photo by: Nick Krug

A visitor arrives at the Douglas County Courthouse Monday, Nov. 6, 2017, during advance voting.

TOPEKA – This close to Election Day, a lot of people may be wondering why their mail boxes are stuffed with particular kinds of political ads, and why their friends or neighbors might be getting completely different mailers, or maybe even none at all.

The answer is, it’s about your “score.” That’s a number, and sometimes a whole set of numbers, that political parties and campaigns use to predict how likely you are to turn out and vote, how likely you are to vote for one political party or another, and even how likely you are to contribute or volunteer for a campaign.

“Scores and targeting is a pretty standard practice across both sides. It’s something everybody uses,” said Lauren Martin, political and data director for the Kansas Democratic Party.

Political parties and candidates have always had access to certain public records about voters from the information contained in voter registration databases — their age, address, how frequently they vote, whether or not they’ve ever contributed money to a campaign and, in some states like Kansas, their party affiliation.

In decades past, that information might be typed, or even hand-written, on an index card and stored away in a file box.

But in recent years, parties have turned to increasingly sophisticated methods of predicting voter behavior by combining that information with other information they can purchase from third-party vendors — consumer information, magazine subscriptions, club memberships and other data — and then using complex statistical models that try to predict how likely someone is to vote one way or another, or even at all.

“What started to happen after the 2000 elections in particular was that parties started to really build out those voter files with additional information, not stuff that they could gather themselves, but that they could buy,” University of Kansas political science professor Patrick Miller said in an interview.

“Things like your generic credit card history, or magazine subscriptions, or charitable donations to certain organizations — a lot of times, that information about you is for sale,” he said. “And it comes in the fine print of credit card contracts and so forth. If you have a mortgage, oftentimes that information can be sold to third parties.”

The practice has become so pervasive, Miller said, that the national Republican and Democratic parties have scores for virtually every registered voter in the United States, and they share that information with state party organizations who, in turn, share it with their candidates.

Officials at the Kansas Republican Party did not respond to a request for an interview for this story. But Martin, at the Democratic Party, said the information helps campaigns be more efficient with their time and resources.

“I always tell my candidates they have limited time and limited people to help them and limited money,” she said. “So when they use scores they’re likely to be more efficient with the voters that they’re contacting instead of spending a whole bunch of time talking to people who are unlikely to vote for you and, in fact, knocking on their door will actually turn them out to vote against you.”

That would explain why people might see a candidate walking door-to-door during a campaign, knocking on some doors, but passing by others.

It also helps explain why some people get inundated with mail and phone calls, while others get none at all.

Martin said that people with high scores leaning toward one party or the other may never receive campaign material, because one party thinks it doesn’t have to worry about that person, and the other party knows it would be a waste of time.

Other people, however, with scores in the middle range of the political spectrum might be seen as more persuadable, and therefore more likely to be targeted from both sides, Martin said.

But the process can also be imperfect.

“The thing about modeling is, you’re always making some assumptions,” Martin said. “For example, we have the (National Rifle Association) membership list. You will assume that people on that list are more likely to vote for Republicans. It’s not perfect, but it’s kind of how you make that assumption.”

Martin said that even a person’s party affiliation can be misleading if it’s not looked at in combination with other factors.

“For example, women as a whole are more likely to vote for Democrats,” she said. “But if you are a registered Republican woman who happens to live in a house with no other registered Republicans, you’re more likely to vote for the Democrats.”

Miller said there are a lot of other subtle indicators buried in people’s consumer information that indicate how a person is likely to vote.

People who drive Subarus, for example, as a whole tend to lean toward the Democratic Party, he said, while people who drive Chevys and Fords skew more toward Republicans.

But all of those assumptions are based on consumer surveys, and they only yield statistical probabilities about a person’s likelihood to vote one way or another. Often, Miller said, they are completely wrong.

“I’ve seen some scores of people here in Kansas — including, to be blunt, myself — and they are completely off on how I and a few other people I know vote,” he said.

That’s partly due to the fact that parties keep much closer tabs on voters in some states than others.

“The biggest files on voters are going to be in your huge swing states like Florida, Ohio, North Carolina,” Miller said. “When you get to a state like Kansas that doesn’t see as many competitive elections, that may not have as much of an infusion from a national party, there’s a good chance they’re going to have a lot more limited data available to the parties.”

But the amount of information that candidates and parties can get about voters, if they choose to, might astonish some people.

For example, every time someone uses a frequent-shopper discount card at the grocery store or hardware store, Miller said, information about that person’s purchases goes into a database, and depending on the fine print in the contract people sign to get those cards, the store may have a right to sell that information to third parties.

The same is true whenever someone swipes a debit or credit card at the checkout counter to pay for those purchases, Miller said

The information may not be as detailed as identifying what items people buy, or exactly how much money someone spent on a particular day. But Miller said it probably does identify the types of stores a person visits, and other aggregated information about the person’s shopping patterns, which are good indicators of someone’s income and demographic profile, all of which can be used to predict voting behavior.

And while that may be unsettling for some people, Miller said there’s not much that an individual can do to avoid giving away that kind of information.

“If you’re concerned, your only real alternative is to not drive a car, shred your debit and credit cards, don’t subscribe to anything and, much like the Unabomber, go try to live totally off the grid in the woods of Montana, I suppose,” he said.

In the end, though, Miller said there is nothing inherently dangerous or intrusive about the process.

“It’s not like anyone’s going to be able to steal your identity from this information,” he said. “The only difference that this information makes to a party or a credit card company trying to sell you something is, do you get a mailer or not; do you get a phone call or not; is someone going to knock on your door or not.

“That’s not an incredible intrusion on your life. It’s not a risk to you, either.”