Application of Lessons Learned in the Illinois EDC Project Toward Emerging Election Data Standards and System Guidelines

1
Application of Lessons Learned in the Illinois
EDC Project Toward Emerging Election Data
Standards and System Guidelines

• Scott Hilkert, Managing Principal
• NIST Workshop on a Common Data Format for
• Electronic Voting Systems. Oct 29 - 30, 2009

2
Presentatoin Topics

• Background of Illinois Election Data Collection
  project.
• Primer on Election Systems and Data Paths
  Between.
• Challenges Faced in EDC Project.
• Inconsistency in Contest Naming
• Inconsistency in Precinct Naming
• Suitability of Raw Machine Data for Reporting
• Relevance to Common Data Workshop

3
Illinois EDC Project Background

• Sponsored by EAC Grant awarded to 5 states.
• Objective Devise a means of automatic election
  data reporting for the 2008 General Election
• Objective Report election results data at the
  precinct level.
• Applied to the 2008 General Election
• Illinois Project Started late, Limited
  functionality in place for March 2009 reporting
  deadline
• Grant extended, final software to serve as
  proof-of-concept for Electronic Canvass.
• Focused on Data Paths 1 2 on the following
  slide.

4
Election Systems and Data Flow
5
Election Systems and Data Flow
6
Election Data Flow Highlights

• 1 Data exports from vote tabulation systems.
  (VTS)
• All 4 major Vendor systems represented in
  Illinois
• 2 Data exported from VR or election management
  systems (EMS)
• Typically the Statewide VR System
• 3 Data used to configure vote tabulation systems
  for each election.
• Lack of consistent naming and usage was a
  challenge
• 9 EMS / VTS System Boundary
• Inconsistency across boundary was a challenge

7
Election Data Collection Software

• Accepts Standard Export Files from each county
  tabulation system.
• Separate Translaction Plugin developed for each
  of 4 major vendor systems (Hart, ESS, Sequoia,
  Premier)
• Accepts Data Export from Statewide Voter
  Registration System.
• In bottom up states, Statewide VR data is
  derived from separate county systems.
• Local survey data is also combined.
• Data must be mapped and merged as it moves up the
  reporting hierarchy on following slide

8
Election Data Reporting Hierarchy
9
Challenge Inconsistency in Contest Naming

• Data from separate county systems must be
  correlated and merged.
• Requires mapping of contest and choice names to
  state standard conventions
• For example, Illinois 17th Congressional District
  spans 12 whole counties and portions of 2 more
  counties. Example names for this contest
• US House of Representatives
• Representative in the United States Congress,
• or simply Congress.
• Note that none of these identified this as the
  17th.

10
Challenge Inconsistency in Contest Naming

• A name mapping feature had to be developed (See
  screen shot on following slide)
• Also required for Party name and Choice names.
• This required significant staff hours to manually
  analyze and configure the data mapping.
• Had the naming conventions used to program the
  systems (Data Flow 3) been standardized, the
  manual mapping process could have been avoided.

11
Contest Mapping User Interface
12
Remedies? Inconsistency in Contest Naming

• Establish standard codes or short names for major
  contests (Replace old FIPS codes)
• Train and educate local election jurisdictions on
  VTS programming and usage conventions
• Incorporate standards and guidelines into VTS
  Vendor products and documentation.

13
Challenge Inconsistency in Precinct Naming

• Caused by the EMS / VTS system boundary discussed
  earlier.
• Limits ability to correlate voter registration
  data, ballot request data with votes cast data.
• A similar name mapping interface had to be
  developed.
• Very ineffective for 1000s of Precincts.
• Remedies similar to that of previous mentioned
  challenge.

14
Challenge Suitability of Raw Machine Data for
Reporting

• Small discrepancies encountered between VTS
  export files and official canvass reports.
• Provisional and other ballot types added to
  Canvass in external process.
• Date of export uncertain
• Some adjustments are inevitable.
• Remedies?
• Establish standard VTS usage guidelines for
  ballot types.
• Encourage use of VTS exports with Official
  Canvass
• Data standards must track manual adjustments
  separatly from original machine counts.

15
Challenge Suitability of Raw Machine Data for
Reporting

• Small discrepancies encountered between VTS
  export files and official canvass reports.
• Provisional and other ballot types added to
  Canvass in external process.
• Date of export uncertain
• Some adjustments are inevitable.
• Remedies?
• Establish standard VTS usage guidelines for
  ballot types.
• Encourage use of VTS exports with Official
  Canvass
• Data standards must track manual adjustments
  separatly from original machine counts.

16
Relevance to Common Data Workshop

• Inconsistent Data Format between the 4 systems
  was actually the least of the challenges faced.
• Data usage and naming conventions should be
  promoted along with any emerging data standard.
• EAC Sponsored Technology initiatives can help.