Consider this test setup:
CREATE TABLE dept (deptid integer PRIMARY KEY, deptname TEXT);
CREATE INDEX dept_name_idx on dept(deptname);
The dept table contains 1000 rows and the deptname column contains 10 unique values that are evenly distributed.
Which of the following two sample queries would use the index dept_deptname_idx?
1) SELECT deptid from dept where deptname ='SAPA';
2) SELECT deptid from dept where deptname <>'SAPA';
With only 10 distinct values, evenly distributed, chances are that neither query will use the index. A sequential scan of the table is typically faster than involving any indexes when retrieving more than roughly 5 % of all rows. Exact numbers depend on many details.
Also, 1000 small rows like in your example fit on a hand full of data pages. A sequential scan is hard to beat with such a small table.
With a much bigger table and/or substantially more distinct values in deptname, query 1 would be a candidate for using the index, but not query 2 (which retrieves most rows and will always use a sequential scan).
To optimize read performance for query 1 you could then use a multicolumn index on (deptname, deptid) - if preconditions for index-only scans are met.
