What do you think? Leave a respectful comment.

How Data Mining is Causing Analytical Revolution for Campaigns’ Victory Strategy

Increasingly, behavior psychology concepts like ‘social pressure’ can be used not only to get voters out to the polls but getting them to vote for a specific candidate. Hari Sreenivasan talks to journalist Sasha Issenberg about his latest book, “The Victory Lab: The Secret Science of Winning Campaigns.

Read the Full Transcript


    And finally tonight, a different way of looking up close at campaigns and what influences voters.

    Hari Sreenivasan has our book conversation.


    Journalist Sasha Issenberg has spent two years examining the fundamentals of elections, from microtargeted polling to the on-the-ground tactics operatives deploy to make sure people show up at the polls.

    His new book, "The Victory Lap: The Secret Science of Winning Campaigns," takes readers into campaign war rooms for an inside look at the data driving some of the political decisions.

    He joins us now. Thanks being with us.

    SASHA ISSENBERG, author, "The Victory Lab": Thanks for having me, Hari.


    So, you describe almost a scientific revolution that is happening in campaigns. And you say there are experiments happening now on who votes and why that are similar to drug trials. How is that?


    This is basically randomized control trials.

    So, instead of randomly assigning people into groups and giving them a different pharmaceutical and leading a control group or having a placebo, people are sending mail or online ads or robo-calls, randomly assigning them, and then going out and measuring the impact, either looking at who voted after the election to see if it affected turnout or polling before and after to see whether it changed people's opinions.


    It used to be that polling was the science of campaigns. And now there is so much more behavioral science and prediction and analytics that is going into it. When did that happen?


    In the wake of 2000, I think a lot changes.

    You have all sorts of new data coming from the commercial world that is making its way into politics.

    You have academics in political science discovering field experiments. Political science had never spent anywhere near as much time as some of the other social sciences actually testing things in the field.

    People realize they can run these experiments and see what actually moves voters, and when you start to realize that a lot of things we see in polls and in focus groups don't really live up to the way that people experience politics in the real world.


    So this combination of commercial data and political data, give me an example.


    Well, for you, Hari, you are a registered voter, so right away I know on your voter registration record if you are affiliated with a party, which elections you voted in, if you voted in primaries.

    I know your date of birth. I know your age. I know your gender. In some states, I know your race.

    I have all sorts of political information maybe from other groups or campaigns that have contacted you in the past. So I know that someone from John Kerry campaign knocked on your door eight years ago and you told them you weren't interested in voting or you told them you care about the environment. All of that is in a database somewhere.

    Maybe, you know, Planned Parenthood knocked on your door, or the NRA too, and that information is — exists somewhere.

    And then there is all this information that has been gathered by commercial data warehouses, from magazine subscriptions, to things you put on warranty forms.

    A lot of it is first gathered by people who are creating credit ratings. They obviously want to know as much about you as possible and develop predictions. What has happened in the political world is people are doing statistical models with the same goal.

    So, instead of predicting whether you are likely to default on your loan or pay off your bill on time or run up $500 on your credit card in a given month, they're trying to predict how likely are you to vote in November, who are you likely to vote for, what issues are you likely to care about.


    So, give me an example of how I am being, well, the politicians might say targeted, but a voter might feel manipulated.



    So, after they have done these statistical models and are sort of moving you into one of a few categories, are you somebody they want to persuade or are you somebody they want to turn out?

    The science on turnout has gotten a lot better. We don't know a whole lot more about what makes you change your mind or how you change your mind than we did a decade ago. We know a lot more about what can motivate to actually cast a ballot.

    And, increasingly, it is informed by behavioral psychology. And so the most effective tool we have ever seen documented was in this experiment in Michigan in 2006 where some political scientists worked with a pretty quirky mail vendor in Lansing to send voters copies of their own voter history.

    So, this says, in the last eight elections, you voted in the 2010 general election, you didn't vote in the school board election. And it says, here are your neighbors' vote histories, and it has a bunch of people on your block and whether they voted in certain elections.

    And it said, there is an election coming up. After the election, we are going to send everybody an updated set.

    It had a sort of astronomical impact compared to any other tool for getting out the vote. It did — it is what psychologists call social pressure. And people have been sort of working with this idea that voters are concerned with sort of their identity around voting time.


    So, let's talk a little bit about this campaign now. There is so much talk of the huge amount of money that is being spent, a dollar per voter figures, right, especially in battleground states.

    How much of those dollars are spent on traditional kind of negative campaign ads, and how much of it is spent doing this kind of research that you are talking about?


    I think, increasingly, campaigns are recognizing the value of doing turnout, as opposed to just persuasion, partially because the science has gotten better.

    Some of the — a lot of the research isn't necessarily taking place in campaigns. Campaigns still are not likely to spend a dollar now for something that will yield a lesson in December.

    What they are doing is insights that are coming out of academia or coming out of — sort of some of these research institutes that have popped up in politics are affecting everything that goes on, on the ground.

    And so, you know, when you get contacted the weekend before the election by somebody doing get-out-the-vote for the Obama campaign, they are almost certainly going ask you what time do you plan to vote on Election Day?

    They are not going write down the answer, but it's because they have learned through years of experiments that asking people what time they plan to vote, where they are going to come from beforehand, what they will be doing that day makes them develop a plan, which is a sort of behavioral mechanism that actually makes somebody more likely to follow through on an action after they have sort of thought through how they are going to do it.


    So, both of these campaigns seem to be spending a tremendous amount of money on going after either the off-the-grid or on-demand voter, meaning the person who is not watching traditional TV.

    They might have cut the cord for cable. They're only watching online ads.



    One of the big things that has happened this cycle is campaigns are finally able to sort of merge or sync up your offline political identity and your online political behaviors.

    And so we have — there have been lots of analytics basically since 2004 in the online world, but they sort of existed in isolation there. So if you signed up on a website, then I was pretty good at being able to track how you moved around that website.

    But the only things people did online were sign up to volunteer or give money. You still vote in the real world. You vote in the precinct where you live. Your registration is tethered to an address. And we never knew whether the person who signed up on a list was the same person who was registered at a given address.

    That has changed now through cookies online. Campaigns are able to link a voter file record to somebody moving around the Internet.

    And it's given them the ability to sort of sync up their interactions with you online to what they are doing when the volunteer calls you or knocks on your door or they are sending you direct mail at your house.


    And you also in the book get into this cat-and-mouse game, that essentially every campaign learns from its failures and moves forward, and whether it is what Karl Rove did in 2000 or 2004, the Democrats respond, and then the Republicans respond again and again.


    Yes, learn from their failures or overreact from their losses. It is probably a little of both.

    But definitely the 2004 Bush reelection scared Democrats like nothing before, partially for things that the Republicans had really figured out and partially for a mythology that had emerged around Rove and some of the people in Bush's world for things that they had done.

    But I actually think that the — and I sort of tell the story of 2004 in the book. And I think the parties are actually far closer in terms of where they were in unlocking the underlying secret of micro-targeting, which is that you can predict what an individual is going to do.

    One of the advantages Bush had in '04 which Obama has in 2008 is that, when you are a presidential reelection, you have several years to sort of build a system where people actually will follow through on things.

    And the Bush campaign was far better at implementing it in their day-to-day operations than Kerry ever was.


    All right.

    The book is called "The Victory Lap: The Secret Science of Winning Campaigns."

    Sasha Issenberg, thanks so much for your time.


    Thanks, Hari.

The Latest