The impacts of underenumeration and age estimation error on the D0–14/D ratio and palaeodemographic measures

Abstract

The objectives of this study were to evaluate the impacts of infant and elderly underenumeration and age estimation error on previously reported measures derived from the D0–14/D ratio, specifically total fertility rate and the rate of natural increase. This study used data from the United Nations database for the year 1960. A two-step approach was taken: to test the stability of the relationship we examined the correlation between the D0–14/D ratio and population dynamics of interest with data omitted or misclassified to simulate a range of sample underrepresentation and age estimation error scenarios. To evaluate the practical implications, we used our existing equations to estimate total fertility and natural increase rates using the simulated differentially represented samples and calculated the standard error of the estimate. Correlations remained robust until a small number of infants and the elderly were represented. Where both infants and the elderly were underrepresented, as much as 75% of these age categories could be removed before accuracy of the equations was significantly compromised. Where either infant underenumeration or elderly underenumeration is suspected, our palaeodemographic measures maintain accuracy when up to 25% of the sample is missing. Age estimation error had a negligible impact. These measures demonstrated robusticity in a range of sample underenumeration scenarios, particularly for sources of bias that impact both infants and the elderly equally, and age estimation error. Where either infants or the elderly have been significantly underrepresented or omitted from the burial site, alternative measures may be required.