Skip to content

p-value calculation on a joint distribution

2 messages · Leon Yee, Greg Snow

#
Dear R users,

     For a uni-variable distribution represented in a numerical vector, 
we can obtain a distribution function using 'ecdf', and then calculate 
corresponding p-values. But if I have a 2-column dataframe representing 
a bi-variable joint distribution, given a pair of values, how can I get 
the p-value? And how can I plot out the density of the joint distribution?

Best wishes,

Leon
5 days later
#
I don't think that the p-value concept is as well defined for multivariate distributions.  Do you want the area under the curve corresponding to (x < t.x & y < t.y) or (x < t.x | y < t.y) or ( t.x + t.y < C ) or all the area where the height of the density is less than at t.x,t.y?  or possibly others

Do you have the definition of the density? Or is the data frame a representation of the heights of the distribution at given x and y coordinates? If the second, do they x and y coordinates form a grid? Or some random pattern? Etc.?