Counting Dead Voters: Death Registration and Electoral Fraud in Kenya

• In Event: The Politics of Identity Documents in the Global South

Abstract

A recent literature examines how and why states invest in the capacity for vital registration of citizens. The registration of births and, as studied here, deaths may have important implications for political processes like voter registration and boundary delimitation. This paper examines one aspect of the intersection of vital registration and electoral politics: how failures in death registration lead to the accumulation of "dead voters" on the voter register. While academics, election observers, and policy-makers all point to the presence of dead registered voters as a potential catalyst for electoral fraud, no work examines the relationship between dead voters and fraud. In this paper, I demonstrate how dead voters may facilitate election fraud, using data from Kenya's 2002 General Election. Rather than attempt to identify specific dead voters on a voter register, I apply a common demographic technique called a life table to estimate the probability that each voter is dead on election day, and use this to estimate the number of dead voters on the voter register. I provide simple visual tests for examining voter registers for the accumulation of dead voters. Then, I provide descriptive evidence that the prevalence of dead voters at the polling station level positively correlates with a common measure of ballot stuffing in Kenyan elections: the discrepancy in turnout across simultaneous elections at the same polling station. I show that as turnout discrepancies get larger, their relationship with dead voters increases, suggesting those polling stations with the largest turnout discrepancies tend to also have more dead voters. These results are robust to multiple model specifications and tests.

Author

• J. Andrew Harris, New York University Abu Dhabi