This is what my comment below was about.  If you want it to add to 100, then you have to take it out an extra decimal place or two (sometimes even three or four).  When rounding, your decimal place will either go up or stay the same. For example, 29.8% + 70.2% = 100.0%.  Yea!  This is what we want! If we round to whole integers, the 29 will change to a 30, but the 70 will remain the same.  Thus, 30% + 70% = 100%. Yea!  We still got what we want! Now, let's try a different example where both numbers will go up. 29.5% + 70.5% = 100.0%.  Yea!  This is again what we want! Now, let's round to whole integers again.  Now the 29 becomes a 30, and the 70 becomes a 71.  So, 30% + 71% = 101%.  Oh no!  This is not true!  (Was that your thought?)  It is still true because you now estimated instead showing an exact result. Maybe estimation is where your question arises.  When you estimate, you will get close, but not always exact.  Rounding is a way of estimating.  No one wants to see a long stretch of decimal places, so we round to make the display of values look prettier, and we understand that there will be a little variation. So, are you okay with it not showing up as exactly 100%, or do you have people viewing your results that do not understand rounding? ... View more

I don't think your question should be an issue.  When rounding, you will not always add up to 100%.  If you truly want to get 100% by adding up your displayed values, you would have to take it out an extra decimal place or two.  But most people would recognize that there is a rounding situation going on there.  No one will add it up and say, "Wait a second!  These values only account for 99.9% of the data.  Where is the other 0.1%?" This reminds me of the Modern Family quote: "Success is 1% inspiration, 98% perspiration, and 2% attention to detail." ... View more

I just noticed that you tagged this post for help in SAS Enterprise Guide.  I have never used EG, so I don't know that what I just told you will be of much help.  Hopefully the idea can somehow be incorporated, though. ... View more

SAS seems to only put columns in the order that they are created, not numeric or alphabetical order.  If you want to specify the order of the columns, I have used one of two ways: 1.  Use PROC SQL and use the SELECT statement to list out the columns that you want in the order that you want. 2.  Use a DATA step and use the LENGTH statement to specify lengths of the columns in the order that you want (more work and less efficient). This, however, does not seem too enjoyable if you have a lot of columns.  In that case, you could probably try something like: proc sort data=sashelp.vcolumn (where=(libname='AAA' & memname='BBB')) out=sorted;       by name; run; proc sql;       select name into :columns separated by ', '       from sorted       where name not in ('','');  /* fill in the blank single quotes with the names of variables that you are not interested in sorting */       create table sortedColumns as       select &columns.       from AAA.BBB; quit; In the above code snippet, AAA is the library containing your transposed dataset, and BBB is the name of the transposed dataset.  I just created a macro variable that stores the names of the variables that you want to sort (separated by a comma for use in PROC SQL).  Then I created a new table that keeps those columns in that sorted order. I hope this helps! ... View more