There is a recently published paper on election fraud by Andrew T. Little in the Quarterly Journal of Political Science, titled “Elections, Fraud, and Election Monitoring in the Shadow of Revolution.”
Here’s the abstract:
Elections are modeled as a public signal in an incomplete information game of revolution. By changing beliefs about the general level of anti-regime sentiment, elections can make citizens more or less apt to rebel and hence make a successful revolution more or less likely. This effect makes elections valuable to incumbents that are not secure in office as they have more to gain by good results than they have to lose from bad results. Electoral fraud is modeled as a distortion of the public signal, and election monitoring is incorporated as changing the cost of this distortion. In equilibrium, citizens discount the distortion, so the average protest size and probability of revolution are the same as when the incumbent cannot commit fraud. This makes election monitoring valuable to incumbents as it ties their hands and lowers the equilibrium amount of fraud. So, elections may be held that would not occur in the absence of monitoring.