Age Heaping in African Census Data: Prevalence and Implications for Regression Discontinuity Research Designs

Savita Diggs, Yale University
Taryn Dinkelman, University of Notre Dame

Age heaping is a form of measurement error: it occurs when an individual misreports age in a systematic way, ``heaping'' onto focal digits such as 0 and 5. Economic historians have used the changing prevalence of age heaping in Census data as a proxy for improved numeracy at the population level, where direct measures of numeracy are lacking. However, this evidence has not often focused on African countries. Our project has three aims. We first document the extent of age heaping in Census data across a range of African countries from the 1980s, 1990s, and 2000s, highlighting reductions in heaping in some of these countries over time. Then, we show that age heaping is more pronounced among non-literate segments of the population in each country. Finally, we demonstrate that this non-random measurement error in age reporting generates measurement error bias in regression discontinuity research designs that rely on age as a running variable. In particular, studies that estimate the impact of female education on fertility (and other) outcomes using age as a running variable are potentially biased because of this non-random age heaping.

Keywords: Census data, Data and Methods, Economic Demography, Human Capital, Education, and Work

See extended abstract.