"I saw an incredible amount of ads from both campaigns," said Laura Godwin, a senior at the College of William & Mary, reflecting on the 2012 presidential election. Thinking specifically about Internet advertisements, Godwin noted that "there were more ads about women, which would probably make sense by what I search on the Internet. Like I shop for clothes, so it's pretty logical."
But for many voters, the idea that political ads could be targeted toward at them based on something as ubiquitous as Internet purchasing habits is anything but logical.
"I don't like that at all," said Jakob Stalnaker, another William & Mary student, upon learning just how much data targeting is undertaken by political campaigns. "[Having] that kind of information is not okay."
As Americans increasingly conduct their lives online (according to the Pew Internet & American Life Project, 85 percent of American adults use the Internet and 72 percent use social networking sites like Facebook), collecting data on Americans' online purchasing and browsing habits has become a lucrative business. Political campaigning is one industry that is increasingly using this wealth of consumer information to target individual voters with more and more accuracy. As the political use of large voter databases proliferates, the politicians in charge of consumer protection legislation and agencies like the Federal Trade Commission (FTC) are also increasingly the beneficiaries of that very same consumer data during their campaigns.
Presidential campaigns — particularly President Barack Obama's 2008 and 2012 campaigns — have made headlines for their use of technology, but smaller races are also taking advantage of sophisticated targeting opportunities to reach voters in their districts.
Jordan Lieberman, president of Internet advertising company Campaign Grid, which works primarily with Republican campaigns, explained that his firm was active in races large and small in 2013. "We did New Jersey's gubernatorial, Virginia's gubernatorial, Virginia's attorney general," he said. "The New York mayoral, Austin mayoral, Seattle mayoral, Houston mayoral" and "a couple dozen state house races in Virginia and New Jersey."
How much data do campaigns have on voters and how do they use it?
Since the political use of aggregated consumer data is a relatively recent phenomenon and one that is not widely discussed, there exists little information on exactly how political campaigns gather data and how much data they have. Generally, campaigns contract with so-called political data brokers, companies which have spent time and money collecting data from various sources and combining it with voter registration records. These companies then consult with campaigns or sell this data for political use.
"I think it's hard to answer in terms of the average campaign, but certainly the big, well-financed presidential campaigns have either collected data themselves or rely on [data] firms like Aristotle or Catalist," explained Ira Rubinstein, an adjunct professor of law at New York University School of Law and a senior fellow at the Information Law Institute.
"I think they have a great deal of data, and if you look at some of the websites of those firms, they all seem to boast about the amount of data that they offer as one of their selling points. They talk about having records on 200 million Americans or more and having hundreds or even thousands of different attributes as part of those records."
Indeed, data broker Experian's website has a category for "Life-Event Triggers" and advertises the company's ability to predict when individuals are new parents, homeowners or have recently moved.
Data broker Datalogix's website asks potential customers, "Need to reach pet owners? SUV drivers? Green consumers?" The firm markets more than 700 "segments" of Americans, divided into categories based on their past purchases, demographics and financial data.
Lieberman explained that his firm combines files of registered voters with consumer data from other firms to match voters with commercially available data sources. This politicized consumer data can then be used by the firm to target messaging to voters that fit certain criteria.
"The targeting that we're doing now is as precise as direct mail or phones," said Lieberman. "Before data-driven online advertising, [advertising on] the Internet was as precise as skywriting."
Perhaps this targeting process is how Godwin experienced more than the usual number of women's issues ads in 2012.
The regulatory situation
According to the FTC, there are few laws regulating the collection of consumer data for political purposes and no requirements that data brokers maintain data privacy unless the data is being used for employment, insurance, or housing purposes.
In a speech to the Forum for E.U.-U.S. Economic Affairs in September 2013, U.S. Federal Trade Commissioner Maureen K. Olhausen said that ensuring online consumer privacy and the exploration of big data are top priorities for the FTC in 2014.
"I think the way to think about this is that there are many state laws, some federal laws regulating the voter registration process. And as a rule, those laws at best have some incidental privacy protection, but they're very limited," said Rubinstein. "Then you combine that information with commercial data, a lot of that commercial data is not especially well-protected."
Members of Congress are also exploring the regulation of consumer data. Rep. Joe Barton, R-Texas, a founder of the Congressional Bipartisan Privacy Caucus has pushed for individuals to be able to consent before a private company or government agency is allowed to disclose any personally identifiable information. The caucus has also pushed for consumers to have the right to be informed when private companies collect or disclose personal information.
In a July 25, 2012 letter to several data brokers, the caucus requested more information regarding the company's data collection practices, writing that "consumers, regardless of age often have little or no knowledge of about the identity of brokers, how they collect personal information, and to which outside parties they sell or otherwise provide the information."
In its 32-page response on August 15, 2012, data broker Acxiom cited the benefits of consumer data and emphasized its attention to privacy, as well as provided more detail regarding their data-gathering processes. They failed, however, to provide information regarding the sources of much of their data.
Several lawmakers expressed disappointment in a joint statement, noting, "The data brokers' responses offer only a glimpse of the practices of an industry that has operated in the shadows for years."
So far, the caucus has not directly addressed the use of consumer data in political campaigns.
Free speech?
Lieberman said regulation of data used for political purposes is a difficult issue because of political speech protections. "I think there are serious constitutional questions in terms of free speech," he said. "I also think…that data-driven online advertising is more secure than any kind of direct mail or telemarketing because we're not using any kind of personally identifiable information."
Nicco Mele, a lecturer at Harvard University's Kennedy School of Public Policy and webmaster for Howard Dean's 2004 presidential campaign, had a different take. "I'm sure political applications [would be affected] – a lot of online fundraising is done using data targeting techniques," he said, explaining that technology developed for uses beyond politics (such as "cookie" technology that allows websites to track user behavior) would likely end up uniformly regulated under any new consumer protection laws.
Rubinstein agreed, describing data brokers' First Amendment protection arguments as an "oversimplification" of the issue. Laws that have exceptions for political speech generally involve direct political communication, such as the National Do Not Call Registry, which has exceptions for both political campaign calls and non-profits. Privacy laws regulating the collection and distribution of data, on the other hand, would broadly impact all consumer data collection practices. Data used for targeting is several steps removed from actual voter contact and political speech, and would therefore be unlikely to require political exceptions.
"There are some ways in which you can kind of thread that needle and at the very least require more disclosure and transparency about the collection and use of voter data," he said.
But are members of Congress likely to regulate the use of political data? Rubinstein described it as "extraordinarily unlikely," since politicians benefit from it.
"The notion that [campaigns] will inhibit their own ability to take advantage of these technologies seems pretty unlikely, especially when from what I can gather, the data brokers and the online advertising firms are beginning to use this as a counter-argument against regulating them at all. Going into offices and saying, "Don't you realize that your members are dependent on this data for their reelection campaign?"
A question of privacy
When asked how much of her personal data she thought had been collected by marketing companies or third parties, Godwin was resigned. "Way more than I wish, probably," she said.
Stalnaker viewed the distribution of some personal data as part of doing business on the Internet. "I'm not comfortable with it, but I sort of view it as part of the world," he said. "I don't get upset about it."
When informed about the aggregation of multiple sources of data for political use, however, Stalnaker was adamantly against it. "I would not know any person who would be okay with some outside group having access to that much personal data," he said.
Most Americans aren't aware of how much consumer data is collected about them. Rubinstein thinks they would be even more surprised that political data firms are collecting consumer data. "[Companies are] combining records from voter registration and records purchased from consumer data brokers and cookie-based profiles into very large troves of data about individual voters and their preferences and attitudes that are all things that are used for targeting purposes."
Lieberman, of Campaign Grid, was quick to emphasize that while personal information was used to match voters with their consumer records, these personally identifiable pieces are removed from Campaign Grid's voter file once each record has been combined with consumer data.
"I might use your vote history, meaning what elections did you participate in, are you registered with one of the parties, what's your age, and what's your income, things like that," he said. "And that information is combined with commercially available data and takes into account your buying patterns, lifestyle patterns, demographics, all kinds of other data."
Even without names or addresses, Internet data can reveal a surprising amount of information about individuals. A Proceedings of the National Academy of Sciences paper found that a range of behaviors and beliefs could be accurately predicted using just Facebook "likes" (records of pages Facebook users follow). The authors used "likes" to predict party affiliation, race, personality characteristics, and even sexual orientation.
Godwin seemed unfazed by the privacy issues raised by the aggregation of large amounts of data. "Everyone gets really mad that sites like Facebook personalize ads, but I like it. I am happy that they're actually going to advertise something I might click on," she said. "Sure, great, you can know my profile and know I'm a girl, that's fine. I don't actually care that much."
On the other hand, she described the political uses of this data as "kind of creepy."
"Because then it's like, 'I'm going to tailor what I show you,' and the thing I don't like about that is that you only see that aspect of the campaign, rather than 'Let me show you my entire platform.'"
Rubinstein agrees that the use of data by political campaigns constitutes a violation of privacy. "I think that it suffers from a lot of the same issues that arise in the commercial use of data: secondary uses that are not consented to, security issues, a sense of powerlessness and vulnerability about whether you're being monitored and whether you can do anything about it."
He also argued that political uses of data raises "political privacy issues" and that a strong democracy needs to allow voters to discuss and learn about politics free from surveillance.
"Citizens need some space…in which to develop their own thoughts and reflect on the issues and do so without being monitored and without their data being routinely collected by any online site that they might interact with in order to find out about the candidates and the issues," he said. "So I think that's a pretty serious problem and I do think it needs to be addressed."