Howard_Schreier@ITA.DOC.GOV (Howard Schreier) wrote in message news:<200409051947.i85JlhK4003124@listserv.cc.uga.edu>... > Do you have SAS/IML (Interactive Matrix Language) licensed? It provides the > most straightforward solution. Quoting from the manual: "The division > operator (/) divides each element of matrix1 by the corresponding element > of matrix2, producing a matrix of quotients." I don't think so, my sas version is 8.2 student, so unless it's standard... > > The same thing can be accomplished using either a DATA step or PROC SQL (never used it), > but it may be awkward. That depends largely on how many variables 49 there are > and how they are named. Are the variable names in A and B the same? Are the > names formed with a common stem and numeric suffixes, like VAR1, VAR2, ... > VARp? Var in A and B are the same, individuals are the same just the values change since they have been measured under another condition. They don't really have a common name stem, but that could be changed .

> Depending on what you have to do after calculating the ratios, mainly plots or frequencies distribution charts you may want > to start by reshaping your datasets so that each has n x p observations and > three variables (N_KEY, P_KEY, and VALUE). That would make computing the > ratios a snap in either a DATA step or PROC SQL.