Last Week in Tech Law & Policy, Vol. 35: Microtargeting and the Use of Voter Data to Win Elections

(By Sean Doran, Colorado Law 3L)

Both major political parties in the United States currently gather and aggregate massive amounts of data on American voters. Over the last several election cycles, with the advent of advanced data analytics and advances in data storage and processing, campaigns have gained the ability to learn and track a surprising amount of data about voters. This creates a level of precision that allows campaigns to build advanced models for identifying and targeting individual voters to receive (or not receive) individual messages (microtargeting).  Parties are building “political dossiers” on American voters which are some of the largest, unregulated aggregations of personal data that currently exist.

How does microtargeting work?

First, data is gathered from a diverse range of sources. For example, many states have voter files that are public record and which can be obtained by paying a nominal fee. Additionally, both political parties have been collecting and maintaining records of their supporters since as far back as William Jennings Bryan.

A more sophisticated modern source is personal consumer data, provided from “Big Data” vendors like Acxiom, Experian, and Dun and Bradstreet, just to name a few.  As one reporter put it:

[Politicians] know not just your name, address, and voting history but also your age and the age of your children, whether you smoke cigars, where you shop, where you attend church, what kind of car you drive, how old it is, whether you’re on a diet, and what type of pet you have.

Next, the data is analyzed and leveraged to help a campaign build a winning coalition of voters. A winning coalition will likely consist of a base of support from partisan voters and other voters that may be persuaded to vote for the politician. Voters are given a score that indicates likelihood of participation and identifies undecided voters. These two groups of people are considered “persuadable” by political parties and therefore worth the expense and time consuming effort to contact personally.

However, the effects of mircrotargeting are unclear and, some argue, unproven. One staffer for President Obama estimated that microtargeting could perhaps swing an election by 2%.

So, why do campaigns use microtargeting?

It is important to understand that microtargeting evolved to further traditional campaigning practices such as canvassing, phone banking, attack advertising, and fundraising. Microtargeting can make all of these traditional means of activating a political base, or persuading undecided voters much more efficient.

For example, instead of a campaign sending volunteers to knock on all the doors of registered Democrats in a neighborhood, to talk to voters about their candidate, volunteers can go straight to the voters the models have identified as “persuadable”.  This prevents a campaign from wasting time on people that are 1) solidly partisan, who will vote for their candidate regardless of personal messaging or 2) people that are not likely to be receptive to their message.

In addition to the clear efficiency improvement, political campaigns argue that this type of data collection can promote democratic vales by helping politicians to better understand their constituents and can also raise political participation be engaging traditionally inactive voters.

Privacy  Concerns

None of the patchwork of privacy laws in the United States apply directly to voter data. For example, data that the campaigns collect themselves through surveys or door-knocking are not covered by consumer protection laws. Similarly, information collected through a campaign’s web site will likely be unaffected by Federal Election Commission oversight because the commission does not have relevant rules about on-line privacy practices.

This leaves the use of voter data at the discretion of campaigns and their staff. Already this election cycle, campaigns have engaged in questionable uses of data that implicate voter privacy concerns.

For example, just before the Iowa primary, the Ted Cruz campaign infamously sent out a “voter violation” card that supposedly reported the voting participation rate of not only the recipient but also of that person’s neighbors. Presumably the violation card was an attempt to use peer pressure to increase voter turnout. Regardless of the validity of the information—it is speculated that the grades were made up and not based on actual data—voter data should be protected by more than the judgement of a campaign that’s desperate to stay competitive.

Political Harms

The reason microtargeting exists is to separate groups of people and help campaigns communicate with those groups disparately. One concern with this practice is that a smaller amount of diverse information will reach the potential voter, or, worse yet, no information at all (political redlining).

One of the efficiency gains by political parties is not spending resources on voters that are not likely to participate in the election. These are often low-income or less educated voters. While this makes sense from a campaign economics perspective, it disregards the fact that these voters are still members of a democratic society. Ignoring them only increases the political knowledge gap that may contribute to low turnout rates.

Tailored messaging is also a concern because it is very effective to supply a voter with information about a candidate’s stance on an issue that is relevant to that voter. However, this too leads to a dearth in general information about politics. For example, a campaign may screen out information or a stance on an issue the candidate has taken that a voter is likely to disagree with, while highlighting an area that the voter may favor.

Is microtargeting different in-kind or in degree from traditional campaigning techniques? If in-kind, do we then need new regulation to address this new technology? Regulation of voter data may be prudent, but how would regulation square with the First Amendment’s protection of political speech?  Lastly, these voter data, databases are in overdrive every 2 to 4 years, but what happens to the data between election cycles?