It's Always The Semicolon

If you've ever taught SAS or even used SAS much you will recall that the most common error is to leave out a semicolon. Its always the semicolon. I wish I had remembered this fact today when I was working with a client who was doing a power analysis.

This was not a difficult question at all. It involved finding the power of a correlation for a given sample size. SAS's Proc Power is generally very straight forward and easy to use. I hadn't used it for correlation before so I looked the command up in the online SAS documentation. There was a simple example that I modified for my client. The problem was, it didn't work. I tried a bunch of things but it just wouldn't run.

Finally, I copied the example straight from the documentation, pasted it into the SAS editor and ran it. Except, it didn't run either. Here is what the example looks like:

proc power;
onecorr dist=fisherz
null = 0.15
corr = 0.35
ntotal = 180
power = .
run;

Now remember, this is a straight cut-and-paste from SAS' own documentation. There was one hint though, SAS kept on claiming the Proc Power was still running. But once it encounters a run statement with a semicolon, it should stop running. This means that there was a missing semicolon. Looking at Proc Power for multiple regression showed that there needs to be a semicolon before the run statement (at the end of the power statement).

There last two lines should look like this:

power = . ;
run;

So, if you're having so problems with SAS, just remember, it's always the semicolon. Well, almost always.

pbe