Since SAS keeps only one "row" of data in memory at a time, you can't exactly replicate access to row 19 when calculating a value in row 20.
But you can use the "lag" function, as in:
data want;
set have;
if termyear=:'Fall' then pct_change=divide(dif(degreecount),lag(degreecount));
run;
Whenever termyear starts with 'Fall' the lag function updates a fifo queue, by extracting the last time it was invoked (i.e. the previous "Fall") and storing the current value into the queue. In this case the lag function is a queue of size 1 ("lag2" would have a size of 2, etc.).
The dif function in nothing more than current degreecount minus the lagged value of degreecount (i.e. the change in degreecount.
Now if I had done this:
data want;
set have;
if termyear=:'Fall' then pct_change=dif(degreecount)/lag(degreecount);
run;
I would get the same numeric results, but also an error message because the first instance of "Fall" would retrieve a missing lagged value, which would cause this log note:
NOTE: Missing values were generated as a result of performing an operation on missing values.
Each place is given by: (Number of times) at (Line):(Column).
1 at xxx:42 1 at xxx:59
Using the DIVIDE function instead avoided the note.
