Identifying fraud in democratic elections: a case study of the 2004 presidential elections in Mozambique

摘要

The 2004 Presidential Election in Mozambique was marred by allegations of fraud. We assess the validity of these allegations by testing whether or not qualitative descriptions of the methods and locations of misconduct are consistent with a series of simple quantitative tests. Most studies of electoral misconduct are based on ecological regression analysis or on comparing a different data set with the electoral data in question – past elections, exit polls, etc. In the case of Mozambique this is impossible due to data restrictions. Instead we use qualitative evidence to inform a quantitative identification strategy. The concordance between specific complaints and the statistical evidence suggests that ballot box stuffing, improper ballot nullification, and (intentional) organisational failure took place. While the overall election result was unaffected by the fraud, our analysis demonstrates a method of assessing allegations of misconduct and points to areas of concern for those managing or observing future elections in Mozambique and elsewhere.