Electoral Districting in the City of Edmonton

Vol. 41, No. 6, November–December 2011, pp. 534–547 issn 0092-2102 eissn 1526-551X 11 4106 0534

http://dx.doi.org/10.1287/inte.1110.0544 © 2011 INFORMS

Designing New Electoral Districts for the City of Edmonton
Burcin Bozkaya
Sabanci School of Management, Sabanci University, Orhanlı-Tuzla, 34956 Istanbul, Turkey, [e-mail]

Erhan Erkut
Ozyegin University, 34662 Istanbul, Turkey, [e-mail]

Dan Haight
Centre for Excellence in Operations, University of Alberta, Edmonton, Alberta T6G 2R3, Canada, [e-mail]

Gilbert Laporte
HEC Montreal, CIRRELT, Montreal, Quebec H3T 2A7, Canada, [e-mail]

Every few years, the city of Edmonton, Canada must review and evaluate changes to its electoral district boundaries. The review process that was completed in 2009 resulted in modifying the district plan from a six-ward system with two council members in each to a single-member 12-ward system. The authors of this paper designed the redistricting plan. This paper describes the algorithm we applied to solve the problem and the decision support system we used. The algorithm is based on a multicriteria mathematical model, which is solved by a tabu search heuristic embedded within a geographic information system (GIS)-based decision support system. The resulting district plan meets districting criteria, including population balance, contiguity, compactness, respect for natural boundaries, growth areas, and integrity of communities of interest. This plan was formally approved as a city bylaw and used in the municipal elections in 2010. Key words: political districting; decision support system; GIS. History: This paper was refereed. Published online in Articles in Advance July 11, 2011.

T

he city of Edmonton, Canada is governed by a city council whose members are elected through a district-based municipal election system. In most North American cities, the elected representatives are from single-member

References: Bozkaya, B. 1999. Political districting: A tabu search algorithm and geographical interfaces. Doctoral dissertation, School of Business, University of Alberta, Edmonton, Alberta, Canada. Bozkaya, B., E. Erkut, G. Laporte. 2003. A tabu search heuristic and adaptive memory procedure for political districting. Eur. J. Oper. Res. 144(1) 12–26. City of Edmonton. 2000. City of Edmonton Bylaw 11388 (as amended)—A Bylaw to Set Council Composition and Ward Boundaries. Accessed October 25, 2010, http://webdocs .edmonton.ca/bylaws/c11388.doc. City of Edmonton. 2009. Ward boundary design policy C469A. Accessed October 25, 2010, http://www.docstoc.com/docs/ 26342469/Report-for-City-Council-February-17-2009-meeting. Duggal, S. 2009. Council doubles wards to a dozen. Accessed October 25, 2010, http://www.edmontonjournal.com/news/ Council+doubles+wards+dozen/1818789/story.html. Glover, F. 1986. Future paths for integer programming and links to artificial intelligence. Comput. Oper. Res. 13(5) 533–549. Li, Z., R. S. Wang, Y. Wang. 2007. A quadratic programming model for political districting problem. Accessed October 25, 2010, http://www.aporc.org/LNOR/7/OSB2007F49.pdf. Alayne Sinclair, City Clerk, City of Edmonton, 3rd Floor City Hall, 1 Sir Winston Churchill Square, Edmonton, Alberta T5J 2R7, Canada, writes: “I am pleased to write about an Operations Research implementation that took place in the City of Edmonton, in Alberta, Canada, to create electoral wards for electing members to the City Council of Edmonton. I am writing this letter to support the publication of our work in your journal, for which a manuscript has been submitted electronically using the online submission system. “Electoral districting is a crucial part of the City of Edmonton’s election planning responsibilities. The division of the wards can be a political sensitive issue but is required to ensure equal democratic representation is maintained as shifts occur in the City’s population demographics between elections. In January 2009, we made contact with the University of Alberta in Edmonton, where we knew previous academic studies on automated electoral districiting methodologies had been carried out. In response, a team of academics from the University of Alberta as well as other universities were gathered to help out election planners work on our ward boundary review. “The main task before the team of academics and the planners was to develop a new ward boundary map of Edmonton which would be used for the upcoming city council elections. The challenge was to modify the existing six-ward plan where 2 councillors are elected in each ward, into a 12-ward plan where 1 councillor is now to be elected in each ward. The districting process itself is typically a manual one that could take several months to complete before the results are presented to the public, but in this case, council had given a deadline of roughly one month. Bozkaya et al.: Designing New Electoral Districts for the City of Edmonton Interfaces 41(6), pp. 534–547, © 2011 INFORMS 547 To ease the process, council amended the districting bylaw to allow a wider deviation between the new wards—from 10 percent to 25 percent. “Using the GIS-based districting decision support system developed at the University of Alberta as a result of Dr. Burçin Bozkaya’s dissertation, we were able to develop several map alternatives in a matter of hours. Each of these alternatives respected the relevant criteria, and were then refined to reflect additional qualitative factors such as expected growth or communities of interest. The manual district-building tools available with the decision support system made these final refinements exceptionally easy. “The result of this process was a great success! By the end of March 2009, we completed the “mapmaking” activity which resulted in the new 12-district plan of Edmonton. The map solution exceeded all of the relevant criteria by a significant margin. Rather than requiring the more relaxed 25 percent population deviation, the ward populations deviated by at most 6 percent. Additionally, although in the past councilors had expressed reservations about maps that allowed boundaries to straddle the river, this solution respected the river with one small deviation consisting of a zero population industrial area. The new boundary maps were presented to City Council, and received first reading in April, 2009. The final approval was issued by the City Council in July 2009 and the 12 ward district map that was drawn by the team will be in effect from the 2010 General Municipal Election. “To list the overall benefits of the structured ORapproach that was used in this project: “(a) Objective criteria have been successfully considered and used by the decision support system, which left little further manual work on the resulting maps. The system even provided alternatives for ensuring solution quality such as the use of different measures of district compactness. “(b) Many months of our election planners’ time was saved due to the computer-based approach, without sacrificing any of the criteria relevant to the Council. “(c) The resulting maps were of much higher quality than we expected—exceeding all of the relevant criteria. “(d) The system was extremely easy to use, intuitive, and interactive, requiring no training for our planners. “(e) The decision support system is available to us for further use as needed. This makes our job much easier, considering the effort needed each time to create election maps. “To summarize, I would like to emphasize how an OR implementation such as this has had a profound effect on how we carry out one of our critical tasks at the City of Edmonton. “We sincerely hope this information is useful in pursuing the publication of our project work in your journal. If you need additional information about this project and the facts presented in this letter, please do not hesitate to contact me.”