"Michael G. Martin" <michael(at)vpmonline(dot)com> writes:
> I just ran a vacuum analyze with the specific column. Still get the
> same explain plan:
Did the pg_stats data change noticeably?
ANALYZE is a statistical sampling process in 7.2, so I'd expect the
results to move around somewhat each time you repeat it. But if it
changes a lot then we have a problem.
You could also try
ALTER TABLE symbol_data ALTER symbol_name SET STATISTICS n
for larger values of n (10 is the default) and then re-ANALYZE
to see if the stats get any more accurate. The default of 10
was more or less picked out of the air ... perhaps it's too small.
regards, tom lane