My frequency table is not showing all variable categories

[Banke]

Hello all,

i want to create a frequencyt table for my newly categorized variables. there is no error in log but when i run the proc freq , it doesnt show all the categories i created for the first variable which is race. it showed just 2 instead of 5

here is my syntax

DATA MEPSIS.RACEDU;
SET MEPSIS.TEMP5;

IF RACE = 1 THEN RACECAT = 'HISPANIC';
IF RACE = 2 THEN RACECAT = 'NON-HISPANIC WHITE ONLY';
IF RACE = 3 THEN RACECAT = 'NON-HISPANIC BLACK ONLY';
IF RACE = 4 THEN RACECAT = 'NON-HISPANIC ASIAN ONLY';
IF RACE = 5 THEN RACECAT = 'NON-HISPANIC OTHER RACE OR MULTI-RACE';
IF RACE < 1 THEN DELETE;


IF EDUCATION = 1 THEN EDUCAT = '=<8TH GRADE';
IF EDUCATION = 2 THEN EDUCAT = '9TH TO 12TH GRADE';
IF EDUCATION = 13 THEN EDUCAT = 'GED OR HIGH SCHOOL DIPLOMA';
IF EDUCATION = 14 THEN EDUCAT = '>HIGH SCHOOL BUT WITHOUT 4 YEAR DEGREE';
IF EDUCATION = 15 THEN EDUCAT = 'COLLEGE OR BACHELORS DEGREE';
IF EDUCATION = 16 THEN EDUCAT = 'GRADUATE OR PROFESSIONAL DEGREE';
IF EDUCATION < 1 THEN DELETE;

RUN;


/** PRINTING TO CHECK**/
PROC PRINT DATA = MEPSIS.RACEDU (OBS = 5);
VAR RACECAT EDUCAT;
TITLE "CROSSCHECKING QUESTION 4";
RUN;

[Shmuel]

Is it possible that one of the two conditions is true:

- IF RACE < 1 THEN DELETE;

- IF EDUCATION < 1 THEN DELETE;

Check the missing observations, maybe they satisfies above situation.

I propose to replace the DELETE with '***';

[Astounding]

These statements define RACECAT as 8 characters long, the number of characters in "HISPANIC". Other values get truncated. To fix this, add just after the SET statement:

LENGTH RACECAT $37;

[Reeza]

1. DATA MEPSIS.RACEDU;
   SET MEPSIS.TEMP5;
   
   length RACECAT EDUCAT $50.;
   
   IF RACE = 1 THEN RACECAT = 'HISPANIC';
   ELSE IF RACE = 2 THEN RACECAT = 'NON-HISPANIC WHITE ONLY';
   ELSE IF RACE = 3 THEN RACECAT = 'NON-HISPANIC BLACK ONLY';
   ELSE IF RACE = 4 THEN RACECAT = 'NON-HISPANIC ASIAN ONLY';
   ELSE IF RACE = 5 THEN RACECAT = 'NON-HISPANIC OTHER RACE OR MULTI-RACE';
   *ELSE IF RACE < 1 THEN DELETE;
   ELSE RACECAT="CHECKME";
   
   
   IF EDUCATION = 1 THEN EDUCAT = '=<8TH GRADE';
   ELSE IF EDUCATION = 2 THEN EDUCAT = '9TH TO 12TH GRADE';
   ELSE IF EDUCATION = 13 THEN EDUCAT = 'GED OR HIGH SCHOOL DIPLOMA';
   ELSE IF EDUCATION = 14 THEN EDUCAT = '>HIGH SCHOOL BUT WITHOUT 4 YEAR DEGREE';
   ELSE IF EDUCATION = 15 THEN EDUCAT = 'COLLEGE OR BACHELORS DEGREE';
   ELSE IF EDUCATION = 16 THEN EDUCAT = 'GRADUATE OR PROFESSIONAL DEGREE';
   ELSE IF EDUCATION < 1 THEN DELETE;
   ELSE EDUCAT = "CHECKME";
   
   RUN;
   
   
   
   proc freq data=mepsis.racedu;
   table race*racecat education*educat racecat*educat;
   run;

2. You only printed 5 observations, how do you know that all categories weren't created?
3. You say PROC FREQ but are showing a PROC PRINT, did you forget to include some of your code? You should check recoding with PROC FREQ not PROC PRINT. 
4. I would recommend coding the categories to something else, instead of deleting, checking your output and then once you were satisfied, deleting the records. Delete only after your selection is verified. 

---

@Banke wrote:

Hello all,

i want to create a frequencyt table for my newly categorized variables. there is no error in log but when i run the proc freq , it doesnt show all the categories i created for the first variable which is race. it showed just 2 instead of 5

here is my syntax

DATA MEPSIS.RACEDU;
SET MEPSIS.TEMP5;

IF RACE = 1 THEN RACECAT = 'HISPANIC';
IF RACE = 2 THEN RACECAT = 'NON-HISPANIC WHITE ONLY';
IF RACE = 3 THEN RACECAT = 'NON-HISPANIC BLACK ONLY';
IF RACE = 4 THEN RACECAT = 'NON-HISPANIC ASIAN ONLY';
IF RACE = 5 THEN RACECAT = 'NON-HISPANIC OTHER RACE OR MULTI-RACE';
IF RACE < 1 THEN DELETE;


IF EDUCATION = 1 THEN EDUCAT = '=<8TH GRADE';
IF EDUCATION = 2 THEN EDUCAT = '9TH TO 12TH GRADE';
IF EDUCATION = 13 THEN EDUCAT = 'GED OR HIGH SCHOOL DIPLOMA';
IF EDUCATION = 14 THEN EDUCAT = '>HIGH SCHOOL BUT WITHOUT 4 YEAR DEGREE';
IF EDUCATION = 15 THEN EDUCAT = 'COLLEGE OR BACHELORS DEGREE';
IF EDUCATION = 16 THEN EDUCAT = 'GRADUATE OR PROFESSIONAL DEGREE';
IF EDUCATION < 1 THEN DELETE;

RUN;


/** PRINTING TO CHECK**/
PROC PRINT DATA = MEPSIS.RACEDU (OBS = 5);
VAR RACECAT EDUCAT;
TITLE "CROSSCHECKING QUESTION 4";
RUN;

---

[Banke]

ok. thank you. i uploaded the wrong picture, sorry about that. It actually created all the categories i wanted, i scanned through the dataset. i just observed that it grouped the categories into 2 instead of 5 in the frequency table whereas there are 5 categories in the created dataset. what does the "CHECKME" mean please?

[tarheel13]

She recoded instead of deleting. Also I think you should use if else if instead of just a bunch of if statements.

[Reeza]

---

@Banke wrote:

ok. thank you. i uploaded the wrong picture, sorry about that. It actually created all the categories i wanted, i scanned through the dataset. i just observed that it grouped the categories into 2 instead of 5 in the frequency table whereas there are 5 categories in the created dataset. what does the "CHECKME" mean please?

---

It literally means go check those records. That is because they're falling out of your IF/ELSE conditions which means you're not accounting for some conditions or deletions. Coding it to a value rather than delete allows you to verify the logic before you delete it. This style of programming helps you to find errors in your code before they propogate.

[tarheel13]

The proper way to check this variable derivation is like this:

proc freq data=mepsis.racedu;
tables race*racecat / list missing;
run;