Adjusting for underenumeration in the 1991 census

EDITOR,--The problem of non-random under-enumeration in the 1991 census has recently been highlighted in the BMJ.1 2 This problem may be substantial for certain groups in the population. Glover pointed out that the population figures for black Caribbeans, especially young men, may be substantially underestimated.3 Such issues are important for those concerned with estimating the differential prevalence of diseases among minority groups. We, for example, are estimating the prevalence of psychosis among different ethnic groups in two inner …