Counting Records By Group - if first/last / how to get valid counts

[Clo13]

Can anyone help me identify the problem in code below? I think it might be the first set of if statements creating _ct (count) variables. Those values are very off though in general the code runs and even the final indicators will likely be correct once I can get appropriate values for the _ct variables. 

Thank you for your suggestions. 

[PaigeMiller]

Explain the desired output.

And: show us the desired output for a small example data set (and show us the small example data set as well).

Please provide code as text, not as screen captures, pasted into the window that appears when you click on the "little running man" icon.

[Clo13]

Hi Paige. Agreed. I work on SAS via a remote desktop and I have to get a system down for transferring all the helpful info out of the SAS environment.

[average_joe · Accepted solution]

The issue seems to be how you initialize the _ct vars. Try this:

if first.scrssn then do;
    likely_ct = 0;
    high_likely_ct = 0;
    acute_ct = 0;
end;

[Clo13]

Thank you so much! That works! I need to study more. 🙂

[ballardw]

The code on the if first. is almost certainly wrong unless you want to do a logical comparison between one value and the results of two other logical comparisons.

When you do something like this (and you can copy this as it is text and run in your session)

data example;
   x=3;
   if x=3 then var1=10 and var2=0 and var3=0;
run;

What SAS does when it encounters Var1 = it assumes that EVERYTHING after the = is involved with assigning the value to Var1. This gets coupled with SAS returning 1/0 for true/false from comparisons. So VAR2 is compared to 0,. returning either a 1 or 0. In this case since I do not have a value for VAR2 missing is compared to 0 and returns "not equal" or false, 0. Same for Var3. Var1 = (the result of comparing) 10 and 0 and 0. SAS treats any numeric value other than Missing or 0 as False. So that resolves to Var1 = True and False and False, which is False (true only if all are true) and hence 0.

To assign non-missing values to multiple variables you need each assignment as a single statement. Like this:

data example;
   x=3;
   if x=3 then do;
     var1=0;
     var2=0; 
     var3=0;
   end;
run;

You might be asking why did I use Var1=10? With your code you may have thought that the Likely_ct=0 was correctly assigned and trying to figure out why the rest weren't. In fact, the result of zero for that was the result of a logical comparison.

So your counts EXCEPT Likely_ct were not resetting at First.Screen.