How can we better evaluate the effectiveness of our programming efforts?