An Evaluation of North Carolina's 1998 Congressional Districts by Professor Gerald R. Webster

    R A 

An Evaluation of North Carolina's 

1998 Congressional Districts 

Professor Gerald R. Webster 
Department of Geography 
University of Alabama 

Tuscaloosa, AL 35487-0322 
Phone (205) 348-1532 
Fax (205) 348-2278 

GWebster@ualvm.ua. edu 

On December 9, 1997, I met with representatives of the North 

Carolina Attorney General's Office and was asked to evaluate the 

state's twelve congressional districts intended for use in the 

1998 elections (hereafter referred to as the "1998 districts"). 

For this purpose I was provided maps of the 1992-1996 (hereafter 

referred to as the "1992 districts") and 1998 districts, and 

access to data from Election Data Services, one of the country's 

most prominent providers of data on elections and districts. 

From Election Data Services I secured the majority of the 

compactness indicators used in this report, and shape files of 

the congressional districts in selected states including North 

Carolina. 

The purpose of this report is to evaluate the twelve 

congressional districts now in place for the 1998 elections 

(Figure 1). Where pertinent, these districts are compared to the 

1992 districts (Figure 2). This report uses a set of selected 

traditional districting criteria on which to base its evaluation. 

These criteria are discussed in the published work of Professor 

1  



Figure 1 

North Carolina Congressional Districts, 1998 

  
  

  
        
    

   



  

        
  

    
  
       



  

Richard Morrill (Department of Geography, University of 

Washington), and Professor Bernard Grofman (School of Social 

Sciences, University of California, Irvine). 

Professor Morrill is a political geographer and the foremost 

geographic authority on districting procedures in the United 

States today. His 1981 monograph, Political Redistricting and 
  

Geographic Theory, includes a chapter entitled "Criteria for   

Redistricting” which was consulted in the preparation of this 

report. Professor Grofman is a political scientist and is among 

the most recognized authorities from that discipline on 

redistricting issues. Professor Grofman's "Criteria for 

Districting: A Social Science Perspective" published in the UCLA 

Law Review in 1985 is perhaps the most exhaustive evaluation of   

redistricting criteria published in the past twenty-five years. 

(Additional sources detailing districting criteria include Dixon 

1982; Morrill 1982, 1987 and 1994; Cain 1984; Butler and Cain 

1992; Grofman et al., 1992; Grofman 1993; Pildes and Niemi 1993). 

REDISTRICTING CRITERIA - 
  

What follows is an evaluation of North Carolina's twelve 

congressional districts on the basis of selected criteria 

including equal population, contiguity, compactness, the 

integrity of local government boundaries, and continuity of 

representation. It should be noted at the outset that it is 

virtually impossible for a redistricting plan to simultaneously 

satisfy all criteria perfectly or completely. In some cases, the 

4 

 



  

ov) ® ® 

real-world application of two criteria may contradict one 

another. In other cases, data and real-world constraints limit 

the degree to which one or more criteria may be simultaneously 

satisfied. The preeminent criterion for evaluating redistricting 

plans is population equality which stems from the United States 

Constitution. It therefore carries far greater weight in the 

evaluation of districts than secondary criteria such as 

compactness or the non-subdivision of local political units. 

Thus, a congressional redistricting plan which includes 

exceptionally compact districts with widely varying populations 

would be found unacceptable. 

Some explanation of each criterion's purpose precedes the 

North Carolina specific evaluation. 

1)Equality of Population. The "equality of population" criterion   

stems from the U.S. Constitution and it therefore must carry 

substantial, if not preeminent, weight in the development of all 

redistricting plans. Since the 1960s the legally permissible 

variations between district populations have declined 

substantially. Concomitant improvements in technology and the 

detail of the data provided by the U.S. census have allowed ever- 

greater adherence to the principal of one-person-one-vote. 

Population equality between districts may be measured by 

multiple indicators. The "total deviation" (also referred to as 

the overall range) is defined as the population difference 

between the smallest and largest districts (see Grofman 1985: 

5 

 



® » 
175, for the total deviations of districts existing in 1983, and 

Rayburn and Leib 1994: 23, for the total deviations existing in 

1993). The "maximum deviation" is defined as the largest 

absolute (+ or =) population deviation among a jurisdiction's 

districts relative to the optimal population (total 

population/number of seats) to be allocated to each district. 

Also pertinent is the mean or average of all district deviations 

from the ideal population. These measures may be expressed in 

terms of absolute numbers and percents. 

The maximum permissible population deviation of districting 

plans depends on their geographic scale - whether, for example, 

the districts are used to elect members of the U.S. House of 

Representatives, a state legislature, or a city council. Due to 

the limits of both the census of population data and geography, 

maximum allowable deviations are generally greater the more local 

the jurisdiction being evaluated. But congressional districts 

are expected to have total deviations of less than 1% (Morrill 

1981; Grofman 1985; Grofman et al., 1992). As determined by 

Rayburn and Leib (1994: 23), the districting plans in all states 

in 1993 were in compliance with this criterion with the largest 

total deviation determined for Georgia at 0.94%. 

The ideal population for each congressional district in 

North Carolina is 552,386 (1990 state population of 6,628,637 / 

12 districts = 552,386.42) (Table 1). The average absolute (+ or 

-) deviation of North Carolina's twelve congressional districts 

is 361 individuals, or 0.065% of the ideal population to be 

6  



  

Table 1 

Population Equity Between North Carolina 
Congressional Districts Existing in 1998% 

Absolute Percent 
District # Total Population Deviation? Deviation? 
  

  

  

l 552,161 225 -0.041% 

2 ; 552,152 234 -0.042 

3 552,622 236 +0.043 

4 551,842 544 -0.098 

5 552,084 302 «0.055 

6 552,171 215  =0.039 

7 552,382 4 -0.,001 

8 553,143 757 +0.137 

9 552,615 228 +0.041 

10 553,333 947 +0.171 

11 552,089 297 -0.054 

12 552,043 343 -0,.062 

Total 6,628,637 -— -— 

Mean 552,386 361 0.065% 

Total Deviation® 1,491 0.270% 

Maximum Deviationd 947 0.171% 

  *Based upon the 1990 Census of Population. 
2 Absolute difference between district and mean of 552,386. 
Percentage difference between district and mean of 552,386. 

© The total deviation is the difference between the largest (#10) 
and smallest (#4) district. 

The maximum deviation constitutes that district most deviating 
from the ideal population. 

SOURCE: Calculated by author from data provided by State of North 
Carolina and reflecting the 1990 Census of Population. 

- 

 



allocated to each district. Among the districts, District 7 has 

the smallest deviation at a mere 4 individuals (0.001%), and 

District 10 the largest at 947 individuals (0.171%). District 

10's deviation thus constitutes the Maximum Deviation for the 

state's congressional districts. The Total Deviation for the 

twelve districts is 0.270%. 

In terms of the population equality criterion, North 

Carolina's present congressional districts are well within 

accepted guidelines. Their level of population equality is 

therefore evaluated favorably on this criterion. But it should 

be noted that 1990 census of population data is now eight years 

old. Based upon population projections by the North Carolina 

Office of State Planning, the state's 1997 population was nearly 

12 percent greater than enumerated in 1990. That office further 

projects the next Census, to be undertaken in a little over two 

years, will enumerate 7.7 million North Carolinians, or 16.4 

percent more individuals residing in the state than were counted 

in 1990. Thus, while the constitutional purpose of the census of 

population is to aid the reapportionment and redistricting 

process and the 1990 census should be used for such comparisons, 

malapportionment is clearly a result of using eight year old 

data... It. dls thebetire true that any subsequent delineation of 

North Carolina's 12 congressional districts prior to the 2000 

census will be hampered in achieving compliance with the 

preeminent criterion of the districting process, that of equal 

population between districts.  



  

2)Contiquity. 
  

A majority of states (37), including North Carolina, require 

that legislative districts be contiguous (Grofman 1985). A 

district is typically defined as contiguous if every part of the 

district is accessible to all other parts without traveling into 

a second district. This requirement has rarely generated 

controversy in the past. When controversy has emerged, the focus 

of contention has generally pertained to areas dissected by water 

features and whether bridge connections are sufficient to 

constitute contiguity (see Grofman 1985: 84). 

All of the North Carolina's 1998 districts are contiguous, 

and are thus eviduated favorably on this criterion. Second, on 

this criterion the present plan compares very favorably with the 

state's 1992 congressional districts which generated controversy 

pertaining to the contiguity of the 6th and 12th districts 

(Grofman 1993: 1261). The practical contiguity of Districts 1, 

2, 5 and 10 was also substantially improved. For example, the 

proruption of District 2 into District 1 in Halifax County in the 

1992 plan no longer exists. In the 1998 plan the mutual boundary 

of Districts 1 and 2 is the Halifax County-Nash County boundary. 

A second example pertains to the no longer existing proruption of 

District 5 into District 10. The boundary between these two 

districts now entirely follows county boundaries. 

 



  

3)The Allocation of Local Government or Electoral Units. 
  

The non-division of local political units in their 

allocation to districts has traditionally been viewed as a 

legitimate though secondary goal of the districting process. 

Counties, for example, may constitute political systems in and of 

themselves. Political parties frequently use counties as 

subdivisions for their organization at the "grassroots level." 

And citizens may have strong allegiances to their respective 

counties, most particularly in the South (Webster 1984). But it 

is also true that the equal population criterion carries 

substantially greater weight in all evaluations of districting 

plans than efforts to limit county subdivisions. Thus, any 

county which has a population above the ideal population per 

district will almost certainly be divided into two or more 

districts in a system which employs single-member districts. And 

at times it may be appropriate and/or necessary to subdivide 

other counties with lesser numbers of residents to achieve 

compliance with the equal population criterion. 

The 1992 congressional districts in North Carolina divided 

44 off the state's 100 counties (Webster 1995). The 1998 

congressional districts reduce this number to 23, a 48% percent 

decline in the number of county divisions. Thus, from the 

perspective of county divisions, the 1998 districts are superior 

to the 1992 districts on this criterion. 

The building block units used to delineate congressional 

districts differ between states. While some states rely on 

10 

 



  

census units such as blocks or tracts, others use voting 

districts or precincts (Raburn and Leib 1994: 23). Seventy-nine 

of North Carolina's 100 counties are subdivided by voting 

precincts which follow the boundaries of other units of census 

geography such as blocks or tracts. It should be a goal of 

districting plans to avoid subdividing voting precincts wherever 

possible. If these units are subdivided by district boundaries, 

additional polling units or ballot forms may be required. Such 

changes typically lead to voter confusion and frustration. 

As of October 1, 1996, there were a total of 2,531 election 

precincts in the state of North Carolina. The 1992 districts 

divided 80 precincts. The 1998 districts reduce this number to 

only 2. Thus, the 1998 districts are superior to the 1992 

districts in their minimal number of voter precinct divisions. 

Of the 31 counties which do not include census voter 

precincts, only one (Beaufort County) is not allocated to a 

congressional district in its entirety. This division employed 

township boundaries which are vastly superior to the use of 

census blocks or tracts which may have little significance to 

human patterns of interaction or governance. 

4) Geographic Compactness. 
  

Background 

Unlike the equality of population criterion, the geographic 

compactness criterion has no foundation in the U.S. Constitution. 

But it has long been viewed as a legitimate criterion for 

11 

 



evaluating redistricting plans, and is included as a goal in 

  

approximately half of all state constitutions. The state of 

North Carolina is not among those states legally requiring that 

the members of legislative bodies be elected from compact 

districts (Grofman 1985). But given recent Supreme Court 

decisions, the compactness criterion has received renewed 

attention. 

There is substantial legal and academic disagreement over 

the value of mandating districts be compact. First, requiring 

compact districts does not guard against the political or racial 

manipulation of electoral space. The geographic resolution and 

quality of the data now provided by the census in conjunction 

with the increasing sophistication of computer technology may 

allow compact districts to be delineated which are also directly 

discriminatory to a population group. Thus, while irregularly 

shaped districts may suggest manipulation, highly compact 

districts may also be intentionally detrimental to a population 

group (see both Morrill 1981, and Grofman 1985). 

Second, in the real-world perfectly compact districts are an 

impossibility. Most compactness indices assume that an optimal 

district will be a perfect circle, the most compact of geometric 

shapes. But if circles were employed to subdivide the space of a 

jurisdiction, some of the jurisdiction's area would not be 

allocated to districts, but be left in the gaps between circles. 

Thus, circular districts are an unrealistic abstraction without 

direct application to real-world circumstances. 

12 

 



Third, the building blocks of redistricting plans, blocks, 

  

block groups, tracts, or voting precincts, are frequently 

delineated by streets and are oftentimes square or rectangular in 

shape. The shapes of these building blocks therefore largely 

precludes circular districts from being formed. 

A fourth concern limiting real world compactness is the 

constitutional criterion of equality of population discussed 

above. It is of substantially greater legal necessity to comply 

with this criterion than to create perfectly compact districts. 

To meet this constitutional criterion map makers may be forced to 

create districts of less than perfect compactness. As stated by 

Professor Morrill (1981: 22): 

A too simplistic application of . . . geographic 

compactness measures is foolish, especially where the 

distribution of the population is uneven, perhaps strung 

out along roads or railroads. Travel may be easier and 

cheaper in some directions than others, such that an 

elongated district astride a major transportation 

corridor might in fact be the most compact in the sense 

of minimum travel time for a representative to travel 

around the district. 

In short, it must be stressed that legislators represent 

people, and the distribution of the population on the landscape 

is substantially uneven. Given this unevenness it is largely an 

impossibility to create districts which are perfectly compact and 

13 

 



include equal numbers of people. 

Finally, there are a host of different geographic 

compactness measures available (see Niemi, et al., 1990 for a 

description of 24 different measures). In general these methods 

concentrate on a district's perimeter, areal dispersion, or 

population distribution in their design. 

Two compactness measures are used in the present report, 

both of which are now among the most commonly recognized and 

applied by legal and academic scholars. Their elevated 

recognition is due largely to their calculation and application 

in a 1993 Michigan Law Review article by Richard Pildes and   

Richard Niemi. Both authors are among the most recognized 

authorities on redistricting and the courts in the United States 

today. The purpose of their article was to measure the 

compactness of all congressional districts in the United States 

existing in 1993 in such a manner as to parallel the discussion 

in the Supreme Court's decision in Shaw v. Reno (1993). Adding 

to the relevance of this article and its methods was its citation 

in Bush v. Vera, 1996, as supporting evidence for the Supreme 

Court's findings that three congressional districts in Texas were 

unconstitutional. 

The first measure is based on the geographic "dispersion" of 

a district. To calculate this measure a circle is circumscribed 

around a district. The reported coefficient is the proportion of 

the area of the circumscribed circle which is also included in 

the district and ranges from 1.0 (most compact) to 0.0 (least 

14  



  

compact). 

The second measure is based upon the calculation of the 

"perimeter" of the district. The reported coefficient is the 

proportion of the area in the district relative to a circle with 

the same perimeter and ranges from 1.0 (most compact) to 0.0 

(least compact). | 

Pildes and Niemi( 1993: 564) provide some guidance on the 

evaluation of both measures. With respect to the dispersion 

compactness measure, they suggest "low" is equal to or less than 

0.15. On the perimeter compactness measure they suggest that 

"low" is equal to or less than 0.05. By suggesting these 

guidelines they caution that "we do not imply that all districts 

below those points, or only those districts, are vulnerable after 

Shaw." These cutoff points are therefore best characterized as 

general guidelines and they should not be employed as absolute 

indicators of acceptable or unacceptable levels of compactness. 

Supporting their statement is the fact that in 1538 ten 

congressional districts in the U.S. are below the 0.05 benchmark 

on the perimeter measure (e.q., New York's 12th (0.021) and 

Texas' 6th (0.027)), and 13 are below the 0.15 benchmark on the 

dispersion measure (e.g., Florida's 22nd (0.0331) and 

California's 36th (0.042)). 

Instructive Comparisons 

In addition to the compactness scores for each of North 

Carolina's congressional districts, additional scores were 

15 

 



calculated for hypothetical and real-world units. The purpose 

  

for calculating these additional scores is to provide comparisons 

for the evaluation of the compactness scores for North Carolina's 

congressional districts. Thus, Figure 3 displays a square 

circumscribed by a circle. While most would visually evaluate a 

square district as a highly compact, the geographic dispersion 

score for the square is .640, and its perimeter compactness score 

is .785. Similarly, Figure 4 displays a rectangle circumscribed 

by a circle. Again, while most viewers would visually evaluate a 

rectangular district as substantially compact, the dispersion 

compactness of the rectangle is .431 and its perimeter 

compactness is .641. 

Table 2 provides further comparisons in tabular format. 

Among those units for which the two compactness indicators were 

calculated are Camden, Davie and Swain counties, the cities of 

Greensboro, Charlotte and Winston-Salem, and two precincts each 

in the cities of Greensboro, Charlotte, and Winston-Salem. These 

additional comparisons are indicative of the contrast in levels 

of the two compactness measures for the same jurisdiction, the 

perimeter compactness measure generally being lower in magnitude 

for real world geographic or political units. Thus while 

Charlotte's dispersion compactness measure is quite high at .571, 

its perimeter compactness coefficient is much lower at 0.079. 

Similar contrasts in the magnitudes of the two coefficients were 

also found to characterize both Greensboro and Winston-Salem. 

The purpose for calculating the selected compactness 

16 

 



Figure 3 
Compactness of a Square 

  

  

  

{1
 

      

Area of Square = 16 square inches 
Perimeter of Square = 16 inches 
Area of Circle = 25 square inches 
Perimeter of Circle = 17.741 inches 
Dispersion Compactness = .640 
Perimeter Compactness = .785       

 



Figure 4 
Compactness of a Rectangle 

  

  

  

8T
 

      
Area of Rectangle = 10 square inches 
Perimeter of Rectangle = 14 inches 
Area of Circle = 23.21 square inches 
Perimeter of Circle = 17.074 inches 
Dispersion Compactness = .431 
Perimeter Compactness = .641       

 



  

Table 2 

Comparator Compactness Indicators for Geographic 
Units in North Carolina 

  

Geographic Dispersion Perimeter 
Unit Compactness Compactness 

North Carolinax* 0.268 0.382 

Selected Counties: 

Camden County* 0.173 0.242 

Davie County=* 0.540 0.410 

Swain County#* 0.321 0.265 

Selected Cities: 

Greensboro 0.444 0.113 

Charlotte* 0.571 0.079 

Winston-Salem 0.548 0.075 

Precincts in Greensboro: 
Number 119 0.297 0.173 
Number 139 0.289 0.207 

Precincts in Charlotte: 
Number 10X2 0.179 0.212 
Number 195 0.154 0.213 

Precincts in Winston-Salem: 
Number 1439 0.211 0.199 
Number 1446 0.184 0.114 

  *Units for which there are accompanying figures. 
Source: Compactness measure coefficients calculated by University of Alabama Cartography Lab using U.S. Bureau of the Census TIGER Files. 

19 

 



Figure 5 

North Carolina 

  
  
Dispersion Compactness = .268 
Perimeter Compactness = .382     
  

  

   



  

IC
 

  
  
  Dispersion Compactness = .173 

Perimeter Compactness = .242 

Figure 6 

  

~ Camden County, North Carolina. 

  

  

  
 



  

<C
 

  Dispersion Compactness = .540 
Perimeter Compactness = .410   

Figure 7 

‘Davie County, North Carolina 

  

  

  
 



Figure 8 

~ Swain County, North Carolina 

  

ee 

  

      

£C
 

  
  

    Dispersion Compactness = .321 
Perimeter Compactness = .265 
mtr et —— i ———— 

      
  

 



Figure 9 

_ City of Charlotte 

  

     

    

    
  

ve
   

        
Dispersion Compactness = .571 
Perimeter Compactness = .079 

  

  

 



  

® » 
coefficients of precincts in the three cities is to demonstrate 

the constraints posed by the building-block units used by the 

State of North Carolina in delineating its congressional 

districts. On the dispersion compactness measure the 

coefficients range from a low of 0.154 to a high of 0.297 (Table 

2). On the perimeter compactness measure the scores range from a 

low of 0.114 to a high of 0.213. These measures are significant 

because building-block units of relatively low compactness will 

generally result in districts of less compactness than districts 

composed of highly compact building-block units. 

Compactness of North Carolina's Congressional Districts 

In 1992 the mean dispersion compactness of North Carolina's 

twelve congressional districts was 0.280, with the range being 

from a low of 0.045 for district 12 to a high of 0.440 for 

district 6 (Table 3). The mean level of dispersion compactness 

for the 1998 districts increased to 0.354. While the 12th 

District continued to be the lowest among the state's 12 

congressional districts, it increased to 0.109. District 7 had 

the greatest level of dispersion compactness at 0.622. It should 

be noted that District 7's level of compactness is nearly that of 

the square displayed in Figure 3. 

In 1992 the mean level of perimeter compactness for North 

Carolina's twelve districts was 0.095 (Table 3). The perimeter 

compactness coefficients ranged from a low of 0.014 for the 12th 

district to a high of 0.319 calculated for the 4th district. The 

25 

 



Table 3 

Compactness of North Carolina Congressional Districts 
as Existing in 1992 and 19982 

1998 

Disper.®? Pperim.€ 

1992 
District Disper.®? perim.C 
  .254 

«247 
«350 
.398 
.136 
.440 
.287 
.334 
.281 
«299 
.288 
. 045 

.028 

.061 

. 055 
+319 
.080 
.092 
+052 
.166 
.070 
.061 
«143 
.014 

317 
.407 
.407 
.450 
.206 
.362 
.622 
«345 
«292 
.402 
«335 
.109 

«107 
+176 
129 
«277 
«130 
«113 
+325 
+255 
+193 
«259 
.306 
.041 

Mean .280 . 095 .354 .192 

  SOURCE: Compactness measures for 1992 districts were calculated by Election Data Services. Compactness measures for the 1998 districts were calculated by the University of Alabama Cartography Lab. 2 Both the "dispersion" and "perimeter" compactness measures are discussed in Pildes and Niemi (1993). This article was cited in the Supreme Court's plurality decision in Bush v. Vera (June, 1996) as providing substantiation for its ruling that Texas congressional 
districts 18, 29 and 30 were unconstitutional. The comparable indices for the three Texas districts prior to the decision were as follows: District 18: dispersion .3s, perimeter .01. District 29: dispersion .19, perimeter .01. District 30: dispersion .24, perimeter .02. 
Perimeter measures are typically of lesser magnitude than dispersion scores. In 1993, 13% of all congressional districts had perimeter scores below .10. In 1998 9% of all districts are below .10 on the perimeter compactness measure. 

This measure ranges from 0.0 (least compact) to 1.0 (most compact). To calculate this measure the district is first circumscribed by a circle. The coefficient is the percentage of the area in the 
circumscribed circle which is also contained in the district. 
© This measure ranges from 0.0 (least compact) to 1.0 (most compact). To calculate this measure the perimeter of the district is first calculated. The area of a circle with this perimeter is then calculated. The reported coefficient is the proportion of the area in the district relative to the circle. The equation used is (((4 X Pi) X Area of District) / (District's Perimeter YX)  



  

mean level of perimeter compactness for the 1998 districts is 

0.192. These coefficients tages from a low of 0.041 for the 

state's 12th district to a high 0.325 for the 7th district. 

Table 4 presents the absolute and percentage change in the 

levels of compactness on both the dispersion and perimeter 

measures for North Carolina's 12 congressional districts. The 

average district in North Carolina increased its level of 

dispersion compactness by 0.075 or 39.1%. The increase in the 

level of District 12th's dispersion compactness was greatest at 

142.2%. District 6's level of dispersion compactness actually 

fell by nearly 18 percent. This finding underscores that changes 

in one district very frequently lead to changes in other 

districts, and the direction of such successive impacts will not 

necessarily be in the desired direction. 

The average level of perimeter compactness for the state's 

1998 districts is also well above what existed in 1992 (Table 4). 

The mean district's level of compactness rose 0.097 or nearly 

172%. District 7's increase nas the greatest at 525%. District 

4's perimeter compactness fell by over 13%. 

North Carolina Compared to Other States 

In addition to North Carolina, other states have made 

adjustments in their congressional districts since 1992. Table 5 
compares the means for both compactness scores in 1992 and 1998 

for the state's of North Carolina, Florida, Georgia and Texas. 

These coefficients are invaluable because they suggest the degree 

27 

 



Table 4 

  

Change in Compactness of North Carolina Congressional Districts 
as Existing in 1992 and 19982 

  

  

  

District Change in Compactness 
No. Dispersion Perimeter 

Absolute Percent Absolute Percent 

1 +.063 +24.8% +.079 +282.1% 

2 +.160 +64.8% ++3115 +188.5% 

3 +,057 +16.3% +.074 +134.5% 

4 +.052 +13.1% -.042 -13.2% 

5 +.070 +51.5% +.050 +62.5% 

6 -, 078 17.7% +.021 +22.8% 

7 +.335 +116.7% +.273 +525.0% 

8 +,011 +3.3% +.089 +53.6% 

9 +.011 +3.9% +.123 +175.7% 

10 +.103 +34.4% +.,198 +324.6% 

11 +.047 +16.3% +.,163 +114.0% 

12 +.064 +142.2% +.027 +192.8% 

Mean +.,075 +39.1% +.007 +171.9% 

  

SOURCE: Calculated by author from data ‘in Table 3s 

@ Both the "dispersion" and "perimeter" compactness measures are discussed in Pildes and Niemi (1993). See Table 3 for definitions of measures. 

28 

 



  

Table 5 

Increase in North Carolina Congressional District 
Compactness in Comparison to Florida, Georgia and Texas 

  

Mean Dispersion Mean Perimeter 
Compactness Absolute Percent Compactness Absolute Percent State 1992 1998 Change Change 1992 1998 Change Change 

North Carolina .280 .354 © +0.074 26.4% . 095 .192 +0.097 102.1% (N=12) 

Florida .314 «323 +0.009 2.9% .203 .207 +0.004 2.0% (N=23) 

Georgia .347 .400 +0.053 15.3% .181 .272 +0.091 50.3% (N=11) 

Texas .310 322 +0.012 3.9% «131 .164 +0.033 25.2% (N=30) 

  

SOURCE: Calculated by author. 

29 

 



  

of change impacting all districts within each state. In short, 

the redrawing of a single district may impact most other 

districts in a state. Thus, while North Carolina's 12th district 

was a primary focus of legal challenges, no district in the state 

was left untouched by the changes to its design. 

North Carolina's mean dispersion compactness coefficient is 

0.354 for the 1998 districts, higher than both Florida and Texas 

(Table 5). Its mean dispersion compactness rose by 0.074, or 

26.4% between the 1992 and 1998 plans. This mean increase in 

compactness was the greatest of the four states examined. 

North Carolina's mean perimeter compactness coefficient is 

0.192 for the 1998 districts, higher than that for Texas (0.164) 

and similar to that calculated for Florida (0.207) (Table 5). In 

terms of absolute change, North Carolina's mean level of 

perimeter compactness rose from 0.095 to 0.192, or by 0.097. In 

terms of both absolute and percentage change, this level of 

increase was the greatest among the four states examined, 

doubling the percentage increase of second place Georgia. 

Table 6 specifically compares the levels of compactness for 

North Carolina's 12th congressional district with other 

successfully challenged districts in Florida, Georgia and Texas. 

As can be ascertained from the table, the increase in the 12th 

district's level of dispersion compactness is second only to 

Georgia's 11th district among the seven examined. While the 12th 

district experienced the smallest level of increase among the 

seven districts examined on the perimeter compactness measure, 

30 

 



State 

Table 6 

Increase in Congressional District Compactness 
of Selected® Congressional Districts in Florida, Georgia and Texas 

Dispersion 
Compactness 

1992 1998 
Absolute 

Change 

Percent 

Change 

Perimeter 
Compactness 

1992 1998 
Absolute 

Change 
Percent 

Change   

North Carolina 

District 

Florida 
District 

Georgia 
District 

District 

a2 

3 

2 

11 

0.045 

0.111 

0.323 

0.169 

0.136 

0.541 

0.444 

+0.064 

+0.025 

+0.218 

+0.275 

+142.2% 

$22.5 

0.014 

0.011 

0.088 

0.065 

0.041 

0.050 

0.411 

0.259 

+0.027 

+0.039 

+0.323 

+0.194 

+192.9% 

+354.5 

+367.0 

+298.5 

Texas 

District 18 0.356 0.335 -0.021 0.011 0.151 +0.140 + +1,272.7 

District 29 0.194 0.384 +0.190 0.008 0.178 +0.170 '+2,125.0 

District 30 0.243 0.383 +0.140 +57.6 0.016 0.180 +0.164 +1,025.0 

  SOURCE: Calculated by author. 
8 North Carolina's 12th District was ruled unconstitutional on June 13, 1996. Florida's 3rd District was ruled unconstitutional on April 17th, 1996. Georgia's 11th District was ruled unconstitutional on June 29th, 1995, and its 2nd District was found likewise on October 30, 1995. Texas' 18th, 29th and 30th Districts were ruled unconstitutional on 
June 13th, 1996. 

 



its increase was substantial at 192.9%. 

The levels of compactness for North Carolina's twelve 

districts are substantially increased in the 1998 districts as 

compared to the 1992 districts. Thus, the system as a whole 

experienced a substantial increase in compactness. At present 

eleven of the state's twelve districts are above the benchmarks 

suggested by Niemi and Pildes (1993) as indicative of "low" 

compactness. While the 12th congressional district's level of 

compactness remains below the benchmarks, its level of increase 

is comparable to the redrawn districts in other states. 

Secondly, and in reference to the quote from Professor Morrill on 

page 13, evaluations of the 12th district's geographic 

compactness should also consider its travel time compactness. 

Travel times are rarely a direct function of straight line 

distances. Rather the available travel modes and directness of 

traffic corridors between points must also be considered. Thus 

it is highly probable that the time needed for the 12th 

district's representative to travel to meet constituents at 

opposite ends of the district is substantially less than in many 

other more geographically compact districts. 

5)Continuity of Representation. 
  

It is counterproductive to fundamentally alter the system of 

districts in a state each time new districts are delineated. 

Wholesale change may lead to voter confusion and frustration, and 

thus non-participation. While the goal of preserving the 

32  



  

integrity of the district system may be difficult if a state has 

experienced substantial population growth in the decennial 

districting cycle, boundary "stability helps to develop and 

maintain a sense of identity with districts" (Morrill 1981: 27). 

The goal of maximizing the continuity of representation is 

particularly important when changes occur in the middle of the 

normal decennial redistricting cycle. Thus the voters of the 

state of North Carolina have now experienced two significant 

changes to their district association in the past seven years, 

with a third necessarily coming for the 2002 election cycle. 

It is clear that the 1998 congressional districts are 

significantly different from those employed in the elections of 

1992, 1994, and 1996. It is also clear that effort was expended 

to maintain the geographic cores of the 1992 districts in the 

1998 remap. Table 7 presents the proportion of each district's 

area in 1992 which remains in the 1998 districts. On average 

76.4% of the area in each of the state's twelve districts in 1992 

was preserved in the 1998 districts. This proportion ranges from 

a high of 96.7% for the 11th district to a low of 41.6% for the 

12th district. Thus, the 12th district was more dramatically 

redrawn than any other district in the state. By comparison, 

when Florida's 3rd district was redrawn under similar 

Circumstances, 48.4% of the area of the early 1990s district was 

preserved in the current district (see Figures 10 and 11). 

The changes created by the redrawing of North Carolina's 

congressional districts also shifted one-quarter of the state's 

33 

 



  

Table 7 

Proportion of Area in 1992 Districts That 
Remains in the 1998 Districts 

Percentage of 1992 
Area That Remains in 

  

  

District # the 1998 Districts 

3 65.3% 

2 70.3% 

3 75.2% 

4 75.5% 

5 66.3% 

6 81.8% 

7 88.3% 

8 86.9% 

o 90.0% 

10 79.4% 

il 896,7% 

32 41.6% 

Mean | 76.4% 

  

SOURCE: Calculated by the University of Alabama Cartography 
Laboratory. 

34 

 



Figure 10 

Florida District 3, 1992 

  

7% 
3   

        Dispersion Compactness = .111 
Perimeter Compactness = .010       

    

 



  

9¢
 

  

  

      

Figure 11 

Florida District 3, 1998 
fer ih = 

A ] 

  

       

   

  

  

Dispersion Compactness = .136 
Perimeter Compactness = .049 

  

  
   



o » 
population to a different district, a very substantial movement 

of the population between districts even for a decennial 

redistricting (Table 8). The impact on individual districts 

ranged from a low of 7.6% of District 11's population being 

shifted, to a high of 40.5% for District 2's population. 

Districts 1,2, 4, 3, and.12 all had more than 30% of their 

residents shifted to another district in the 1998 remap. 

Clearly, such substantial shifts from one representative to 

another can be unsettling for many constituents. 

Attention to balancing the legal necessity for change with 

the interests of those represented is appropriate in all 

districting cycles, but most particularly for those changes in 

the middle of the normal decennial cycle. The changes in the 

1998 districts as compared to the 1992 districts cannot be 

characterized as minimal. Rather I would characterize the change 

as moderate because attention was paid to maintaining the 

geographic cores of the 1992 districts (I would characterize 

extreme change as a plan that paid no heed to the existing 

geographic cores of the former districts). Thus, on this 

criterion I evaluate the efforts of those producing the remap 

favorably because of their balance in complying with the June 

1996 court order while preserving a moderate level of continuity 

of representation for residents of the state's twelve 

congressional districts.  



Table 8 

Population Shifts from 1992 Districts in the 
Drawing of the 1998 Districts=* 

Number of People Percentage of 1992 
Shifted from District Population 

District # 1992 District Shifted 
  

1 180,984 32.8% 

2 223,685 40.5% 

148,886 26.9% 

194,395 35.2% 

177,687 32.2% 

143,065 25.9% 

155,585 28.2% 

81,410 14.7% 

53,587 

118,426 

42,110 

174,471 

  

Total 1,694,291 (25.6% of state's 

Mean 141,191 

  

*Based upon the 1990 Census of Population. 
SOURCE: Provided by North Carolina Attorney General's Office. 

 



  

Conclusions 
  

The purpose of this report was to examine the twelve 1998 

congressional districts in the state of North Carolina from the 

perspective of five traditional redistricting criteria. Where 

appropriate the current districts were compared with those first 

used in the 1992 congressional elections. The findings of this 

report are summarized below. 

1) Equal Population. The total deviation of the 1998 

districts is 0.270%, well below the generally accepted guideline 

of less than 1.0%. Thus the present districts are evaluated 

positively on this criterion. 

2) Contiguity. All twelve of North Carolina's 1998 districts 

are contiguous and are therefore evaluated positively on this 

criterion. Secondly, the level of contiguity in the 1998 

districts is substantially increased over that existing in the 

1992 districts, particularly as this criterion relates to the 6th 

and 12 districts. Thus, when considering the change between the 

1992 and 1998 districts, the current plan is evaluated very 

favorably. 

3)The Allocation of Local Government or Electoral Units. 

The 1998 districts reduce the number of divided counties from 44 

in 1992 to 23 at present. This is a substantial reduction and 

the present plan is therefore evaluated favorably from this 

perspective. Secondly, the 1998 districts effectively employed 

voting precincts as building blocks with only 2 of the 2,351 

39 

 



  

census defined precincts in .the state being divided. Of the 31 

counties not subdivided by census defined voting precincts, only 

one was divided in the districting process. In this case 

(Beaufort County), township boundaries were used, which are far 

better building blocks than census blocks or tracts. 

4) Geographic Compactness. The level of compactness 

determined for the 1998 districts is substantially improved over 

that existing in the 1992 districts. At present only the 12th 

congressional district is marginally below the suggested 

guidelines for judging "low" compactness. But such judgements 

should also consider the rate of change for the 12th, which was 

found here to be very substantial and in line with the magnitude 

of change experienced for other districts ordered redrawn by the 

court system. Secondly, the 12th's level of compactness is 

similar to other existing districts not under challenge (e.q., 

Florida's 22nd, see Figure 12). Finally, Professor Morrill's 

1981 argument that elongated districts which center upon 

transportation corridors may be highly compact in terms of travel 

time is pertinent. 

5)Continuity of Representation. The redrawing of North 

Carolina's congressional districts in 1997 introduced substantial 

change in the state's system of representation. On average, 24% 

of the area in the 1992 districts was allocated to a different 

district on the 1998 map. Most dramatic was the impact upon the 

12th congressional district - the 1998 district includes only 42% 

of the original area of the 1992 district. Similarly, nearly 26% 

40 

 



  

17
 

Figure 12 

Florida District 22, 1998 

       
      

    

ST 5 oi : ge 

Dispersion Compactness = .033 
Perimeter Compactness = .047 fo a yg   
            

 



  

of North Carolina's population was moved to a different district 
in the 1998 plan as compared to the 1992 map. The magnitude of 
this change is substantial, most particularly in a mid-decennial 
redrawing of the state's congressional districts. In short, 
those undertaking the remap appear to have sought balance between 

the necessity of district boundary change and the preservation of 
representational continuity. 

42 

 



  

References 

Butler, D. and Cain, B. (1992), Congressional Redistricting: Comparative and Theoretical Perspectives, New York, NY: 
MacMillan. 

  

  

Cain, B. (1984). The Reapportionment Puzzle. Berkeley: University of California Press. 
  

Dixon, R. G. (1982), "Fair Criteria and Procedures for Establishing Legislative Districts," in B. Grofman, A. Lijphart, R. B. McKay, and H. A. Scarrow eds., Representation and Redistricting Issues, PP. 7-19. Lexington, Mass. : Lexington Books. 

Duncan, P. D., and Lawrence, C. C. (1997), Politics in America, 1998: The 105th Congress, Washington, D.C.: Congressional Quarterly. 

  

  

Grofman, B. (1985), "Criteria for Redistricting: A Social Science Perspective," UCLA law Review, 33: 77-184.   

Grofman, B (1993), "Would Vince Lombardi Have Been Right If He Had Said, 'When It Comes to Redistricting, Race Isn't Everything, Its the Only Thing'?," Cardozo Law Review, 14: 1237-1276.   

Grofman, B., Handley, L., and Niemi, R. GC. (1992), Minority Representation and the Quest for Voting Equality, Cambridge, NY: Cambridge University Press. 
  

Morrill, R. L. (1981), Political Redistricting and Geographic Theory, Washington, D.C.: Association of American Geographers. 
  

Morrill, R. L. (1982), "Redistricting Standards and Strategies After Twenty Years," Political Geography Quarterly, 1: 361-369.   

Morrill, R. L. (1987), "Redistricting, Region, and Representation," Political Geography Quarterly, 6: 241-260.   

Morrill, R. L. (1994), "Electoral Geography and Gerrymandering: Space and Politics," in G. J. Demko and W. B. Wood eds., Reordering the World, PP. 101-119, Boulder, CO: Westview Press. 
  

Niemi, R. G., Grofman, B., Carlucci, C. and Hofeller, T. (1990), "Measuring Compactness and the Role of a Compactness Standard in a Test for Partisan and Racial Gerrymandering," Journal of Politics, 52(4): 1155-11381. 

  

Pildes, R. H. and Niemi, R. GC. (1993), "Expressive Harms, "Bizarre Districts," and Voting Rights: Evaluating Election- District Appearances After Shaw V. Reno," Michigan Law Review, 92 (3): 483-587,   

43 

 



Raburn, R., and Leib, J. I. (1994). "Congressional District Building Blocks: Choice and Impact in the 1990s," Comparative State Politics, 15(2): 17-27. 

Webster, G. R. (1984), "The Spatial Reorganization of County 
Boundaries in Kentucky," Southeastern Geographer, 24(1): 14-29.   

Webster, G. R. (1995), "Congressional Redistricting in the Southeastern U.S. in the 1990s," Southeastern Geodrapher, 35(1): 1-21. 

 



  

APPENDIX 

45 

 



| Figure A-1 

North Carolina District 1, 1992 

  

—_— 
  

  

    
   
  

Dispersion Compactness = .254 
Perimeter Compactness = .028     
    

   



  

Figure A-2 

North Carolina District 2, 1992 

  

  

    Dispersion Compactness = .247 
Perimeter Compactness = .061   
    

    

 



Figure A-3 
North Carolina District 3, 1992 

  

    

        
  

   
  

      Dispersion Compactness = .350 
Perimeter Compactness = .055 
 



  

  
Dispersion Compactness = .398 
Perimeter Compactness = .319     

        

Figure A-4 

_Norih Carolina District 4, 1992 

    

    
  

    
 



Figure A-5 

~ North Carolina District 5, 1992 

i] 

        

    
Dispersion Compactness = .136 
Perimeter Compactness = .080 

  
  

   



Figure A-6 

_ North Carolina District 6, 1992 

  

    

    
    

      
Dispersion Compactness = .440 
Perimeter Compactness = .092   
    

    

 



  
  

  
Dispersion Compactness = .287 
Perimeter Compactness = .052 

Figure A-7 

North Carolina District 7, 1992 
    

  

  
  

        

 



Figure A-8 

_ North Carolina District 8, 1992 

  

irr ~——— — 

  
  

  
          Dispersion Compactness = .334 

Perimeter Compactness = .166   
 



Figure A-9 

North Carolina District 9, 1992 

  

  

  

        Dispersion Compactness = .281 
Perimeter Compactness = .070 

  

      

  
 



Figure A-10 

North Carolina District 10, 1992 

      
  

        

  
Dispersion Compactness = .299 
Perimeter Compactness = .061      



Figure A-11 

North Carolina District 11, 1992 

  

          
  
Dispersion Compactness = .288 
Perimeter Compactness = .143 

  
   



Figure A-12 

North Carolina District 12, 199 

  

— ——— — —— —_— 

Lert pd — 

  
  

  
  

        
Dispersion Compactness = .045 
Perimeter Compactness = .014       

   



Figure A-13 
‘North Carolina District 1, 1998 

  

  

  

  
   
    

Dispersion Compactness = .317 
Perimeter Compactness = .107     
  

      

 



  

  

    

  
Dispersion Compactness = .407 
Perimeter Compactness = .176 

  

  

N 

  

Figure A-14 

    

i 

  

orth Carolina District 2, 1998 

a
 

———————————————————— 

    

    

  

 



Figure A-15 

_ North Carolina District 3, 1998 

  

Lh ar Te — 
rr so 

og 

   

    

        
  
Dispersion Compactness = .407 
Perimeter Compactness = .129     
   



  

    Dispersion Compactness = .450 
Perimeter Compactness = .277 

  

Figure A-16 

North Carolina District 4, 1998 
I = a 

ne 

Ei 
~~ _ 

   

  

  

    

    

   

  

  

  
 



  
Dispersion Compactness = .206 
Perimeter Compactness = .130   

North Carolina District 
Figure A-17 

Fi NE tes 

  

b A998. 

  

    

  

  

  
 



  

Figure A-18 

  

  

  

  Dispersion Compactness = .362 
Perimeter Compactness = .113 
  

  

  
 



Figure A-19 
. North Carolina District Zz 1998 

  

  

  
      

Dispersion Compactness = .622 
Perimeter Compactness = .325   
  

        

 



Figure A-20 

  

te i 2 gh 

      

  
  Dispersion Compactness = .345 

Perimeter Compactness = .255 
————     
            

 



Figure A-21 
North Carolina District 9. HE gh 

  

1908. 
TE Se, 

   

  

      
Dispersion Compactness = .292 
Perimeter Compactness = .193     
      

 



    
  

Figure A-22 

_ North Carolina District 10, 1998 

  

pe 
a il 

  
        
  

  Dispersion Compactness = .402 
Perimeter Compactness = .259   
            

    

  

  

  
  

   



  
  

Figure A-23 

North Carolina District 11, 1998 

  

  

  

  Dispersion Compactness = .335 
Perimeter Compactness = .306 
    

    

  

  

      
  

 



Figure A-24 

North Carolina District 12, 1998 

  
   
  

— oe 

    

  

  

  

        
  Dispersion Compactness = .109 

Perimeter Compactness = .041   
    

—— ee 

 



Qualifications, Publications and Compensation. 

  

My vita accompanies this report, and lists my qualifications and 
all publications. My hourly fee is $100. 

Testimony in Court or by Deposition in the past five years: 

Dillard v City of Greensboro, Alabama (M.D. AL) 

Cannon v Durham County Board of Elections, (E.D. NC) 

Dillard v Baldwin County Commission (M.D. AL) 

Stovall et al., v City of Cocoa, et al. (M.D. FL) 

Rice v Bennett (Montgomery (AL) Co. Cir. Ct.) 

Signed: 

Lewd dL tL 
Gerald R. Webster, Ph.D. 

Professor of Geography 
University of Alabama 

  

Date: 

fer Wd #4