Each stage of the Census is subject to stringent quality assurance measures. However, in a Census there are recognised sources of error which may survive in the data produced. Some of these are overcome or 'repaired' by careful processing procedures and quality management of the processing itself. The effect of those that remain is generally slight, although it may be more important for small groups in the population. The main kinds of error to keep in mind are:
Partial non-response - in some cases where an answer was not provided to a question an answer was imputed or derived (often from other information on the form). In other cases a 'not stated' code was allocated.
Processing error - while such errors can occur in any processing system, quality management is used to continuously improve the quality of processed data, and to identify and correct data of unacceptable quality.
Random adjustment - cells containing small values are randomly adjusted to avoid releasing information about particular individuals, families or households. The effect of random adjustment is statistically insignificant.
Respondent error - processing procedures cannot detect or repair all errors made by persons in completing the form, therefore some may remain in final data.
Undercount - although the Census aims to count each person once, there are some people who are missed and others who are counted more than once. A post enumeration survey is conducted soon after the Census to measure the undercount.
Want more information on Data Quality?
A series of Census Working Papers have been produced to assess and report on various aspects of 2001 Census data quality. More Information