Proc Tabulate - Order of Class Variable

[eer_seer]

All,

I have a class variable: time_to_graduation

Time_to_graduation has the following possibilities:

'3-Year and Under Graduate'
'4-Year Graduate'
'5-Year Graduate'
'6-Year Graduate'
'7-Year Graduate'
'8-Year Graduate'
'9-Year Graduate'
'10-Year Graduate'

They are not tabulating in order I have listed. I numbered them intuitively hoping I wouldn't have to do any type of format step. Those hopes, however, were dashed when I tabulated the variable.  Let's just say, they are not in order.  I've attempted various fixes, but I can't seem to get it to work.

Any advice on getting the tabulation to display the time_to_graduate class variable to tabulate in the order i have above?

Thanks!

[Reeza]

Replace them with codes, 1, 2, 3 etc in the correct order and then use a format to have the value displayed instead.

---

@eer_seer wrote:

All,

I have a class variable: time_to_graduation

Time_to_graduation has the following possibilities:

'3-Year and Under Graduate'
'4-Year Graduate'
'5-Year Graduate'
'6-Year Graduate'
'7-Year Graduate'
'8-Year Graduate'
'9-Year Graduate'
'10-Year Graduate'

They are not tabulating in order I have listed. I numbered them intuitively hoping I wouldn't have to do any type of format step. Those hopes, however, were dashed when I tabulated the variable.  Let's just say, they are not in order.  I've attempted various fixes, but I can't seem to get it to work.

Any advice on getting the tabulation to display the time_to_graduate class variable to tabulate in the order i have above?

Thanks!

---

[eer_seer]

I follow on the first part of your suggestion, don't follow on the second.

[Astounding · Accepted solution]

As a character string, the values are alphabetized.  Since "1" alphabetizes before other digits, "10" will come out first.  A simple fix would be to add a blank before all the single-digit values:

' 3-Year and Under Graduate'

[eer_seer]

This was a great and simple solution to the problem and worked just as you described.  Thank you.

[BrunoMueller]

Check out the CLASSDATA= option it allows you to provide the data in a specific sequence. Additionally it provides a way to provide class values for which there is no data.

Here is an example:

data have;
  infile cards dsd;
  input 
    gradText : $30.
  ;
  seqNr + 1;
  random = rand("uniform");
  someValue = ceil( random * 100 );
cards;
'3-Year and Under Graduate'
'4-Year Graduate'
'5-Year Graduate'
'6-Year Graduate'
'7-Year Graduate'
'8-Year Graduate'
'9-Year Graduate'
'10-Year Graduate'
;

data classOrder;
  set have;
  keep gradtext;
run;

proc sort data=have out=have2;
  by random;
run;

proc tabulate data=have2 classdata=classorder;
  class gradText / order=data;
  var someValue;
  table 
    gradText, someValue
  ;
run;

[eer_seer]

Ok, great!

This is helpful as well.