unique observations of subject id in longitudinal data

[ccp27]

Hello.

I am very new to sas and am hoping someone can help me. I have a longitudinal survey data set and want to conduct simple descriptive statistics. When I run proc surveyfreq, sas is counting each ID as an observation. Is there an easy way to make sure sas only counts the unique subject IDs?

For example this is a snippet of my data :

SUBJECT_ID  GENDER  AGE  

1007197           1               45

1007197           1               45

1007197           1               45

1007813           2               62

1007813           2               62

1007925           1               53

1007925           1               53

So if I ran a freq on gender it would show total of 7 obs when in reality it should only be a total of 3. 

Would it be correct to do a proc sort with nodupkey to create a separate dataset to conduct only the descriptive stats?

Thanks in advance!

[novinosrin]

"Would it be correct to do a proc sort with nodupkey to create a separate dataset to conduct only the descriptive stats?"- I really like this idea.

[SteveDenham]

Other ways might be to sort on subject_id, and SET the data using first. or last. criteria.  Pretty much the same thing as nodupkey, though.

SteveDenham

[novinosrin]

Sir @SteveDenham  What a pleasant surprise. I didn't know God of statistics is still active? Have you incarnated again? Believe it or not, Many of stat mates were missing you so badly. 

[ghosh · Accepted solution]

Lots of great suggestions, here is an SQL approach

data have; input 
SUBJECT_ID  GENDER  AGE;  
cards;
1007197           1               45
1007197           1               45
1007197           1               45
1007813           2               62
1007813           2               62
1007925           1               53
1007925           1               53
;
proc sql;
  create table want as
  select distinct SUBJECT_ID, GENDER, AGE, count(SUBJECT_ID) as n_ID 
  from have
  group by SUBJECT_ID
  ;
 quit;

[Tom]

SQL is a simple way to do this without modifying the data.

You can use the DISTINCT keyword inside the COUNT() function. Examples:

proc sql;
select count(distinct subject_id) as nsubjects from have;
select age,count(distinct subject_id) as nsubjects
  from have
  group by age
;
quit;