Mar 28, 2008

Statistics question

If x follows a Chi square with k degrees of freedom and y follows a Chi square with m degrees of freedom, can we say anything about the distributionf of x-y???

2 comments:

sergio said...

a chi square is a sum of standard normals, therefore a xhi square(m) is a sum of m standard mormals. If you subtract the sum of n other standard normals you'll be left with m-n standard normals. Hence the distribution will be a chi square(m-n). However you have to check that the standard normals included in the first chi square are independent from those included in the other chi square

Pierre-Louis said...

you confuse me...a chi square is a product of normals, not a sum??? no?? what the (&)(%)&((?