Date:         Mon, 29 Jul 2002 09:10:05 -0400
Reply-To:     Quentin McMullen <QuentinMcMullen@WESTAT.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Quentin McMullen <QuentinMcMullen@WESTAT.COM>
Subject:      Re: Memo:  Proc Freq Question
Comments: To: "jassimrahma@HSBC.COM" <jassimrahma@HSBC.COM>
Content-Type: text/plain; charset="iso-8859-1"

Jassim Rahma [mailto:jassimrahma@HSBC.COM] writes:

> Why I'm not getting the format NO INFORMATION in my proc freq and just
> getting it as Frequency Missing. My Format is :
>
> proc format;
>      value $emp_score
>
>      'A' = 'Armed Forces/Ministry'
>      'G' = 'Government/Public Sector'
>      'M' = 'Manufacturing'
>      'O' = 'Others'
>      'P' = 'Private Sector'
>      'T' = 'Trading'
>      'W' = 'Wholesale'
>      'Q' = 'Not Asked'
>      other = 'No Information';
> run;
>
>
> and proc freq is :
>
> proc freq data = MEFCO.MEFCOmerge;
>      tables employment_status;
>      **-- var PRIM_SCD_FNL_SCR;
>      format employment_status $emp_score.;
>      **-- class PRIM_SCD_FNL_SCR Decision;
> run;

Hi,

I see Dennis Diskin already referred you to the missing option.

Another issue here might be that your category 'No Information' may contain
out-of-range values ('X','Y', etc.) as well as missing values (' ').  When
this happens, PROC FREQ will treat all values in the category as missing.
Cody & Pass _SAS Programming By Example_ has a good
demonstration/explanation of this on pages 206-207.  In general, I think it
is good practice that whenever you have other= as part of a format, you
should have separate categories for the missing data.  Below is an example.

Kind Regards,
--Quentin

193  data a;
194    do sex=1,2,3,.;
195      output;
196    end;
197  run;

198
199  proc format;
200    value badsexF
201      1='Male'
202      2='Female'
203      other='No Information'
204    ;
NOTE: Format BADSEXF has been output.
205  run;

206
207  proc freq data=a;
208    tables sex;
209    format sex badsexF.;
210  run;

NOTE: There were 4 observations read from the data set WORK.A.
NOTE: PROCEDURE FREQ used:
      real time           0.04 seconds

/***********
Freq shows two missing, but really there is one missing and one bad value.

                                Cumulative    Cumulative
sex    Frequency     Percent     Frequency      Percent
     -----------------------------------------------
Male         1       50.00             1        50.00
Female       1       50.00             2       100.00
                    Frequency Missing = 2
**********/

211
212  proc freq data=a;
213    tables sex/missing;
214    format sex badsexF.;
215  run;

NOTE: There were 4 observations read from the data set WORK.A.
NOTE: PROCEDURE FREQ used:
      real time           0.11 seconds

/*********
           sex    Frequency     Percent     Frequency      Percent
-------------------------------------------------------------------
No Information           2       50.00             2        50.00
Male                     1       25.00             3        75.00
Female                   1       25.00             4       100.00
**********/


216
217  proc format;
218    value oksexF
219      1='Male'
220      2='Female'
221      .='Missing'
222      other='Out of Range'
223    ;
NOTE: Format OKSEXF has been output.
224  run;

NOTE: PROCEDURE FORMAT used:
      real time           0.00 seconds


225
226  proc freq data=a;
227    tables sex;
228    format sex oksexF.;
229  run;

NOTE: There were 4 observations read from the data set WORK.A.
NOTE: PROCEDURE FREQ used:
      real time           0.00 seconds

/*************
                                         Cumulative    Cumulative
         sex    Frequency     Percent     Frequency      Percent
-----------------------------------------------------------------
Male                   1       33.33             1        33.33
Female                 1       33.33             2        66.67
Out of Range           1       33.33             3       100.00

                   Frequency Missing = 1
***********/