Mining “Big Data” for Political Donors

Red states and blue states and often predictable in their political allegiances. But when looking for likely campaign contributors, expect the unexpected.big-data-for-politics

Consider that the top five blue (Democratic) states—based on voting in presidential elections between 1992 and 2008—are District of Columbia, California, Massachusetts, Rhode Island and New York. However, all of these states’ residents are, per capita, far above average when it comes to making contributions to Republican Party causes and candidates.

How can this be?

The answer lies in the real value of Big Data: it’s all in how you analyze it.

First the easy part. Obtain a random sample of people across all 50 states who have donated money to Republican or Democratic Party candidates at the state, local or national level. Now for the heavy lifting. If you compare the various non-political attributes of these consumers to 40 different consumer databases, you can identify the critical factors that most precisely predict a likely Republican or Democratic contributor.

What emerges from this exercise are predictive models that incorporate more than 150 unique data factors to construct a targetable audience of likely contributors. Call them “lookalikes” to actual contributors.

What are some of these predictive factors? Things like home values, frequency of home relocation, whether people live in apartments, number of children in the household, financial health and use of premium gold and platinum credit cards. Other factors include above average expenditures on personal care products, food and beverages, art, children’s products and home furnishings.

The fact that someone is wealthy really doesn’t tell us a whole lot about their inclination or desire to make political donations to any particular party. (Fabulously successful capitalists like George Soros write checks to Democratic causes, while fabulously successful capitalists David and Charles Koch donate to Republicans.)

In the predictive analysis described above, only 27% of the available consumer records deal with personal finances. This helps to explain, for example, why residents of the true blue state of Rhode Island, on a per capita basis, are 79% more likely to donate to Republicans even though Rhode Island ranked No. 16 on the list of wealthiest states (by median household income) as recently as 2009.

Adults in Utah and Nebraska, both traditional red states, are 9% and 16% less likely to contribute to Republicans, respectively.

An analogy from the non-political realm further underscores the value of meaningful Big Data analyses. Not every U.S. adult will choose to buy a life insurance policy with a death benefit payout of $1 million, even though such coverage can cost just $2,200 annually in premiums.

Many adults will easily drop $2,200 a year on family vacations, but for reasons known only to themselves they would never consider spending that amount of money on a $1 million life insurance policy. So how do you locate those who will? By examining the attributes of people who are existing holders of $1 million life insurance policies and combing consumer databases for people with similar characteristics.

With recent advancements in applying predictive analytics to the online population, it is now possible to precisely target likely political contributors. In fact, the analysis above indicates there are approximately 14 million U.S. adults who have a high propensity to contribute. Modeling with Big Data allows advertisers to focus online ad dollars on prospective Republican contributors deep with a Blue state (or vice versa for Democratic donors), while protecting the anonymity and privacy of the individual.

Published 6/5/2012 on MediaPost.com

This entry was posted in In the News and tagged , , . Bookmark the permalink.

Comments are closed.