An experiment with census-type age enumeration in Nigeria.

Abstract Censuses were held and data subsequently published in Nigeria in 1952/3 and 1963. Age data and other aspects of the enumerations gave rise to considerable doubts. In 1969 the Demographic Training and Research Unit of the University of Ife carried out a census of ages amongst 10,000 persons in the Western State of Nigeria employing in succession orthodox methods, supporting historical records, and two approaches involving the identification of single-year cohorts of contemporaries. Certain other data were also collected for explanatory purposes. The project confirmedthe tendency for age misstatement to exaggerate the number of females aged 20-44 and the number of older males, and to understate the number of persons aged 0-9 and females aged 50-59. It was shown that amongst females, aged 15-24, there is a significant tendency for the ages of the unmarried to be understated and those of the married, expecially those of higher parity, to be overstated. Such tendencies may be reduced by any type of enumeration if more time and care is allowed but anomalously this may increase the underenumeration of the very young. The more elaborate methods of enumeration are examinedto see whether they yield more accurate data and whether their use would be possible in a national census. Age statement was examined in some detail and it was shown that the majority of data originate neither with respondents nor enumerators but with third parties, who may wellbe continuing and untrainable sources of biassed error. The project provided more evidence than anticipated on the validity of the 1963 Census in the area, thus leaving the relative validity of the 1952/3 and 1963 censuses and hence the likely population of Nigeria still an open question.