On Sat, 27 Sep 2003 09:18:05 +0000 (UTC), Jonathan Neve wrote:

>> ...
>>
> This is _extremely_ confusing! :-)
> The reason I called it "high" selectivity, is simply because the value
> given by the index stats is... high! This value,
> from what I understood, corresponds to the average number of records
> given by each value of the index, which explains
> why unique indexes always have value 1. Anyway, never mind about
> terminology.

The values you find just _represent_ selectivity.
I think of selectivity (for indexes) as the ratio of "unique index values"
to the "total number of rows" (which conveniently gives 1 for indexes on
primary keys).


--
Aage J.