topic Re: Missing data in SAS Programming

Missing data

kristiepauly1 — Wed, 10 Mar 2021 03:00:45 GMT

Hello,

I am manipulating my data using the following code:

/*Convert motherheight to height in inches*/
DATA LabAnal.study_cohort_2;
SET LabAnal.study_cohort;
IF MotherHeight = '' THEN Ht_Inches = .;
ELSE IF MotherHeight ='?' THEN Ht_Inches ='';
ELSE Ht_Inches=(INPUT(SUBSTR(MotherHeight, 1, 2), 2.)*12) + (INPUT(SUBSTR(MotherHeight, 4, 2),2.));
RUN;

^^this works great

/*Convert priorweight to a numeric variable*/
DATA LabAnal.study_cohort_2;
SET LabAnal.study_cohort;
Priorweight2 = INPUT(PriorWeight, 3.);
RUN;

^^this also works great

/*Calculate BMI*/
DATA LabAnal.study_cohort_2;
SET LabAnal.study_cohort;
IF Priorweight2= "." or Ht_inches= "." THEN BMI= ".";
ELSE BMI= (Priorweight2*703)/(Ht_Inches*2);
RUN;

^^but when I run this, it sets all the values from priorweight2, ht_inches and BMI to missing. I've tried with and without quotes around the periods in the IF statement for BMI code. It doesn't change anything.

Re: Missing data

Astounding — Wed, 10 Mar 2021 03:25:26 GMT

As you might have guessed, there are problems with your final step. The main problem is here:

/*Calculate BMI*/
DATA LabAnal.study_cohort_2;
SET LabAnal.study_cohort;

You are using the wrong data set. This step replaces LabAnal.study_cohort_2, by reading in the data rom LabAnal.study_cohort (which does not contain Priorweight2).

There are other, minor issues. For example, to refer to a missing value for a numeric variable, do not use quotes. Change the next line to:

IF Priorweight2= . or Ht_inches= . THEN BMI= .;

Re: Missing data

japelin — Wed, 10 Mar 2021 03:34:26 GMT

The story changes depending on the attributes of the variables, so please provide reproducible data.

Without seeing the data, I can't say exactly what you're seeing.
Perhaps,

IF Priorweight2= "." OR Ht_inches= "." THEN BMI= "." ;

this statement is the problem.

Not . which means missing but "." which is string is stored in Priorweight2, Ht_inches, and BMI.

IF Priorweight2= . . or Ht_inches= . THEN BMI= . ;
would improve this point.

Re: Missing data

kristiepauly1 — Wed, 10 Mar 2021 04:09:15 GMT

I completely overlooked the DATA SET. That was the fix. Thank You!!