Question

I’m struggling with a task that could be very simple but I’m having a hard time figuring it out. I’ve two database table:

options_table

option_id | option_name | question_id |
    1     |   opt one   |      4      |
    2     |   opt two   |      4      |
    3     |   opt three |      4      |
    4     |   opt four  |      4      |

and votes_table

question_id | survey_id | option_id |
    4       |    1      |     1     |
    4       |    1      |     2     |
    4       |    1      |     2     |
    4       |    1      |     1     |
    4       |    1      |     3     |
    4       |    1      |     2     |

From this example, we have 2 votes on option 1, 3 votes on option 2 and 1 vote on option 3, but none on option 4.
I’d like to retrieve all the options name with the relative count, but I can’t get the option without votes. I’m sure I’m overlooking something, it’s workday’s end and I must be tired.
I tried something like:

   SELECT o.option_name, count(*) as cnt 
   FROM votes_table AS v 
   JOIN options_table AS o 
   ON o.option_id = v.option_id 
   WHERE v.survey_id = 1 
   GROUP BY o.option_name

but I just get

option_name | option_id | cnt
 opt one    |   1       |  2
 opt two    |   2       |  3
 opt three  |   3       |  1

How do I get also “opt four”? So I can properly display all the survey results including the option noone voted. I also tried using OUTER JOINs, but to no avail.

I’m ashamed for this might sound a stupid question, but I still have a lot to learn on databases and relationships

Answer 1

Use LEFT instead of INNER join – put in the Left side the table with all the options, and in the Right side the table with missing options.

Move any condition that uses a column from the Right table, like the v.survey_id = 1 one, from the WHERE to the ON clause.

Use COUNT(RightTable.column) instead of COUNT(*), so you get 0 for options not voted:

SELECT o.question_id
     , o.option_name
     , COUNT(v.option_id) as cnt 
FROM 
    options_table AS o  
  LEFT JOIN 
    votes_table AS v
      ON  v.option_id = o.option_id 
      AND v.question_id = o.question_id
      AND v.survey_id = 1 
WHERE o.question_id = 4            --- if you want results for one question only 
GROUP BY o.question_id
       , o.option_id

I suppose that (option_id, question_id) is the Primary (or a Unique) Key of the options table. That’s why you either need to GROUP BY both of them or WHERE o.question_id and GROUP BY o.option_id.

Otherwise, if you only GROUP BY o.option_id, the query for every option, will count votes for all questions that were voted with this option.