No One’s Looking

Will Americans Vote Differently on a Touchscreen?

Scientists call it online disinhibition. It’s why you’re clicking to put extra bacon on your pizza—and might elect more fringe candidates when ballots move away from paper.


One of the basic insights of behavioral science is that the format of a choice set—how the options are arranged on a page—can significantly shape our decisions. This is true when ordering a meal at a restaurant, but also at the ballot box, as the design of a paper ballot can influence which candidates we end up voting for.

For example, studies led by Jon Krosnick at Stanford University have shown that candidates at the top of the ballot get, on average, about 2 percentage points more votes than they would have if listed farther down. This primacy effect even holds in national elections, when voters are more familiar with the candidates. When the name order of candidates was randomized on the California state ballot in the 2000 election, George W. Bush’s vote total was 9 percentage points higher in districts where his name appeared first versus last, Krosnick says. To deal with this bias, many states have begun randomizing the order of candidates, taking steps to ensure that a cognitive quirk doesn’t determine the winner of the election.

In recent years, however, the paper ballot has begun to be replaced by the touchscreen, with approximately 30 percent of U.S. voters using some sort of electronic device to choose their preferred candidate. While this transition has real advantages—there are no “hanging chads” in voting machines—it also comes with potential downsides, such as vulnerability to hackers and malware.

As a behavioral economist, I’m interested in how the simple act of voting on a screen (and not on paper) might shift our political preferences. Although we typically assume that our desires and choices are unaffected by technology—we’ll pick the same things regardless of whether we’re in a retail store or on a website—there’s preliminary evidence that these digital devices might be shifting the patterns of our behavior, altering how we process information and make decisions in all sorts of subtle and interesting ways.

Just look at our pizza decisions. A team of scientists led by Avi Goldfarb recently analyzed more than 160,000 orders placed over four years at a North Carolina pizza chain. Because the chain introduced an online ordering system in the midst of the study period, the researchers were able to conduct a field experiment into how the introduction of technology changed the content of customer orders. According to the data, presented recently in the journal Management Science, online customers chose pizzas that were more complicated, expensive, and idiosyncratic, containing 33 percent more toppings, 20 percent more bacon, and 6 percent more calories. In terms of popularity, online pizza orders contained 9 percentage points fewer top-10 items.

What caused this shift? The scientists argue that people ordering pizza on a screen are more likely to express their true desires, since a screen (unlike a human cashier) won’t judge their selection. As a result, they order fewer traditional pizzas, leading to a far greater variety of pizza configurations. This is known as the online disinhibition effect and it applies in many different situations. It’s why we tend to be more honest when asked questions by machines and why digital pornography features more “exotic” content than its analog competition.

I can’t help but wonder if such disinhibition might influence our voting decisions on screens. More research is definitely needed, but one implication is that people might be more comfortable voting for a fringe party or candidate that seems to represent their true political preferences if they’re voting on a screen. (It’s the political equivalent of extra bacon on a pizza.) Although every voter is granted anonymity, I wonder if we are more likely to feel that anonymity when using digital devices.

But disinhibition is not the only difference that influences our digital lives. Another crucial factor involves changes in the nature of attention. For instance, numerous studies have found that people read worse on screens and comprehend less of the content. My own research, conducted with John Payne, shows that roughly half of people fail basic tests of attention when asked questions on screens.

I’m certainly not the only scientist who gets these results. In a recent study led by Daniel Oppenheimer, the scientists found that anywhere between 14 and 46 percent of subjects failed a test of attention when taking an online survey. In one instance, Oppenheimer and colleagues could only replicate the results of classic behavioral science experiments on a computer if they disregarded those who failed the attention filter. We are a scatter-brained species, and probably more so when thinking on screens.

This research has major implications for the future of voting. In my state of California, citizens are often asked to decide on complicated propositions; making a good decision involves carefully reading descriptions of the law. My concern is that voters using screens will have a tougher time mustering the attention necessary to make prudent choices.

Even the size and type of screen might matter. John Payne and I conducted a pilot study that asked people various financial questions on smartphones, laptops, and paper. Interestingly, subjects did significantly worse when answering questions on smaller screens. While only 45 percent of people got a basic question about inflation right on an iPhone, nearly 57 percent of people got it right when answering on paper. The scores from laptops were somewhere in between. Although the underlying cause of these differences remains unknown, and far more research is needed, I think they highlight the importance of understanding how screens affect voting. After all, even seemingly irrelevant factors, such as the dimensions of the display, can influence the ability of people to process information.

People have been voting on paper for hundreds of years and we’re only beginning to understand the ways in which the format of a ballot can sway an election. In some instances, this has led to important reforms, such as randomizing the order of candidates. However, as modern democracies increasingly rely on screens during the voting process, we should conduct more research into the relevant digital variables that shape our choices. Right now we have far more questions than answers, but given the closeness of many American elections—12 presidential contests have been decided by a margin of 1 percent or less—it’s imperative that we learn how technology can impact our democracy.