Here are the values in thsoe tow vars

[Tom]

Since GENDER_SP is length $2 it might be that the single digit codes are in the second character instead of the first. Or that the second character has some other non-graphic character other than a space.

Since it looks like the test you are making are non-overlapping it might make sense to do them as a series of ELSE clauses. Then you can include a catch all for any cases that do not fit.

DATA NovDec19.MERGED_NUMPARTS_1;
  SET NovDec19.MERGED_NovDec19_NUMPARTS;
  length mensex $3;
  if gender_sp ="1" AND NUM_PRT_3M="1" THEN MENSEX= "1";
  else if left(gender_sp) ="1" AND NUM_PRT_3M="2" THEN MENSEX="2";
  else if left(gender_sp)="1" AND NUM_PRT_3M="3" THEN MENSEX="3";
  else if left(gender_sp)="1" AND NUM_PRT_3M="5" THEN MENSEX="5";
  else if left(gender_sp) in ("3" , "9") AND NUM_PRT_3M in ("1" , "2" , "3" , "5") THEN MENSEX="999";
  else MENSEX='XXX';
run;

And most import make sure you are checking the NEW dataset and not the old one when check whether the re-code worked.

[Dhana18]

Alphabetic List of Variables and Attributes

Variable

Type

Len

Format

Informat

Label

NUM_PRT_3M

Char

1

gender_sp

Char

2

gender_sp

[PaigeMiller]

As requested earlier by @ballardw , you need to provide us with a portion of this data set, using the instructions in the link he provided. We do not want screen captures, we do not want Excel files or attachments, we need to see a portion of this data in that format and not any other format.

[ballardw]

---

@Dhana18 wrote:

gender_sp	NUM_PRT_3M	mensex	Frequency
1	0	999	128
1	1	999	128
1	2	999	70
1	3	999	8
1	5	999	16
2	0	999	193
2	1	999	144
2	2	999	107
2	3	999	57
2	5	999	40
3	0	999	17
3	1	999	4
3	2	999	13
3	3	999	11
3	5	999	6
9	0	999	44
9	1	999	3
9	2	999	3
9	5	999	3

---

One thing that we learn eventually about SAS is that it tries to help us make "nice" output. Sometimes that interferes with diagnosing problems when using output from procedures to examine values. One of things that happens is that most procedures will strip leading spaces from values. So it is possible that the 1, 2, 3 and 9 above are actually " 1", " 2", " 3" and " 9" for the gender_sp variable so comparisons of equality such as ="1" fail.

Here is a brief example you can run. Note that the row heading for Y in the second table now is indented because the display Style instruction says to use the leading space:

data example;
   x='2';
   y=' 2';
run;

proc tabulate data=example;
title "Default display removes leading blanks";
   class x y;
   table x y , n;

run;title;
proc tabulate data=example;
title "ASIS option shows leading blanks";
   class x y;
   classlev y / style=[Asis=on];
   table x y , n;

run;title;

Which is one of the reasons that we ask about actual data an not proc output.

[ballardw]

---

@Dhana18 wrote:

gender_sp	NUM_PRT_3M	mensex	Frequency
1	0	999	128
1	1	999	128
1	2	999	70
1	3	999	8
1	5	999	16
2	0	999	193
2	1	999	144
2	2	999	107
2	3	999	57
2	5	999	40
3	0	999	17
3	1	999	4
3	2	999	13
3	3	999	11
3	5	999	6
9	0	999	44
9	1	999	3
9	2	999	3
9	5	999	3

---

You are showing values of gender_sp and Num_prt_3m not in the code with results.

So I have to assume that you have some other code affecting your MENSEX variable.

Here is an example data set that just uses character values of those shown in your freq output above:

data example;
   length gender_sp $ 1 Num_prt_3m $ 2 mensex $ 3;
   do gender_sp='1','2','3','9';
      do Num_prt_3m = '0','1','2','3','5';
         if gender_sp ="1" AND NUM_PRT_3M="1" THEN MENSEX="1";
         if gender_sp ="1" AND NUM_PRT_3M="2" THEN MENSEX="2";
         if gender_sp ="1" AND NUM_PRT_3M="3" THEN MENSEX="3";
         if gender_sp ="1" AND NUM_PRT_3M="5" THEN MENSEX="5";
         if gender_sp in("3" , "9") AND NUM_PRT_3M in("1" "2" "3" "5") THEN MENSEX="999";
         output;
         /* reset result before next loop iteration*/
         call missing(mensex);
      end;
   end;
run;

Note that we get every possible combination of the one character length values of the two input variables.

And the result of:

proc freq data=example;
  tables gender_sp*Num_prt_3m*mensex/missing list nopercent nocum;
run;

is

The FREQ Procedure

gender_sp	Num_prt_3m	mensex	Frequency
1	0		1
1	1	1	1
1	2	2	1
1	3	3	1
1	5	5	1
2	0		1
2	1		1
2	2		1
2	3		1
2	5		1
3	0		1
3	1	999	1
3	2	999	1
3	3	999	1
3	5	999	1
9	0		1
9	1	999	1
9	2	999	1
9	3	999	1
9	5	999	1

Which does not have some of the output values you show for the same input. So either 1) your input is different than implied by the Proc Freq output and/or 2) your code as executed includes something you have not shown.

Such as a value set in a previous data step, read from the source file, or manipulated after assignment in this data step.

I suggest that you carefully search your code for every instance of MENSEX and see if you have another assignment or possibly and array using that variable.

I had removed the commas in the last comparison just incase your pasted code had contained a non-visible character that was causing problems.