Separating the Sheep from the Goats with Political Data Mining

In the 2012 U.S. presidential election we witnessed an unprecedented usage of data mining to increase the effectiveness of political campaigning. The incumbent President Obama soundly defeated the challenger Romney, and the campaign team and pundits widely credited the Obama campaign’s strategic and highly integrated data mining tools with widening the gap.

ComputerWorld did a story yesterday (Obama and Romney big data experts continue the battle as businesses) on how Obama’s data analytics team had re-formulated itself as a business and was most recently seen helping Cory Booker win the NJ Democratic Senate primary. The article talks about how the team determined, down to the individual, the likelihood of each person participating in the primary election. It is no secret that Obama’s campaign team used such information to determine which voters received phone calls, visits, or perhaps even the tone of mailings they may have received.

But assessing the likelihood of any one individual’s participation in an election cycle is of limited value. Encouraging an apathetic voter who doesn’t see things with your worldview can easily backfire. The last thing you want to do is “get out the vote” for individuals who won’t support you. Reading between the lines, it is clear that the winning data mining team must be going beyond assessing participation and on to assessing the probable outcome, expected value if you will, of each person’s vote using models trained on income levels, age, race, zip code, and the like.

In essence, political data mining gives a campaign the ability to separate the sheep from the goats. The sheep, the masses moving along mindlessly with the objectives of the campaign, are assumed to be a known quantity, a give-me of sorts. The goats are the wildcards who ensure victory or defeat and require intensive manipulation. They are the margin by which elections are won or lost. They will also become the scapegoats on whom all blame for a party’s failure to win is placed, much as we saw with the mainstream Republican party blaming the conservative wings following the Romney loss.

There is something inherently offensive about this use of data mining. The known quantities, or sheep, aren’t spoken to and aren’t listened to. The goats get all the attention and become the focus of expert manipulators. When taken to its most extreme, you end up with the Kevin Costner film Swing Vote in which it becomes clear that neither side really believes in any of their core principles and will only say and do that which will get them the win. Governor Chris Christie of New Jersey publicly announced his support for such strategies yesterday as well (Chris Christie lays out argument for 2016).

1 Comment

  1. The terms “data mining” and “data-driven” makes me crazy! Data is consistently manipulated to produce desired results. I’m not a fan. I tend to use my own experience and common sense to make decisions – all the while constantly challenging my own beliefs.
    If I have to sit through one more power point presentation where NOPD show how they use data and studies to prove how their stupid initiatives work I may have to kill myself!
    Maybe they should mine your blogsite LOL!

Comments are closed.