Presentation to the Committee on Future of Voting: Accessible, Reliable, Verifiable Technology at the National Academies of Science, Engineering, Medicine (NASEM)
1. Defects by design | 1
Defects by design
Ballots that fool voters
Whitney Quesenbery
Center for Civic Design
@civicdesign | @whitneyq
2. EAC / Design for Democracy Effective Designs for the Administration of Federal Elections (2007)
Better Ballots (2008), Better Design, Better Elections (2012) and Design Deficiencies and Lost Votes (2011)
Field Guides to Ensuring Voter Intent, Vol 01 Designing Usable Ballots
Anywhere Ballot
3. A failure of intent caused by
Typography
Ballot layout
Marking mechanisms
Instructions and messages
and election codes that require poor design
5. Defects by design | 5
43% of adults in the US read
at basic or below basic levels
U.S. National Assessment of Adult Literacy http://nces.ed.gov/naal/kf_demographics.asp
14% 29% 44% 13%
30
Million
63
Million
95
Million
28
Million
6. Defects by design | 6
If you can’t read the ballot effectively,
you can’t vote accurately
7. Defects by design | 7
If you can’t read the ballot effectively,
you can’t vote accurately
8. Defects by design | 8
If you can’t find the correct marking location,
you can’t vote accurately
9. Defects by design | 9
If you don’t know how voting works,
you can’t vote accurately
17. Defects by design | 17
Clear, minimal, helpful instructions
in the right place
18. Defects by design | 18
Clear, minimal, helpful instructions
in the right place
19. Ballot design can help
voters be successful
Follow proven election design principles
Meet voters' needs for easy interaction,
plain language, and clear design
Test with voters
Discuss: design defects that lead to votes not being counted and offer recommendations that might be implemented to correct for this.
Thank all the work
All of it. Ballots are riddled with simple basic design problems from the overall layout of the ballot to instructions.
I’m not even going to talk about accessibility much, because others are going to cover that. Except to say that almost all of the problems I’m going to discuss affect people with disabilities, low literacy, low English proficiency and most of all civic literacy gaps disproportionately
So they not only don’t meet all of our various laws intended to ensure equal access to the ballot, but they exacerbate the disparities.
What makes this so aggravating is that the solutions we can recommend are not new and not even cutting edge design. We have known about some of the solutions for at least 17 years – but really much more. Let’s start with typography.
U.S. National Assessment of Adult Literacy http://nces.ed.gov/naal/kf_demographics.asp
Below basic – only the most simple and concrete reading skills
Basic – able to manage everyday tasks
Intermediate – moderately challenging activities like consulting reference material
Proficient – interpreting text, comparing viewpoints
Below Basic: 14% or 30 million people
Basic: 29% or 63 million people
Intermediate: 44% or 95 million
Proficient: 13% or 28 million
Text size – complaints galore in 2010 when NYC ballots didn’t even reach the level of the VVSG
Clutter – party emblems, all capital letters, row and column identifiers, instructions in the wrong place, voting ovals to the right of names
Our demonstration of a ballot that meets proposed Voter Friendly Ballot Act
Shorter, clearer instructions at the top of the ballot where they will be read
Reduced clutter leaving room for larger text
Shading and lines help separate contests, even with the full face restriction
Ovals to the left and immediately next to the candidate name
2008 testing with 100 people
More than 14,000 missing votes for CD 13 in Sarasota County (14% of voters).
Charlotte County had 2.5% residual votes
Consistency, headings, one contest per page if that’s the usual standard
Of 8.5M voters, 236K overvotes – 3%
Los Angeles: 2M votes 150K overvotes – 7%Madera: 27K votes, 850 overvotes – 3%
Santa Cruz: 95K votes, 985 overvotes – 1%
Why the fetish for marking on the right, where it has an uneven relationship to the names.
Single columns better!
Ballot A – 116 errors - 9 overvotes
Ballot B – 92 errors – 4 overvotes
Not significant, but a clear trend, also stronger preference.
2008 testing with 100 people