In this case you have two dichotomous variables: status (amateur/prof) and succes (succes/unsuccess). So there are (2*2=) four possible response patterns. Create one observation for each pattern and fill in its frequency in a separate variable. Then weight by
this variable. Now each single record 'counts for' N records where N is the value on the frequency variable.
I'd probably code your example as follows:
data list free/status success freq.
begin data
0 0 1544
0 1 14
1 0 17
1 1 7
end data.
weight by freq.
value labels status 0 'Amateur' 1 'Professional' / success 0 'Failure' 1 'Success'.
set onumbers labels.
crosstabs status by success
/statistics chisquare.