Nigeria’s recent population censuses: a Benford-theoretic evaluation

Context: Population censuses in Nigeria have been plagued with under- or over-enumeration, as well as outright manipulation. This paper examines the claim of manipulated results of Nigeria’s 1991 and 2006 population censuses.
Data Source & Method: Data on both censuses were obtained from the National Bureau of Statistics and analyzed via fitting Benford’s probability distribution. The overall census data, as well as aggregate data for the six geopolitical zones of the country were examined to determine the level of conformity with Benford’s distribution, using the Chi-square goodness of fit test.
Findings: The conformity analyses showed that the overall counts differed significantly from Benford's in both censuses. The North-West region had the highest deviation in both censuses, while the North-East and South-West had the lowest deviation in 1991 and 2006 censuses, respectively. Significant conformity was observed in the sizes of the local government areas and the population density for the 2006 census.
Conclusion: Some datasets with built-in minimum and maximum values may still conform to Benford’s distribution provided the range of values of the first significant digit span digits 1 to 9. Census results should be scrutinized on the basis of Benford’s distribution as an additional check on the quality.

Keywords
Benford’s distribution; demography; population census; fraud-detection.