Analysis of Selected 2008 Voter Registration Databases

Last updated: Oct. 18, 2008

Who are those new people being registered to vote? Who are the people being removed from the voter rolls? Some states provide easy access to their statewide voter registration files that permit answers to these questions. An analysis for Nevada, North Carolina and Ohio and government-generated statistics from Kentucky and Wisconsin are presented below. I hope to prepare a similar analysis for Florida and New York when I receive the voter files I have requested from these states.

Background

States vary on what information they provide on their voter files, as I discuss in a peer-reviewed article "The True Electorate" published in Public Opinion Quarterly ($). Most states provide registration date, voting history and other election administration related information. IN addition, Ohio provides year of birth and party registration (as per their record of voting in the last party primary) and North Carolina provides date of birth, party registration, sex and race.

Florida

I still do not have the Florida file, but statistics through August, 2008 posted on the Florida Division of Elections website indicates there were 562,621 new registrations and 116,034 purged registrations.

Kentucky

Kentucky provides a report of new registrations by party registration. The last month reporting was Sept. 2008.

Nevada

The file analyzed here is the Oct. 17 file with the latest new voter registration listed as Oct. 16. Nevada's voter registration deadline for the 2008 general election is Oct. 14 for in-person registration and an Oct. 4 postmark for mailed applications. I do not know if the 19 new registrations with a post-Oct. 16 date are for persons ineligible to vote or if the state identifies new mail registrations by the date received rather than the postmark.

The Nevada voter registration file contains a record of birthdate and party. The state generates reports of all registered voters by party and age. Here, I report these demographics in 2008 for the 257,157 persons registering in 2008. For age, I calculate a person's age on Nov. 4, 2008.

The number of new registrations by party favor the Democrats by over a 2:1 margin. Indeed, Republicans barely outpaced "Non-Partisan" registrants. Interestingly, the age distribution is less skewed towards the youth than in other states. Presumably, this is a consequence of the substantial migration into this fast-growing state by older persons.

It may seem odd that there are persons under age 18 by the Nov. 4 election. These younger persons are permitted to register to vote. However, they are flagged with a "P-17" code in their registration status which presumably prevents them from voting until they become eligible to vote.

I further plot out the number of new registrations by date to get a sense of when people were registering. The single highest day of new registrations was the 7,701 filed on the day of the Jan. 19 caucus. The Democrats permited Election Day registration for their participants. The number again ramped up as the Oct. 4 mail-in registration deadline approached, peaking at 7,617 on that date. The number peaked again to 6,084 on the last day of the in-person registration. Again, note that it is possible that all new registrations have not been processed.

North Carolina

The most current file available at the time of this writing is dated Oct. 11, 2008. According to the North Carolina State Board of Elections calendar, the registration deadline is Oct. 10. Similar to the Ohio file, the most recent new registration was recorded on Oct. 9, 2008, suggesting that the localities have not finished processing and uploading new registrations into the statewide file. Indeed, the state accepts mail-in applications postmarked by Oct. 10. Additionally, the state permits what it calls "One-Stop Voting" where an eligible voter may register and vote early in-person during the period of 19 to 3 days prior to the election. So far, a total of 674,724 new registrations have been filed with the state in 2008.

North Carolina provides a wealth of information on their statewide voter registration file. I have choosen to present statistics for new registrations by party registration, age, sex, ethnicity/race, and a voter status field.

The demographics suggests that the new registrants fit a profile of Sen. Obama supporters. On party registration, 47.9% of new registrants affiliated themselves with the Democrats and Unaffilated voters outpaced Republicans 30.2% to 21.7%. Nearly half - 47.9% - of all new registrants are under the age of 30. Women outpace men by seven percentage points, 51.9% to 44.8%.

North Carolina tracks race and ethnicity of registrants to provide data to the federal government to demonstrate compliance with the Voting Rights Act. Similar to the United States Census Bureau, North Carolina asks separate questions for race and Hispanic/Latino ethnicity. I have classified responses into four categories: non-Hispanic White, non-Hispanic Black, Hispanic of any race, and a residual other category. Here, 30.2% of new registrants are non-Hispanic Black. As a point of comparison, discussed in "The True Electorate" the 2004 exit polls found that 25.8% of North Carolina voters were Black. A voter file I obtained in 2005 after the 2004 voting history had been updated indicates 18.6% of all 2004 North Carolina voters were non-Hispanic Black and the Census Bureau's Current Population Survey reports 21.6% of all 2004 North Carolina voters were non-Hispanic Black. The higher proportion of African-Americans among new registrations may be an indicator of a shift in the composition of the electorate; however, these are registrations, not voters, so it remains to be seen if these voters indeed show up at the polls.

The voter status code is taken directly from a field in the statewide voter registration file entitled voter_status_reason_desc. My understanding of North Carolina procedures is that upon registering, local officials send out a "verification" notice. If the notice does not come back, the voter is "verified." If the notice is returned as undeliverable, a second mailing is sent out. If this second notice is returned undeliverable, then the voter's registration is denied. There are thus two stages in the verification process where a registrant may be "unverified" because no mailing has been sent out or "pending" because the first or second notice is in the mail. The 152,937 new registrations that are pending verification speak to the high volume of new registrations that are still being processed.

Separate from the verification process is a confirmation process, which is triggered whenever a mailing to a registrant (separate from the verification process) is returned as non-deliverable. Here, 184 records indicate a confirmation was not returned or undeliverable. In this situation, a voter is moved into an "inactive" status and will be required to fulfill additional procedures before being permitted to vote. If these inactive voters do not cast a ballot for two federal general elections, they are removed from the rolls.

North Carolina 2008 New Registrations Summary Statistics

To date, the single highest day of new registrations - 32,234 - occured on the April 11 registraton deadline for the May 6 primary. The next highest number - 12,692 - occured on the date of the May 6 primary. Note that most likely, all new registations have yet to be processed and the One-Stop registration and early voting period begins on Oct. 16 and extends through Nov. 1.

Ohio

The most current Ohio statewide voter file available - at the time of this writing - was last updated on Oct. 6, 2008, the day of registration close for the Nov. 4 presidential election. The breakdown of new registrations by age and voting in last party primary is as follows:

Ohio 2008 New Registrations Summary Statistics

The party numbers may seem surprising since Democrats and Sen. Obama's campaign have been credited with a well-organized voter registration effort. These are the number of people who voted in the March 4 party primaries, not those who registred with a particular party. Of the 141,395 people who registered by the Feb. 4 registration deadline, 134,468 or 95.1% voted in the primary election. Those who were new registrants in 2008 prior to the primary were only slightly more likely to vote in the Democratic primary over the percentage of those overall who turned out to vote in the 2008 Ohio primary. Sixty-five percent of all voters who participated did so in the Democratic primary while 70% of new registrants participated in the Democratic primary.

The age distribution of new registrants appears to favor Obama, with 422,048 or 49% of new registrants being between the ages of 18-29. However, these are exactly the sorts of people who are not registered in the first place. For many, this will be the first election that they are eligible to participate in and others must register since they have recently moved into the state.

When the number of new registrations are plotted out by day, the the Feb. 5 deadline stands out as the single highest date of new registrations. The second highest number of 22,500 new registrants who registered on the March 4 primary date are likely persons who tried to vote but discovered they must be registered first.

Ohio has a period where early voting and registration overlap, between Sept. 30 and Oct. 6, making it possible for voters to register and vote at the same time. The new registrations here suggest that a modest number of people at best took advantage of this provision in state law. 42,355 new registrants were recorded during this period.

In general, the pattern suggests that there was not a significant spike in new registrations as the registration deadline approached, as typically happens in previous elections and happened for the Feb. 4 primary. However, we most likely do not have a full picture of the last minute registrations. The file analyzed here is for Oct. 6, the last day of registration, and these files include no registrations filed on Oct. 6. Election officials have most likely not completely processing their new registrations. I will update these numbers and charts as new data become available.

It may still be possible that no last minute spike in new registrations occurred even after the registration numbers are finalized. This would be consistent with the greater attention paid to this campaign by the American citizenry. People are interested and aware of the election and have likely already registered on their own or approached by a campaign or other third party organization conducting a registration drive.

Purging

Purging is the term used by election administrators to refer to records that are removed from the voter rolls. A recent analysis conducted by the New York Times for the story States’ Actions to Block Voters Appear Illegal found purging in excess of new registrations and improper use of the Social Security Database to verify new voters' identities in some battleground states. Ohio was of the latter types:

Although purging was not alleged to be a problem in Ohio, I can assess who was purged over the course of the last year by comparing the Jan. 20, 2008 voter registration database to the Oct. 6, 2008 database. I obtained a Feb. 1, 2008 file from the same Ohio Secretary of State website, though the file itself is no longer available. A record is considered purged by matching the SOS_VOTERID field across the two files, which is an unique identifier assigned to each record in the statewide voter registration file. While in theory voters should be assigned the same SOS_VOTERID if they move within Ohio, a 'purged' record indentified by this method could be one that was removed by one county and an entirely new record initiated by another county.

In all, there were 327,323 records purged over the course of most of the year to date. With 859,734 new registrations added, new registrations outpaced purging by 532,411 records. Unlike the new registrations, the purged registrations appear to be a little more evenly distributed on party affiliation - in terms of primary voting - and age (note the last primary a voter participated in may not be the 2008 presidential primary). For age, both younger and older voters tend to be removed for different reasons, younger people because they have moved and older people because they have moved on to a better place, and all voters may be removed under the National Voter Registration Act of 1993 (a.k.a. 'Motor Voter') if a post card sent by an election office is returned undeliverable and a voter has not voted in two federal general elections or if a voter requests their record be removed, typically by registering in a new jurisdiction.

Wisconsin

Provides statistics for new applications since January 1, 2008. Keep in mind that Wisconsin is an Election Day registration state and that the age cetagories are not consistent with the other states analyzed here. With that in mind, the age distribution of new registrations appears simialr to North Carolina and Ohio.