Political Campaign Data Analytics

The field of applied political science and political campaigning is one the right must familiarize itself with in order to have an effective grassroots strategy going forward.

Having previously worked on many different kinds of political campaigns before, I have already had the exposure to this technical field of political science, so I wanted to relay some of that information for my benefit of my readers. It was very interesting to see how scientific some of the actual practices from my campaign jobs really were and to see these technical tactics applied in the field.

Learning the statistical and data related side of political science is important because — in my opinion — the political science department at my university and probably at other schools as well is not heavy enough in the quantitative material related to math and technology that allows political science to be practically applied in the real-world setting.

I want to provide an extensive assessment of how a team of academics, scholars, statisticians, and strategists are reshaping the modern-day state of political campaigns in America, and explores war room strategies based in behavioral psychology and randomized experiments. The multidisciplinary nature of the scientific and technological techniques and methods used in political campaigns are highly interesting.

Political elections, when it comes down to it, are based almost entirely on mathematics and the electoral rules and laws can dictate the best mathematical approach to solving the problem — in this case, how to win the election.

In America, elections come down to a few important factors, but to put it simply one third of voters will support your candidate enthusiastically, one third will never vote for your candidate, and it is the last third that are the most important: the undecided voters who are up for grabs from either side of the aisle.

One of the most popular political campaigning methods is called micro-targeting. This method skyrocketed in popularity after Barrack Obama’s campaign used it to reach the White House in 2012. Micro-targeting involves using data to target voter groups down to families and individuals.

For example, a campaign will purchase data from a company who does data mining, such as Facebook or Amazon. With a list of products purchased from your Amazon account, or a list of pages you like on Facebook, the campaign can begin building a profile on you.

Whichever party you are registered with is public domain, so the campaign already knows who its registered and dependable voters are, so they will not waste campaign resources on someone who is already likely to vote for your candidate.

So back to micro-targeting; suppose you are trying to elect a Republican candidate. You want to use your resources to convince Republican leaning voters to come out and vote for your candidate. If someone’s Amazon data says they recently purchased a yoga mat or own a hybrid car, your campaign is probably going to ignore them outright. But if someone’s consumer data shows they own a gun, or have donated to their church, then the data suggests based on their lifestyle they lean Republican.

In addition to data analytics, there are also numerous examples of political psychology experiments and other sociological research which allows campaign managers and data gurus to basically read the minds of voters.

Anything that can be found out through data mining is available to political campaigns, who use them just as often and in almost the exact same way a corporate advertising firm does.

Using the data, the campaign will make lists of voters based on what kind of messaging works best for them. This messaging can be broad or even just a single issue.

For example, if you are working for a pro-life candidate, you may have a list of voters who — for every other reason would vote for a Democrat — but could be liable to vote for your candidate based on just the pro-life messaging. The campaign can then send campaign literature individually tailored to the specific voter they’re trying to reach.

Additionally, the data is used for Get Out the Vote efforts. Get out the Vote is a tactic, normally used by nonprofit group outside of an official campaign, to encourage voters to hit the polls on election day.

Campaigns and like-minded non-profit groups toe the line of legality and ethics to advance their causes: nonprofits are not allowed to work specifically for the campaign, so they do get out the vote campaigns, which counts as public service since it is technically nonpartisan. Then the nonprofit just uses the data to only drive out the vote for people who are likely to vote the way they want them to but just might need a little push or reminder to actually go out and vote.

Campaigns and nonprofits can also gather data in the field through canvassing, which makes up a bulk of political campaign work. Canvassing is the process of going out into the field to collect data personally. Field workers will receive lists of houses to go to, even broken down by who lives there, and then they are sent to go collect more data.

For example, I did this at one of my old jobs. I was a field activist for a nonprofit group that advocated for conservative issues, but not candidates (though we worked around the rules to indirectly support candidates who advance our groups goals). We had lists of voters who, based on our data, were likely to vote Republican. My job was to go door to door to interview people so we could begin to aggregate that data into a strategy.

Basically, the company I worked for was trying to figure out a strategy to begin implementing anti-labor union laws and wanted to gauge the support for or against their ideal policies. Hoping to find voters who opposed labor unions, we used data to find Republican-leaning voters and ask them about specific issues to see how realistic it would be to get those laws passed in my state. We also did the exact same thing, but by robocalling using an iPad app instead of actually going out door knocking.

Lots of software is used to compile voter data like this. The software I used was called i360 and it had a few components to it, including door knocking maps, rob dialing lists, and technology to remotely sync this data to a larger national database so that our company could aggregate and compile this data on a national scale.

Another app used by Republican campaigns is called Advantage and it works in a similar fashion to i360 to allow volunteers and staffers in the field to track door knocking and phone calls.

They do this because any data on Republican leaning voters can be sold off to any other group or campaign. When a candidate loses, and ends their campaign, they are usually in a lot of debt and they often sell their voter data, fundraising email lists, and such to pay their way out of the debt. This is why in a presidential primary, you may support one candidate, but when they drop out, you start receiving emails from the winning candidate.

Data analytics met its match in one very significant event in political statistics: the 2016 presidential election. After a massive upset victory, almost every pollster and pundit had been proven wrong. Even the statistics wonks in the Las Vegas betting market were so wrong on the probability of the outcome.

Eventually, the consensus was reached that too much faith had been put into polling and data. Inside looks at Hillary Clinton’s campaign revealed her campaign staff relied almost entirely on political data, to the point of tunnel vision — the numbers told them what they wanted so they blindly accepted them and ignored everything else instead of trying to run a multidisciplinary campaign that may have been more in touch with politics applied in the real world.

While I think political campaign data analytics have completely modernized our electoral process, I still think there can be problems when data is put on an altar like it was in Hillary’s losing campaign. I think further developments in behavioral psychology can also help adjust and fix some of the problems associated with data driven campaigning.