How to calculate the average of a variable per subject with multiple observations and missing data?

[meghanrm]

Hi,

I am trying to create a new variable that is the average of a variable V1 for every subject. Each subject has 3 rows containing values for V1 that were taken at 3 different time points. I can't calculate the average using a data step: average=(total_V1 / 3) because some subjects have 1-3 missing values for V1. Therefore, for those subjects that are missing V1 values it should be divided by 1 or 2 for example not 3. I can't figure out how to create a variable to go in the denominator of my average calculation to make the numbers accurate, please help!

Here is an example of how the data exists:

id.    V1

1.      3

1.      4

1.      5

2.      6

2.      .

2.      7

3.       .

3.       .

3.       2

[ballardw]

You should always show what you expect an output data set to look like. Do you need a data set for further manipulation or a report for people read?

One way to get a data set:

proc summary data=have nway;
   class id;
   var v1;
   output out=summary (drop=_:) mean=;
run;

[meghanrm]

Sorry I am new to posting on here. I need a data set for further manipulation. I am trying to create a new average variable and I have already written the code to get the total for the numerator of the average calculation but I can't figure out what code would work for the denominator being that subjects with missing data should not be divided by 3

[PaigeMiller]

---

@meghanrm wrote:
Sorry I am new to posting on here. I need a data set for further manipulation. I am trying to create a new average variable and I have already written the code to get the total for the numerator of the average calculation but I can't figure out what code would work for the denominator being that subjects with missing data should not be divided by 3

---

Use PROC SUMMARY, as explained by @ballardw and then do the "further manipulation" in the output from PROC SUMMARY.

Which brings up the question: what "further manipulation" do you want to do? Perhaps this is already available in SAS (for example, PROC STDIZE) and so maybe PROC SUMMARY isn't needed.