PG, Rick, Steve, Thank you all for your valuable inputs, when I get back to work tomorrow (I'm on Australia time zone!) I'll try the PGperm solution, I think it'll give me what I want. I'm designing an experiment so I want to have all 8 treatment groups starting with similar average Y (which is a variable representing process stability). Random allocation of groups did not give very similar group averages for Y, maybe my assumption of normally distributed Y was wrong. Regards, Fethon Naoum
... View more