NVRA requires the states to provide citizens with the opportunity to register to vote for federal elections by three important means:
1. At the same time that they apply for or renew driver's licenses
2. At the same time that they apply for Public Assistance (PA) programs, such as food stamps or welfare
3. By mail
NVRA also requires the states to collect data regarding how many voter registration applications are generated via each of these means, and to report summaries of those data to the U.S. Election Assistance Commission (EAC) every two years. EAC, in turn, is required to submit a summary report to Congress biennially (the most recent such report covers the 2011-2012 period; the 2013-2014 report is due by June 30th of this year).
From 1995 through 2012, the North Carolina State Board of Elections (SBOE) published on its web site annual summaries (in the form of Excel spreadsheets) of its NVRA compliance data. But, beginning in 2013 (when McCrory took office), that practice appears to have come to a halt, and no annual summaries are available there for McCrory's term (2013-2014). Monthly data files (for the period May 2010 through March 2015) are still made available on the SBOE web site, but these are comma-delimited (CSV) machine-readable data sets, not easily understood by average readers.
Curious to know how NVRA compliance was faring in North Carolina under McCrory - and troubled by the absence of recent annual summaries by which this could be judged - my colleagues and I recently unpacked and data-mined the monthly data files available at the SBOE web site. And what we've found is nothing short of alarming.
Finding 1: A systematic sharp decline in new voter registrations originating from Public Assistance (PA) programs began on or about January 2013 and continues to this day
Figure 1, below, summarizes statewide new voter registrations originating from PA programs, by month, and compares these with new voter registrations originating from the Department of Motor Vehicles (DMV).

Fig. 1: North Carolina new voter registrations originating via Public Assistance programs (top panel) and via the Dept. of Motor Vehicles (bottom panel) from May 2010 through March 2015. Red and green horizontal lines indicate overall averages for the periods May 2010 through December 2012 (green lines; “Pre-McCrory Average”) and January 2013 through March 2015 (red lines; “McCrory Average”). Months for which reports are missing, or contain incomplete data, are excluded from these averages (5/2010, 9/2010, 3/2011, 5/2011, 8/2011, 5/2012, 6/2012, and 3/2015).
As Fig. 1 illustrates, new voter registrations originating via PA programs have declined substantially and continuously during the McCrory administration (from a pre-McCrory baseline averaging 2,187 per month to a McCrory-era average of 736 per month - a 66% decrease). This difference is highly statistically significant (P << 0.05). By contrast, new registrations originating from the DMV have declined only a statistically insignificant 17% during the same period. Taking the average monthly difference in pre-McCrory and McCrory registrations (-1,451 per month) and multiplying this by the 27 months of the McCrory administration covered here, suggests an overall deficit of 39,177 ‘missing voters’ (i.e., NC citizens who would have been registered had this decline not occurred).
Finding 2: The 'McCrory Deficit' is unlikely to be due merely to declining PA applications
A potentially benign explanation for what I shall from now on call the 'McCrory Deficit,' illustrated in Figure 1, might simply be that over the past two years applications for Public Assistance have fallen in NC, perhaps due to improving economic conditions, which would in turn naturally lead to a similar decline in the number of voter registrations originating via PA programs. In order to test this possibility, we downloaded NC Dept. of Health and Human Services data regarding the number of NC state Food & Nutrition Services (FNS; mostly food stamps) applications received, by month. FNS applications greatly outnumber other PA applications (for example, those for NC’s Work First welfare program), thus making this the best data set to use to test the hypothesis.
Figure 2, below, illustrates the time course (by month) of new voter registrations originating via PA, expressed as a percentage of the total number of FNS applications for each of those same months. This way of looking at the data controls for any possible effect that declining PA application numbers might naturally have on PA-originating voter registrations.

Figure 2: North Carolina new voter registrations originating via Public Assistance programs, expressed as a percentage of Food & Nutrition Services applications, from May 2010 through March 2015. Red and green horizontal lines indicate overall averages for the periods May 2010 through December 2012 (green line; “Pre-McCrory Average”) and January 2013 through March 2015 (red line; “McCrory Average”). Months for which reports are missing, or contain incomplete data, are excluded from these averages (including all of the months listed in the legend to Fig. 1, plus months for which FNS application numbers were underreported during the switchover to the Dept. of Health and Human Services new NC FAST reporting system: 9/2012 through 11/2012 and 1/2013 ).
As Figure 2 illustrates, McCrory-era PA voter registrations are equivalent to just 1.7% of all FNS applications, very substantially lower than the pre-McCrory baseline value of 4.2% (i.e., 60% below the pre-McCrory application-adjusted baseline). The McCrory Deficit is, therefore, not a simple result of declining PA applications during his administration. Finding 3: The ‘McCrory Deficit’ occurs statewide, not merely in a handful of counties
Another potentially benign explanation for the significant drop in voter registrations originating via PA programs during the McCrory administration might be that only a handful of 'rogue' county PA offices (and their managers) are to blame for discouraging PA applicants from registering to vote (a situation which might potentially exonerate the McCrory administration itself). Figure 3, below, illustrates that this is not the case: declines in PA-originated new voter registrations during the McCrory administration have occurred in 99 of North Carolina's 100 counties, presenting the possibility that the McCrory Deficit may be the result of statewide administrative policy mandated by the McCrory administration itself.
attribution: DocDawg at Daily Kos

Figure 3: Percent change in new voter registrations originating via Public Assistance programs (from the pre-McCrory baseline period to the McCrory period), by county. Months for which reports are missing, or contain incomplete data, are excluded from these averages (including all of the months listed in the legend to Fig. 1).
Conclusion Today, North Carolina's voter registration rolls are missing some 40,000 or more poverty-level citizens (and still counting), due to what may prove to be systematic actions by the McCrory administration, in violation of the National Voter Registration Act of 1993. To put that number of voters in perspective: in the 2014 election, Thom Tillis (R) beat incumbent Kay Hagan (D) for his current U.S. Senate seat by roughly the same number - just 45,608 votes.
Prior to publication of this diary, I have shared our findings widely with a number of organizations whose interests include defending voting rights, as well as with the Raleigh News & Observer. These revelations have met with considerable interest; expect to hear much more regarding this story in the weeks and months to come - including right through McCrory's 2016 re-election bid, the upcoming oral arguments in NC-NAACP v. McCrory, and the massive demonstration by voting rights supporters from around the nation which the NC NAACP will conduct in Winston-Salem, NC during that trial (and which NAACP's Rev. William Barber has called "Our Selma").