THOSE MISSING MILLIONS - IMPLICATIONS FOR SOCIAL STATISTICS OF NONRESPONSE TO THE 1991 CENSUS

This article highlights some of the issues raised by incomplete covera ge of the 1991 census in Britain, illustrating how far-reaching are it s implications across the spectrum of social statistics. The origins a nd the current state-of-play of the debate over the distribution of th e non-response are summarised and the implications then assessed. For many census variables, non-response reduces the observed variation bet ween areas, with consequences for the distribution of resources for lo cal services. The implications for the reliability of the most basic o f health statistics are raised through mortality rates. A rough estima te is made, and supporting evidence presented, of the possible numbers of households which were missed by the census. Some implications for housing needs assessments are illustrated along with the effect of the sudden upwards revision of official estimates of the dwelling stock a nd the resulting overall great uncertainty over the number of vacant d wellings in the country. Finally, recent research on the electoral imp lications of people's choice not to be registered is discussed as this may well help to understand census non-response. Such a pervasive deg ree of uncertainty across almost all figures concerning people has not been experienced in Britain since the last century. There is little r eason to expect this uncertainty and the importance of its social poli cy implications to diminish in the future, without significant changes to census procedures and to the relationship between government and g overned.