A Thurstonian model for ranking scientific journals