2001 Census Variables
2001 Census Variables
3. Method of Analysis
For each item (a single census question or a group of related questions generating a combined census item), the categories to be analysed were taken as presented in the relevant Univariate table. The number in each category were assembled from:
the output database;
the final input database which distinguished persons and households added by the One Number Census (ONC); and
the audit database that identified enumerated records with values that had been edited or imputed.
There were a few inconsistencies between the audit database and the other two. These lay chiefly in the fact that one of the methods of disclosure control applied to the 2001 Census, that of record swapping, was applied to the input database from which the output database was extracted, but not to the audit database. The analyses have been adjusted to remove inconsistencies resulting from this mismatch. Otherwise they would appear to show that an edit that automatically set the 'carer' variable to 'not a carer' was apparently applied to persons who were in fact carers. Leaving these discrepancies could be disclosive of the record swapping process (by revealing rates of swap). The conclusions about the way Edit and Imputation affect variables are not materially affected.
Certain items were composites of responses to several questions on the form. For example, ‘activity last week’ was derived from responses to 5 questions covering whether the person was working in the week before the census, whether looking, or available, for work, whether waiting to start a job already obtained, or whether economically inactive in some way.
The analysis of questions on destination and method of travel to place of work or study has been divided between persons travelling to either type of destination (depending on whether the person is working or a full-time student). The analysis of relationship has been tackled in a unified way, even though there was a separate response for each possible pair of persons in the household. Otherwise, each item is taken on its own and not in combination with other items. For example, there is no analysis of the extent to which values were edited or imputed for the various categories of ethnicity according to the person’s age and sex. Such analyses may be produced in a later exercise.