multiple left joins

[SAS_inquisitive]

Hello,

I was trying to understand the mechanism of multiple left joins in SQL.

I wonder how it works? I assume each left join creates an intermediate table and that combines with subequent table. In my example dat1and dat2 yields an intermediate table and it joins with dat3.

Is the total number of observations (records) in multiple left joins equal to the number of observations in first left join?

data dat1;
 input id $ x;
 cards;
 a  1
 b  2
 c  3
 ;
 run;
 
 data dat2;
  input id $ y;
  cards;
  a  1
  ;
 run;
 
 data dat3;
  input id $ z;
  cards;
  b 2
 ;
 run;
 proc sql;
   create table t1 as
    select a.*,b.*,c.*
    from dat1 as a 
        left join dat2 as b
          on a.id = b.id
          left join dat3 as c
          on a.id = c.id;
 quit;

[LinusH]

Yes, for #1 you are correct.
But for any SQL join no of observations in the result set is dependent on the source data, key values and join criteria. Joins creates (at least logically - physical optimization haooends behind the scenes) a product which is filtered by the join criteria.

[SAS_inquisitive]

@ LinusH, since the first left join preserves the number of observations in table a (dat1), this eventually will be preserved in later left joins, right?

[PGStats · Accepted solution]

---

I wonder how it works? I assume each left join creates an intermediate table and that combines with subsequent table. In my example dat1and dat2 yields an intermediate table and it joins with dat3.

True. Unless you alter join operations order with parentheses.

Is the total number of observations (records) in multiple left joins equal to the number of observations in first left join?

Not necessarily, if there are multiple matches on the right side, there can be more records in the joined table than there were in the left table.

---

[LinusH]

Not necessarily.
As per my previous answer, ech join could be 1 to many, or even many to many which would expand the no of result observations.