When Are Elections Good Enough? Validating or Annulling Election Results votes were invalid due to irregularities. The court applied the magic number rule and upheld the election, but it left open the possibility that another, “more rigorous,” test might be developed, perhaps using statistical analysis, to determine whether the result of an election has been affected.34 Given this limitation, the magic number rule is most suitable for jurisdictions using the “uncertain outcome” test discussed above. In jurisdictions following the stricter “but for” approach, a challenger will generally be required not only to show that a certain number of votes have been wrongfully cast or voters wrongfully excluded but also to provide some evidence as to how those votes were cast or would have been cast. Courts have accepted various forms of evidence, including circumstantial evidence and statistical analysis in addressing such situations.35 In some cases where the identity of disenfranchised voters is known, the court has allowed voters to give evidence on how they would have voted if they had been allowed to do so. In other cases, courts have deducted tainted votes from candidates in proportion to their share of valid votes cast in the polling places in question (so-called “proportional deduction”).36 However, in one high profile case, a court refused to apply proportional deduction on the ground that it was illogical to assume that tainted votes cast necessarily reflect the proportions of valid votes cast.37 As with approaches to the wider outcome-determinative problem, there is no single correct approach to how to account for the impact of tainted votes on the outcome of an election. The important thing is that the approach to be used is clear, pre-established and consistently applied so that election contestants understand from the outset how a challenge to results will be considered. When Tainted Votes Are Undeterminable When it is not possible to determine the number of votes tainted by irregularity, determining if the outcome of an election has been affected can be extremely challenging for an adjudicator. For example, how can the impact of voter intimidation on Election Day or of voter bribery be determined? In such cases, the determination as to whether the election has been affected will inevitably involve judgment of a qualitative nature. However, decisions of this kind pose a particular danger to the legitimacy of the electoral process because of the risk that an adjudicator will make a weak or ill-founded judgment or because of the possibility that bias will influence, or be perceived to influence, the decision. 34 Following up on this suggestion, Canadian researchers Francis Hane and Heather Emptage have proposed a statistical analysis of the likely distribution of the untainted votes among candidates to determine how likely it was that a losing candidate had in fact won the election. Using this approach, they claimed there was a 0.48 percent chance that the challenger in Opitz would have won the election if there were no irregularities. We are not aware of any case where the proposed analysis has been used in resolving an election dispute. See Disputed Election Results: A More Rigorous “Magic Number Test”, in Constitutional Forum/Forum Constitutionelle, Vol 22, No 3 (2013), 27-34, https://journals.library.ualberta.ca/constitutional_forum/index.php/constitutional_forum/article/view/21082/1 5934. 35 William and Mary School of Law, Election Law Manual, 9-37. 36 Steven F. Huefner, “Remedying Election Wrongs,” 44 HARV. J. ON LEGIS, 282. 37 Borders v. King County, Chelan County Superior Court No. 05-2-00027-3, available at State of Washington, Office of the Secretary of State website accessed July 21, 2005. 9