Combining information from election observation missions with statistical evidence, Hyde (2011) assesses how the electoral results differ between polling stations that were subject to an international election observation and those that were not observed. Her analysis of elections in Armenia is strengthened by the fact that the polling stations observed were selected randomly. While she finds that polling stations with no election observations reported lower vote shares for the opposition, this might not result from electoral irregularities. In particular, it could be an indirect effect of the presence of the observation mission on election behaviour.