Ballot design research agenda: Where is design still disenfranchising voters?

Although Congress complains that they’ve seen little action from the Election Assistance Commission (EAC), it and the National Institute of Standards and Technology (NIST) have actually cranked out loads of excellent, deep, applied research about voting system security, usability, and accessibility over the last 5 years. We know so much more now about the differences plain language makes in instructions, ensuring that voters with disabilities can vote privately and independently, and where the scary parts of the chain of custody are. NIST is now in a cycle of validating test methods developed to fold into the voting system certification program. This undertaking has been huge. Much, much larger than anyone imagined.

Voting is a wicked problem. The work is laudable. (I think that’s what I’m doing, here.) Certainly, we’ve come a very long way from the butterfly punch card ballot in 2000. We have succeed with the Voluntary Voting System Guidelines (1.1) and the test methods being developed now in addressing many, many known problems. The VVSG now cleans up issues that we knew about in 2005. And, as voting systems catch up in their system architecture and user interface design, we’re going to see subtler problems surface.

There are still a number of unanswered questions. I’m not sure whether NIST is the body to answer them. But I would dearly love to see some money made available to look into ballot usability and accessibility issues that affect disadvantaged and minority voters. Here is my personal research agenda.

Ranked choice voting

RCV or Instant Run-off Voting (IRV) is very popular these days, as margins get tighter in top-of-ballot races at the state and local level. Election administrators like it because it saves the costs of conducting run-off elections. The problem is that voters don’t know how it works and don’t understand how their votes are counted. While this technique is used in many other countries, usually in parliamentary systems, the idea of having in mind a second and third choice for mayor or supervisor is indeed foreign to most American voters. I want to know how well voters understand RCV and what ballot design features work best to help voters use RCV most effectively.

Older adult voters

In case you haven’t noticed, there is a massive cohort of people turning 50 — the largest in history. For the next 50 years, Americans over age 50 will outnumber younger people. They will live much longer than their parents and grandparents. And, they will vote. But we don’t know a lot about this group of voters and how they perform with voting systems and different ballot designs. Most of the research has been with younger voters or people under age 60. How well do best practice ballots support older voters? What issues do older voters have using touch screen voting systems?

Multi-language optical scan ballots

Linguists and the news media will tell you that there are fewer and fewer languages spoken all the time. Election officials have got to be wondering where that data comes from, as jurisdictions urban and rural add one, two, or three languages to their ballots every few years. How many languages can you have on a ballot before all voters begin to have issues? What’s the optimum number of languages to control printing costs and keep voter mistakes to a minimum? What combinations of languages work best?

Vote by mail

Washington State and Oregon have already discovered that all-vote-by-mail elections are much less expensive than traditional elections with polling places and poll workers. There’s a case for increased participation. Voters like it because it leaves them more control about when they vote. But, as Minnesota learned in the last US Senate election, absentee ballots can make the difference in a very close race. But guess what: The error rate in many states is higher on absentee ballots than on ballots voted at the polls. There are a few theories about why that would be: there’s no verbal instruction from poll workers; there’s no one to ask questions of; there’s no one to model what you’re supposed to do; written instructions in absentee ballot packages are often not in plain language. What can design do to improve the usability of vote-by-mail packages? How can we ensure that voters who aren’t in the polling place vote the way they intend?

Poll worker training

Three million temporary election workers staffed the polls across the US for the presidential election in 2008. The average age of poll workers in the US is 72. In every election, the media gives us stories of election workers who break security seals, offer invisible ink to voters using touch screen voting systems, and give incorrect instructions to voters. Most of these mistakes are just that — mistakes. But the mistakes come from poor training. Most election workers never touch the voting system they’re supposed to be helping people use until Election Day. Other researchers are studying how best to train poll workers. I’m wondering, how does the training out there compare to training practices in industry? How well does the typical poll worker training meet best practice for training and procedural documentation? What does the best poll worker training material look like?


The usability and accessibility of voting has improved, but there is still so much to do

When the standards and the test methods are in place for certifying voting systems, we won’t be done designing. The voting experience, and the experience of election workers still has a lot of holes. And then perhaps we can begin to work on the back-end election administration systems. In the meantime, every Election Day, I pray for wide margins.


Leave a Comment