By Michael Crowell, Professor of Public Law and Government at the University of North Carolina:

Hotly debated before its enactment and challenged in lawsuits since, North Carolina’s new photo identification requirement for voting goes into effect in 2016. Is it a genuine bulwark against fraud, providing confidence in elections, or is it a partisan attempt by one political party to suppress the vote of the other? That argument doubtless will go on for some time. A more practical question is how many voters will be affected because they do not have an acceptable ID.

The State Board of Elections took a first bite at answering that question in April 2013 but its report has received surprisingly little attention. Especially surprising because the board has identified by name and address each voter who may not have the necessary identification.

The State Board was attempting to get a general idea of how many people might be affected by the photo identification requirement. To do so, they matched the full statewide list of about 6.5 million registered voters with the more than 12 million Division of Motor Vehicles’ records of state driver licenses and DMV-issued non-driver ID cards. (For simplicity, let’s use “driver license” for both cards.) The goal was to see how many registered voters had a driver license and how many did not. Voters who do not have a driver license may have another form of acceptable voter ID, such as a passport or military ID, but the license is the most common identification and the non-matches should give a good picture of the number of people who could be affected by the new law.

In the first run the State Board matched exact first and last names and driver license or Social Security numbers of voters with the DMV list. (Driver license and Social Security numbers were not required for registration until 2004, and still are not found on older registrations.) But sometimes names do not match exactly because of misspellings, misplaced initials, and similar minor discrepancies, so the State Board went through more than two dozen additional cuts of the data, looking for similar first names, similar last names, matching birth dates, identical street addresses, and a variety of other combinations. In each instance the board was trying to find enough matches to offer reasonable assurance that the voter and driver license holder were the same person even if not every data point was an exact match.

What the State Board found was that it could confidently match over 95 percent of its 6,425,820 voter registration records with a driver license — but could not match 318,643 voters. At first blush, then, it would appear that there could be over 300,000 North Carolinians who are not prepared to meet the voter ID requirement when it takes effect in 2016. Further analysis from the State Board, though, reduces that number significantly.

First, keep in mind that the State Board was attempting to find matches for all the names in its voter registration list, some of whom surely are no longer eligible voters. Voters move away, for example, but their names may remain on the rolls. Since the mid-1990s, federal law has halted purging the names of voters who have stopped voting on a regular basis. To comply with that law, the state now has an active voter list and an inactive voter list (the inactive list is generated primarily by election mailings returned as undeliverable) but all are considered registered voters.

Over 60,000, or about 20 percent, of the 318,000 non-matching names were from the inactive list. Additionally, the board found that 115,000 of the 318,000 voters for whom no driver license could be found had not voted in any of the last five general elections or any election in between. That result suggests that over a third of the voters who appear not to have a driver license are no longer voting in North Carolina, and in a perfect world they would not be on the voter rolls at all.

In the most useful part of the report the State Board looks at 138,425 registered voters who in fact voted in the 2012 general election but do not match as having driver licenses. There is no question these are active voters, and they appear likely to not have a photo ID. Of those 138,000 non-matching voters, 81,000 were registered as Democrats, 30,000 as Republicans, and nearly 27,000 unaffiliated. Thus, of the people who voted in 2012 but may not have a driver license, about 59 percent were registered Democrats, 22 percent were Republican, and nearly 20 percent were unaffiliated. Democrats made up 43 percent of registered voters in November 2012, whereas nearly 31 percent of registered voters were Republican and 26 percent were unaffiliated. A majority, about 44,000, of the 81,000 registered Democrats who did not match DMV lists were African American; about 30,000 were white.

From those raw numbers it would appear that the photo ID requirement is likely to affect Democrats more than Republicans and unaffiliated voters, and is likely to affect black Democrats more than anyone else. A word of caution is in order, however. The data itself raises many questions and makes any conclusion shaky.

For one thing, there is an odd disparity between female and male non-matches. Of the total 318,643 non-matches the State Board came across nearly twice as many were women as men. Similarly, 85,263 or nearly 62 percent of the 138,425 non-matching voters who voted in the 2012 general election were female. It may be that women really are that much less likely than men to have driver licenses — or it may be that disproportionately more female voters failed the State Board criteria for reliable matches because they are more likely than males to change names when marrying or divorcing, resulting in different names on voting and DMV records. Or there may be some other explanation.

Another kind of disparity that raises questions is the difference in the percentages of non-matches from one county to another. Mecklenburg and Wake Counties, for example, both have around 640,000 registered voters, but the State Board found over 40,000 non-matches for Mecklenburg and fewer than 25,000 for Wake. It appears that in most counties about four to six percent of registered voters did not match the DMV database, but a few counties stand out with higher rates of non-matches. In Durham, for example, 19,000 out of 200,000 voters did not square up with driver license records, almost 10 percent. Cumberland and Robeson both had non-matching rates of nearly nine percent, and Orange and Onslow were over seven percent.

What might explain those differences from one county to another? Perhaps the higher rates of voters without matching driver licenses in Cumberland and Onslow are somehow related to the number of military personnel there, and that might even be a factor for neighboring Robeson. And maybe the high number of non-matches in Orange is affected by the large student population. There is no obvious explanation. And what explains Durham? And why the big difference between Wake and Mecklenburg?

The State Board of Elections report did not attempt to dig deeper. It was looking only for a big picture of how many voters might be affected by the photo ID requirement. A closer look at the results — the number of inactive voters, the disproportionate number of women, the differences between counties, etc. — shows that one should not take at face value the suggestion that there are 318,643 voters who do not have a photo ID. And the mix of people who actually do not have IDs, the proportions of women and men, Republicans and Democrats, may be different than in that group of 318,643. Still, one reasonable conclusion from the study is that about 150,000 to 200,000 active voters in North Carolina may not have a driver license, and if they do not have a driver license it seems unlikely they will have one of the other forms of acceptable identification.

Most importantly, the State Board’s study has resulted in a list by individual name of registered voters who may not have IDs and may need assistance in getting one between now and 2016. The consequence of the April 2013 study is that the State Board knows the name, address and precinct of each registered voter whose name does not clearly match a driver license. Election officials, political parties, candidates and civic groups can take that list to contact those individuals and assist them, if needed, in meeting the new identification requirement.