 |
 |
> The 'guide to accuracy' claim is difficult to defend. It is possible to > believe that 'the ONC results remain the best central estimates' as the > website says later on. But it is evident that the main source of error is > deficiency of coverage. To talk about confidence levels based > on sampling > error is to parade statistical method in a way that can impress only the > unintelligent and the statistically naive. The sampling error for coverage is based on statistical analysis of the Census Coverage Survey, an interview survey of 300,000 households. The ability for the first time to put confidence intervals around census data does impress me. Information about the coverage survey is available on the Census website and has been mentioned on this list. There is real debate to be had on the Census results and the information needed to be assured of their reliability. The debate includes (a) the disclosure of evidence to judge the possible additional people not estimated by the coverage survey (the 900,000 emigration versus undercount debate), (b) the disclosure of evidence to judge the success of the Coverage Survey. Regarding the latter, modelling from the Coverage Survey of the numbers missed in different areas and amongst different types of people inevitably provides smoother patterns of non-response than the reality. The amount of smoothing depends on how well the coverage survey did in each local authority and the extent to which it failed in some areas or age groups and had to be supplemented with survey information from other areas or age groups. This is information not yet released. The robustness of the census in small areas as well as large ones is important to social analysis using health, employment, and other rates that assume an accurate population. Nonetheless, the evidence on emigration vs undercount is of prime importance, as the statistical reliability of sub-population data is somewhat dependent on knowing the accuracy of the total numbers I think. RadStats and other organisations can contribute to an open debate by putting these questions in a way that encourages ONS to release more information. Maybe some press release will be in order, hopefully based on some consensus. Ludi [log in to unmask] - Centre for Census and Survey Research at the University of Manchester (currently Tues, Wed, Thurs). Direct telephone: +44 (0)1274-642838. Post reaches me most quickly when sent to 41 Park Crescent, Bradford BD3 0JZ, UK. For Bradford Council matters, please contact me on tel. +44 (0)1274-754252 (Mon, Fri). ****************************************************** Please note that if you press the 'Reply' button your message will go only to the sender of this message. If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to [log in to unmask] *******************************************************