In 2002, the Help America Vote Act required (among other things) that states must maintain a "computerized statewide voter registration list". These lists (henceforth "voterfiles") contain information about every registered voter and their voting history.
When people who have not worked with voterfile data before hear about voterfiles, their first response is almost always "But in my 8th grade civics class, I learned that a critical component of American democracy is the secret ballot! How can states have a list of how you vote?" Voterfiles do NOT include information about how an individual voted. They report whether an individual voted in a specific election.
The exact format and contents of a publicly available voterfile differ from state to state. At a minimum, a file will contain:
- the name and address of a voter
- polling place
- date of registration
- status (active or inactive)
- which elections the voter has voted in
In some states, this history goes back decades, in others it include only a few recent elections.
Other common fields on voterfiles include:
- age, full date of birth or just birth year
- additional contact information (phone or email, these are usually pretty sparsely filled out)
- party registration
- gender
- race (this is common in states that were covered by section 5 of the voting rights act)
- voting method for each election (in person, by mail, absentee?)
Depends on the state.
In North Carolina and Ohio, the files are available for free download online.
In Pennsylvania, the files are available for online download for a small fee.
In New York and Florida, the files are available for free by request on CD.
In Minnesota, only voters registered in the state may request a voterfile (there is a small fee).
In New Hampshire and Virginia, only political organizations such as candidates, parties and PACs may request a voterfile.
Requirements tend to vary by state and change over time, so the above are only a guide of availability the last time I checked.
In most states, the use of these files is limited to non-commercial activity. Many states require that the file be used for political purposes. I have found that most of these states include journalism in this category, as journalists are informing the public about elections and voting.
In general, if you can't find information about a state's voterfile policies online, the best bet is to call the election administration department, which is usually housed somewhere in the state's department of state.
A variety of vendors maintain full national voterfiles based on the state files. They are not cheap. In my experience, national vendors are sometimes willing to generate a few aggregate stats for reporters.
Voterfiles contain at least one row for every registered voter in a state. That means if you request a voterfile, you should expect to get hundreds of thousands or millions of rows, depending on the size of the state. You will need to be able to use some kind of data processing software more powerful than Excel to deal with the results. Most people I know either dump voterfile results in a SQL database or process them with R or Python.
In many cases, you can get data about registration and voting directly from the state. Most states publish registration and voting statistics broken down by party, geography, and sometimes demographic features. If that is what you are looking for, I strongly recommend you stick with the aggregate statistics. Sometimes states or counties will cut the data along a different axis if you ask for file a Freedom of Information request (many charge for this, it can be complex and time-consuming). Here is a great example of a Tampa Bay Times graphic about the 2018 Florida primary that can be made based on aggregate data released by the state combined with census data.
Here are some examples of types of stories I've seen where raw voter data is truly essential:
Let's say you want to talk to registered Democrats under the age of 40 who voted in 2016 but not in 2014 in a specific precinct. You could knock on doors and hope, but with a voterfile, you can find those people directly. Similarly, I've used a voterfile to background a source. If someone claims to be a Republican who voted in a given election, I can confirm that before quoting that person if I've got the voterfile.
Sometimes you want to look at registration or turnout among voters in a very specific demographic. For example, in this piece we wanted consistent age buckets across a variety of states. Some states did not provide registration data by age at all, others included buckets that were wider than we wanted.
Reports of specific voters being turned away at the polls are often best investigated using voterfile data. Reporters can confirm whether the voters were registered, whether they were listed as inactive, and/or whether there were patterns of voters being made inactive on the file, although it will depend somewhat on the exact fields a state makes available.
Many pollsters survey only registered voters or likely voters (generally people who have voted in the past). If you're conducting a poll, knowing who is registered can be one helpful filter.
One reason journalists occasionally use voterfiles is to look up the voting history of a political candidate or other public official. The voterfile can be a good source, but I would avoid using it as the only source, as typos are not unheard of. Many states have a voter lookup portal that is much easier to use than downloading an entire voterfile. For example, North Carolina requires only a person's name to look up their vote history. Other states might require date of birth or city (usually findable for public figures), or other harder to find identifying data. It is worth checking for a lookup site before downloading the entire
Here are a few things reporters I've worked with often request that cannot be done with a voterfile.
Sorry, no can do. As discussed above, the voterfile tells you whether someone voted, not who they voted for.
A voterfile is a snapshot in time. It tells you information about people who are currently registered to vote in a given state. Anyone who was registered but has since died or moved is no longer on the file. So any analysis of registration leading up to, say, the 2014 election is not possible using a current voterfile. Thankfully, many states provide these stats historically, assuming you are happy with the geographic and demographic groups they use. Only if you had the forethought to save a 2014 snapshot from a similar time will you be able to run a year-over-year comparative analysis from a voterfile.
We can try, but it probably won't work. Since the voter has moved, we can't count on address to identify them, so we're probably limited to name and year of birth. While that will uniquely identify some voters, the risk of mismatches is quite high, especially among people with common names. If you are looking to talk to a few people who have moved, this approach will likely find you some. If you're looking for an aggregate statistic, I'd recommend caution. Also movers will eventually be removed from the old voterfile, so your match rate will also depend on how quickly the first state cleans movers off their file.
Individuals who have reason to fear for their safety can have their information removed from the public voterfile in some states. The categories of people who can be removed also vary by state and may include law enforcement officials; victims of domestic abuse, human trafficking or stalking; individuals who have an active restraining or protective order; workers in the reproductive health field. The paperwork burdens are typically rather high. More information here (note, this link will download a PDF)
Voter registration laws vary from state to state. So does the information a state collects. And the format of the information that the state provides. For these reasons, I recommend some caution in doing cross-state comparisons using voterfiles without having a very good understanding of the states' regulations and data policies. For example, the deadline to register before an election or the date of a primary election could have large impacts on turnout and registration, so a spike in one state compared to others needs to be thoroughly investigated in that context. In addition, you should be aware of any changes in state law (this is true for within-state comparisons, too). For example, in 2017, the Governor of Virginia restored voting rights to some 150,000 felons. North Carolina allowed 16- and 17-year-olds to pre-register to vote from 2010-2013, then eliminated the program, and then was ordered by a court to reinstate the program, which they are fighting on appeal.
- A house divided: Central Texas families who cast opposite votes, Austin American-Statesman, May 17, 2018
- Young People Keep Marching After Parkland, This Time to Register to Vote, The New York Times, May 20, 2018