American citizens are inundated with political messages – on social media, in their news feeds, by email, texts and phone calls. It is no coincidence that people are bombarded: political groups prefer a “multimodal” electoral contact strategy, where they use many platforms and multiple attempts to persuade a citizen to engage with their cause or his candidate. An announcement is followed by an email, which is followed by a text message, all designed to reinforce the message.
These strategies are used by political campaigns, political action committees, advocacy groups and nonprofit organizations. These different groups are subject to very different rules and regulations, but they all rely on capturing and devouring data on millions of people in America.
Who is in these datasets?
Almost everybody. Most campaigns get their voter information from a handful of data providers, non-partisan or partisan. These companies try to provide data on all American adults, whether they are registered or not. An individual supplier is unlikely to have complete files on all eligible U.S. voters, but the Pew Research Center, which published a report on commercial voters files in 2018, found that more than 90% of people in its own sample of U.S. adults could be found in at least one registry.
What data is collected and where does it come from?
The main source of voter data is the public vote register, which includes voters’ names, addresses and party affiliation. But voter data is very uneven and decentralized: each state has its own database, and it often has different attributes. Suppliers therefore supplement it with other sources, such as telephone directories and credit data.
It’s difficult to get a complete picture of everything that goes into supplier databases: the recipe each one uses is generally considered a trade secret. The Pew study explained that the registers are “a fusion of state administrative data on registration and voting, modeled data on partisanship, political engagement and political support provided by vendors; and demographic, financial and lifestyle data drawn from a wide variety of sources. ”
Data providers attempt to match and reconcile these different data sets to create a complete record for every person in the United States based on key identifiers such as name, address, gender, and date. of birth.
L2 is one of the largest companies that market this information and claim to have over 600 data attributes drawn from census data, emails from business sources, donor datasets, etc. Experts say most providers provide hundreds of data points on each voter.
How accurate are these electoral databases?
We have to debate. Some data points are very precise, but others are really just predictions or guesses. Party and race, for example, are often inferred based on a person’s name and location. Someone with the last name Ryan is assumed to be white, while someone in a strongly Republican district is assumed to be a Republican voter.
The accuracy of specific attributes varies widely: Pew found race to be accurate 79% of the time, education 51%, and religion 52%. Household income, on the other hand, was only accurate 37% of the time. There was also a measurable bias, with higher error rates for younger, highly mobile, unregistered, and Hispanic voters.