Friday, November 18, 2005

No Complaints

I'm in Washington, D.C., at the moment, attending the annual meeting of the African Studies Association, but I was excited to find something in my email box late last night and wanted to bring it to the attention of my readers.

The Hugo Commission (formally known as the General Assembly's Joint Subcommittee to Study the Certification, Performance, and Deployment of Voting Equipment) will be meeting for the last time next Monday, November 21, when it is expected to make its recommendations regarding Virginia's voting equipment of the future.

I have written about the Hugo Commission's proceedings before (most notably on August 22, August 26, indirectly on August 30, and September 6), so this might be -- depending on the joint subcommittee's recommendations -- the last time I revisit this issue.

What I have to offer today is the result of a survey undertaken by the staff of the Charlottesville Electoral Board. And I think the survey speaks for itself, so I will reprint here, in full, the report that was submitted to the Hugo Commission in anticipation of Monday's meeting. (My formatting here is somewhat different than that of the printed report that the subcommittee received.)

For those who don't want to wade through the whole thing, here's how I read it: Virginia voters are, on the whole, satisfied with the electronic voting systems currently in use. Only a very few have issued any complaints to local election officials. Out of 2 million voters who went to the polls on November 8, only a few dozen, at most (based on the results of our survey), expressed reservations about the current machines to their local officials. Now that doesn't mean that others may not have complained to their friends, or that they didn't express their views to advocacy groups or political party officials -- it only means they took only a minuscule number of complaints, compared to the universe of active voters, to their General Registrars or other election officials.

Are the results of this survey dispositive? No, but I'd say they provide a strongly persuasive indication that Virginia voters are fundamentally satisfied with the electronic voting equipment their localities use.

If my arithmetic is correct, in voting jurisdictions constituting just over 50 percent of localities that use DREs (the rest use optical scan systems, except Highland County, which uses paper ballots, and Virginia Beach, which used its old-fashioned lever machines for the last time in this election), there were 26 complaints about DRE equipment and somewhat fewer complaints about, questions about, or requests for a VVPAT (voter-verifiable paper audit trail). (The latter category is a bit vague, since a couple of registrars responded to the survey question with "several complaints" or "several questions"; a more formal survey would have ferreted out firmer numbers.) The largest number of complaints came in Falls Church, which was using the Hart eSlate system for the first time in this election. I am certain this number will fall in subsequent elections in that small city.

This suggests that even if we surveyed all the DRE-using counties and cities in Virginia, the number of complaints would hover around 50 to 60 out of perhaps a million or more voters in last week's statewide elections. That's what I would call a vote of confidence (no pun intended) in the voting systems Virginia localities use.

I'd be interested in hearing responses from the people who did complain and ask: Should their minority opinion prevail in an issue that will require millions of dollars of investment in new voting equipment (should the General Assembly legislate such a requirement) in an era of government budget austerity?

The full report follows:

Report to Members of the Joint Subcommittee to Study the Certification, Performance, and Deployment of Voting Equipment

November 17, 2005

Results of a Survey of General Registrars Following the November 8, 2005 General Election, Regarding the Performance of Direct Recording Electronic (DRE) Voting Equipment


Beginning on November 14, 2005, the staff of the General Registrar and Electoral Board of the City of Charlottesville conducted an informal survey of the Registrars of the Commonwealth of Virginia regarding the performance of DRE voting equipment, and the nature of feedback from voters regarding the use of such equipment. In many localities across the state, this was the first major election in which such equipment had been utilized, due to the imminent January 1, 2006, deadline placed on localities by the Help America Vote Act (HAVA), by which time all voting equipment must meet federal guidelines.

The general level of feedback both from the electorate, and from registrars and election officials was overwhelmingly positive, with few serious complaints regarding the performance of DRE voting equipment or the lack of a Voter-Verifiable Paper Audit Trail (VVPAT) used in conjunction with such equipment. Across the state, disabled and elderly voters responded very positively regarding the DREs “ease of use,” and their capability to allow blind and other disabled voters to cast a secret ballot, for the first time in the case of many voters. Most of the complaints regarding the DRE machines were due to operator errors or to calibration errors in systems utilizing touch-screen technology. There were complaints regarding the lack of a paper “receipt” from voters after casting their votes electronically, but in nearly all cases these were quelled after the Officers of Election or staff of the Registrar’s Office explained the many redundancies built into DRE machines and their ability to print a paper record of each vote cast if necessary.

The following is a summary of the feedback from the 42 General Registrars who responded to the survey. Their responses are organized by region of the state, and they include information on the type of voting equipment in use and the number of registered voters in each locality. Each respondent was asked to provide feedback on the following three questions:

(1) Were there any voter complaints on or after Election Day 2005 regarding DRE equipment? If so, how many complaints were there?

(2) Were there any voter complaints on or after Election Day 2005 regarding DRE performance? If so, how many?

(3) Were there any voter requests on or after Election Day for adding a voter-verifiable paper audit trail (VVPAT) to current equipment?

Tidewater

Accomack County
Type: Rural
Registered Voters: 20,232
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Caroline County
Type: Rural
Registered Voters: 14,832
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Charles City County
Type: Rural
Registered Voters: 4,742
Voting Equipment: Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Essex County
Type: Rural
Registered Voters: 6,198
Voting Equipment: Sequoia Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Middlesex County
Type: Rural
Registered Voters: 7,111
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT

New Kent County
Type: Rural
Registered Voters: 10,269
Voting Equipment: iVotronic – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE
No Requests for VVPAT


Northampton County
Type: Rural
Registered Voters: 8,379
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Suffolk City
Type: Urban
Registered Voters: 46,516
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Sussex County
Type: Rural
Registered Voters: 6,569
Voting Equipment: Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Williamsburg City
Type: Urban
Registered Voters: 6,086
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE
No Requests for VVPAT

Central Virginia

Albemarle County
Type: Rural
Registered Voters: 59,167
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Appomattox County
Type: Rural
Registered Voters: 9,234
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Bedford County
Type: Rural
Registered Voters: 41,975
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
4 Complaints about DRE
1 Request for VVPAT


Botetourt County
Type: Rural
Registered Voters: 21,074
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Campbell County
Type: Rural
Registered Voters: 31,002
Voting Equipment: Patriot – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE
Several Requests for VVPAT


Charlotte County
Type: Rural
Registered Voters: 8,143
Voting Equipment: Patriot – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Dinwiddie County
Type: Rural
Registered Voters: 15,276
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE
No Requests for VVPAT


Franklin County
Type: Rural
Registered Voters: 30,130
Voting Equipment: Sequoia – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Henrico County
Type: Suburban
Registered Voters: 172,471
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT

Henry County
Type: Rural
Registered Voters: 34,113
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Lexington City
Type: Urban
Registered Voters: 3,295
Voting Equipment: Patriot – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE
Several Questions about VVPAT


Louisa County
Type: Rural
Registered Voters: 17,492
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE
No Requests for VVPAT


Lunenburg County
Type: Rural
Registered Voters: 7,330
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Madison County
Type: Rural
Registered Voters: 7,975
Voting Equipment: Patriot – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT

Richmond City
Type: Urban
Registered Voters: 100,989
Voting Equipment – WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
2 Complaints about DRE
No Requests for VVPAT


Waynesboro City
Type: Urban
Registered Voters: 11,280
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Northern Virginia

Arlington County
Type: Suburban
Registered Voters: 125,296
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE
5 Requests for VVPAT


Fairfax City
Type: Urban
Registered Voters: 14,100
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE
1 Question about VVPAT

Falls Church City
Type: Urban
Registered Voters: 7,404
Voting Equipment: eSlate – Non-Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
Approx. 5-10 Complaints about DRE equipment
3 Requests for VVPAT


Fauquier County
Type: Rural
Registered Voters: 39,186
Voting Equipment: Edge/Accuvote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Frederick County
Type: Rural
Registered Voters: 40,087
Voting Equipment: iVotronic – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Manassas Park City
Type: Urban
Registered Voters: 5,272
Voting Equipment: Optech Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Prince William County
Type: Suburban
Registered Voters: 188,034
Voting Equipment: Optech Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Warren County
Type: Rural
Registered Voters: 20,595
Voting Equipment: WinVote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Winchester City
Type: Urban
Registered Voters: 13,847
Voting Equipment: iVotronic – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Southwest Virginia

Bland County
Type: Rural
Registered Voters: 4,396
Voting Equipment: Patriot – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Buchanan County
Type: Rural
Registered Voters: 16,936
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Dickenson County
Type: Rural
Registered Voters: 11,505
Voting Equipment: Optech Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT

Lee County
Type: Rural
Registered Voters: 15,882
Voting Equipment: AVC – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Montgomery County
Type: Rural
Registered Voters: 45,703
Voting Equipment: WINvote – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
4 Complaints about slowness of Audio ballot
No Complaints about VVPAT


Pulaski County
Type: Rural
Registered Voters: 20,256
Voting Equipment: iVotronic – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT


Radford City
Type: Urban
Registered Voters: 7,442
Voting Equipment: Edge – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
1 Complaint about DRE touch screen calibration
No Requests for VVPAT


Roanoke City
Type: Urban
Registered Voters: 55,869
Voting Equipment: iVotronic – Touch Screen DRE

Feedback on the Nov 8, 2005 Election:
No Complaints about DRE or Requests for VVPAT
Unless I have something to say about the final meeting of the Hugo Commission next Monday, my next election-related blog posting will most likely be about the Deeds-McDonnell recount, which should begin around December 19.

6 comments:

  1. Butbutbut ... if even one voter thinks his vote hasn't been properly tallied, then shouldn't we scuttle the entire election process to honor the request? After all, it's not a real election until absolutely everybody is 100% satisfied.

    ReplyDelete
  2. While I disagree with the Anonymous commenter that we need to satisfy everyone, I take issue with this statement: "There were complaints regarding the lack of a paper 'receipt' from voters after casting their votes electronically, but in nearly all cases these were quelled after the Officers of Election or staff of the Registrar’s Office explained the many redundancies built into DRE machines and their ability to print a paper record of each vote cast if necessary."

    I took issue with there being no paper receipt, but it was a fait accompli; the choice given to me was either to vote on the electronic system, without paper trail, or not vote at all. In that circumstance, sure, I'm going to give in, stiffle my protest, and vote.

    I believe that there were many who didn't complain to election officials, but who complained to others. Of course, it's difficult to gather and credibly present anecdotal evidence.

    ReplyDelete
  3. I love my car. By Rick's logic, I'll never have to take my car in for an oil change, get my tires rotated, or even wash it. The sheer affection and good feelings I have about my car will be more than enough to keep it working reliably.

    The right metrics to use are reliability and accuracy. By these measures, paperless machines in Virginia failed in the last election. See

    http://www.cs.duke.edu/~justin/voting/dat/VA/2005/per_race.html

    last updated the evening of November 17th.

    If you examine the undervote rate -- that is, the number of people who went to the polls but didn't vote in a given race -- for the Lieutenant Governor and Attorney General you see that the average rate for optical scan systems was 2.8% for each race. The undervote rate using DRE machines, though, was 50% higher in each case: around 4.0 - 4.2%. A statistical analysis of varaince shows that the only significant factor to explain this increase in voter apathy is the technology used, NOT "voter satisfaction" or even locality size.

    In short, it is likely that touchscreen machines missed about 19,000 votes for Lieutenant General, which is about the margin of victory, and about 15,000 votes for Attorney General (over 40x the margin of victory).

    But, boy, aren't we lucky that people aren't complaining about that?

    -Justin Moore

    ReplyDelete
  4. Funny, Justin, I've been looking at these same figures you've cited. They don't prove what you think they do -- if anything, they prove the opposite. For instance, when you compare the results of Bedford Country, which uses electronic voting machines, with Brunswick County, which uses optical-scan ballots, you find that Brunswick's optical scan has a slightly larger percentage of undervotes.

    Yet in each precinct report, regardless of the type of machine, a small but statistically significant number of Virginians (three to four percent) appear to have voted in the Governor's race and abstained from the others. This is perfectly consistent with known voter behavior.

    In other words, Justin, you've drilled a dry hole.

    ReplyDelete
  5. Funny, Justin, I've been looking at these same figures you've cited. They don't prove what you think they do -- if anything, they prove the opposite. For instance, when you compare the results of Bedford Country, which uses electronic voting machines, with Brunswick County, which uses optical-scan ballots, you find that Brunswick's optical scan has a slightly larger percentage of undervotes.

    Wow. I mean, just wow. That's an absolutely amazing strawman argument. Find one optical scan-using locality that has a relatively high percent of undervotes, and use it to disprove your point.

    Fine, let's look at Brunswick County. Of the 31 localities using optical scan machines, it has the 6th-highest undervote rate for Attorney General, at 3.40 percent. That means 16% of localities using optical scan machines had an undervote rate higher than 3.40%.

    What about DRE machines? A staggering 52 out of 92 localities using DRE machines -- a full 56 percent -- had an undervote rate for Attorney General higher than Brunswick County. And while Bedford County had an undervote rate of 2.14%, Bedford City -- also using DRE machines -- had an undervote rate of 5.21%.

    For the full picture of the undervote rates and their relation to technology, see the graphs at

    http://www.cs.duke.edu/~justin/voting/dat/VA/2005/Attorney_General.png

    and

    http://www.cs.duke.edu/~justin/voting/dat/VA/2005/Lieutenant_Governor.png

    Yet in each precinct report, regardless of the type of machine, a small but statistically significant number of Virginians (three to four percent) appear to have voted in the Governor's race and abstained from the others.

    It's funny that you use the phrase "statistically significant" in your post. I can also use that phrase. For example: There is a statistically significant difference between the voting behavior of Virginia localities using optical scan machines and localities using paperless machines. And a large amount of my "day job" work involves analyzing the performance of different technologies and modeling the behavior of these technologies and finding statistically significant (and insignificant) distinguishing factors.

    I'm willing to put my analysis of state-wide voting trends up for peer review by professional statisticians. Are you?

    -jdm

    ReplyDelete
  6. Rick, while I am sure you are sincere (pun intended) in reporting that there were no complaints, I find it interesting that the survey didn't include Roanoke County, where a lot of complaints were recorded by the local television station and newspaper.

    Here's what WDBJ-7 reported:

    "November 8, 2005
    Voters report problems with voting machines in Roanoke Co.


    Election day voting is going on across the Old Dominion, but everything has not gone smoothly in Roanoke County.

    News 7 has received calls from several voters in at least four different precincts who say their votes for Tim Kaine were not recorded or took several attempts to go through.

    They contend the electronic touch screens repeatedly indicated they were voting for Republican candidate Jerry Kilgore instead of registering their intended vote for his Democratic opponent Tim Kaine.

    Roanoke Co. Registrar Judy Stokes says she doesn't want to say the problem is operator error on the part of the voters, but she points out the touch screens are sensitive. She says anyone who is having difficulty voting should ask one of the poll workers for assistance.

    State election officials have been told of the problem. They believe if there is a problem, it could have been caused by the way the machines were stored.

    The Kaine compaign is reportedly watching the situation in Roanoke County."

    http://www.wdbj7.com/Global/story.asp?S=4089899

    Even if one voter in each of the four precincts complained that would be more complaints per capita than the registrars reported!

    As Justin concludes in research on his his website, "In short, even if a paper ballot is used infrequently and verified correctly a small percentage of the time, is leads to an incredibly high likelihood that machine errors will be caught. The cost-benefit tradeoffs of a voter-verified paper ballot trail are tremendous."

    As the Association for Computer Machinery commented, after polling its membership, "voting systems should enable each voter to inspect a physical (e.g., paper) record to verify that his or her vote has been accurately cast and to serve as an independent check on the result produced and stored by the system. Making those records permanent (i.e., not based solely in computer memory) provides a means by which an accurate recount may be conducted. Ensuring the reliability, security, and verifiability of public elections is fundamental to a stable democracy. Convenience and speed of vote counting are no substitute for accuracy of results and trust in the process by the electorate."

    Golly, I always thought libertarians wanting liberty!

    Beth
    Roanoke, Virginia

    ReplyDelete