There is a wide range of possible responses to Census questions, and to make these responses more manageable for output purposes, it is necessary to group together like or related responses. Each response is allocated an index number, and these index numbers are then reformatted into classes or classification levels of similar response.