A right join returned more rows that were joined than actually read from the right table

[george7899]

Both the left and right tables have been standardized. There are 63 rows in the left table, and 31 rows in the right one.

A Data Join node (right join) was applied successfully. But the log shows there were 33 rows joined. Does it make sense to have 33 rows joined instead of 31?

[Astounding · Accepted solution]

Things to consider ... perhaps ...

• The left table contains a few records that match the same record in the right table
• The joining provisions are not specific enough, so that the same record in the left table matches multiple records in the right table.

It's almost certainly a result of what's in the data, combined with the joining criteria.

[george7899]

Yes, the left table does contain some records that match the same record in the right table.

[novinosrin]

A test to understand

data left;
do key_l=1 to 61;
 if key_l=31 then do _n_=1 to 3;
 output;
 end;
 else output;
end;
run;

data right;
 do key_r=1 to 31;
 output;
 end;
run;

proc sql;
create table w as
select *
from left  right join right 
on key_l=key_r;
quit;

[george7899]

Thanks for your codes. very helpful!