Yes, but that is because the order you are thinking of is not a *logical* order.  Say you have: 1     abc 2     def 3      efg and 1     efg 2     abc What order should efg appear at?  What about abc, it is order 2 in the second example, but 1 in the first, the two merge but don't have the same order.  What I would suggest is that you think of the logical order for display, something you can put in a proc sort.  Alternatively have a look at the specifications, that should be your base, i.e. if the specs state: 1=abc, 2=def... then you can code that in and then merge the two sets of variables to that as you then have a predefined list.  If you haven't got specs, well... ... View more

Yes, that is a consideration.  To control this, you may wish to adjust the p values. Please look at The MULTTEST Procedure Example 61.5 Inputting Raw p-Values for an example of how to adjust for the comparisons you choose. Steve Denham ... View more

Dear LarryWorley, chriismtster,Hal.kuo, Thank you for a lot of responses. proc transpose is good and thank you for giving me goog advice. Since sdataset is big and code value is flexible, the answer using proc sql doesn't work... My specific task is to get the mean value of test data each site. data: valuecode  value sitecode 1                 1.11      3 2                  2.22     3 1                  1.1       4 2                  2.2       4 ... result dataset ( and then write to excel) mean1  mean2 ... mean10 1.11        2.22          2.34 The script I thought: %do I=1 %TO 10;    proc sql;       create table have&I as       select value&I from alldata where code = I;    data meandata;        merge meandata have&I;    run; %end; proc means data=meandata;    var value1-value10;    output=result mean(value1)=mean1 ...; run; It works good.But I thought it doesn't good script, so I want to know how sas masters think. Hide ... View more

Using a regular expression with prxchange() is for sure a very powerful way of modifying a string. Considering that it's such a simple text pattern and the question is asked in the "Getting started" forum using a function like TRANSTRN() is also acceptable. data _null_;   myVariable='name=j** ney';   myVariable=TRANSTRN(myVariable,'**','*');   put myVariable=; run; ... View more

Dear Michelle and Patrick, I appreciate your useful advice. Based on your advice, it'll be done in Base SAS. My environment is in SAS 9.1, no use PROC FCMP. If it'll have updated 9.2, i want to understand feature of PROC FCMP. And i will learn the macro with url which of michelle advices. Best Regard, ... View more

Look at your results more carefully and you will see that PROC SQL does not support MEDIAN as a summary statistic. You can use the SAS MEDIAN function in PROC SQL, but with one argument all you get is the trivial result that the median of a single number is that number. proc means median data=sashelp.class ; var age ; run ; proc sql ; select median(age) from sashelp.class ; quit ; Celko's SQL median has been implemented in SAS (http://www.sascommunity.org/wiki/Celko's_Median). That should get you started if you really have to do this in SQL. ... View more