Electoral manipulation undermines the function of elections as a mechanism of representation and accountability, and erodes public trust in government institutions; however, our theoretical understanding of its causes is still limited. Research has focused on the blunter forms of electoral manipulation. Less attention has been given to the more subtle forms, although these are more common. This paper investigates one type of subtle electoral manipulation: miscounting, meaning election officers who selectively reject ballots during the counting. It suggests that miscounting (one of the ways in which political candidates can rig elections) is characterised by low risks and high direct costs. On one hand, it is almost invisible, embedded in the sociocultural norms and practices surrounding elections in many African countries. On the other, it is expensive, requiring large amounts of patronage to co-opt election officers. This makes miscounting attractive only to incumbents who fear electoral defeat and have deep clientelist networks. The paper tests this argument against data from the 2008 and 2012 parliamentary elections in Ghana. It shows that the number of ballot rejections is positively correlated with the number of years the party of the incumbent MP has held the constituency seat, but negatively correlated when this variable is interacted with the win-margin in the last parliamentary election. This pattern is consistent with the model, supporting the argument that the effect of electoral uncertainty on miscounting is conditional. When MPs do not have resources at their disposal, they choose other types of electoral manipulation or opt out of electoral rigging altogether.

Electoral manipulation undermines the function of elections as a mechanism of representation and accountability, and erodes public trust in government institutions; however, our theoretical understanding of its causes is still limited. Research has focused on the blunter forms of electoral manipulation. Less attention has been given to the more subtle forms, although these are more common. This paper investigates one type of subtle electoral manipulation: miscounting, meaning election officers who selectively reject ballots during the counting. It suggests that miscounting (one of the ways in which political candidates can rig elections) is characterised by low risks and high direct costs. On one hand, it is almost invisible, embedded in the sociocultural norms and practices surrounding elections in many
African countries. On the other, it is expensive, requiring large amounts of patronage to co-opt election officers. This makes miscounting attractive only to incumbents who fear electoral defeat and have deep clientelist networks. The paper tests this argument against data from the 2008 and 2012 parliamentary elections in Ghana. It shows that the number of ballot rejections is positively correlated with the number of years the party of the incumbent MP has held the constituency seat, but negatively correlated when
this variable is interacted with the win-margin in the last parliamentary election. This pattern is consistent with the model, supporting the argument that the effect of electoral uncertainty on miscounting is conditional. When MPs do not have resources at their disposal, they choose other types of
electoral manipulation or opt out of electoral rigging altogether.