PS 120

1
Lecture 1

• PS 120

2
Outline

• Distortions and political representation
• Measuring errors in the electoral process
• The VTP and a briefing of the work so far
• Introduction to issues of voting technology

3
Political Representation

• What does it mean to represent?
• Direct v. indirect
• Substantive v. descriptive
• Key assumption is that preferences are translated
  into public policy in a transparent and direct
  manner

4
A Naïve Model

• Utility for candidate J
• Uij F(Vi - Cj)
• Voters cast ballots for J if utility for J is
  greater than utility for other candidates
• Election process is to count all votes, the
  candidate with the most votes wins office
• Note the assumptions here

5
Failures of Representation?

• Citizens do not register
• Citizens do not vote
• Citizens not informed, candidates obfuscate
• Citizens who try to vote have their preference
  distorted or not counted by the process (e.g. bad
  voting machines)
• Distortion of preferences through institutions
  (e.g. Electoral College)

6
Lets Start With Institutions The Electoral
College

• Origins of Electoral College
• Founders wanted indirect election of president.
• Original system allowed state legislators to
  determine method of selecting delegates to
  Electoral College
• Each state allocated one delegate for each U.S.
  Senator and House of Representatives member
• Winner became president, runner-up got the door
  prize (becoming vice president)

7
Evolution of the Electoral College

• 12th Amendment forced Electoral College to cast
  separate ballots for president and vice
  president House of Representatives to break
  ties.
• During 1800s most states move to direct
  selection of electors and to winner take all
  rules.

8
Current System

• Each state gets 1 elector per Senator and House
  member
• Members of Congress and federal employees cannot
  be electors
• Political parties select lists of electors
  voters cast ballots (directly or effectively) for
  electors.

9
Current System

• Whichever partys slate of electors gets most
  votes wins (exceptions are Maine and Nebraska
  where electors are selected by combination of
  statewide outcome (2) and popular vote in each
  Congressional district)
• On Monday following second Wednesday of December
  state electors meet in state capital to cast
  electoral votes (at least one of their votes must
  be for a person from outside their state)

10
Pros and Cons?

• Cons
• Produces minority presidents
• Does not reflect popular will
• Nothing stops elector defections
• Decreases voter turnout, disenfranchises voters
  in some states
• Enhances odds of two-party system

• Pros
• Maintains federal checks/balances
• Increases minority power
• Requires geographic distribution of support
• Enhances odds of two-party system

11
Should We Keep It?

• Many calls after the 2000 and 2004 presidential
  election to eliminate the Electoral College
• Would require Constitutional amendment
  difficult!
• Current proposals intra-state compact for
  popular vote winner California to break
  winner-take-all tradition?
• What kind of system would we replace it with, and
  would that necessarily be superior to what we now
  have?

12
Other Distortions Eligibility and Registration

• Who voted in the 2004 Presidential election?
• 215,694,000 voting-aged (VAP)
• 197,005,000 CVAP
• 142,070,000 registered (72.1 of CVAP)
• 125,796,000 voted (63.8 of CVAP)
• Are the preferences of those who vote identical
  to those who do not? To those who are not
  registered? To those who are not eligible?
  Should we care?

13
Distortion Errors and Lost Votes

• How can we measure the extent to which voters try
  or want to vote and are foiled
• By mistakes early in the process?
• Mistakes when they go to vote?
• Voting machine problems?
• Critical research question!

14
How Can We Measure Machine Errors?

• Residual vote analysis the deviation between
  votes counted and votes cast.
• Most jurisdictions report these data, and such
  data goes back in time.
• While consistent across time and space,
  imperfect do not necessarily partition machine
  from voter errors.

15
How Can We Estimate Other Errors?

• Voters denied service
• Those who show up to vote, but dont cast a
  ballot because of long lines or polling place
  problems
• Those who try to vote, but cant because of
  registration problems
• Census CPS data asks registered non-voters why
  they didnt vote

16
Lost Votes Technology
17
Lost Votes Registration
18
Lost Votes Polling Places
19
Lost Votes
20
Lost Votes?

• Due to problems with absentee voting UNKNOWN
• Due to fraud or other irregularities UNKNOWN
• Due to voter suppression and intimidation
  UNKNOWN

21
The Starting PointFlorida 2000
Why did this crisis occur?

• It was a close election in a pivotal state in a
  Presidential race
• No plan for recount
• Poorly designed ballots? Yes, in a few counties
• Poor performance of technology? Yes Many lost
  votes
• Was this typical? Yes
• Can this happen again? Yes (Ohio 2004?)
• Could the situation be improved? Yes
• Have federal and state reforms since 2000 helped?
  MAYBE

22
Florida, By the Numbers

• There were 6,137,937 million ballots cast, and
  5,959,038 Presidential votes counted
• That means there were 179,855 ballots that were
  blank, spoiled or otherwise were not counted ---
  2.93 of ballots cast
• Bush won Florida by only 537 votes!

23
Florida Optical Scans and Overvotes
                                                
             Close

• Voter here made a variety of marks on an op-scan
  ballot
• This is an example of an overvote

24
Florida Punchcards and Undervotes
                                                
             Close

• Here is a Votomatic punchcard seen illuminated
  from behind
• Clear votes (the entire box punched)
• Possible votes, or undervotes (the boxes are
  partially punched)

25
Florida Uncounted Ballots and Voting Systems
26
But there were other problems in Florida

• Long lines at polling places
• Severe voter registration issues
• Florida had hired an outside company to examine
  its voter registration rolls before the election
  for duplicates, deceased persons, and felons
  182,000 identified, at least 8,000 of these were
  incorrect and many were purged.

27
The Caltech/MIT Voting Technology Project

• Began December 15, 2000
• Multi-campus, multi-disciplinary effort
• Produced major report in July 2001, Voting What
  Is, What Could Be
• Report and research product provided foundation
  for passage of federal election reform, Help
  America Vote Act and legislation in many states

28
Caltech-MIT Voting Project
Purpose of the project
Assess and evaluate current state of
system Develop recommendations for
improvement Why? Restore voter confidence and
reduce the likelihood of another election crisis.
What we are actually doing?

• Study of current system performance,costs,
  financing, and practices
• Short term fixes
• Longer term recommendations

29
Voting Equipment in the 1980 and 2000 Elections
30
1 million lost votes In 2000
31
Why Uncounted Votes?

• Voting machines
• machine failures
• poor maintenance
• poor ballot design
• poor design of voter interface

• Voting behavior
• apathetic voters (poorly informed, not
  connected, unfamiliar with voting technology)
• politicized voters (minor party members,
  protest voters)

32
Voting Machines are Just the Tip of the Iceberg!

• Many of the voting systems in use today are
  problematic
• High uncounted vote rates
• Evidence of variation in uncounted vote rates,
  holding voting machine type constant, is
  normatively and legally problematic
• But, there are other pressing problems

33
Remember Lost Votes!
34
Transition 1 Voting Machines

• Nationwide phase-out of poor performing voting
  machines
• Movement from punchcards, paper and lever
  machines to optical scanning and electronic
  voting systems
• Transition has reduced uncounted votes

35
Uncounted Votes in Seven Largest Florida Counties
Diamonds indicate counties transitioning from
punchcards to optical scan other counties
transitioned from punchcard to DRE
36
Georgia, 1998-2002
37
The Great E-voting Debate

• Concern rising in certain sectors about the
  security and integrity of electronic voting
• Calls for integration of voter verified paper
  audit trails (VVPAT) with electronic voting
  machines
• Source of controversy and debate unclear whether
  VVPAT is the solution, or even a partial
  solution.

38
Current Environment
Source EDS, October 17, 2008
39
Lost Votes due to Problems withRegistration and
Polling Place Practices

• CPS 3-8 of registered non-voters had
  registration problems (1-3 million lost votes in
  2000)
• CPS 1-3 of registered non-voters had polling
  place problems (long lines and other problems
  lost 1 million votes in 2000)

• Recommendations
• Computerized registration information at
  precincts
• Provisional ballots
• More pollworkers and training

40
Transition 2 New VR Practices

• Help America Vote Act 2002
• Statewide VR files
• First-time registrations provide ID
• Files checked against SSA, DMV lists
• Potential for better accessibility AND security
• But what will REALLY happen in 2006 and 2008 when
  they are implemented across the nation?

41
Transition 3 Conveience voting Transition
3 Absentee, early and supercenter voting

• On the rise
• On-demand vs. cause
• Nearly 30 in California this year
• Oregon went to 100
• Very little effect on turnout (small and
  positive)
• Supercenter experiments
• Security vs. access argument
• Also true in voter registration debates
• Internet voting, other convenience voting ideas
  (telephone, SMS)
• In the future
• System security issues
• Field testing
• Too early for widespread implementation

42
Rise of Convenience-Based Absentee Voting
Absentee ballots as a of votes cast
43
A Tradeoff Between Accessibility and Fraud?
California Secretary of States Election Fraud
Investigation Unit Database, 1994-2002.
44
Election Fraud Convictions In California,
1994-2002
There were 60 election fraud convictions,
1994-2002
45
Where Have We Made Progress Since 2000?
Maybe?
Some progress
?
46
Non-precinct Voting
Bottom line Rate of non-precinct voting is rising
rapidly Absentee voting for those who have no
reasonable alternative is necessary Absentee
voting on demand is a convenience,
BUT... Significant potential for large-scale
fraud and vote-buying detection problem Early
voting is a better alternative to on-demand
absentee voting