Compliance Review: Final Report and Recommendations
Addressing the Causes of Non-Compliance
Errors and Expectations
The Supreme Court's majority decision in the Etobicoke Centre appeal case recognized that, given the nature of Canadian federal elections and the fact that ordinary citizens conduct them, some degree of human error is inevitable.
At paragraph 46 of their judgement in Opitz v. Wrzesnewskyj,Footnote 26 Justices Deschamps, Abella, Rothstein and Moldaver JJ. state:
The practical realities of election administration are such that imperfections in the conduct of elections are inevitable. . . . A federal election is only possible with the work of tens of thousands of Canadians who are hired across the country for a period of a few days or, in many cases, a single 14-hour day. These workers perform many detailed tasks under difficult conditions. They are required to apply multiple rules in a setting that is unfamiliar. Because elections are not everyday occurrences, it is difficult to see how workers could get practical, on-the-job experience.
However, the Supreme Court was not suggesting that election officials are excused from ensuring they perform their duties to anything less than the highest of standards. At paragraph 69 in the same decision the Justices state:
In recognizing that mistakes are inevitable, this Court does not condone any relaxation of training and procedures. . . . The CEO has an obligation to ensure, as far as reasonably possible, that procedures are followed. . . . Failure to live up to this mandate would shake the public's confidence in the election system as a whole and render it vulnerable to abuse and manipulation.
One crucial question this review has struggled with is what error rate might be publicly acceptable for Election Day registration and voting procedures. Most review participants, and a cross-section of citizens the Reviewer surveyed informally, said that this should be a very low rate indeed.Footnote 27 The majority indicated the acceptable percentage as "zero". None thought the error rates established in the contested Etobicoke Centre election, or in the figures indicated in the national sample audit, to be even remotely acceptable to an average Canadian.
The assumptions behind this — that serious errors in election procedures simply should not be made and that the only acceptable errors might be purely clerical or unrelated to procedural safeguards — raises the bar for procedural improvement. Public trust is at risk if the rate of error is not significantly reduced by the next general election.
However, reducing the current rate of serious errors during registration and vouching transactions forms an immense challenge that should not be underestimated.
Moving to a New Model of Voting Services
A core finding of this review is that election officer error rates will not be much improved just by enforcing existing procedures more rigorously. Findings from the audit of the November 2012 by-elections reinforce this: adding a compliance advisor role to monitor activities at voting sites led, at best, to only modest reductions in error rates.Footnote 28
There is broad consensus among participants in this review that comprehensively addressing the existing causes of error requires moving to a new voting services model. That model needs to reduce, consolidate and streamline individual tasks and change who those tasks are assigned to. This level of change is not merely a "tweak" of the existing system with better enforcement of rules, or more in-depth training. The nature of the change required is fundamental and structural and cannot be achieved without changes to the framework of electoral legislation.
Even before the Etobicoke Centre court challenge, Elections Canada was researching alternative voting service models that might be used to restructure and improve the current approach set out in the Canada Elections Act.
In recent years, many senior election managers across Canada have come to regard the "polling division" model as increasingly unsuited to 21st century realities. Further, there is broad consensus in the electoral administration community that significant opportunities for improved voter service, procedural compliance, as well as efficiency gains, would come from applying modern information technology. Particularly compared to humans hired to work at an unfamiliar task for a single day, properly programmed computers are exceedingly capable of executing detailed instructions and complex processes accurately. However, to use technology in the voting process inevitably involves a "re-design" of the voting services model, and would require restructuring the legal framework that defines the process.
For several years, Elections Canada has been working to identify a simple and streamlined model for voting services, motivated by a desire to increase efficiency, improve voter experience and improve access to the electoral system.
The context of the Etobicoke Centre legal case and this review's own findings suggest strongly that moving to a new model carries potential to significantly reduce rates of error by election officers. However, the new model must be developed in such a way that compliance is a core design requirement. That design must take into account the realities of the election context. The new model's design must explicitly strive for accurate procedures and record-keeping. It must enable a typical election officer — a non-expert, with limited training, who works in the role only occasionally but for long hours, under intense pressure and public scrutiny — to achieve full compliance with all rules and procedures.
The "New Brunswick" Model
As part of its research in finding an alternative model to propose to Parliamentarians, Elections Canada examined whether the "New Brunswick" model for providing voting services could be "re-engineered" to meet federal electoral standards and national delivery requirements.
A modified approach to providing voting services was pioneered in New Brunswick during that province's 2008 local government elections. The Chief Electoral Officer of New Brunswick, Michael Quinn, describes the genesis of the design for the model being inspired by a recommendation made by the province's "Commission on Legislative Democracy":
That the electoral process be updated and streamlined to make it easier to cast a vote; make the rules around elections more clear and understandable; and provide a more flexible and efficient process for administering elections. . .Footnote 29
The New Brunswick model entirely removes the concept of polling stations from the voting equation. Instead, each voter "checks-in" to prove eligibility, then obtains a token and exchanges it for a ballot, marks their ballot and deposits it in a ballot box. At each stage, voters simply present themselves to the "next available officer" — there's no need to go to a specific table. The use of networked computers for voters list look-ups at the "check-in" stage, and an automated scanner-tabulator at the ballot deposit stage, provides substantial efficiency gains. Footnote 30
Because Elections New Brunswick is mandated to manage provincial elections as well as local government elections, it quickly saw benefits in applying the same model in a provincial election. It made a case for moving in this direction, proposed legislative changes, and was permitted to use the new voting services model in the 2010 provincial election. That provincial election was successfully held, successful local government elections followed in 2012, and the model is now firmly established as the standard for local and provincial elections in New Brunswick.Footnote 31
The "New Brunswick model" features important improvements over the traditional "polling division" model, still used federally, and in all other provinces and territories in Canada.
Elections Canada conducted a careful analysis that demonstrated that a "re-engineered" implementation of the New Brunswick model could meet all major electoral process requirements contained in the Canada Elections Act. However, the national agency's approach to an envisioned implementation of the model differs from New Brunswick's in one fundamental way — it assumes all voting locations would be equipped with "real time" access to a national computerized voters list, utilizing Internet data communications. This capability would allow for "live" voter registrations and updates, and automated list "strike-off" processing in all locations.
(Diagrams of the flow of voters through voting sites under the current "polling division" model and Elections Canada's proposed "re-engineered" voting operations model are shown on the following two pages.)
Current "polling division" voting services model
Text version of "Current polling division voting services model"
IO – Information Officer
CPS – Central Poll Supervisor
PS – Polling Station
DRO – Deputy Returning Officer
PC – Poll Clerk
Proposed "re-engineered voting operations" model
Text version of "Proposed re-engineered voting operations model"
There are many indications that a move toward a streamlined "re-engineered" model, along the lines of the one used in New Brunswick, has the potential to significantly improve procedural accuracy by resolving the major causes of non-compliance with voting day procedures identified during this review.Footnote 32
The re-engineered model breaks down election officers' duties into more manageable, specialized tasks. It also provides a "live" electronic voters list and a possibility of ballot scanners for automated tabulation of voting results.
The model includes the potential to manage complex "exception" cases as a separate stream. That complexity could be reserved for specialized and experienced staff, reducing the potential for errors.
Automated support for "striking-off" voters' names on the list could be built in such a way that candidate campaign offices receive information on which voters have turned up to cast a ballot in "real time". This would alleviate the considerable administrative burden currently experienced by election officers who need to continually fill out Statements of Electors Who Voted ("bingo sheets") and provide copies of them to all candidate representatives every half hour.
The model combines potential for fast, accurate automated vote tabulation with the insurance of paper ballot audit trails. This would likely shorten and simplify closedown procedures for the vast majority of staff, with the polling site as a whole responsible for providing results rather than each individual polling station unit.
Supporting improved supervision
Re-allocating tasks by function would make most roles under this model simpler. Supervisors could focus their priorities on staff dealing with complex cases. In smaller voting sites, supervisors could be trained to deal with "exception" cases, in addition to supervising.
Incorporating automated system features carries potential to add valuable information about compliance. Key compliance indicators could be captured and used as a "real time" management tool by supervisors on Election Day.
The new model does not tie tasks to particular individuals, creating the potential for better working conditions because election officers could be replaced with other officers, in order to allow proper breaks.
It would also provide potential for a less labour-intensive, more cost-effective use of human resources; for example by "scaling" services to meet levels of demand. This would likely reduce overall person-hours required. At the same time, improved working conditions would also remove difficulty in recruiting the number of individual staff required.
It would also make dealing with difficult cases more effective, by assigning a handful of most experienced and capable staff per voting site to the complex roles, while making the majority of roles considerably easier to perform.
Allocating tasks by function, and streaming "exception" cases to specialist staff, would enable training to target individual roles, and reduce the amount of information most staff would need to learn.
With most staff learning fewer procedures, training could focus on practice sessions demonstrating simpler, rote tasks, rather than assimilating complicated and detailed information about how to deal with many different types of "exceptions".
Most staff would therefore likely require less training while some specialist staff, such as those dealing with "exception" cases, would likely require more targeted training.
Improving the voters list
The live, electronic voters list would no longer be polling division based for front-end users.
Data for voters already on the register, but whose information must be revised in some way, could be found and updated on this electronic list. Persons who presented themselves at an incorrect voting location could be efficiently redirected to their assigned voting site — or possibly they could be accommodated by "vote anywhere" features designed into the re-engineered model. Anyone already registered would not need to be treated as a new voter, allowing for more efficient processing in an "update" stream.
Potentially, voters could check their new registration information, or updates to existing registration data, on-the-spot to verify accuracy. This could be one way to help ensure that the list becomes more accurate going forward.
System controls could ensure consistent and accurate addressing formats, and links to street index and postal code lookup tables would help improve the quality of address data captured. This, in turn would allow for better maintenance of registration records between elections.
Changing the culture of service provision
Handling "exception" cases with a much smaller subset of experienced staff, with assistance from supervisory staff, carries the potential for more accurate and consistent application of the rules.
Requiring that system rules be met for all cases is likely to improve compliance and consistency, and carries the potential for better reporting without additional overhead.
Each type of "check-in" process could have a step-by-step script on the screen, allowing the officer to process each voter in a fully compliant manner. This would likely lead to extremely high levels of compliance in the enforcement of voting eligibility rules.
Elections Canada recognizes that moving to a new model, similar to the approach in New Brunswick, will require fundamental changes to election legislation and substantial redefinition of election officers' roles. This would constitute a paradigm shift, with far-reaching impacts on all the support structures underpinning election delivery. Changes would include new voting procedures, instructional materials, training methods and content, materials and supply requirements, human resource and financial management procedures, management information collection, and IT support systems.
Realistically, in the time available before the 42nd national general election, a new voting services model cannot be implemented without unacceptable risk. Yet, moving to a new voting services model is ultimately necessary to fully enable temporary election officers to easily, systematically and consistently comply with more efficient procedures and rules.
The most prudent approach would be to define and rigorously test a pilot implementation of a new model in the next election, and then implement it nationally before the 43rd general election, now scheduled for 2019.
Elections Canada senior management approved a comprehensive business case for "Re-engineering of Voting Process" in the summer of 2012 and a project charter for a "Re-engineered Voting Operations" pilot project was subsequently drafted. The approval of that charter, and the final preparations of a detailed project plan for the model's development and pilot implementation, now depends on how the organization decides to improve compliance. Appropriately, management has purposely held off on finalizing the design of the alternative voting services model until it resolves the priorities and scope of the action plan it will prepare in response to this report's recommendations.
Re-engineering federal voting services using the New Brunswick model as the foundation for change holds great promise for comprehensively addressing the many outstanding problems of the current "polling division" format. Serious efforts should be made to prepare and demonstrate workings of the re-engineered model to the various Committees of Parliament that must approve it for a pilot implementation before, or during, the 2015 general election.
Pilot testing a re-engineered model before or during the 2015 election will be a significant commitment for Elections Canada. It will aim to prove that there is a comprehensive longer-term solution to improved procedural compliance by election officers, as well as provide a number of other benefits, including greater access and improved services to voters.
Addressing Compliance for 2015
The compliance challenge remains for the 2015 general election, despite the fact that a new voting services model has been identified which holds considerable promise as being a sustainable solution that addresses the causes of serious error by election officers.
It remains critical that the 42nd general election has a very low incidence of serious errors by election officers. Even though the preferred final solution will not be available, maintaining Canadians' trust in their electoral system requires that the compliance problem be effectively addressed to minimize error rates.
Modest changes to existing legislation, as well as more extensive changes to administrative procedures and support mechanisms, are required to minimize the rates of serious error while continuing to use the "polling division" model for the general election of 2015.
This review has identified a list of pragmatic legislative amendments and administrative modifications that would help to achieve significant compliance gains for the 2015 general election. Please see the following section of this report titled 'Recommendations' for details.
Beyond the New Model
This review's primary finding is that a fundamental redefinition of how voting services are provided is necessary for Canadian federal elections. An alternative model has been identified and could be available nationwide for the 2019 general election. In coming years, Elections Canada must stay fully alert to the high probability that any re-engineered voting approach is unlikely to endure as long as the veteran "polling division" model has.
But, what's next? Most compliance review participants thought voting via the Internet will be the next logical, evolutionary step in voting methods. Most expected Internet voting in federal elections to become a reality within their own lifetimes. Some indicated they were already "I-voting" in their local government elections.
In recent years, Elections Canada studied the ways and means to provide Internet-based voting. At one point it publicly stated its intent, contingent on Parliamentarians' approval, to pilot I-voting in a by-election after 2013.
However, during the summer of 2012 Elections Canada's senior management decided to scale back efforts on Internet voting and delay any I-voting pilot project until after the next general election. Reasons given for this decision included high costs and risks during a period of fiscal restraint, a critical lack of publicly available user "authentication" methods, and the fact that expected benefits and gains in efficiency are likely several years away. Modest research efforts are being continued.
Current Internet voting systems carry with them serious, valid concerns about system security, user authentication, adequate procedural transparency, and preserving the secrecy of the vote. However, evolving technology and societal expectations seem very likely to modify this equation in coming years.
A prudent, yet visionary approach will manage the inevitable evolution of voting arrangements to include electronic interactions between electors and institutions of democracy. During that evolution, it is best that proactive assessments are made of the compliance implications of proposed voting methods.
Return to source of Footnote 26 Supreme Court of Canada, Opitz v. Wrzesnewskyj, 2012 SCC 55; available online at:
Return to source of Footnote 27 See collected responses to question number four appearing in Annex H of the Compliance Review – Interim Report; available online at: http://www.elections.ca/content.aspx?section=res&dir=cons/comp/crir&document=h&lang=e
Return to source of Footnote 28 See Annex C to compare the "2011 General Election" column percentage error rates to those listed for the Victoria and Durham by-elections. Note that Calgary Centre did not have on-site Conformity Advisors and was treated as a statistical "control" group.
Return to source of Footnote 29 The Commission on Legislative Democracy, Province of New Brunswick, 2004, Summary of Recommendations, page 15. This document can be accessed at: http://www.electionsnb.ca/pdf/cld/CLDSummary-e.pdf
Return to source of Footnote 30 See Annex D for a fuller description of how the "New Brunswick model" works.
Return to source of Footnote 31 Preparations are currently underway to use the model in ten New Brunswick local government by-elections scheduled for May 13, 2013. The model will also be used in the next scheduled New Brunswick provincial general election — Election Day will be September 22, 2014.
Return to source of Footnote 32 Elections New Brunswick has expressed a willingness to partner with Elections Canada to accurately measure compliance levels experienced under their model. They believe the 10 pending by-elections of May 13, 2013 offer an opportunity to capture "live" measures of compliance as well as perform a post-election detailed audit review.