Problems with traditional rating scales