International Foundation for Electoral Systems to consider restricting the discretion of adjudicators and allowing annulment only where there is clearcut evidence that the outcome of an election has been affected by fraud. The difficulty in determining whether a particular irregularity has affected the outcome of an election was undoubtedly part of the reason for the recent amendments to the Mexican electoral law, which provide for annulment if certain irregularities are found and the result of the election is close (i.e., the margin between candidates is less than five percent of valid votes cast). While this approach might be characterized as mechanistic and potentially unfair, it does provide a straightforward and easily verifiable way of dealing with challenges to election results that avoids the difficult and potentially divisive analysis required by the outcome-determinative test. When Tainted Votes Are Determinable An important consideration in determining if an election result was impacted is whether the number of tainted votes can be determined. Determining the number of tainted votes can be relatively straightforward; for example, when ineligible voters have been allowed to vote or eligible voters have been wrongfully turned away from a polling place. Where the number of votes affected by an irregularity can be calculated, the most common and simplest approach is to compare the number of votes tainted by irregularity to the number of votes separating the top two candidates. Where the number of tainted votes exceeds the margin of victory, then a finding that the irregularity may have affected the election is supported. The Constitutional Court in Austria applied this approach, sometimes referred to as the “magic number” rule, when it annulled the results of the second round of the 2016 presidential election due to various violations of the Federal Presidential Elections Act.32 Annullment of Austria's 2016 Presidential Election 100000 77,000 80000 60000 40000 30,863 20000 0 Margin of victory Number of affected votes However, as the Supreme Court of Canada observed in Opitz v. Wrzesnewskyj,33 the “magic number” rule is problematic since it does not address the fact that tainted votes do not necessarily all go to the winning candidate but are most likely to be distributed among the various candidates. In Opitz, there was a 26-vote difference between the winner and the loser, and the court found that no more than 20 32 See the Organization for Security and Co-operation in Europe’s Office for Democratic Institutions and Human Rights (OSCE/ODIHR) Observation Report, 3, https://www.osce.org/odihr/elections/austria/264241?download=true. 33 2012 SCC 55. 8