North Carolina's Shell Game of Electoral Apartheid:
How the Board of Elections Shoved Black Voters Away From the Ballot Box in 2014
In the election of 2012, North Carolina Republicans achieved full control of state government for the first time since 1896.
Soon after, in a move which opponents charge targeted African American voters (among whom Democrats outnumber Republicans by more than 30-to-1), the Republican legislature and Republican Governor McCrory rushed into law the sweeping Voter Information Verification Act (VIVA), rolling back past election reforms which black voters disproportionately rely on — lopping a full week off the Early Voting period and ending both same-day voter registration at the polls and out-of-precinct voting — while imposing strict voter ID requirements which many poor and minority voters find difficult to meet. The U.S. Dept. of Justice and co-plaintiffs including the League of Women Voters and the North Carolina chapter of the NAACP are currently challenging VIVA in federal court (and we're proud to have contributed evidence to their case).
But laws like VIVA, and more familiar legislative shenanigans like gerrymandering, are not the only voter suppression tools available to protect a ruling party's position of power. Administrative actions by agencies — such as North Carolina's state and county boards of elections — can likewise pack powerful punches. By law, boards of elections in North Carolina are dominated by the party in power. And in 2014 Republican-dominated county boards of elections changed the locations of one third of all Early Voting sites in North Carolina, citing "efficiency" and "cost-savings."
insightus researchers have just completed the first-ever quantitative study to determine how that tsunami of polling place relocations which swept across North Carolina in 2014 impacted black versus white Tar Heel State voters. We did this by acquiring six and a half million current voter registration records (which include racial and home address information) and using those data to calculate each voter's distance from her home to the nearest Early Voting location in her county, for both 2012 and 2014. And our findings are a call to action.
Time is the principal cost of voting: time to register, to discover what parties are running, to deliberate, to go to the polls, and to mark the ballot. Since time is a scarce resource, voting is inherently costly.
- Anthony Downs (1957). An Economic Theory of Democracy
Why Distance-To-Poll Matters
Both common sense and a large body of scientific research1 indicate that a registered voter's probability of casting a ballot depends, in part, on his distance-to-poll (usually expressed as the straight-line distance between his home and his assigned polling place). So distance-to-poll matters to friends of democracy because turnout at the poll matters, and more distant polls equal lower election turnouts.
Distance-to-poll also matters to friends of an equitable society because it represents a cost of voting. That cost is measured not merely in the gasoline consumed, or the bus fare, or the shoe leather invested in traveling to the poll, but also in what an economist would call the opportunity cost involved. Time spent traveling to the poll and waiting in line is time not spent at work, and therefore income not earned. That's a cost which weighs far more heavily on low income voters, like the counter worker at McDonald's, than it does on affluent voters such as the owner of a fast-food franchise. Viewed in this light, inequitable distance-to-poll changes amount to a kind of poll tax — something which an abundance of case law has established is unconstitutional.
Finally, distance-to-poll matters to friends of civil rights because the consequences of four hundred years of systematic oppression don't just magically disappear after a mere forty years of half-hearted relief. The economic effect of that lingering legacy on African Americans' mobility today is something we can quantify: 16% of black North Carolina householders do not have access to an automobile, while only 4% of whites do not, and blacks who work outside the home are three times more likely than whites to rely on public transportation.
Tear gas and riot police envelop voting rights marchers in Selma, Alabama, 1965
How Black and White Tar Heels' Distance-To-Poll Is Changing
Between the 2012 and 2014 elections the total number of Early Voting sites operated by North Carolina county boards of elections, statewide, increased modestly: from 363 to 366. But that same period witnessed substantial changes in those sites' locations. According to Voting Information Project data, 114 sites operating in November 2012 were no longer open in November 2014, replaced by 117 different sites. The impact on the average voter (across all races) was fairly insignificant: an increase from about 3.5 miles in 2012 to 3.6 miles in 2014 — a difference of only about 300 feet (roughly one city block).
But a look at the aggregate impacts by race reveals a startlingly different picture (Figure 1, below). While the average white voter's distance to his or her nearest Early Voting site increased by just 26 feet in 2014, the average black voter's distance increased by a quarter of a mile. Summing that up over the members of each race, that's an aggregate increase in distance-to-poll of just 21,000 miles for white voters (71% of the electorate), but more than 350,000 miles for black voters (22% of the electorate). That latter distance is the equivalent of a trip from the Earth to the Moon, and half way home again.
Figure 1: North Carolina's Early Voting polling place relocations in 2014 moved black voters significantly farther from the polls than white voters. Top panel: Total change in distance-to-poll from 2012 to 2014 for all African American voters ("BLACKS"; brown bar) and all white voters ("WHITES; beige bar). Bottom panel: Average individual change in distance-to-poll from 2012 to 2014.
Data Sources Used in This Project:
- Voter addresses & races: NC State Board of Elections voter registration files
- Voter residence geocodes: NC State Board of Elections data files (supplemented by a commercial geocoding service)
- Early Voting site addresses: Voting Information Project data feed
- (Early Voting site geocodes were provided by a commercial service)
- Supplemental demographic data: U.S. Census & American Community Survey data
- See, for instance, Gimpel and Schuknecht (2003) DOI; Haspel and Knotts (2005) DOI; Brady and McNulty (2011) DOI.
- During the Early Voting (EV) period in North Carolina, a voter may vote at any EV location in his or her county. For this report we call the straight-line distance between a voter's home address and the nearest EV site in the same county that voter's 'distance-to-poll.'
Six Counties (A Sampler of Shame)
Each of North Carolina's one hundred counties has its own board of elections, which is responsible for proposing the numbers and locations of polling places in its own jurisdiction (the State Board simply approves or denies these proposals). Thus, ignoring the possibility of collusion between state lawmakers, the State board, and county boards, it is to the individual counties alone which we must look in order to understand how the pushing back of black voters away from their Early Voting polling places was achieved in 2014, and what it accomplished. Here follow a few examples. The six counties discussed below did not, themselves turn the tide for white North Carolina voters. Rather, they simply provide good illustrations of the range of polling place shenanigans our study discovered — games which played out across the majority of the Tar Heel State's counties in 2014.
(For summary data on all 100 counties, see Supplemental Figures 1 and 2, at right)
Orange County: Where Horses Vote
Orange County (population 140,000) is largely rural/suburban, but with one important population center in the southeast — Chapel Hill, home to a large and prosperous campus of the University of North Carolina — and the much smaller city of Hillsborough (which is something of an affluent artist's colony) in the central region. Due to the overwhelming influence of these two communities, Orange is a dependably Democratic county.
Figure 1: Orange County, NC. Top panel: Map of Orange County displaying the distribution of white voters (green clouds) and black voters (orange clouds). Also shown are the locations of Early Voting sites open in 2012 only (i.e., closed for 2014; green circles), sites first opened in 2014 (red circles), and sites open in both years (green+red circles). Bottom panel: The aggregate (upper chart) and individual average (lower chart) changes in distance-to-poll for white voters (beige bars) and African American voters (brown bars).
Orange relocated two of its five Early Voting sites in 2014. A site on the University of North Carolina campus in Chapel Hill was moved off-campus and into the city, and the county's northernmost site, serving a small multi-ethnic community, was shuttered and replaced with a new site far to the east — an area of extremely low population density most notable as the home to a handful of affluent gentlemen's horse farms. In the aggregate, these changes added most to whites' total distance-to-poll (a 6,000 mile increase, versus less than 5,000 for blacks). But at the individual level — where the rubber meets the road, so to speak — blacks paid the heaviest toll: an average increase of about 0.4 miles per person, versus less than a tenth of a mile for whites.
Hertford County: In a Poor Land, Draconian Poll Closings Hurting Everyone
Hertford County is majority black, mostly rural, has a high poverty rate (26% versus 17% statewide), and is reliably Democratic.
Figure 2: Hertford County, NC. See Fig. 1 legend for the definitions of symbols.
In 2012, Hertford County voters were served by three Early Voting sites, one in each of three multi-racial communities. But in 2014 the two westernmost of those sites (serving the county's largest communities) were closed, leaving just a single site to serve the entire county. As a result of those closings, in 2014 Hertford County voters found themselves a total of about 75,000 miles further from the last remaining ballot box. Collectively, African American voters bore the largest share of this added burden — nearly 50,000 miles — compared to about 25,000 miles for all white voters, because the two now-unserved western communities are majority black. At the individual level however, the average black and white voter each saw his distance-to-poll increase by the same margin: just over 5 miles. Hertford County holds the record for the largest increase in average distance-to-poll imposed on its voters by the county board of elections.
Closing the majority of a county's polling places, as Hertford did in 2014, amounts to passing on to voters one of the single most fundamental costs of democratic government: the cost of administering elections. The county saves money because it has fewer sites to staff and operate, while voters pay the cost of increased travel. At the IRS-accepted mileage allowance for auto travel in 2014 (56 cents per mile) this amounted to an unfunded mandate imposed by Hertford County on its voters (but not on non-voters) of about $41,000.
Moore County: More, Or Less?
Demographically, Moore County is the opposite of Hertford County: whiter and wealthier than the state overall (only 13% black, versus 22% for the state), and reliably Republican. The income gap between the races is particularly wide here: blacks' per capita income is just 40% that of whites (versus 62% statewide).
Figure 3: Moore County, NC. See Fig. 1 legend for the definitions of symbols.
Like Hertford's, Moore County's board of elections also imposed a substantial cut in Early Voting sites, down from four in 2012 to just two in 2014. And here too the puzzling effect was to move Early Voting sites away from the county's major population center (and particularly far from its only large black community). Because most Moore voters are white, the aggregate burden fell mostly on them (a 175,000 mile increase, as opposed to 40,000 miles for the county's small black population). Black voters bore the heaviest individual burdens however, with an average increase in distance-to-poll of nearly 5 miles, while whites saw an average 3 mile increase.
Caswell County: Even When They Win, Blacks Lose
Tiny Caswell County (population 23,000 and shrinking), at 34% African American and 64% white, is poised demographically between predominantly black Hertford and predominantly white Moore counties. But only 9% of adults have college degrees (versus 27% statewide), the poverty rate is high (26%), and blacks and whites are more nearly equally impoverished (black per capita income is 74% of white).
Figure 4: Caswell County, NC. See Fig. 1 legend for the definitions of symbols.
Caswell is an example of a county which significantly increased the number of its Early Voting sites in 2014 — to 4 from just one in 2012. That single 2012 site, in the county's only city, Yanceyville, was central to Caswell's black community. But two of the new 2014 sites (in the northwest and east central parts of the county) serve predominantly white populations (93% and 82% white, respectively), while the swelling black community in the county's northeastern corner (45% black) remains unserved. As a result, while both aggregate and individual distance-to-poll declined for Caswell blacks and whites alike, whites enjoyed the greatest benefits — 71% percent of the total reduction in distance-to-poll, while comprising just 64% of the population. African Americans moved closer to the polls in Caswell County in 2014, but not as much closer as did their white neighbors.
Catawba (population 155,000) is unusually white (85% versus 77% statewide), and its poverty rate (17%) is comparable to that of the state as a whole. The county's modest black population (typical of western North Carolina) is highly concentrated on the south side of its largest city, Hickory.
Figure 5: Catawba County, NC. See Fig. 1 legend for the definitions of symbols.
Like Caswell, Catawba added Early Voting sites in 2014. Two of those three new sites serve nearly exclusively white communities, while the southside Hickory black community remains without a site of its own. These changes produced a very large decline in white Catawba voters' aggregate distance-to-poll (a 160,000 mile reduction, versus 8,000 miles for blacks) as well as a two-to-one advantage in individual distance-to-poll (a 2 mile decrease, versus a 1 mile decline for African Americans).
Mecklenburg County: Balanced On A Knife-Edge
North Carolina's largest county by population (just over 1 million), this region is a study in contrasts: home to major corporate headquarters like Bank of America, Lowes, and Duke Energy (and the well-paid executives and technologists they bring), but likewise to large minority communities in deep poverty. While the overall poverty rate is low (15%), that statistic hides great income disparities: blacks' per capita income is just 50% that of whites. The population is diverse (49% non-Hispanic white, 32% black, 13% Latino), but large swaths in the north and south are predominantly white, while the county's heart is predominantly black. Its politics are mostly purple: Bush edged out Gore by the thinnest of hairs (just a few hundred votes) in 2000, while Kerry topped Bush by only a few thousand votes in 2004. Obama, however, won the county by wide margins (61% and 62%), demonstrating the powerful potential of African American turnout.
Figure 6: Mecklenburg County, NC. See Fig. 1 legend for the definitions of symbols.
Mecklenburg went from 22 Early Voting sites in 2012 to 21 in 2014, and four sites saw relatively small changes in their locations between those years. By far the two most impactful changes, however, were the net loss of one site from the African American community in the central county, and the net gain of one site in the predominantly white southernmost region.
The net effect of all of these changes was the addition of nearly 40,000 miles to blacks' aggregate distance-to-poll — three times that of whites — and a small increase in blacks' average distance (about 0.2 miles) which was, nonetheless, an order of magnitude larger than whites' average change (0.03 miles).
Can two tenths of a mile really matter? Research suggests that, in an urbanized and too-close-to-call environment like Mecklenburg county, it just might. In their 2011 study of distance-to-poll effects in a Los Angeles County election, Brady and McNulty1 concluded:
For each one tenth of a mile increase in distance up to 0.4 of a mile, polling place voting decreases by 0.50%
A more useful way to ask the question, then, would be: could a 1% decrease in black voter turnout (about 2,100 voters) in precariously balanced Mecklenburg County really matter? And the answer is yes: that small number of additional Democratic voters turning out in 2000 would have been more than enough to move Mecklenburg County from the Bush (R) to the Gore (D) column. And anyway, voter suppression, like boiling a live frog, is a game of tiny increments — a voter ID requirement here, a paring of Early Voting days there, the confusion everywhere of discovering that last election's site has disappeared, and much more. Add to the mix that two tenths of a mile increase in distance-to-poll...almost beneath notice in isolation...and a knife edge-balanced county like Mecklenburg can almost be relied upon to fall the way the party in power wishes it to.
Where We Go From Here
There are many potentially good reasons to change polling place locations between elections: to improve accessibility, to keep up with population growth trends, to simplify election administration and reduce waste or error being just a few of them. Always lurking in the background, however, is the potential terrible motivation: to protect the party in power by making it more difficult for the 'wrong' voters to cast a ballot. But even when changes are made for all the right reasons, failure to carefully consider their likely impacts on voters of color may still have markedly undemocratic — and unlawful — impacts on the electoral process.
In the real world, it isn't realistic to demand that election administration changes should have absolutely no measurably different impacts on voters based on race. But in the case we have uncovered here in North Carolina — where a flurry of polling place relocations added 350,000 miles to African American voters' aggregate distance-to-poll, but just 20,000 miles to that of white voters — it is painfully obvious that an unacceptable disparity was (either intentionally or unintentionally) the terrible result.
Both the U.S. Constitution's 15th Amendment and Section 2 of the Voting Rights Act provide that voting rights "shall not be denied or abridged by the United States or by any state on account of race, color, or previous condition of servitude." To "abridge" is to reduce, and we think it is obvious that to move the ballot box dramatically farther away from voters of one color than from those of another hue is a stark abridgement of the former's voting rights. Bold violations of the 15th Amendment are nothing new in post-Civil War southern politics, which previously introduced to the world such innovations as poll taxes, literacy tests, and even state-sponsored terror, all with the aim of suppressing the African American vote. And here we are, once again. Each of those past outrages was battled against, and those battles were won, by people of goodwill. And it is time to fight once more.
In the coming months, insightus will work with like-minded organizations to fight — in both the court of law and the court of public opinion — for redress of North Carolina's obviously unlawful shell game of electoral apartheid. But while that state is our first focus, we also intend to use the methods developed for this study to closely scrutinize other states' poll-shifting gambits too, and to take the fight to the worst of those offenders as well. Because where there's smoke, there's fire.