New York Civil Liberties Union Reports 'Stingray' Usage By New York City Police
FCC Seeks $29.6 Million Fine Against Phone Carriers For Alleged Cramming And Slamming

Voter Tracking, Data Collection, Analysis, And Privacy

While the New Hampshire primary and Iowa caucuses have passed, there are many more upcoming primaries this year before the general election in November. These primaries represent data collection opportunities for companies to learn more about voters. Marketplace reported:

"One company is tracking voter characteristics through some likely sources — their phones. Dstillery is a big data intelligence company that sells targeted advertising information about consumers to big companies like Microsoft and Comcast. But in the Iowa primary, the company tried its hand at compiling voter traits... people who loved to grill or work on their lawns overwhelmingly voted for Trump in Iowa... people who watched and supported NASCAR also happened to support Donald Trump and Hillary Clinton..."

Dstillery's has an impressive list of clients: AT&T, Cablevision, Comcast, DirecTV, Hulu, Sprint, T-Mobile, Verizon, Vonage, and many more. If you remember your college statistics classes, then you know that a correlation does not man causation. Things may happen together but it doesn't mean one causes the other. Being a NASCAR fan doesn't mean a voter will vote for certain candidates. Voting for certain candidates does not mean you will be a NASCAR fan.

This "big data" collection is also a reminder of how much we consumers share on social networking sites. All a consumer has to do is "Like" a brand (e.g., NASCAR, one of these top-10 barbeque grills, a particular politician, etc.) on Facebook.com, or "Follow" that brand (or politician) on Twitter and it is pretty easy for a big data intelligence company to collect, analyze, and compare voters preferences. (Facebook knows far more about you than you realize.) Even if you didn't "Like" or "Follow" a brand, the data collection is still pretty easy. All a big data intelligence firm has to do is troll through the metadata attached to photos you took with your phone and posted online: racetracks on Instagram, NASCAR cakes on Pinterest, or whatever else. You get the idea. The metadata attached to your photos recorded where and when you were (e.g., geo-location of the racetrack), the background scene (e.g., stands, pits, etc.), and the people (e.g., emblems on their clothes). This blog post explains what happens when you stop "Liking" posts and comments on Facebook.

The data analysis is also pretty easy because many most of you gave your mobile phone numbers to social networking sites so you could use their mobile apps. Both social networking sites and data brokers have two crucial data elements (e.g., your birth date, your phone number) to match, merge, and purge data about you. So, political campaigns (via data brokers and big data intelligence firms they hire) can understand pretty easily who actually voted, and for whom, at a particular voting location.

Is this a good thing? I guess your answer to that depends upon how much privacy you want associated with your voting activity. What you do within the voting booth may be private, but there are many players performing surveillance outside the booth to reveal what you did in the booth. And, if you aren't careful what you say in front of Internet-of-Things devices installed in your home (e.g., toys, smart televisions, smart speakers or search robots, etc.), then the data collection is probably even more extensive.

Is this a good thing?

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Chanson de Roland

I don't subscribe to social media websites or apps, so the campaigns have to work harder to discover who I support and, thus, will vote for. However, they do seems to know this much: That I don't much like any of their candidates and that, for me, the choice is the lesser of evils or at least the choice that is least bad.

Yet, I was recently disabused of my notion that my relative anonymity had made me obscure to the campaigns or at least not worth their efforts. I received a phone call out of blue, where a young lady or a lady who seemed young had the temerity, as a complete stranger to ask me whether I supported pro life measures. This immediately suggested that she was polling for Republicans, though it could have been Democrats. It could have been Distillery. But, whoever it was, my response was why on earth she thought that she, as a complete stranger, could cold-call me and engage me on the phone about a politically controversial topic, as if we were old friends. Her response was to repeat her scripted question, and, when I insisted on an answer to my question, she was for a moment nonplussed and then simply hung up.

One of things that most struck me about the phone call was the young woman's expectation that I would share my thoughts with a total stranger. Her expectation or, rather, her employer's expectation was that enough people will answer a stranger asking these types of intimate question to make it worthwhile to do it. And my personal experience confirms that. I often ask people questions that they must answer on penalty of contempt, but many people will answer and share the most intimate things without such rigor.

And an observation about the stats on this: It is true that correlation isn't causation. But the firms that collect this type of data, which the Editor writes about, supra, rely on two things that make their data useful for drawing inferences about a respondents’ voting behavior. First, they have information about a population of voters, how members of that population will respond to the survey questions, and who members of that population will support and, thus, vote for. Thus, when one’s answers to the survey questions put one in a particular population, the surveyor has with a fairly high likelihood of success determined who that person will vote for. The second thing that there firms have is a lot of solid enough psychological theory on how responses to certain questions define a person's personality, values, and beliefs, so the correlation is informed by a theoretical understanding. So, to the extent that the underlying knowledge about populations of voters and/or theories about voters are valid, the responses and other data collected from voters can be quite useful in determining who a person will vote for.

Lawyer, in big cases, use similar techniques during voir dire in an effort to fashion a jury that will favor their client's position. It works pretty well, especially where opposing counsel, as opposed to the judge, conduct voir dire. So I know that this stuff works pretty well, much better than throwing darts at a dart board.

The comments to this entry are closed.