RCV Minneapolis Method Story
The hand-counting process for determining winners of
single and multiple seat offices in a Ranked Choice Election
Minneapolis MN - November 2009
Background
In 2006, the voters of Minneapolis approved a change from traditional balloting to Ranked Choice voting for municipal elections.
Minnesota Election law requires both federal and state certification of all electronic voting systems. Since there is no certified equipment that can conduct a Ranked Choice Voting election, the City of Minneapolis election staff had to hand-count the 2009 election.
Research and Planning
As part of the 2006 Minneapolis Instant Runoff Voting Task Force, election staff had completed research and reports that would guide the planning.
In December 2006, Minneapolis election staff met with Secretary of State-elect to seek support for creation of the Minnesota Ranked Choice Voting Issues Task Force. This task force had an open membership and included two sub-committees: Technical Advisory and Legislative/Rules Committee.
Minneapolis in 2009
The election planning included a dual-track schedule, as the Council could postpone implementation until a future election.
The 2009 Municipal election would have 22 offices on the ballots. In each precinct, there would be five different offices on the ballot.
During planning of 2009, election staff completed these tasks:
- Officially adopted Ranked Choice Voting as the name of the voting method to more accurately reflect the process voters use to rank candidates in single and multi-seat offices. In addition, "Ranked Choice" did not imply "instant" results from the process.
- Reviewed the newly created Ranked Choice Voting city ordinance for housekeeping changes needed
- Determined the best method to count the multiple seat offices that would comply with Minnesota law was the Weighted Inclusive Gregory Method (WIGM), which could produce the same election results in a recount.
- In May 2009, a "test election" was conducted for several purposes
- Develop the first-draft ballot design
- Work with different draft versions of materials to be used by election judges in the polling place to help voters
- Kick-off our voter outreach efforts by inviting various groups to experience a Ranked Choice Voting experience & share their feedback on the experience and the ballot itself
- Develop the method for hand-counting the single seat and multiple seat offices to determine the winner(s). Ballots were counted by combining all of the ballots for an office. For a turnout of 70,000, it was estimated that the hand-count for the 22 offices could take between 24 and 129 8-hour shifts of 39 counters.
- In June 2009, the council confirmed the Ranked Choice Voting election schedule
- The ballot design was improved based on the feedback from the Test Election and other community feedback
- In August 2009, the hand-count process was redesigned. A one-week "work-out" session developed the Minneapolis Method of hand-counting the ballots at the precinct level and using the precinct level data for analysis by office. Based on the Minneapolis Method, with a 70,000 voter turnout, it was estimated hand-counting the 22 offices would take 37 8-hour shifts with 102 election judges serving as counters and data entry staff. This new method would assure seating elected candidates on time.
- The training plan was designed to use at least one-half of the class-time on explaining Ranked Choice Voting to the election judges
- Hired an organization to conduct a impartial survey of voters, candidates and election judges concerning implementation
- Recruited a Historian to document the implementation
- In addition to the traditional precinct staffing, election judges were recruited and scheduled to do counting and data entry
The Minneapolis Method
The Minneapolis Method combines a hand-count with data analysis that avoids using an uncertified ballot counting program.
In the future, if certified equipment is developed and implemented for Ranked Choice Voting, the Minneapolis Method would be an efficient method for conducting a recount. In Minnesota, a recount must be conducted by hand.
Overall, determining winners based on the ballot data rather than sorting and re-sorting the actual ballots was easier and saved time. Some advantages of the Minneapolis Method include
- Ballots are counted by precinct rather than combining all ballots for the office. This avoids the problem of candidate rotation precinct by precinct that complicates sorting ballots.
- By precinct and office, ballots are sorted down to the unique 3-choice combination (including any possible write-in), counted and then documented on Precinct Ballot Summaries. For an office with 11 candidates, there can be up to 990 different 3-choice combinations…not including the write-ins.
- Counting offices by precinct allowed multiple offices to be counted simultaneously. Combining all of an office’s ballots together for counting would have only allowed one city-wide office to be counted at a time.
- When the counting of all offices on the precinct ballots is completed, the ballots can then be sealed and stored by precinct as required by MN law.
- Providing a means to verify that the same number of votes was counted for each of the five offices on the precinct ballots
- Counting by precinct meant that many precincts could be counted simultaneously which allows expanding the counting process if necessary.
Implementing the Minneapolis Method
The Tabulation Center
The Minneapolis Elections Warehouse was converted to a Tabulation Center for counting, data entry and data analysis. Amenities included new vending machines in the break room, improved heating, ergonomic chairs, a cleaning crew and nametag racks to hold color-coded nametags to indicate political party affiliation.
Human Resources
Counters and Data Entry judges were selected from among Chair and Assistant Chair Judges as well as top performing Team Judges as recommended by Chair Judges. Every day as judges arrived at the Tabulation Center, they picked up their name tags and timesheets, signed in with staff and were directed to a precinct pod seated next to a Counter with a different color-coded nametag.
Supply and Transport
A Supply and Transport Crew was responsible for ballot security and delivering color-coded supplies to each Precinct Pod. The supplies were color-coded to help with organization and visual management.
Some highlights of color-coding of supplies
- A different color was used for each of the five offices for both the name placards and also the Ballot Summaries.
- Beige was used for Precinct Supply Lists, Duty Cards and timesheets.
- The only white paper allowed at the Precinct Pod was the actual ballots.
Other notes on organization
- Tables were taped off to create different spaces.
- Each pod had three sets of name placards with the candidate names to label their sorting area.
- A three-letter abbreviation of each candidate name was taken from the first three letters of a candidate’s last name. Using the 3-letter abbreviation saved time for Counters writing and the abbreviations were also built into the Data Entry documents.
- Pods had two color-coded slips used to silently request assistance with supplies or process questions, which helped to reduce the background noise.
Sorting and Counting
Precinct pods for counting were designed using a combination of tables to hold the ballot length. Each pod was staffed with six Counters, three teams of two judges of different political parties. A crew of up to six roamed the floor to help with on-going training and to answer questions.
Counters at each precinct pod
- Staged the ballots for the precinct (sorted them all the same direction)
- Inspected each ballot for voter errors specific to Ranked Choice Voting and accounted for these errors
- Sorted the ballots for each office down to the unique 3-choice combination (including all write-ins), counted the ballots with that combination and completed a Ballot Summary for each unique combination in the precinct
When a precinct office was completely counted, the Supply and Transit Crew would review the Ballot Summaries for completeness and then deliver them to the Data Entry Teams.
Counting each precinct took between 5.5 hours to 8.0 hours, depending on the number of ballots and ballots with voter errors. Counting began Wednesday November 4 and was completed Friday, November 13.
Data Entry
Data entry judges working at computers as a team of two judges of different parties, entered the precinct level data from the Ballot Summary sheets into the computer. The team also double-checked their work. A data analysis team then verified the data.
With six teams of two judges each, data entry of the ballot summaries for a precinct office took an average of one-half hour, depending on the number of ballot summaries. Data entry began Wednesday November 4 and was completed Friday November 13.
Data Analysis
Data Analysis was conducted using a dual track system. Each of the two teams consisted of a lead analyst and an observer. Both teams did analysis on the same office, performing the exact same steps and calculations, and then verified their results with each other.
Data analysis of council offices (which have between 8 to 11 precincts) took between 50 minutes to 1 ½ hours. Analysis of the Park District offices (which have between 19 to 24 precincts) took 50 to 70 minutes. Determining the winning candidate for the city-wide office of Mayor (131 precincts) took 4 hours and 20 minutes for one round.
Data analysis for the two city-wide multiple-seat offices with five or six rounds took over eight hours each.
Last updated Sep. 27, 2011