Count problem...'IN' and 'NOT IN' problem?

Question

I'm not sure what I missed here. Can anyone help?
I have an address_main table, and a case_main table. Case_main attach case to address by using address_id, which is a unique number.
SQL> select count(address_id) from address_main;
COUNT(*)
39576
SQL> select count(address_id) from address_main
where ADDRESS_ID in (select address_id from case_main);
COUNT(ADDRESS_ID)
39080
I'm expecting the next statement come back a count result at 456 (39576-39080), but it returns '0'.
SQL> select count(address_id) from address_main
where ADDRESS_ID NOT IN(select address_id from case_main);
COUNT(ADDRESS_ID)
0
So 'IN' and 'NOT IN' can not cover all the posibility?

Answer 1

Check for NULL values, they often affect results just the way you describe.
Consider:
SQL> create table address_main(x) as
  2  select rownum
  3    from dual
  4  connect by level <= 5;
Table created.
SQL> create table case_main(x) as
  2  select rownum
  3    from dual
  4  connect by level <= 5;
Table created.
SQL> -- We'll update just one row in CASE_MAIN with NULL value
SQL> update case_main set x = NULL where x = 5;
1 row updated.
SQL> select * from address_main;
         X
         1
         2
         3
         4
         5
SQL> select * from case_main;
         X
         1
         2
         3
         4
SQL> select count(x) from address_main;
  COUNT(X)
         5
SQL> select count(x) from address_main
  2   where x in (select x from case_main);
  COUNT(X)
         4
SQL> -- But ...
SQL> select count(x) from address_main
  2   where x not in (select x from case_main);
  COUNT(X)
         0