[mysql] MySQL Join Where Not Exists

I have a MySQL query that joins two tables

• Voters
• Households

They join on voters.household_id and household.id.

Now what I need to do is to modify it where the voter table is joined to a third table called elimination, along voter.id and elimination.voter_id. However the catch is that I want to exclude any records in the voter table that have a corresponding record in the elimination table.

How do I craft a query to do this?

This is my current query:

SELECT `voter`.`ID`, `voter`.`Last_Name`, `voter`.`First_Name`,
       `voter`.`Middle_Name`, `voter`.`Age`, `voter`.`Sex`,
       `voter`.`Party`, `voter`.`Demo`, `voter`.`PV`,
       `household`.`Address`, `household`.`City`, `household`.`Zip`
FROM (`voter`)
JOIN `household` ON `voter`.`House_ID`=`household`.`id`
WHERE `CT` = '5'
AND `Precnum` = 'CTY3'
AND  `Last_Name`  LIKE '%Cumbee%'
AND  `First_Name`  LIKE '%John%'
ORDER BY `Last_Name` ASC
LIMIT 30 

There are three possible ways to do that.

1. Option

   SELECT  lt.* FROM    table_left lt
   LEFT JOIN
       table_right rt
   ON      rt.value = lt.value
   WHERE   rt.value IS NULL
   

2. Option

   SELECT  lt.* FROM    table_left lt
   WHERE   lt.value NOT IN
   (
   SELECT  value
   FROM    table_right rt
   )
   

3. Option

   SELECT  lt.* FROM    table_left lt
   WHERE   NOT EXISTS
   (
   SELECT  NULL
   FROM    table_right rt
   WHERE   rt.value = lt.value
   )
   

I'd use a 'where not exists' -- exactly as you suggest in your title:

SELECT `voter`.`ID`, `voter`.`Last_Name`, `voter`.`First_Name`,
       `voter`.`Middle_Name`, `voter`.`Age`, `voter`.`Sex`,
       `voter`.`Party`, `voter`.`Demo`, `voter`.`PV`,
       `household`.`Address`, `household`.`City`, `household`.`Zip`
FROM (`voter`)
JOIN `household` ON `voter`.`House_ID`=`household`.`id`
WHERE `CT` = '5'
AND `Precnum` = 'CTY3'
AND  `Last_Name`  LIKE '%Cumbee%'
AND  `First_Name`  LIKE '%John%'

AND NOT EXISTS (
  SELECT * FROM `elimination`
   WHERE `elimination`.`voter_id` = `voter`.`ID`
)

ORDER BY `Last_Name` ASC
LIMIT 30

That may be marginally faster than doing a left join (of course, depending on your indexes, cardinality of your tables, etc), and is almost certainly much faster than using IN.