topic Re: How to include the missing value? in Statistical Procedures

How to include the missing value?

ybz12003 — Tue, 06 Dec 2022 21:51:18 GMT

Hi all,

Please let me know how to proceed if I want to include the missing value in the calculation variables (ED ICU Regular Physician Lab Radiology Pharma Thera).

proc tabulate data=have out=want;
  var ED ICU Regular Physician Lab Radiology Pharma Thera;
  class PT_class;
  table (ED ICU Regular Physician Lab Radiology Pharma Thera), (PT_class all)*((N mean median p25 p75 sum));
  format PT_class PT_class.; 	
run;

Re: How to include the missing value?

Quentin — Tue, 06 Dec 2022 22:09:44 GMT

Do you mean missing values of PT_CLASS?

You can add the MISSING option to the PROC TABULATE statement.

Re: How to include the missing value?

ballardw — Tue, 06 Dec 2022 23:22:25 GMT

How do you want the "missing values" for the Var variables included? You can get a count of the missing values using the NMISS statistic.

proc tabulate data=have out=want; var ED ICU Regular Physician Lab Radiology Pharma Thera; class PT_class; table (ED ICU Regular Physician Lab Radiology Pharma Thera), (PT_class all)*((N mean median p25 p75 sum nmiss)); format PT_class PT_class.; run;

Re: How to include the missing value?

ybz12003 — Wed, 07 Dec 2022 01:21:20 GMT

I have missing values in all the variables to be calculated. To get the average of them, I would like to include the count of the missing values in the denominator. Maybe I should use another step to do this separately cause the code is used to count all the non-missing values. Right?

Re: How to include the missing value?

webart999ARM — Wed, 07 Dec 2022 02:10:49 GMT

To include missing values in the calculation variables in proc tabulate, you can use the MISSING option in the VAR statement. This will tell proc tabulate to include missing values in the calculations for the specified variables. Here is an example of how to do this:

proc tabulate data=have out=want;
  var ED ICU Regular Physician Lab Radiology Pharma Thera / missing;
  class PT_class;
  table (ED ICU Regular Physician Lab Radiology Pharma Thera), (PT_class all)*((N mean median p25 p75 sum));
  format PT_class PT_class.; 	
run;

Note that using the MISSING option will include missing values in the calculations for all of the specified variables, not just for individual variables.

Re: How to include the missing value?

Quentin — Wed, 07 Dec 2022 13:14:36 GMT

@ybz12003 wrote:

I have missing values in all the variables to be calculated. To get the average of them, I would like to include the count of the missing values in the denominator. Maybe I should use another step to do this separately cause the code is used to count all the non-missing values. Right?

Do you mean you want to treat missing values as 0? So this will lower all your statistics? If that's what you want, I would probably use a data step to replace missing values with 0. But it would be an unusual approach to treat missing values as 0.

Re: How to include the missing value?

ybz12003 — Wed, 07 Dec 2022 17:21:51 GMT

Yes, I plan to treat the missing as 0.

Re: How to include the missing value?

Quentin — Wed, 07 Dec 2022 20:38:18 GMT

@webart999ARM @ybz12003

I don't see a /missing option for the VAR statement in the documentation. And when I run in 9.4, I get an error:

1
2    data class ;
3      set sashelp.class ;
4      if _n_=1 then age=. ;
5    run ;

NOTE: There were 19 observations read from the data set SASHELP.CLASS.
NOTE: The data set WORK.CLASS has 19 observations and 5 variables.
NOTE: DATA statement used (Total process time):
      real time           0.01 seconds
      user cpu time       0.00 seconds
      system cpu time     0.00 seconds
      memory              559.21k
      OS Memory           16624.00k
      Timestamp           12/07/2022 03:35:00 PM
      Step Count                        13  Switch Count  0


6
7    proc tabulate data=class ;
8      class sex ;
9      var age /missing;
                -------
                22
                202
ERROR 22-322: Syntax error, expecting one of the following: ;, STYLE, WEIGHT, WGT.
ERROR 202-322: The option or parameter is not recognized and will be ignored.
10     tables sex,age*(n nmiss mean median) ;
11   run ;

Is that accepted answer working? In what version of SAS does it work?