Compared to a DRE voting machine an old fashioned paper ballot or an optical scan ballot seems like it ought to be really simple. What can be a more basic user interface than filling out a paper form. But it turns out that paper ballots have their share of usability problems too and it's been extensively documented that there are problems with the design of paper ballots. Have, have caused, very large changes to, election results. We're gonna look at a few of those, possible, problems now. Looking at the design of ballots, and seeing, what can go wrong. So one problem with, with paper ballots is that they're something that voters interact with only occasionally, only, you know, once every year or so on election day. And sometimes the rules for filling them out can be a little bit technical but voters don't read instructions. So this is just human nature that you come to the ballot and think I know how to fill out a paper form I voted before. So the instructions on a ballot are something however that people in practice really just glaze over or skip. This can be a huge problem sometimes however if the ballot layout is complicated. Let me show you this example. This comes from Washington State in King County in their 2009 election. Look at that left column right beneath the instructions. That's actually the first question on the ballot. But because voters don't read instructions, they just skipped over the whole first column, thinking it was part of the instructions. [laugh] And they missed this first race. In fact, ten percent of voters didn't bother to fill out that question at all, a much larger fraction than on other ballot designs or other questions on the same ballot. Another kind of problem is layouts that are illogical or too complex. Here's a design of a ballot from Wisconsin in 2002. Look at how this first race, the gubernatorial race, is split in the middle of the list of candidates on to the second column. And this was a space saving measure that the designer of the ballot implemented probably. But the result was that about twelve percent of voters didn't cast a valid vote for governor. Probably a lot of them marked a box in each half of that split gubernatorial race resulting in an over-vote condition. They'd marked too many candidates and their vote was completely thrown out. Another problem is if the design of the ballot is internally inconsistent. So take a look at this one, this is a ballot from Los Angeles County in 1979. Observe how the races on the left side and the races on the right side have a fundamentally different layout, there's much more spacing on the right, also the style of the titles has shifted from a, on the left you have a very prominent one on the top, then a tight one on the bottom, and then ones that are rotated 90 degrees on the right side. The problem here is similar to the Sarasota Florida DRE ballot we saw earlier, seventeen percent of voters just completely missed, this senate race on the bottom of the left side, they were blinded to it by the presidential race on top of it, or confused because the headers were different, people just forgot to fill out one of the most important races on the ballot. So all of these are examples of things that can go wrong when the ballot design is poor, and in, in the US since we have different candidates on the ballots, different races in, in, in each local jurisdiction. Ballot designs or, or, or, are very unstandardized. There, there are a range of different designs, they're designed by different people they run on different systems. . There just isn't consistency and problems like this, occur all the time. Here's another kind of problem. This is an example of what's called a ranked- choice ballot, and this comes from Alameda California in the 2010 general election. On a ranked-choice voting system, rather than just voting for one candidate, you get to vote for multiple candidates in a priority order. You have a first choice, a second choice, a third choice. We, we may get to how those ranking systems work and how they, the candidates are counted later in the course. But one problem with ranked choice is it complicates the design of the ballots, since voters don't just have to indicate one candidate, they have to indicate a whole preference order. Even though it's complicated, remember, voters tend not to read instructions. So all o f this area devoted to how to vote here is generally going to be ignored. Instead, voters are just going to concentrate on this part, the actual choices. So, think for a second about what mistakes voters might make here, and then I'll tell you what we find from research in a second. So researchers recently did a usability test on this ballot. And usability testing is hugely important in ballot design. People who do usability testing will tell you that every ballot is going to eventually be tested for usability on election day. So it's a great idea to test it out without it, it, really counting sometime beforehand. And what they do in a usability test usually is they get some volunteers and ask them to fill out the ballot either for certain prescribed candidates or however they like. But they have a researcher watching. And the researcher gets to note whether the voter has any hesitation or ask the voter at the end, to make sure the choices that are actually marked or what they intended, or ask if there was anything they didn't understand. So, usability testing for ballot designs, whenever there's a major change in the design anyway, is a highly recommended practice. When the researchers tested this ballot, they found out something very, very disturbing. Which is that voters look at this ballot and saw these numbers, one, two and three, the most prominent things on here. They didn't bother often to read the text next to them. And voters who didn't understand how this kind of ranked-choice balloting worked, often had a faulty mental model of what they were supposed to do. So they had formed in their mind a picture of how this ballot and rent choice voting worked, that was actually not how the actual rules said it would work. What these voters mistakenly thought was that they were giving points to candidates in each of these positions, that there was one point in the first column, two in the second and three in the third. So they'd vote for the person they most wanted to win in the third column to give them three points. This is exactly backwards. In fact, the person you most want to win is supposed to be marked in the first column. A substantial fraction of voters in usability tests had this faulty conception of how this kind of ballot works. And even though the style had been used for some time. What this means is that, those vote, voters' choices in previous elections very likely were not counted the way they intended. So the last thing I want to talk about with paper ballot usability is some design best practices that come out of usability testing. And there's been extensive research with usability tests on ballot design that have led to the development of some really useful guidelines. This is an example of a ballo t that's been developed that tries to follow those guidelines. And you can see immediately that it's a much clearer, easier to understand at a glance, layout than any of the ballets we've seen so far. Let's look at a few of the features on here. One is that the instructions are simple. And they're kept in plain language avoiding jargon. Avoiding unnecessary detail to the maximum extent possible. There's also meaningful use of color and contrast to try and indicate what's what. The name of each race is shaded in grey, all sets of instructions are shaded in blue. Furthermore the columns and the general layout of the ballot reflect and reinforce the organization. Races aren't split across columns, races follow from the middle column onward. There is, aren't confusing split columns as in that earlier ballot. Another guideline that's reflected here is the use of large text. Much larger than in some of the other ballot designs. And avoiding all capital letters. Almost all the other ballots we looked at used large caps all over the place. But because the have a more square shape than mixed case letters, they're much harder for people who have any kind of visual impairment, who are, who are reading quickly to parse and understand. Finally, one interesting thing this ballot does is it puts instructions very clearly. Where voters need them. If you're, supposed to vote for only one or for two candidates in a particular race. That's indicated right on the race. With a little caution icon in this example on the lower right. On a, a race where. All of a sudden you can vote for two instead of one. All of these things their they're ideas that come out of testing. They're ideas that have been introduced to prevent specific problems. And we really need to encourage jurisdictions to adopt guidelines like these in the design of ballots. To, to avoid all of the many kinds of errors that we've seen. If jurisdictions don't, you can imagine cases where someone designing the ballots might even deliberately try to introduce usability errors. In, in this a, in this way, usability problems can become security problems, as well as a, a kind of problem that reduces voters' ability to cast the votes they
intend.
