A Critical Comparison of Two Methods of Statistical Discrimination