Question

I am having something wrong with my query results. It is not displaying the correct rows with the “QuestionContent” field.

For example in the database table if it states (not included all fields):

QuestionContent     Option Type    QuestionMarks     Answer   ....// other fields

What is 2+2         A-D            1                  D
What is 3+3         A-D            1                  B
What is 4+4         A-E            2                  C

Then why when I create my query that the result display’s this:

QuestionContent     Option Type    QuestionMarks     Answer   ....// other fields

What is 2+2         A-D            1                  D
What is 3+3         A-D            1                  D
What is 4+4         A-D            1                  D

Below is the query I am using, how can this be fixed to be able to display the correct result?

SELECT q.QuestionContent, o.OptionType, q.NoofAnswers, a.Answer, r.ReplyType,
       q.QuestionMarks
FROM Answer a                
INNER JOIN Question q ON a.QuestionId = q.QuestionId
      JOIN Reply r ON q.ReplyId = r.ReplyId
      JOIN Question qu ON r.ReplyId = qu.ReplyId
      JOIN Option_Table o ON qu.OptionId = o.OptionId
GROUP BY q.QuestionContent

UPDATE: BELOW is the schema of the 4 tables:

Question Table:

SessionId (PK)  Varchar(3)
QuestionId(PK)  INT
QuestionContent Varchar(250)
NoofAnswers     INT
QuestionMarks   INT
ReplyId(FK)     Varchar(2)
OptionId(FK)    Varchar(2)

Answer Table

  SessionId (PK)  Varchar(3)
    QuestionId(PK)  INT
    Answer          Varchar(10)

Option_Table Table

  OptionId(PK)    Varchar(2)
  OptionType        Varchar(10)

Reply Table

  ReplyId(PK)    Varchar(2)
  ReplyType        Varchar(10)

In the Query I want to display these fields:

QuestionContent 
OptionType
NoofAnswers
Answer
ReplyType 
QuestionMarks 

Hope that is enough information, if not please comment to me 🙂

Answer 1

Given that the primary key of the Question table is SessionID + QuestionID, and that the primary key of the Answer table is also SessionID + QuestionID, you must specify the join on those two tables on both columns:

FROM Answer   AS a
JOIN Question AS q ON q.SessionID = a.SessionID AND q.QuestionID = a.QuestionID

Without both, you end up with a Cartesian product effect when a question ID appears in more than one session.

Also, since you are not selecting SessionID, you will need to deal with duplicate results where two different sessions have the same question and answer information. I think SELECT DISTINCT is probably better than GROUP BY for the purpose. Reserve GROUP BY for when you have aggregates (such as COUNT(*) or SUM(expression)) and do not use it for general ‘duplicate elimination’.

---

Original commentary

I believe there are are two parts to your problem — one possibly not crucial, the other probably crucial.

Your query is, more or less:

SELECT q.QuestionContent, o.OptionType, q.NoofAnswers, a.Answer, r.ReplyType,
       q.QuestionMarks
FROM Answer a                
INNER JOIN Question q ON a.QuestionId = q.QuestionId
      JOIN Reply r ON q.ReplyId = r.ReplyId
      JOIN Question qu ON r.ReplyId = qu.ReplyId
      JOIN Option_Table o ON qu.OptionId = o.OptionId
GROUP BY q.QuestionContent

For some reason, you’ve listed Question in that twice, but there’s no evidence that you need the second alias for it. On that basis alone, you can simplify the query to:

SELECT q.QuestionContent, o.OptionType, q.NoofAnswers, a.Answer, r.ReplyType,
       q.QuestionMarks
  FROM Answer       AS a                
  JOIN Question     AS q ON a.QuestionId = q.QuestionId
  JOIN Reply        AS r ON q.ReplyId = r.ReplyId
  JOIN Option_Table AS o ON q.OptionId = o.OptionId
 GROUP BY q.QuestionContent

That is the possibly not crucial change; you appeared to have a 1:1 join. The optimizer might be able to ignore the superfluous second reference to the Question table, but not mentioning that which is not used is better.

The probably crucial issue is the GROUP BY clause. In most dialects of SQL, you’d need to list all of the non-aggregate selected values in the GROUP BY clause. In this query, where there are no non-aggregates shown, that would mean listing all 6 result columns. When you don’t do this, MySQL takes things somewhat at random. My suspicion is that you need to reveal whether there are any aggregates in the full query, and explain what the GROUP BY is supposed to do for you. I’m not sure whether simply omitting it will give you the answer you need, or whether you need to do something else. One reason for the lack of certainty is the lack of schema in the question; we cannot tell all that much about the schema of the tables and the primary key, foreign key relationships between them.

So, I’d recommend using (trying):

SELECT q.QuestionContent, o.OptionType, q.NoofAnswers, a.Answer, r.ReplyType,
       q.QuestionMarks
  FROM Answer       AS a                
  JOIN Question     AS q ON a.QuestionId = q.QuestionId
  JOIN Reply        AS r ON q.ReplyId = r.ReplyId
  JOIN Option_Table AS o ON q.OptionId = o.OptionId

If that gives ‘the wrong number of rows’, then you need to explain the schema and the relationships between the rows.