The accuracy of demographic data in the Ethiopian censuses

The quality of age-sex reporting and the presence of net undercount in the Ethiopian censuses were assessed using demographic analysis methods. First, to identify the presence of coverage errors, data organised by birth cohort for the three censuses (1984, 1994 and 2007) were graphically examined. Then age-specific sex ratios and cohort sex ratios were assessed. Second, to identify the pattern of digit preference, the two classical indices of heaping (Bachi’s and Myers’) were calculated. Third, to evaluate the overall accuracy of the data, the United Nations age-sex accuracy indices were calculated at the national and sub-national levels. Fourth, the relationship between quality of data and some socio-economic variables were analysed. The findings showed the presence of coverage and content errors in all of the three censuses. The results of all the indices suggested the inaccuracy of the data. The age-sex accuracy index declined from 66.5 in 1984 to 46.9 in 2007, suggesting a modest improvement in the accuracy but the qualification is still in the ‘highly inaccurate’ category. Finally, stressing the importance of better quality demographic data, some recommendations are made.Keywords: Cohort analysis, age-specific sex ratios, census, data quality, Ethiopi