IF-THEN-ELSE | Multiple Conditions

[_maldini_]

I am trying to divide survey respondents into 4 buckets using IF-THEN-ELSE. I'm not having success.

	IF tobacco_pastmo = 0 AND ETOH_pastmo = 0 THEN user_risk = 0;
	ELSE IF tobacco_pastmo = 1 AND ETOH_pastmo = 0 THEN user_risk = 1;
	ELSE IF tobacco_pastmo = 0 AND ETOH_pastmo = 1 THEN user_risk = 2;
	ELSE IF tobacco_pastmo = 1 AND ETOH_pastmo= 1 THEN user_risk = 3;

	/* 	0 = Current MJ only */
	/*	1 = Current MJ + Tobacco user */
	/* 	2 = Current MJ + ETOH user */
	/* 	3 = Current MJ + Tobacco + ETOH user */

A sample dataset and a screen shot of the results using the syntax above are attached.

Thanks in advance!

[Reeza · Accepted solution]

Your data shows values of 1/2 but your code is checking for 0/1 values.

Try a PROC FREQ maybe to verify your results?

Proc freq data = have;
Table tobacco_pastmo*ETOH_pastmo / missing list;
Run;

[_maldini_]

Duh. Thanks.

[japelin]

The data contains only missing, 1,2, but I think the if statement evaluates it with 0 and 1, so it doesn't get the expected value.

[ChrisNZ]

Sorry, I can't access the website holding your PDF. Please paste an image.

Also describe what is wrong (your code looks fine) and include the LOG too.

[mkeintz]

Your logic if fine, but you can simplify by changing 

	IF tobacco_pastmo = 0 AND ETOH_pastmo = 0 THEN user_risk = 0;
	ELSE IF tobacco_pastmo = 1 AND ETOH_pastmo = 0 THEN user_risk = 1;
	ELSE IF tobacco_pastmo = 0 AND ETOH_pastmo = 1 THEN user_risk = 2;
	ELSE IF tobacco_pastmo = 1 AND ETOH_pastmo= 1 THEN user_risk = 3;

to (assuming etoh_pastmo and tobacco_pastmo have 3 possible values: missing, 0, and 1:

  if nmiss(etoh_pastmo,tobacco_pastmo)=0 then  user_risk=  2*etoh_pastmo + tobacco_pastmo

But since the data are really: missing, 1, and 2, then use:

  if nmiss(etoh_pastmo,tobacco_pastmo)=0 then user_risk=2*(etoh_pastmo-1) + (tobacco_pastmo-1) ;

[ChrisNZ]

> Your logic if fine, but you can simplify by changing 

More compact yes, but simpler? This single line of code is A LOT harder to read and to maintain imho.

[mkeintz]

---

@ChrisNZ wrote:

> Your logic if fine, but you can simplify by changing 

More compact yes, but simpler? This single line of code is A LOT harder to read and to maintain imho.

---

I can appreciate that observation.  But what if there were a 3rd, (or 4th, etc.) dummy variable, doubling the number of categories with each new variable?  At some point, I think the (x + 2y + 4z ...) formulation would look more and more appealing compared to 8, 16, .. IF .. THEN statements.

[ChrisNZ]

True. To each problem its solution.