Protecting the confidentiality of the census data has a number of costs related to data utility. For the main census data, output is limited to predefined tables and to certain geographic thresholds. Though the Office for National Statistics will provide certain bespoke tables on request. Some detail is also lost on what are thought to be more identifying variables such as, for example, ethnicity and large households. The tabular output limits the usefulness of the data for statistical modelling purposes (for example in looking at unemployment in a particular area a researcher or policy maker may want to control for the impact of other factors such as the age profile of the population or education levels).
The disclosure control measures have a particular impact on the anonymised microdata (Samples of Anonymised Records www.ccsr.ac.uk/sars/). These are subject to separate measures and a number of the variables in the data are likely to be removed or reduced in detail and geographic identifiers are likely to be removed. Such measures limit the usefulness of the data and have caused concern amongst users. As Wathan (2002) has argued, suppressing information on households with 7+ members and multi-ethnic households would not only lead to a bias in the data regarding ethnicity, overcrowding and number of children but also would have important policy implications. See the user consultation discussions www.ccsr.ac.uk/sars/2001/consultation.html.