The Human Factors Science Behind Mail-In Voting…The Surprising Complexity Problems Impacting Voter Participation
Electronic Voting Machines, Mail-in Ballots vs. U.S. Democracy: There has been a lot of media buzz over the last few years about how voting tech will save democracy. In this misplaced view of the future, electronic voting systems were pushed and prodded into use across the country at precincts large and small. This was done in the interest of improving overall voting consistency and reliability. The ugly truth about this incursion of tech into the most fundamental aspect of our democracy is that almost none of these electronic voting systems have been developed based on even the most basic aspect of human factors science. They are, in ways large and small, far worse in terms of actual usability than the lever and crank systems of the past. That’s right – the new tech-based voting machines had no transfer of learning analysis, no cognitive task analysis, no interface modeling, no cognitive workload analysis, no human interaction use case analysis, no formative testing, no summative final user testing, no risk analysis, and no structured needs analysis. The human component of electronic voting was just not on the ballot.
How Did We End Up With Broken Voting Systems? As a country, we are in this sorry state of affairs with our voting systems because our elected officials, congress, and state voting boards allowed software and hardware engineers to simply use their personal intuition to design electronic voting systems. For anyone schooled formally in human factors science, such an approach is well understood to guarantee that such systems can be used by only one primary user group: the engineers that designed the machines. The rest of those poor souls who have to actually set-up, test, validate, monitor, fix, update, and teach voters how to use these travesties of tech were left to swing in the wind. This is a fundamental violation of how the basic science of human factors research works. Such a process says usable systems must be designed for the RANGE of users’ cognitive and functional limitations. This did not happen with technology-based voting systems. But there we have it, billions of dollars spent on basically unusable voting machines that require far more on-the-ground support than our old world mechanical lever-based systems. They are also more complex, have more interaction variation, and are less reliable than systems designed well before the discovery of the transistor. We have a massively ineffective electronic voting machine ecosystem that only gets worse. Voting tech did not save democracy. In fact, it likely pushed us further along the continuum toward more complexity and less voter involvement for those individuals in our democracy who may have the greatest problems dealing with technology. The only positive benefit of a live on-site electronic voting experience is that, at actual polling places, there are volunteers to help teach confused voters how to cast their votes. This is a very common exercise seen at every polling location.
Enter The Mail-In Ballot…How Complex Can That Be? Given recent changes in state voting policy brought about by COVID-19, perhaps there will be no need to ultimately rely on machine-based technology, when an apparently far simpler voting methodology suddenly finds itself front and center in this presidential election cycle. That seemingly low-tech solution is of course the mail-in paper ballot. How complex can paper ballots be and to what extent might the actual physical design of such systems impact the democratic process? Is there really any need to employ human factors science in the design of such apparently simple low-tech solutions like the mail-in paper ballot. The answer is surprisingly counter-intuitive. Here is why.
Complexity In Surprising Places: It turns out that a professional usability heuristics assessment of paper mail-in ballots and the system upon which they depend reveals a complex design problem full of potential vote-canceling errors and related risks. In fact, such an analysis shows that there are far more potential errors in a mail-in ballot process than the same voter would encounter when shuffling along on election day to their local polling location. On a pure human information processing level, there are actually far more decisions and likely types of user-induced errors in successfully executing most mail-in paper ballots than one might assume. This is another way of saying that the seemingly simple mail-in ballot is a complex human factor science problem. Some of the mail-in ballot designs proffered by major states are full of major human factors oversights, complications, and outright evidence of design decisions aimed at making mail-in ballots far more complex than need be. This fact is borne out by the execution of a professional heuristics analysis of mail-in ballot design. Usability Heuristics is a methodology widely utilized by human factors scientists to objectively determine the cognitive complexity, confusion, and error structure for all manner of systems ranging from the flight deck of the Boeing 737 MAX to the reasons you can figure out how to control your new Apple AirPods. This process can be and should be applied to the design of the mail-in paper ballot system. One might assume this to be overkill, but a simple examination of the tasks related to the execution of the mail-in ballot reveals a surprising set of insights.
Overall Risk Mitigation And Potential Vote-Canceling Error Categories: In the objective assessment of the cognitive and physical complexity of products and systems, including paper forms and related processes, it is common practice to examine where in a given task flow there are observable error potentials. For example, a simple comparison between traditional on-site voting vs. mail-in ballot casting reveals that on-site voting has far fewer actual vote-canceling potential errors than casting the same vote by mail-in ballot. Here is a simple comparison of error categories for on-site vs. mail-in.
On-site Vote Casting Error Categories: There are essentially four potential vote-canceling errors. These include: 1) Forgetting to register to vote, 2) Not knowing the proper voting location, 3) Forgetting to visit the location during hours of voting, 4) Inability to stand in line for sometimes extended periods. By comparison, mail-in voting is replete with vote-canceling error categories.
Even though modern electronic on-site voting machines are unnecessarily complex, there is one major benefit of on-site voting: the availability of poll workers to help confused voters navigate a new electronic interface before casting their vote. The massive increase in poll workers helping voters comprehend and vote using electronic voting machines is one major reason for dramatically increased voting times and associated waiting times. The cost of voting machine user interface complexity on our voting behavior is significant.
Mail-In Vote Casting Error Categories: The number of potential vote-canceling errors induced by the mail-in ballot system can be largely categorized into 26 factors with many sub-task errors not noted here. The errors present in mail-in voting include: 1) Failure to identify availability to vote using mail-in ballots to begin with, 2) Failure to understand how to obtain a mail-in ballot, 3) Failure to know how to request a mail-in ballot, 4) Failure to actually request a mail-in ballot properly, 5) Failure to request a ballot during the allotted time frame, 6) Failure to understand when the request has not been met, 7) Failure to understand how to follow up on unmet request, 8) Failure to properly identify ballot when it arrives in the mail, 9) Failure to understand the difference between mail-in ballot promotional material and actual legal ballot, 10) Failure to understand how to open the outer envelope of the actual ballot and not damage the ballot or return envelopes inside, 11) Failure to understand mail-in ballot directions based on complex syntax and confusing directions, 12) Failure in overall ability to read any instructions due to the size of the font on the ballot and related envelopes, 13) Failure to properly mark required areas on the ballot as defined in instructions in order to select the desired candidate(s), 14) Failure to provide a signature on the ballot return envelope, 15) Failure to use the same signature on the ballot as appears on their government-validated forms, 16) Failure to properly fold and insert paper ballot into secondary envelope in an orientation that places critical information in scanner window, 17) Failure to insert the ballot into the secondary envelope and use the primary outer envelope for return function, 18) Failure to insert inner secondary envelope into outer envelope, 19) Failure to seal either the inner or outer envelope properly, 20) Failure to include a return address on outer envelope, 21) Failure to actually deposit fully completed mail-in ballot into a valid mailbox or dropbox, 22) Failure to use on-line systems to track the status of their ballots, 23) Failure to receive and comprehend reasons for rejection of a ballot based on one of the error states above in time to correct the error before the voting cut off date, 24) Failure to receive a confirmation at any point after the election closing date that their vote was not counted and the corrective action required prior to the next election cycle, 25) Failure to understand how to change mailing address and party affiliation between election cycles, and 26) Failure to receive changes in mail-in ballot procedures from voting commission that will impact the use of the system during next election cycle.
How Complexity Is Reduced By Human Factors Research and Testing: The point of the analysis above is simply to demonstrate that there are MANY potential vote-canceling errors associated with the process of mail-in ballots. In fact, far more than are present in traditional on-site voting. There is extensive evidence that the complexity of the mail-in ballot process may have a defining impact on the upcoming election, especially in states where the margins are razor-thin. It is important to note that this is a scientific fact, not a matter of political opinion. No one benefits from complexity in our voting processes, as the entire process is rendered dysfunctional in ways that can be avoided and are well-understood. In case it is not clear, the overall process of mail-in voting utterly demands the application of human factors science. To that end, there are well-understood methods employed by professional human factors scientists that utilize cognitive task analysis to build mental models of those who utilize mail-in ballot systems, and then design a system that objectively eliminates, minimizes or guards against design solutions that induce user errors, such as those that voters experience currently in staggering numbers.
A Well-Understood, But Unforgivable Usability Problem: An example of a core usability problem with current ballots is the frequent use of fonts that are objectively too small for a significant range of users to read without errors and critical loss of comprehension. Much of the text size on 2020 mail-in ballots is below the perceptual threshold required for millions of actual voters, thus introducing not only errors in comprehension, but outright failure to complete and return the ballot in the required format and mailing configuration. In several of the mail-in ballot configurations, the signature area is not located on the actual ballot, but on the back of the secondary envelope, thus breaking the mental model that voters have with respect to task closure, a central concept of human factors science. The signature is not where voters expect it to be when completing the ballot before mailing it back, leading to omission errors and rejected votes.
Voter Testing During System Design: Central to the application of human factors science is the concept of scientifically valid user testing based on respondents recruited across the range of capabilities and limitations known to impact the task errors above. Compared to the design of complex software/hardware-based high technology products, mail-in voting process design is, in a surprising way, even more complex. It is complex because any solution to the complexity problem must blend human factors science, public policy, congressional rule-making, and government-mandated usability validation through independent testing. Without rigorous human factors science, it is likely that state officials can and will utilize the complexity of this process to dramatically further restrict mail-in voting, in much the same way that gerrymandering voting districts have been used to crush voter participation and impact. This is not a simple usability problem like you might encounter with your new Apple watch, but a systemic usability problem that impacts our entire democratic process and related willingness to exercise our right to vote.
How Does One Design a Usable Mail-in Balloting System? By applying human factors science and related error analysis, those tasked with the design of an objectively easy to use mail-in ballot system can determine how to mitigate risks and reduce complexity and errors by applying a combination of three well-understood dimensions including: 1) Training the user through better Instructions For Use, 2) Automation in the form of advanced technology and 3) Better ballot design based on an understanding of the cognitive and physical limitations of the U.S. voting population. All complex human-machine interface problems, such as mail-in voting, must create the proper balance between user training, automation, and interface design. Based on this mental model, it is clear that the current mail-in voting system is wildly out of balance. How did we get to this position of offering mail-in balloting that is so plainly defective to hundreds of millions of voters from a human factors science and usability perspective? In this case, the source of the problem is well-understood and consistent with other recent problems related to how technology-based systems are optimized for human use. The problem lays squarely on the federal government and congress, which has almost no formal programs in place to require the design and validation of technology-based systems for usability and risk mitigation related to voting systems. This is not a new problem for the Federal government and related agencies of record. Today, the usability of voting systems, both electronic and mail-in forms, resides with those companies who design, distribute, market, and sell voting solutions. Leaving the validation of usability performance to those who design our voting systems, as opposed to requiring independent professional usability validation, is a massive mistake and has led to where we are today. This problem is systemic.
What Does the 737 MAX Failure Have To Do With The Design of Mail-In Ballots? The answer is actually a lot. For example, it was shown in extensive congressional testimony and analysis that the problems with the Boeing 737 MAX were directly traced to a lack of human factors science oversight by the FAA. The FAA simply allowed Boeing to employ human factors science in the design and validation of the 737 MAX flight deck without oversight. This was a massive mistake. But the FAA is not the only federal agency allowing self-validation of the usability performance of products. For example, the FDA has approved medical devices that reach consumers with far too much complexity, yet pass FDA human factors testing. There is, at the core of the mail-in ballot complexity problem, a much larger failure on the part of Congress to create legislation that mandates the human factors validation of ALL VOTING systems, not just mail-in ballots. It is important to note that optimization of the U.S. mail-in ballot system is not a graphic design problem, but a cognitive science problem that must be driven by knowledge of how to utilize the methods and expertise of human factors science.
Who Suffers Most From Mail-In Balloting Complexity? The short answer is voters with any form of cognitive or physical limitations that result in an inability to manage the complexity of this process. There is clear and convincing evidence that voters of minority and low-income backgrounds are harmed most by problems with mail-in ballots. In human factors science, this group is known technically as the Least Competent User profile or LCU. Voters who fit into this category measure in the 10’s of millions. Of these LCU voters, it is estimated that as many as 1 million will have their votes canceled due to problems with executing their ballot properly or failing to mail it back in time. Realistically this group of Least Competent Users extends well beyond the politically and socially disenfranchised, because cognitive and physical limitations are a function of the normal progression of aging, disease state, language skill level, physical dexterity, visual acuity, anxiety over a new process, and many other limitations well-understood to limit one’s ability to deal with complexity. Understanding such variables is at the core of what human factors science does and helps mitigate. The current mail-in ballot design is essentially unusable by a massive range of voters who would benefit most from such a system. The cognitive and physical limitations described above are the same factors that make on-site vote casting and mail-in balloting basically unworkable BY DESIGN. We are discriminating against a massive portion of the U.S. electorate simply through the design of voting machines and through mail-in balloting. Every decision made that is not informed by human factors science is a design solution that will likely fail to address the range of voter needs and limitations.
Who Benefits From Mail-in Ballots? This is not a complex question. The greatest impact of this unnecessary complexity is felt by individuals who have problems physically navigating to and dealing with a given voting center for a wide range of valid and critical reasons. Many of those who fall into this category do so for problems again related to physical or cognitive challenges or both. There are, of course, millions who are not so challenged, but who simply take advantage of the convenience of a mail-in methodology. However, there are far more valid voters who fall into a category in which traveling to a polling location and standing in line for sometimes hours is not workable. The numbers for this group are far greater than one might assume. Perhaps 100 million U.S. voters fall to a greater or lesser extent into the Least Competent User category, reflecting either physical or cognitive impairment, and therefore cannot navigate the traditional on-site voting. This group will only expand as the U.S. population ages and the cost associated with on-site voting becomes even more apparent. The trend toward mail-in voting is unstoppable and supported by the undeniable demographic trajectory of the U.S. population. This assumes a future in which there is no COVID-19 or a similar malady.
Government failure to integrate human factors science in public policy and corporations’ failure to utilize human factors science in the design of their products and services is increasing. This is a surprising turn, when one considers that there are methods based on actual science that can help solve the complexity problem, not just the voting problem.
Voting Infrastructure Has The Same Human Factors Science Requirements: This short analysis has focused on that portion of the mail-in ballot task structure that involves direct interaction with the ballot acquisition and execution process by the voter at home. It is important to note that human factors science is equally important in the design and testing of all infrastructure-related mail-in ballot systems, including postal service handling, vote counting, conflict resolution, and critical status feedback between the vote processing infrastructure and the voter executing and submitting their vote by mail-in ballot.
Who Signed What? At no place in the actual acceptance or rejection of a voters’ mail-in ballot is the process more broken than in signature matching. This single point in the final approval and counting of a vote is often based on untrained and unvalidated methods applied by individuals who can and do cancel millions of valid mail-in votes, without a science-based assessment of signature validity. This simple process demands robust human factors science and user testing validation. Valid signatures can vary between a voters’ ballot and their reference signature for a multitude of reasons, including hand tremors as a function of aging, a failure to have their reference signature available, use of different writing tools, or tools that run out of ink during signature. There is evidence that some mail-in signatures are canceled because voters cannot actually write in cursive, but use block letter signatures that invalidate their vote when viewed by those examining their ballot. These systems must also be based on the application of sound and robust human factors science. Looking at the larger issues of whole-system voter stability and performance, one can only hope that Congress and designated federal agencies implement a science-based human factors process leading to the creation of an electronic and mail-in voting system that supports the goals of our founding fathers. One person, one vote…counted and honored. I leave you with the following question posed to all who voted by mail-in ballot: did you remember to put a stamp on the outer envelope???
Thank you to Chris Morley, MSc, Director of Research at MAURO Usability Science, for his contributions to this article.
About the Author
Charles L. Mauro is President and Founder of MAURO Usability Science (MUS) established in 1975. Today, MUS specializes in advanced human factors research and optimization for medical devices and a wide range of other complex human-machine systems. Over his 45-year career, Mr. Mauro has managed more than 3,000 human factors research and development projects. He is a certified human factor engineering professional (CHFP) and has received awards and citations for human factors research from Human Factors and Ergonomics Society, Industrial Designers Society of America, NASA, the Association for Computing Machinery, and others. His clients include many leading corporate entities and startups. Mr. Mauro has testified as an expert in product development and human factors science in more than 75 major cases. He has been an invited speaker at the USPTO, FDA, NASA, MIT, Stanford, UPenn, and many other leading academic and research institutions. His firm is best known for the application of advanced neuroscience methods in the resolution of complex design problems.
2020 Election Could Hinge on Whose Votes Don’t Count https://www.pbs.org/wgbh/frontline/article/2020-election-could-hinge-on-whose-votes-dont-count/
Timing, signatures, and huge demand make mail-in voting difficult: https://theconversation.com/timing-signatures-and-huge-demand-make-mail-in-voting-difficult-145084