Also you likely do not want to use CATS as the last "word" of address1 will run into the first "word" of address2 creating a string that is less likely to be "like" your target. Please see this example and a suggested alternate:
data example;
length address1 - address3 $ 25;
address1 = 'Apartment 456 ';
address2 = '123 Main St ';
address3 = 'SomeTown ';
resultcats = cats(address1,address2,address3);
resultcatx = catx(' ',address1,address2,address3);
put resultcats= resultcatx=;
run;
And really, provide the entire code as LIKE is only valid in Proc Sql/fedsql and so may have other issues, plus at least one example of values that this doesn't work for.
You may also want to construct a variable with the contents of "%" || t2.address_line_3_clean || "%"
to see exactly what you are comparing too as there is a good chance when using the || operator that your comparison string has a number of trailing spaces. From the documentation:
Note: When you are using the % character, be aware of the effect of trailing blanks. You might have to use the TRIM function to remove trailing blanks in order to match values.
... View more
