revealed preferences vs. hollow virtue signaling