How to delete empty row observations for data with multiple rows per subject

[Astounding]

Ah, now the plot thickens.  The code looks fine, so the number one suspect is the data.  What looks like blanks for character values may not actually be blanks.  Take a few variables that appear to contain blanks, and print them in hex form.  For example, if a variable has a length of $ 5, print it (for just one observation):

put varname varname $hex10.;

Dollars to doughnuts there will be some strange characters in there (hex nulls, carriage returns .... we'll find out).

[ballardw]

---

@Dbynoe wrote:

Thank you so much for the response! I really do appreciate it. The dataset I'm working with has more than 4000 varialbes, do you perhaps know of a more time efficient ways of acheiving the same goal?

---

That many variables is often a symptom of a poor data structure or process design. If by any chance you have a process that is constantly adding new variables every week/month/ or other period then the process is flawed and should be reconsidered. It is much easier to work with data that has a variable to indicate processing period, date or source with the same variables and then use BY group processing. If you need a REPORT for people to read (4000 columns, really?) then report procedures such as Proc Report and Tabulate are very good at creating such things.

[Dbynoe]

I didn't design the dataset, I just received it from a data coordination center my boss contracted. I'm trying to create 4 subsets of data from the larger dataset that are more accessible. 

This particular dataset is a Social Network Analysis, and provides information on many subjects and up to 48 alters for each subjects' network. For example, r1 is the question: "is this person male or female". There are 48 potential responses per subject. So the person who created the dataset provided one line of row per subject, so they created r1_1-r1_48 as variables that indicate the alter's sex, with each variable linked to a unique alter. There are a lot of questions, so that's one reason why there are so many variables in this dataset. I rearranged the dataset so that r1 now represents all the data from r1_1-r1_48...hence the repeated ID measures. It is a very messy dataset, so it's definitely forced me to learn a lot more about SAS than I expected!