I'm trying to do a big SORT CASES on a machine that's probably small
for it: SPSS 9, 64 meg of memory. The file is about 3 million cases --
only one variable, a 50-character string. (Why sort a file with one
variable? To set up for AGGREGATE.)
I've noticed that SORT CASES tends to go as follows:
A. Two complete passes through the data (i.e., cases go from 1 to the
end), at reasonable speed;
B. Multiple passes, faster, starting at the beginning and each going
part way through the file, with the number reached before starting over
increasing each pass. When the number reaches the number of cases in
the file, the sort is complete.
This time, in the second step, the number of cases per pass got up to
150,000 and seemed to stop increasing -- or, perhaps, increased very
slowly. I finally halted the process.
Is this a known, or understood, phenomenon?