a 95% confidence interval for each 
penalty (easy to do using the bootstrap and limits.bca functions) and 
then plot the intervals on one graph similar to what multicomp does. The 
goal is to tell the client which penalties are different.  Is this 
legitimate to do with the bootstrapped intervals?  Any opinions?
Thanks,

Walt Paczkowski