Study of age misstatement among young children in Ghana

Abstract

Summary In the 1960 Population Census of Ghana, the first attempt to enumerate the population by single years of age was made. The use of these data to estimate fertility levels was rendered somewhat difficult by the pattern assumed by the age statement of children. This pattern was similar to others found widely in developing countries. There appeared to be unexpectedly few children under two years of age and more children aged three years than any other single age. In order to investigate such age reporting, and in the hope of being able to provide some kind of correction to the census data, an attempt was made to trace one thousand children 0–99 months of age in ten different centers, with knoum dates of registered birth. If found, they were approached, as if by census enumerators, to obtain statements of age from the families. The investigation had to be confined to the compulsory registration areas where an eighth of the population of the country lives. It was found that age misstatement occurs partly because of digital preference but even more commonly because of a tendency to approximate age to the nearest birthday. Thus, when age misstatement occurs through the latter cause, it always takes the form of age advancement. Errors in age statement of more than one year fell evenly and randomly above and below the correct age. The age data in years were not appreciably affected by seeking further information on age in years and months. However, it was shown that the standard deviation of stated from real age was approximately proportional to real age and was equal to about one-eighth of it. The findings were then applied to the 1960 enumeration, where some problems such as the small number of one-year-olds, were solved. More work is needed to explain satisfactorily the peaking at three years of age. It was shown that age advancement may have reduced the size of the 0–4 age group by about 5 percent relative to adult female age groups, and this would result in an understatement of fertility as measured by the child/woman age ratio of the same amount.The research described in this paper was program of that post. With the exception of a carried out when the writer held the Population part of the analysis which has been carried out at Council demography post attached to the University of Ghana and formed part of the research program of that post. With the exception of a carried out when the writer held the Population part of the analysis which has been carried out at the Australian National University, the project was financed entirely by the Population Council.