As far as I know, either way works and should give the same answer (if done properly). Easy enough for you to test. ... View more

Lack-of-fit/pure error can only be computed if you have one or more exact replicates of your points (among the x-variables). You have zero exact replicates of your data points. So SAS is giving the correct output. ... View more

PROC SGSCATTER will create such a scatterplot with 3 x-variables and one-y variable. An example is given here: https://documentation.sas.com/?docsetId=grstatproc&docsetVersion=9.4&docsetTarget=p0yb5lnzixgxoxn186t2rff7fg9x.htm&locale=en ... View more

@Kurt_Bremser wrote: @Reeza wrote: It's an bit of a long process, as in how software was installed 20 years ago, Actually, with SAS that was a breeze. Insert the tape cartridge, run the tar command from the letter you got with the tape, wait for it to finish, and you were done. 5 minutes of work and 6 hours of waiting for the tape. Maybe another commandline for the setinit. No wizards with deployment plans, multiple tiers, multiple levels or the like. Just SAS.   😉 Unless you were stuck with Windows and two stack-feet of 3.5in floppy disks which required intervention every 3 to 5 minutes... ... View more