VA Redistricting Report of Special Master (Nov. 2015)

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 1 of 66 PageID# 6054
REPO
ORT OF TH
HE SPECIA
AL MASTE
ER
Novem
mber 15, 20
015
Berna
ard Grofma
an*
Speccial Masterr
*Bernarrd Grofman
n is Professsor of Polittical Sciencce and Jack
k W. Peltasson Endow
wed
Chair off Democraccy Studies at the Uniiversity of C
California,, Irvine, an
nd former
Directorr of the UC
CI Center fo
or the Stud
dy of Demoocracy. Hiss research d
deals with
topics su
uch as votiing rights, electoral rules, theorries of reprresentation
n, behaviora
al
social ch
hoice, and political
p
sccience meth
hodology. H
He is co-author of fivee books (fou
ur
from Ca
ambridge University
U
Press
P
and one from Y
Yale Univerrsity Presss), and coeditor of 23 other books;
b
with
h over 300 research a
articles and
d book chap
pters,
includin
ng ten in th
he America
an Politicall Science R
Review. A m
member of tthe Americcan
Academ
my of Arts and
a Science
es since 2001, he has been a sch
holar-in-ressidence at
universiities and re
esearch cen
nters in the U.S., Can
nada, Fran
nce, Germa
any, Italy,
Japan, the
t Netherrlands, Spa
ain, and the
e UK, and he has an honorary P
Ph.D. from
m the
Universsity of Cope
enhagen (D
Denmark) for
f his reseearch on coomparativee electoral
systemss. He has previously
p
been
b
involv
ved as a con
nsultant orr expert wiitness for
federal courts, the
e U.S. Depa
artment of Justice, booth major p
political pa
arties at a sstate
onal level, and
a civil rights group
ps such as tthe NAACP
P Legal Deefense and
or natio
Educatiional Fund
d and the Mexican-Am
M
merican Legal Defensse and Edu
ucational Fu
und.
As a spe
ecialist on redistrictin
ng, his own
n research,, or chapterrs in bookss he has ed
dited,
has been
n cited in more
m
than a dozen U..S. Suprem
me Court deecisions, m
most recentlly in
Arizona
a State Legiislature v. Arizona
A
In
ndependentt Redistrictting Comm
mission (201
15)
and, perrhaps mostt notably, in
i Thornbu
urg v. Ging
gles, 478 US
S 30 (1986)).
REPORT OF THE SPECIAL MASTER

EXECUTIVE SUMMARY
The eight plans/maps offered pursuant to the Courts September 18 deadline are
fatally flawed in either not remedying the constitutional infirmities in the present
CD3, e.g., re issues of discontiguity and city and county splits that appear race
related, and/or by failing to offer a remedy that is narrowly tailored, e.g., modifying
congressional districts that did not need to be changed to deal with the
constitutional problems in CD3. I cannot recommend any of them to the Court.
The two plans/maps I have created to offer for review by the Court (labeled as
NAACP Plan Modification 6, and Current Congressional Plan Modification 16) are
intended to illustrate versions of CD3 that might remedy the constitutional
violation identified in the majority opinion in Page v. Virginia State Board of
Elections in a narrowly tailored fashion. Both illustrative remedial plans locate
CD3 entirely in the Newport News-Hampton-Portsmouth-Norfolk area of the state.
Both plans place all of Newport News and all of Portsmouth in CD3. The two plans
differ slightly in how CD3 is configured, primarily in whether the cities of Norfolk
and Hampton are split between CD3 and CD2, but both place either all or the
predominant part of each citys population in CD3. In one plan the two cities are
split so as to assure contiguity by bridge between separated pieces of CD2; in the
other both Hampton and Norfolk are wholly within CD3 and contiguity is
2
established in CD2 by the link between Poquoson and the eastern shore created by
the intersection of the water boundaries of Poquoson and Northampton. In the plan
in which Norfolk is kept wholly within CD3, for population equality purposes, the
Isle of Wight is split between CD3 and CD4. In the plan where the city of Norfolk is
split between CD3 and CD2, the Isle of Wight is wholly within CD2, and population
equality balancing occurs within the city of Norfolk. In one plan the City of
Franklin is wholly within CD3; in the other, the City of Franklin is wholly within
CD4.
Each illustrative map I wish to propose to the court has been drawn according to
good government criteria, such as limiting splits of existing cities and counties
between two or more districts, achieves an average higher level of compactness than
the current plan, and follows the legal guidance provided to me by the Court. For
example, while the current CD3 splits six cities or counties between CD3 and other
congressional districts, one of the two plans I propose splits only four cities between
CD3 and another congressional district; while in the other, the version in which the
city of Norfolk is entirely within CD3, only two cities or counties are split between
CD3 and another congressional district. In both versions, by both the Polsby-Popper
and the Reock measure of compactness, the compactness scores of CD3 are between
twice as large and over four times as large as the corresponding compactness scores
of the current CD3.
Both illustrative plans fully remedy identified constitutional infirmities in the

current CD3, avoid retrogression in the realistic opportunity of the minority
community to elect candidates of choice, and do not use race as a predominant
criterion. Both are narrowly tailored to achieve constitutional goals. For example,
subject to the overriding requirement of constructing a constitutional map, the two
illustrative configurations I have drawn provide a least change plan in the various
appropriate meanings of that term identified in the body of the Report. First, they
make no changes in a majority of the present districts, and confine all changes to
the district in which the constitutional defect has been found and the districts
immediately adjacent to that district -- districts in which changes would be required
for purposes of population adjustment. Second, they maintain the residences of
present incumbents in their districts. Third, once we take into account the need for
a major redrawing of CD3 to address constitutional defects in the current version of
that district, redrawn districts other than CD3 generally reflect the present shape
of the equivalent districts in the current map.
In the process of rectifying the constitutional violations found in current CD3, and
without using race as a predominant factor, but simply taking into account the
demography and geography and present political subunits of the State in terms of
standard good government districting criteria, the two illustrative remedial maps
create a second district (CD4) in which African-American voters now possess a
realistic opportunity to elect candidates of choice. This is an opportunity that
minority voters have been denied in the current CD4 due to the ways in which the
present congressional map fragments minority voting strength in south central and
southeastern Virginia. These two illustrative versions of CD4 include the cities of
Richmond and Petersburg in their entirety, as well as either ten or eleven other
whole cities or counties. One of the county splits in each of the illustrative plans
was due to the need to insure that each of the homes of the current incumbents was
located within a district that corresponded in number to their present district. The
two versions of CD4 offered in the two maps that illustrate remedial versions of
CD3 are more compact than the current CD4.
Overall, the two illustrative maps are drawn according to good government criteria.
For example, in the five modified districts as a whole, in both illustrative plans, the
number of city and county splits is less than in the current plan. When we compare
compactness on a district by district basis for each of the five modified districts in
the NAACP Mod 6 version as compared to the current plan, using both the PolsbyPopper measure and the Reock Measure, in 9 of the 10 comparisons, the illustrative
plan achieves superior compactness. When we compare compactness on a district
by district basis for each of the five modified districts in my Modification 16 version
as compared to the current plan, using both the Polsby-Popper measure and the
Reock Measure, in 8 of the 10 comparisons, the illustrative plan achieves superior
compactness, and in the other two comparisons we have a tie.
Full information on each of these two illustrative plans is attached as an Appendix

to this Report. Key elements of each plan are discussed in more detail in the
sections below.
FINAL REPORT OF THE SPECIAL MASTER

I. BACKGROUND
1. Pursuant to my responsibilities as a special master in Personhuballah v. Alcorn

to assist and advise the Court, I have
(a) reviewed the present congressional plan for the State of Virginia, and all of the
plans submitted on September 18, 2015, and the responses submitted on October 7,
2015.1
(b) reviewed demographic information on total population and the racial and ethnic
composition of population at various levels of census geography as well as basic
geographic data for the State (county and city boundaries; location of highways,
bridges and tunnels), with a focus on areas of the state contained in or proximate to
the present 3rd Congressional district.
(c) familiarized myself with the Court opinions in Page v. Virginia State Board of
There were seven submissions that contained plans and maps that could be
analyzed, which I reference in short form as Defendant-Intervenors, Plaintiffs,
Governor of Virginia, NAACP, Petersen, Rapoport, and Richmond First. The Bull
Elephant submission did not contain the shape file required for detailed analysis
and was dropped from further consideration. The submission by One Virginia did
not include a map. One submission, that by Defendant-Intervenors, contained two
plans and accompanying maps. Thus, there were eight proposed maps/plans whose
features I reviewed.
1
Elections, 2015 WL 3604029 (E.D. Va. June 5, 2015), especially with respect to the
majority opinions identification of constitutional infirmities in the present
configuration of the 3rd Congressional district that need to be remedied.
(d) obtained (pursuant to an Order of the Court) technical assistance in map

creation from staff of the Division of Legislative Services of the Virginia State
Legislature (Jack Austin, Kent Stigall, and Julie Smith), each of whom has signed
an oath of confidentiality drafted by the Court in consultation with the parties.
(e) created a large number of illustrative congressional maps in an iterative fashion

to allow me to explore options for drawing CD3 in a fashion that might remedy its
present constitutional defects. I have considered modifications of the present map
and modifications of maps presented by plaintiffs, defendant intervenors, and
others, such as the NAACP.
2. There are a number of different criteria that can be used to evaluate a

(congressional) redistricting plan as a whole, or used to evaluate the configuration
of one or more individual districts. These include
(a) conformity to a standard of one person, one vote;
(b) avoiding either fragmentation or packing of geographically concentrated

minority populations that might have the effect or purpose of minimizing or
8
diluting the voting strength of constitutionally protected minorities, and/or lead to

retrogression in the ability of minority communities to realistically have an equal
opportunity to elect candidates of choice;
(c) avoiding use of race as a predominant criterion for redistricting;
(d) avoiding the creation of districts which are divided into two or more
discontiguous parts;2
(e) avoiding splits (partition into two or more congressional districts) of long
standing political subunits such as cities or counties,3 unless these splits become
obligatory or near obligatory by the need to satisfy other criteria such as population
equality;
(f) avoiding unnecessarily ill-compact districts, i.e., ones which are elongated or
have irregularly shaped perimeters.
See below for further elaboration.
In Reports prepared by the State of Virginias Division of Legislative Services,

political entities which are either cities or counties are described as localities. Note
also that some political entities that have city in the title, such as Charles City
more closely resemble what in other states would be labeled as counties.
3
See below for further elaboration.

9
3. In situation such as that applying in Personhuballah v. Alcorn, where a court is

drawing a map to remedy a constitutional infirmity, there is a seventh criterion
that may be relevant:
(g) avoiding changes in existing district boundaries that are not required to create a
constitutional map, i.e., what is sometimes referred to a least change criterion.
4. Considerations that enter into line-drawing:
(a) In drawing illustrative maps for consideration by the Court that in my view
would serve to remedy the constitutional infirmities identified in the majority
opinion in Page v. Virginia State Board of Elections, I have sought to take into
account all of the criteria enumerated above.
(b1) In general, however, there are tradeoffs among the various criteria. In practice,
when there are so many distinct criteria to be balance off against one another, it
may be impossible to satisfy all criteria fully. For example, strict adherence to a
population equality standard may lead to the necessity to split some political
subunits, while undue deference to existing district lines may lead to fragmentation
of minority voting strength.
(b2) The first three of the criteria listed in Section 2. above, 2.(a), 2.(b), and 2.(c), I
10
treated as of highest priority since they are grounded in provisions of the U.S.
Constitution, as these have been interpreted by the U.S. Supreme Court. However,
because the indicia used by the majority in the Page opinion to infer predominant
racial motive included discontiguities in district boundaries that picked up isolated
pockets of minority population, the stretch of CD3 between separated areas of the
state in a fashion that did not appear in any way compelled by the demography of
the state, disregard of city and county lines that appeared linked to race, and the illcompactness of CD3, I was especially attentive to issues of contiguity, compactness
and maintenance of existing political subunit boundaries within the district in
drawing illustrative remedial configurations of CD3.
5. I now discuss briefly how I dealt with each of the seven criteria in my
preliminary illustrative map drawing:
(a). Since the present map and all of the maps offered in briefs submitted on or
before October 7 specified district configurations which were either exactly at ideal
district size, 727,366, or plus or minus one person, the illustrative maps I have
offered to the Court also provide this level of strict population equality.5
In 2016, because of births, deaths, and migration in and out of the districts, the
2010 census figures can only be regarded as approximations to the present
population in the various congressional districts in Virginia. Nonetheless, the 2010
Census provides the best information available about Virginias population
demography and is the appropriate data to use.
11
(b1) In seeking to reach professional judgments as a political scientist in

redistricting concerning issues of potential retrogression, the realistic opportunity
for the minority community to elect candidates of choice, and potential issues of
minority fragmentation or packing, a necessary starting point is a review of the
demography and geography of the State of Virginia, especially in areas of the state
proximate to the present Congressional District 3. I found that there are two areas
of eastern Virginia with substantial black population concentrations:
i. one encompassing the area in the southeastern area of the state around Newport
News, Hampton, Portsmouth, and Norfolk;
ii. one encompassing the cities of Richmond, Petersburg, and Hopewell, each of
which is a majority black city, along with immediately proximate black population
concentrations in Henrico County and Chesterfield County that are also
substantial.
(b2) Neither of these areas of concentrated minority population concentration

standing alone is sufficient to make up the necessary population for a congressional
district. However, the geographically proximate cities of Newport News
(189, 719 persons), Hampton (137,436 persons), Portsmouth (95,535 persons), and
Norfolk (242,803 persons), taken as a whole, do already contain 90% of ideal
congressional district population. These cities in whole or part form the basis for
12
the illustrative remedial plans for constitutional infirmities in the current CD3
discussed in the recommendation section below
(b3a) As shown in the plan submitted by Mr. Rapoport, it is possible to draw an

ideal-sized congressional district that is majority black in voting age population
(labeled District 3 in that plan) based around Newport News, Hampton,
Portsmouth, and Norfolk, that does not include areas such as Richmond or
Petersburg. However, the configuration of the plan that includes this district
makes changes in districts beyond those immediately proximate to the current CD3.
(b3b) As shown in the plan submitted by the NAACP, it is possible to draw an idealsized congressional district that is majority black in voting age population (labeled
District 4 in that plan) that incorporates area around Richmond and Petersburg and
then picks up additional population needed to attain ideal district size by largely
extending south and east, and which does not include Newport News, or Hampton
or Portsmouth or Norfolk. However, the configuration of the plan that includes this
district makes changes in districts beyond those immediately proximate to the
current CD3.
13
(b4) Because there is some overlap among the areas included in District 3 in the
Rapoport configuration and those included in District 4 in the NAACP configuration
it is impossible to draw both districts simultaneously within the same plan.6
(b5) I reserve for later in this report the discussion of the implications of these
geographic and demographic facts for the potential to draw a non-retrogressive plan
that provides minority voters in CD3 a realistic opportunity to elect candidates of
choice, and also neither fragments nor packs minority population concentrations
elsewhere in central or southeastern Virginia. Here I simply note that it is my
professional judgment that configurations for a CD3 that remedies the
I have been able to draw within a single illustrative plan two districts whose
voting age population is majority-minority in composition, though not majority
black in voting age population. One of these districts picks up population from
areas such as Richmond and Petersburg and does not include any part of Newport
News, Hampton, Portsmouth, or Norfolk (it is labeled District 4 in an illustrative
plan I have labeled as Modification 13 to the current map); and one is based
around Newport News, Hampton, Portsmouth, and Norfolk, and does not include
any part of Richmond or Petersburg (it is labeled District 3 in my illustrative
Modification 13 to the current map). However, because Modification 13 to the
current map modifies district boundaries in districts that are not directly
contiguous to the current CD3, and also results in the pairing of two incumbents
(the home of the incumbent in CD4 and the home of the incumbent in CD3 are both
located within CD2 in this plan), and because it provides minority populations in
excess of what are necessary to create a minority opportunity to elect district in
CD3, I have not considered this plan further, and do not wish to recommend it to
the attention of the Court. Because of that fact, I have not included further
information about Modification 13 to the current map in this Report. It was merely
drawn to demonstrate what might be possible in a situation where no form of least
change constraint was legally relevant. However, as discussed below, even when a
least change constraint is operative and even when there is no incumbency pairing,
it is possible to provide constitutionally drawn and narrowly tailored remedial
configurations of CD3 that realistically provide the African-American community
with an equal opportunity to elect candidates of choice.
14
constitutional defects found in current CD3 do not require the creation of a CD3
that is 55% black in voting age population, or even one that is majority black in
voting age population. (See discussion below.)
(c1A) In drawing illustrative maps for a federal court to remedy constitutional

defects identified in the current CD3, because of my concerns for contiguity and the
avoidance of unnecessary splitting of existing political subunit boundaries, and my
attention to compactness, the various illustrative maps I have drawn that create a
heavily minority district, CD3, in the Newport News/Portsmouth area do not reflect
in any way a predominant concern for race. Rather, these maps naturally reflect
the underlying demography of the state, i.e., the fact that there are two distinct
minority population concentrations in central and southeastern Virginia: one
involving a compact contiguous minority population concentration around Newport
News, Hampton, Portsmouth, and Norfolk that is already almost large enough for a
congressional district; and one involving a minority population concentration
anchored at one end by Richmond and Petersburg that can add population south
and east to pick up the additional population needed to constitute an ideally sized
congressional district.
(c1B) Once a decision has been made to remedy the constitutional defects in the
present CD3 in the most obvious way, vis-a-vis locating CD3 in the Newport News
area, the substantial minority population we find in CD4 in the two plans that I
15
wish the Court to consider seriously arises virtually automatically. In particular, in

each of the two plans, between twelve and thirteen cities and counties, including the
City of Richmond and the City of Petersburg, have been maintained wholly within
CD4. Despite the fact that CD4 has been drawn largely as a residual district,
when lines are drawn as described above, the demography of the state assures that
the compact contiguous minority population bloc in the Richmond-Petersburg area
and eastward and south does not have its voting strength minimized or cancelled
out; it is neither being fragmented nor being packed. Rather, a more substantial
minority population is found in CD4 in these plans than in CD4 in the current plan
because the fragmentation of minority population in the Richmond area and
eastward and south, due in large part to the tortuous way CD3 is presently
configured, is remedied when CD3 is redrawn in a constitutional way and the
remaining districts are drawn to reflect this change. Thus, the way in which CD4 is
configured in the two illustrative maps I present to the Court in no way reflects race
as a predominant motive. (See further discussion below about how a decision to
draw CD3 in the Newport News area affects other congressional districts in the
state.)
(c2) It is my professional judgment that there are alternative ways to reconfigure

current CD3 that fully address the constitutional violations in that district without
in any way raising issues of undue reliance on race as a predominant motive in line
drawing. (See further discussion below.)
16
(d) As noted above, in my illustrative line drawing I have been attentive to issues of
contiguity because this issue was important in Page. For redistricting, the standard
(mathematical) way to define contiguity is in terms of the ability of voters to move
from any one part of the district to any other part of the district without leaving the
district. Special issues of interpretation of this definition arise when district
boundaries include substantial bodies of water in whole or in part. In such cases,
sometimes contiguity is interpreted in pragmatic terms as connection from any land
part of the district to any other land part of the district via land, bridge or tunnel.
Alternatively, when the boundaries of political, voting, or census units encompass
water areas along with land areas, contiguity by water might also be established
when legal boundaries touch, even if the areas that are joined in this way have
water at each edge of the boundary. In Virginia, contiguity has also been
interpreted as occurring when there is a direct line of sight connection over a body
of water between two pieces of land. To avoid any possible dispute about whether
the district, CD3, that has been found to be unconstitutional has been drawn in a
contiguous fashion, in the two illustrative maps I have drawn, the land parts of CD3
are contiguous to one another by land, bridge or tunnel.
(e) In the two illustrative maps I have drawn that I believe deserve consideration by
the Court, I have been able to minimize city and county splits in CD3 to an equal or
greater extent than in the present congressional plan. In particular, CD3 in one of
17
the two illustrative maps discussed in my recommendation below has only 2 cities
or counties split between CD3 and another district, and the other illustrative plan
has only 4 cities or counties split between CD3 and another district, while the
current map splits 6 cities or counties between CD3 and another district.
(f) Compactness numbers are very difficult to interpret without some context, and it
is virtually impossible to compare compactness values across jurisdictions in
different states, or sometimes even within a single state across different parts of the
state. Because the feasibility of drawing compact districts varies with the
geography (e.g., the density of populations, and the degree to which the political or
other subunits which are being aggregated are themselves compact, and the
existence of natural boundaries such as state lines or large bodies of water),
compactness is best understood by comparing plans both for the same geography
and for the same types of districts (e.g., congressional, lower chamber, upper
chamber).7 However, even here a note of caution is required. The two main types of
compactness, areal compactness (e.g., Polsby-Popper) and perimeter irregularity
(Reock), measure two rather different things and they do not necessarily move in
parallel when district lines are changed.8 Nonetheless, in each of the illustrative
The potential to draw compact districts needs to be understood in the light of the
average population size of the type of districts that need to be drawn.
See Richard Niemi, Bernard Grofman, Carl Carlucci & Thomas Hofeller,
Measuring Compactness and the Role of a Compactness Standard in a Test for
Partisan and Racial Gerrymandering, 52 J. Pol. 1155 (1990).
8
18
maps I have drawn that I believe deserve consideration by the Court (see below),
CD3 is far more compact than the current CD3, and the average compactness of the
set of changed districts (CDs 1, 2, 3, 4, 7) is higher in these maps than in the
current map. Indeed, as shown later, by both the Polsby-Popper and the Reock
measure of compactness, the compactness scores of CD3 in my illustrative remedial
maps are between twice as large and over four times as large as the corresponding
scores of the current CD3 with higher numbers representing more compact
districts.
(g1A) In the context of a plan that is being drawn to remedy a constitutional

violation found within a particular district, here CD 3, the concept of least change
has potentially five meanings.
(g1B) The most direct of the five possible meanings of least change would be to say
that a plan is a least change plan to the extent that changes in the map are
restricted to the set of districts in which changes are required to remedy the
constitutional violation.9 In all my illustrative line drawing, the plans have clearly
been least change plans in that sense. It is my professional judgment as a
political scientist that the need to address the constitutional infirmities identified
by the decision in Page does not require any changes in district lines that extend
See, e.g., Upham v. Seamon, 456 U.S. 37, 43 (1982).
19
beyond the population and the geography now contained in current CD3 and the
districts immediately contiguous to it that would be most affected by changes in the
configuration of present CD3. In other words, the only plans I recommend to the
Court are "least change" plans in the sense of making no changes in the present
configurations and populations of the majority of the present congressional districts,
namely Districts 5, 6, 8, 9, 10, and 11, while confining changes to the areas and
populations in current congressional districts 1, 2, 3, 4 and 7.
(g2A) A second potential meaning of least change is that the changes in CD3 itself
be minimal. That definition I reject as inappropriate. As Special Master I am
making a recommendation to a federal court and, as Defendant-Intervenors note,
faced with the necessity of drawing district lines by judicial order, a court, as a
general rule, should be guided by the legislative policies underlying a state plan
even one that was itself unenforceableto the extent those policies do not lead to
violations of the Constitution or the Voting Rights Act. (Intervenor-Defendants
Brief in Support of their Proposed Remedial Plans (Intervenors Br. in Supp.)
(ECF No. 232), at 2) (citing Perry v. Perez, 132 S. Ct. 934, 941 (2012) (quoting
Abrams v. Johnson, 521 U.S. 74, 79 (1997)). It is the last clause of this statement
that is critical. Other criteria must bow to the need for a constitutional plan, and
such a plan must avoid using race as a predominant consideration, and avoid
minority vote dilution in violation of the Equal Protection Clause of the
Constitution or the still relevant portions of the Voting Rights Act.
20
(g2B) Given the demography and geography of the State of Virginia discussed in
5.(b) above, as political scientist and expert on redistricting, in my view, although
other options (such as redrawing CD3 as a Richmond area based district) might be
possible, the obvious way to remedy the constitutional violation in CD3 is to redraw
CD3 as a Newport News-Hampton-Portsmouth-Norfolk based district that is
contiguous, highly compact, and has few city splits. Such a district begins already
well on its way toward having the necessary population for an ideally sized
congressional district and incorporates a substantial portion of the population found
in the current CD3. This is certainly a major change from the current
configuration, but it is a change needed to craft a narrowly tailored remedy for the
constitutional violation identified in the majority opinion in Page.
(g3A) A third way to think about least change has to do with the nature of the
reconfiguration of the districts that must be changed in some fashion because of the
constitutionally mandated changes in CD3, which cause spillover effects in the
proximate districts (here, CDs 1, 2, 4, and 7). In the two illustrative maps I have
presented, I began by drawing a Newport News-Hampton-Portsmouth-Norfolk
based CD3 that is already very nearly large enough to constitute an ideal size
district, and I then added proximate population to create a compact contiguous CD3
with a minimal number of city or county splits. That is, I began with the
congressional district where a constitutional violation has been found and remedied
21
that violation before I moved any further. Beginning with CD3 as a Newport NewsHampton-Portsmouth-Norfolk based district that is contiguous, highly compact, and
has few city splits means that the four other congressional districts in the eastern
portion of the state (1, 2, 4, 7) need to be reconfigured for population purposes, and
should be drawn in a fashion that is generally reflective of the geographic
configurations of the current form of these districts.
(g3B) Of course, when the configuration of CD3 is so substantially changed, with

nearly half of the population removed from the district and the district redrawn to
assure both contiguity and compactness and the avoidance of unnecessary
city/county splits, there are obvious limits to how similar the remaining four
changed districts can be to their present exact geographic configurations.
Nonetheless, in the two plans that I have presented to the Court, CD2, as in the
present plan, continues to be configured to run from the eastern shore (Accomack
and Northampton) to Virginia Beach, including these cities/counties in their
entirety, and then adding some additional proximate population for one person, one
vote purposes; CDs 1 and 7 continue to be configured, as in the present plan, as
somewhat more northern and western districts, with CD1 generally above CD7;
while District 4 is largely drawn as a residual district after the other four districts
have been redrawn, and it includes a substantial portion of current CD4 as well as a
significant portion of current CD3 (the part outside the Newport News/Portsmouth
area).
22
(g4A) A fourth meaning of least change has to do with how incumbents in the
current congressional districts are treated. In particular, in alternative maps, are
the homes of all incumbents retained within the districts whose numbers
correspond to those of the incumbents present districts? In my initial exploratory
map drawing, though I was fully aware of the present configuration of congressional
districts, I deliberately operated in ignorance of the location of the home of the
present incumbents so as to examine how map drawing would proceed without such
considerations, simply following good government criteria. I also alerted the Court
to the possibility that, in so doing, I might inadvertently pair some incumbents.
Having completed the drawing of illustrative maps that did not take incumbency
into account, I have now been instructed by the Court (as noted in the Order filed
October 22, 2015 (ECF No. 263)) to take into consideration the residences of
incumbents in assessing alternative plans.
(g4B) In some of the earlier illustrative maps I drew, Congressman Forbes, the
present incumbent in District 4, who resides in Chesapeake, and Congressman
Rigell, the present incumbent in District 2, who resides in Virginia Beach, were
paired in District 2. Though living in different political entities, these two members
of Congress have homes that are only 17.4 miles apart, and they reside in an area of
the state where redistricting options are constrained by proximity to Chesapeake
Bay and the North Carolina border. In the two illustrative plans that I have
23
redrawn following the Courts Order, I have been able to redraw the plans to
eliminate the pairing without substantial change to the prior iterations of these
plans, drawn considering only good government criteria. In general, except for
remedying the pairing, which required a slight adjustment in lines (primarily in
Chesapeake and Suffolk), the differences in configuration are minimal between
illustrative plans that paired two incumbents and the revisions of those same
illustrative plans that did not pair any incumbents, although in one of the
illustrative plans there is a very slight reduction in compactness and an increase of
one in the number of city cuts, as compared to the version of the plan that did not
pair incumbents.
(g5) A fifth approach to least change was offered by Defendant-Intervenors in

their brief submitted for the Courts September 18 deadline. They proposed that a
new plan should faithfully adhere to the Legislatures undisputedand
inarguabl[e], political goals of implementing an 8/3 incumbency protection
plan, and thus preserve a plan that elected 8 Republicans and 3 Democrats to
Congress. (Intervenors Br. in Supp. at 12-13). 10 While protecting incumbent
seniority may be a legitimate goal, it is one that must give way to the need to draw
From my perspective as Special Master, the only aspect of this case that is indeed,
undisputed and inarguable is that federal courts have an obligation to choose
policies that do not lead to violations of the Constitution or the Voting Rights Act.
Perez, 132 S. Ct. at 941 (quoting Abrams, 521 U.S.at 79). The characterization of
the motivation of the legislature in creating the current map has already been
argued and litigated and decided in Page. I also note that I have neither reviewed
the trial evidence bearing on the question of legislative intent nor formed an opinion
about this issue.
10
24
a constitutional plan. As indicated above, I have addressed the incumbency issue

by assuring that the homes of all present incumbents are located within the district
in the new illustrative plans that has the same number as their current district. I
have addressed state preferences for least change by drawing plans in which a
majority of the congressional districts in the state were left unchanged, and others
changed only as a result of a constitutionally needed reconfiguration of CD3, which
impacted four other districts. Moreover, given the fact that the major changes in
CD3 required substantial changes in all proximate districts, as discussed above, the
changes made in the impacted districts in the illustrative plans I present,
nonetheless preserved, roughly, the same general configurations in these impacted
districts as in the current plan. I gave no weight to the notion that it was obligatory
for me to propose to the Court plans that were intended to freeze into place partisan
political outcomes such as an 8-3 Republican to Democratic ratio in Virginias
congressional delegation. I have addressed considerations of electability or of
partisan politics only in areas of the state where this issue could not be avoided
because of the need to assess the potential for the creation of a district in which the
minority community realistically had an equal opportunity to elect candidates of
choice. Elsewhere I have drawn districts in a fashion that was blind to the partisan
politics, while cognizant of the shape of the current districts.11
In Gaffney v. Cummings, 412 U.S. 735 (1973), the Supreme Court did allow for
consideration of partisan impact in a plan that was otherwise acceptable and
where the political purpose achieved is to provide districts that would achieve
political fairness between the political parties. On the one hand, the present plan
has been declared constitutionally invalid and, on the other hand, there has been
11
25
6. Assessment of submitted plans in the light of the criteria enumerated above (but
with discussion of issues connected to the creation of a district that avoids
retrogression and provides the African-American community a minority
opportunity to elect candidates of choice largely postponed until a later section):
(a) The revised plan proposed by Plaintiffs (ECF Nos. 229, 230) in the September 18
round of briefs fails to satisfy the first and most minimal definition of least change
given above, as do most of the other submissions. Given the population and
demography of central and eastern Virginia, I cannot recommend any plans to the
Court that make changes in districts beyond the four districts contiguous to CD3
and CD3 itself, since any such plans do not offer a remedy for the constitutional
infirmities of CD3 that is narrowly tailored. This failure to offer a narrowly tailored
remedy immediately rules out six of the eight submitted plans, namely plans
submitted by Plaintiffs, the Governor of Virginia (ECF No. 231), the NAACP (ECF
absolutely no evidence presented that it would achieve political fairness between

the political parties. Support for a plan by incumbents elected in the past under
very similar configurations, who can thus anticipate the high likelihood of their
reelection under the plan, is, in my professional judgment, not evidence of political
fairness between the parties. As McGann et al. note in their forthcoming 2016
book, Gerrymandering in America: The House of Representatives, the Supreme Court
and the Future of Popular Sovereignty, [I]n general it is difficult to get politicians
to vote against the system that elected them. (Cambridge University Press
(unpublished manuscript) p. 252).
26
No. 227), Senator Petersen (ECF No. 219), Mr. Rapoport (ECF No. 228) and
Richmond First (ECF No. 218).
(b) The plans introduced by Defendant-Intervenors, which make only minimal

changes in the present configuration of CD3, continue to split Newport News,
involving a three way split of the city in ways that appear race related,12 and they
do not satisfactorily address contiguity issues in CD3 in terms of providing a
justification for the arbitrary splitting of that city. Thus, from a political science
standpoint, the question of whether these versions of CD3 in their still tortuous
form are anything other than gerrymanders inextricably linked to race is not
satisfactorily dealt with. Moreover, the brief in support of these plans does not
provide justification for the choices made about minority population/voting age
population share in CD3 vis--vis potential for the African-American communitys
realistic opportunity to elect a candidate of choice, and my own analyses, described
in detail below, demonstrate that these plans, which have a 50.2% and 50.1%
African-American voting age percentage, respectively, do not deal with the
constitutional infirmities in CD3 via a narrowly tailored remedy. Either one of
We may think of the three pieces of Newport News that are split between
congressional districts 2 and 3 in each of the Defendant Intervenors two maps as a
leftmost piece, a rightmost piece, and a middle piece. The leftmost piece and the
rightmost piece are placed in the district labeled CD3; the middle piece is placed in
the district labeled CD2. The pieces that are retained in CD3 are the portions with
greater black voting age population (44.6% in the leftmost piece of the city, and
74.6% in the rightmost piece of the city). In toto, they are majority black in voting
age population (56.4%). The population that is moved into CD2 is only 24.1%
African-American in voting age.
12
27
these defects, standing alone, would make the plan unacceptable, and so I cannot
recommend the Court adopt either of Defendant-Intervenors (very similar) plans.13
(c) In dealing with all the various criteria identified above, my recommendation is
that the Court adopt a plan of its own that draws on the best elements of plans that
have been submitted to the Court.
Moreover, the plans proposed by the Defendant-Intervenors remain highly illcompact in the ways in which CD3 is configured, and have a relatively large number
of total city/county splits in that proposed congressional district in comparison, for
example, to the NAACP plan for CD3. These are among the kind of potential
indicia of predominant racial intent to which the decision in Page called attention.
Furthermore, configuring CD3 in the general form that is configured in the present
plan, and in Defendant-Intervenors plans, and in some other submitted plans,
arguably fragments some minority population concentrations in central and eastern
Virginia in ways that raise equal protection concerns.
13
28
II. REALISTIC OPPORTUNITY TO ELECT MINORITY CANDIDATES OF

CHOICE
7. Measurement issues in ascertaining the realistic opportunity of a minority
community to elect candidates of its choice, i.e., to create what is sometimes
referred to as a minority opportunity district or a minority opportunity to elect
district.
(a1) As previously noted, the starting point for any inquiry into the potential to
create a district offering the minority community an equal opportunity to elect
candidates of choice is the demography and geography of areas where such a district
might be constructed. I have done such analyses (see discussion above and further
discussion below).
(b) The second element in considering the potential to create a minority

opportunity to elect district involves the study of elections in the relevant areas of
the state.14 I have also done such analyses (see further discussion below).
For a more general discussion of this and related issues involving measurement
of aspects of elections related to minority representation, see Bernard Grofman,
Lisa Handley & Richard Niemi, Minority Representation and the Quest for Voting
Equality. (Cambridge Univ. Press, 1992). Though the discussion in that volume of
relevant court cases is now largely outdated, its discussion of social science
methodology remains pertinent.
14
29
(c) In looking to specify the set of elections that it useful to analyze, there are
several principles of best practice:
(c1) The elections analyzed should be ones where a viable minority candidate is a
contestant.15 Usually we examine election results involving contests where there
are both minority and non-minority candidates, and where there is a least one
viable candidate of each race.16 Information can, however, also be gleaned from
contests where only minority candidates are involved, or where there is an election
involving a minority candidate in which that candidate wins uncontested.
(c2) The elections analyzed should be recent.
(c3) The elections analyzed should be in the parts of the state where the proposed
remedial district or districts are to be created or, if the election being analyzed is
statewide, it needs to be possible to report results of that election for areas of the
state that (in whole or part) comprise actual or hypothetical districts, i.e., what are
commonly called recompiled elections. The nature of the districts sufficient to
provide the minority community a realistic opportunity to elect candidates of choice
Looking at contests where there is no minority candidate can be misleading if

white voters are less likely to vote for a minority candidate (of a given party) than
they are to vote for a non-minority candidate (of that same party).
15
Election results where candidates of one race are not viable can be misleading if
projected into contexts where we might expect there to be viable candidates of more
than one race. However, essentially uncontested contests can still be useful sources
of information.
16
30
can vary across different areas of a state. Looking at data on recompiled elections
across different potential districts allows us to take into account local variations in
voting behavior and demography.
(c4) The elections analyzed should be of the same or very similar type as the type of
elections at legal issue. Here a particularly important distinction is between
partisan and non-partisan elections: Partisan elections offer voters a partisan cue,
and are more likely to trigger partisan attitudes and loyalties on the part of voters
to the candidate of whichever party they are most attached to. Another difference is
that partisan elections are typically a two stage process in which there is a contest
for party nomination and then a general election.
(c5) If elections are of a partisan nature, then the realistic analysis of potential to
elect minority candidates of choice must consider both the likely outcomes at the
primary election phase and at the general election phase of the election process. To
put it simply: in a partisan election contest, to win, you must first be nominated (in
a party primary) and, once nominated by a party, be able to go on to win the general
election.17
A more formal way to express this insight is in terms of what statisticians refer to
as the Law of Conditional Probability. That Law states that the probability of the
joint outcome (A and B) equals the probability of the outcome A if the outcome B
has occurred, multiplied by the probability of obtaining the outcome B. In the
partisan election context, what this means is that the probability of a (minority)
candidate of choice of the minority community being elected is the product of the
probability that a (minority) candidate of choice of the minority community wins the
17
31
(c6) Analyses should be attentive to whether or not there is an incumbent in the

election contest, and to the race or ethnicity of that incumbent and, for partisan
contests, they should be attentive to the party of the incumbent.
(c7) Sometimes limitations in terms of the availability of data on elections of an

appropriate type will not make it possible to follow all the rules of best practice
identified above. In such cases, decisions as to choice of elections to analyze
requires an intense appraisal of local circumstances.18
(d) I have also reviewed data on general and Democratic primary elections in CD3,
and also on recent Democratic primary and general election outcomes in CD4.
(d1)For all elections, for voting rights purposes, analyses must usually be attentive
to the (expected) racial composition of the districts; for partisan contests it is
important to be attentive to the expected racial composition of the electorate at both
phases of the election process, primary and general.
general election if that candidate is the nominee of a given political party multiplied
by the probability that a (minority) candidate of choice of the minority community
wins the primary of that party, summed over all parties.
For further discussion of this and related issues see Grofman, Bernard, Lisa
Handley and Richard Niemi. Minority Representation and the Quest for Voting
Equality, (Cambridge Univ. Press, 1992).
18
32
(d2A) In the contemporary U.S., based on survey data, exit poll data, and ecological
inference techniques of aggregate election returns, the vast majority of African
voters tend to support Democratic Party candidates. This is especially true in
contests where the Democratic candidate is himself or herself African-American.
For example, in the 2012 Presidential election, based on New York Times exit
polls,19 a majority of white Virginia voters (61%) voted for former Massachusetts
Governor Romney, while the vast majority of African-American voters (93%) voted
for President Obama.20
(d2B) Ceteris paribus, voters who vote for Democratic (Republican) candidates in
general elections are more likely to vote in the Democratic (Republican) primary
than those who do not support Democratic (Republican) party candidates in general
elections, if they do vote in a party primary. Because African-American voters are
more likely to vote Democrat than Republicans in general elections, while white
voters are considerably more likely to be Republican voters in general elections than
is the case for African-American voters, ceteris paribus, the expected proportion of
African-American voters is going to be higher among voters in Democratic primaries
19
Elections.nytimes.com
Lower proportions of the Hispanic and Asian-American voters in Virginia voted

for President Obama in 2012 than was the case for black voters, but Obama support
among these groups was still well above 60%.
20
33
than the proportion of African-American voters among all voters in a general

election. Conversely, the expected proportion of white voters is going to be lower
among voters in Democratic primaries than the proportion of white voters among
all voters.
(e1A) For studying the realistic opportunity of an African-American candidate of

choice to win elections in different congressional district configurations I have made
use of election data at the voter tabulation unit (precinct) level that can be
projected into alternative district configurations to create a recompiled election
within any given proposed congressional district in the State. For recompilations
to examine the realistic opportunity of an African-American candidate to win a
general election in some particular hypothetical (remedy) district I have made use
of data from both the 2008 and 2012 general elections for U.S. President, where an
African-American candidate (Barack Obama) faced a non-African candidate. For
recompilations to examine the realistic opportunity of an African-American
candidate to win a Democratic primary election in some particular hypothetical
(remedy) district I have made use of data from the 2013 Democratic primary to
select a candidate for Attorney General of the State of Virginia. In this biracial
contest, the African-American candidate, Justin Fairfax, was not an incumbent, and
his principal opponent was a white candidate with a strong background who went
on to win the Democratic primary, statewide, and to subsequently be elected
Attorney General of the State of Virginia.
34
(e1B) In studying elections in order to assess the realistic opportunity for the
minority to elect a candidate of choice, I have followed best practices as indicated
above. (1) I have looked only at contests involving an African-American candidate;
(2) I have looked at recent elections, with the oldest from 2008 and most
considerably more recent; (3) I have looked at contests taking place in the area of
the state where there is substantial black population in or proximate to current
CD3; (4) I have looked only at partisan contests; (5) I have looked at both primary
election contests and general elections, (6) I have been attentive in my analyses to
whether or not there was an incumbent in the contest and to the party of that
incumbent. Finally, while I have examined elections in congressional districts 3 and
4 involving an African-American candidate, in order to consider the realistic
opportunity to elect potential in hypothetical districts whose configurations are
quite different from current CD3, of mathematical necessity I have examined
outcomes in compiled (statewide) elections projected into these hypothetical
districts. In so doing I have been attentive to what we can learn from such compiled
elections about potential congressional elections within the same geography.
(e2A) For election contests chosen in line with the best practices above, to the extent
made possible by the limitations of available evidence, data from compiled elections
can allow us to decide whether or not a given proposed congressional district offers
the minority community an equal opportunity to elect a candidate of choice. It is my
35
professional judgment that, to put it straightforwardly, if a minority candidate of

choice wins a compiled Democratic primary election in a proposed remedial
district by an overwhelming margin, at or above 60%, using actual election data;
and in that same district a minority candidate of choice can also be shown to win a
compiled general election by an overwhelming margin in the general election, at
or above 60%, again using actual election data, unless that election can be shown to
be unrepresentative of usual voting patterns, we can be confident that the district
gives minorities in realistic terms, an equal opportunity to the minority community
elect a candidate of choice. In such a district the minority candidate of choice can be
expected to have a realistic opportunity to win the Democratic primary and then go
on to win the general election as the nominee of that party.
(e2B) Note that this 60% figure for the compiled results I mention above is an upper
bound in that, once we are at or above 60% in projected results of appropriate
elections for both the primary and the general election, we can be confident beyond
any reasonable doubt that the district in question provides minority voters a
realistic opportunity to elect candidates of choice, and thus creates what is often
called a minority opportunity district or a minority opportunity to elect district.
I did not use this 60% figure as a bright line test. It may well be the case that a
realistic opportunity to elect district could be created in which compiled relevant
election results from actual elections for the Democratic Party primary and for the
general election projected into the district were each between 50% and 60%. What I
36
have done by examining a large set of alternative configurations of CD3 is to first

establish that a 55% black voting age population was not necessary to create an
opportunity to elect district in CD3, and then to establish that even a 50% black
voting age population was not necessary to create a opportunity to elect district in
CD3, and then to establish that a district with somewhat above a 40% black voting
age population would provide a opportunity to elect district in CD3. The
illustrative districts I drew to remedy the constitutional infirmities in CD3 were
drawn as very compact contiguous districts incorporating whole cities to the extent
feasible, as part of a plan that is drawn along good government lines, and the
percentage of black voting age population that is found in these illustrative districts
is already eleven to fourteen percentage points lower than the black voting age
population percentage in the present CD3. As a consequence of these facts, I did not
find it necessary to seek to determine the absolute minimum percentage of AfricanAmerican voting age population needed to create an opportunity to elect district
for minority voters, since my illustrative versions of CD3 and of the overall plan
are already narrowly tailored to remedy the constitutional violation in CD3. 21
In general, there will be a many parameters whose expected values affect the
likelihood that a minority candidate of choice might win both the Democratic
primary and the general election in a given district for a given type of election.
These parameters include black Democratic share of the eligible electorate, white
Democratic share of the eligible electorate, Republican share of the eligible
electorate, white Democratic support levels for minority candidates of choice in a
Democratic primary among those who vote in the primary, black Democratic
support levels for minority candidates of choice in a Democratic primary among
those who vote in the primary, white Democratic support levels for minority
candidates of choice running as Democrats in a general election among those who
vote in the general election, black Democratic support levels for minority candidates
21
37
Moreover, in seeking to craft a constitutional remedy, recognition of uncertainty

and the need for caution are called for, since election candidates and contests do
differ.22
(f) I have also reviewed the analyses provided by Dr. Lisa Handley in her expert
witness report attached as an Appendix to the Governor of Virginias filing with the
of choice running as Democrats in a general election among those who vote in the
general election, white turnout rates in the Democratic primary, black turnout rates
in the Democratic party primary, white turnout rates in the general election, and
black turnout rates in the general election. As discussed further below, Dr.
Handleys analyses in her Report attached as an Appendix to the Governors
September 18 brief deals with exactly such factors. A realistic appraisal of local
circumstances might also include yet other factors, e.g., investigating the pool of
available minority candidates and the campaign resources open to them. Here I
simply note that there are viable minority candidate who have already won the
Democratic Party nomination in CD3 or in CD4 or in state legislative districts in
the same or proximate areas of the state.
Also, even if it were to be determined beyond any reasonable doubt that, for some
particular district, a lower black voting age percentage, say with black voting age
population below 40%, would still create a realistic opportunity to elect district, in
that a minority candidate of choice had a realistic chance to win the primary and, if
winning the primary, to go on to win the election, it is highly like that drawing a
district with exactly that black voting age population percentage would require
violating other criteria, such as seeking to avoid city and county splits. Finally, I
note that, while configuring CD3 with a black voting age population slightly above
40% as a minority opportunity to elect district also allowed for the creation of a
reconfigured CD4 becoming a second minority opportunity to elect district, further
reducing the black voting age population in either CD3 or CD4 would not make
possible the creation of a third minority opportunity to elect district in the
relevant part of the state. In my judgment, based on an extensive consideration and
analysis of alternative plans, given the demography and geography of the area
immediately proximate to the present CD3, it is not possible to draw three
minority opportunity to elect districts in that part of the state. In contrast,
drawing two such minority opportunity to elect districts follows naturally from the
population demography and geography of minority population concentrations in
that part of Virginia and from the decision to remedy the constitutional violation in
current CD3.
22
38
Court pursuant to its September 18, 2015 deadline. These analyses are done using
methods that have become standard in political science analyses of cases involving
voting rights issues.23 Most of her analyses focus on the potential for providing
minorities an equal opportunity to elect candidates of choice by drawing CD3 in the
area around Hampton, Newport News, Portsmouth and Norfolk. She looks at issues
such as minority political cohesion and minority turnout. I find her analyses to be
complementary to my own analyses, and leading to very similar conclusions.
(g1A) My expectation that black voters would be disproportionately represented
among the voters in Democratic primaries in Virginia relative to the overall

African-American share of the potential electorate (those of voting age) is confirmed
by Dr. Handleys analyses. She finds that, in the Democratic primary for U.S.
President in 2008, blacks opted to vote in the Democratic primary at a much higher
percentage than whites did: approximately 18% of the black voting age population
compared to approximately 11% of white voting age population cast a vote in the
Democratic Primary in 2008 (Handley Report p.11). As she correctly notes: The
implication of this analysis is that the percent black voting age population needed
to produce an effective black district tends to be lower for Democratic primary
elections than for general elections (Handley Report p.11). Ceteris paribus, this
Some of those methods are ones that I have myself developed, or adapted from
the work of others to make then better applicable to the redistricting context. These
methods, e.g., ecological inference to infer turnout levels and levels of political
cohesion of African-American and white voters (or of Hispanic and non-Hispanic
voters) are discussed in my own published research, including the work that I have
done jointly with Dr. Handley that she cites in her Report.
23
39
finding clearly indicates that there can be a realistic opportunity for a minority
candidate of choice to win the Democratic Party nomination even in a district that,
overall, is less than majority black (majority minority) in voting age population.
(g1B) The likelihood that a minority candidate of choice will win an election
depends in part upon relative turnout levels of minority and non-minority voters in
that election. Dr. Handleys analyses (Table 4 in her Report) demonstrate that
African-American turnout in general elections is only slightly below white turnout
in the general election in the current CD3 (34% versus 37%). Ceteris paribus,
relatively similar levels of white and black turnout make it more likely that a
minority candidate of choice has a realistic opportunity to be elected.
(g1C) The likelihood that a minority candidate of choice will win an election
depends in part and upon the degree of cohesion of minority voters in their voting
support for the minority candidate of choice, and the willingness of non-minority
voters to vote for the candidate of choice of the minority community (what is often
called cross-over voting). Dr. Handleys analyses demonstrate that, in the current
CD3, black voters are almost perfect cohesive in their voting behavior (giving an
average of over 97% of their votes to a particular candidate (the Democrat) in
partisan statewide and congressional contest on candidate choice, and are much
more homogenous in voting patterns than white voters in the current CD3 (see
Table 5 in her Report). In particular, even when whites and blacks support
40
different candidates, a substantial proportion of white voters vote for the minority
candidate of choice. For example, Dr. Handley finds that, in the current CD3, a
(bare) majority of white voters supported Barack Obama in the 2012 presidential
general election, while white support for him in the 2008 general election was also
high, somewhere between 43 and 46% (see Table 5 in her Report). Even in the
2008 Democratic primary election she estimates white support in (then) CD3 was
60.1%. In the primary election for the Democratic Party nomination for the Office of
Attorney General, where white cross-over voting was low, she estimates that at
least 32% of the white voters in that primary cast a vote for Justin Fairfax, the
African-American candidate (Handley Report, p. 13).Ceteris paribus, high levels of
minority political cohesion and substantial levels of white cross-over voting make it
much more likely that a minority candidate of choice has a realistic opportunity to
be elected, even in contests in Virginia where voting is polarized along racial lines,
as long as minority population is large enough to allow for a party nomination and
subsequent election with cross-over support from non-minority supporters of that
party.
41
III. ILLUSTRATIVE MAPS OF POTENTIAL CONFIGURATIONS OF REMEDIAL

DISTRICTS
8. Illustrative maps
(a1A) As indicated earlier, I have drafted two illustrative maps to show how the
constitutional defects in the present CD3 can be remedied. Those plans draw CD3
as a contiguous, highly compact district, with few city splits, and with a substantial
minority population. Also noted earlier, once CD3 has been redrawn in this fashion
as a constitutional district, there is no need for changes to any districts other than
the four districts directly contiguous to CD3, namely CD1, CD2, CD4 and CD7. In
each of these plans, CD2, as in the present plan, is a district drawn beginning in the
eastern shore and picking up the bulk of its proximate population in the city of
Virginia Beach. Similarly, as in the present plan, CD1 and CD7 are districts
further to the west, with CD7 below CD1, with changes in configuration from the
current plans being ones necessary for population equality. Also, as noted earlier,
the illustrative maps both place the homes of all present incumbents in districts
that bear the same number as their present districts.
(a1B) Furthermore, the major change in the configuration of District 3, concern for
avoiding unnecessary splits of cities and counties, and the mathematics of
population equality balancing, has led me, virtually automatically, to create the
remaining district whose boundaries have been changed, CD4, as a district that
42
includes the whole cities of Richmond and Petersburg and which then moves south
and east for population equality purposes. CD4 as so drawn, with a substantial
minority population, neither packs nor fragments minority voting strength in the
eastern portion of the state. Indeed, appropriate remediation of the constitutional
infirmities identified in CD3 has had the additional benefit of creating a second
minority opportunity to elect district in CD4 that remedies previous
fragmentation of the black population in eastern central Virginia.
(b) The two illustrative maps are labeled (using my original mnemonics for the
iterative line drawing process), as NAACP plan Modification 6, and Current
Congressional Plan Modification 16. The key difference between them vis--vis CD3
is that NAACP plan Modification 16 keeps whole in CD3 the cities of Newport
News, Hampton, Portsmouth and Norfolk, and splits the Isle of Wight; while
Current Congressional Plan Modification 16 keeps whole in CD3 the cities of
Newport News, Portsmouth, and the Isle of White, with splits in the city of Norfolk
that largely follow the lines of the split of that city between CD2 and CD3 found in
the current congressional map, and splits in Hampton to assure contiguity by bridge
between northern and southern shore components of CD2. In one illustrative map
the Franklin City is entirely in CD3; in the other it is entirely in CD4.24
Both illustrative remedial maps keep the entire eastern shore and the Virginia
Beach whole, and in the same district, CD2. In one illustrative plan keeping
Virginia Beach whole forces a split in Gloucester for population purposes. In the
other illustrative plan, keeping Virginia Beach whole creates population balancing
issues that affect the nature of the split in Norfolk. Both illustrative plans keep
24
43
(c1) Table 1 below shows key data for CD3 in each of the illustrative plans. A
portion of the current CD3 becomes the largest component of the newly configured
CD3 in each of the illustrative remedial plans I have proposed (NAACP plan
Modification 6, and Current Congressional Plan Modification 16). To facilitate
comparisons, similar data for the current plan, the Defendant Intervenors plans,
and the Plaintiffs September 18 plan are also shown in Table 1. These three
additional plans are shown only as indicative of the types of maps that have been
proposed by parties and Amici, not as maps that I am proposing for consideration by
the Court as remedial plans. For reasons previously indicated in this report, I do
not regard these three maps, nor any of the other eight maps that were submitted
with full supporting data on September 18 pursuant to the Courts Order, as
appropriate remedies for the constitutional violations in CD3 identified in the
majority opinion in Page.
whole Richmond, Petersburg, and the Tri-Cities area in CD4 and also keep whole in
CD4 almost all of the cities and counties of southeastern Virginia from Richmond to
Chesapeake. However, both illustrative maps split Suffolk in the process of making
a connection to Chesapeake, where the home of Congressman Forbes is located, in
order to place him in his home district of CD4 -- with the exact nature of the splits
related to population balancing issues. One of the illustrative remedy maps also
splits Chesapeake into two congressional districts (CD3 and CD4) for population
balancing purposes, with the part of Chesapeake in CD4 including the home of
Congressman Forbes. Given the changes in the configuration of CD3 to remedy the
constitutional violation, avoiding the pairing of Congressman Forbes (CD4) and
Congressman Rigell (CD2) in the same district requires CD2, for population
purposes, to extend into Poquoson and elsewhere on the northern shore so as to
avoid an extension of CD2 into Chesapeake in a way that would include
Congressmans Forbes home.
44
Table 1
Comparisons of CD3 Among Illustrative/ Proposed Remedial Plans
Plan(district3)
White
VAP
Black
VAP
Hispanic
VAP
2012
Dem.
President
2008
Dem.
President
Current
37.30%
56.30%
4.50%
79.50%
77.60%
Plaintiff's
42.20%
51.50%
4.90%
72.70%
70.00%
IntervenorPlan
1
43.10%
50.20%
5.00%
75.60%
IntervenorPlan
2
43.00%
50.10%
5.00%
NAACPPlan
modification6
50.6%
42.3%
Modified
Current
Congressional
plan
modification16
48.3%
45.3%
Split
localities
Polsby
Popper
Reock
Ehrenberg
0.08
63.8%
0.12
0.16
74.10%
59.9%
0.08
0.12
75.10%
73.60%
60.6%
0.10
0.16
5.2%
67.5%
64.8%
60.5%
0.36
0.36
4.3%
67.6%
65.4%
63.0%
0.22
0.31
45
2013
Dem
A.G.
Primary
J.Fairfax
62.4%
0.12
(c2A) In what I have called illustrative NAACP Plan Modification 6, the

reconfigured CD3 is contiguous by land, bridge or tunnel. As demonstrated in Table
1, it is far more compact than CD3 in the current plan, and far more compact than
CD3 in any of the submitted plans provided in Table 1 solely for comparison
purposes, whether we look at the Polsby-Popper measure (based on the area of a
circumscribing circle relative to the area of the district) or at the Reock measure
(based on the degree of irregularity/jaggedness in the district perimeter). The
current CD3 scores .08 on Polsby-Popper and .12 on the Reock measure; the
comparable scores for CD3 in my illustrative modification 16 of the present map are
.36 and .36 i.e., between three and four times as high in each case. Moreover, in
the NAACP modification reconfiguration of the current map, the reconfigured CD3
splits only 2 cities or counties, as compared to 6 that are in the current map.
(c2B) In what I have called illustrative modification 16 to the current map, the
reconfigured CD3 is contiguous by land, bridge or tunnel. As demonstrated in Table
1, it is far more compact than CD3 in the current plan, and far more compact than
CD3 in any of the submitted plans shown in Table 1 for comparison purposes,
whether we look at the Polsby-Popper measure (based on the area of a
circumscribing circle relative to the area of the district) or at the Reock measure
(based on the degree of irregularity/jaggedness in the district perimeter). The
current CD3 scores .08 on Polsby-Popper and .12 on the Reock measure; the
46
comparable scores for CD3 in my illustrative modification 16 of the present map are
.22 and .31 i.e., about two and a half time times as high in each case. Moreover, in
illustrative Modification 16 to the current map, the reconfigured CD3 splits only 4
cities or counties, two fewer than are split in the current map.
(c3A1) As shown in Table 1, the black voting age population in CD 3 in both NAACP
plan Modification 6 and Current Congressional Plan Modification 16 is in excess of
40% but still below 50% (42.3% and 45.3%, respectively), and the white voting age
population in CD3 in each plan is very close to 50% (slightly above in one plan,
slightly below in the other).
(c3A2) As also shown in Table 1, projecting into these illustrative configurations of

CD3 the vote share for Barack Obama in his 2008 statewide contest in the general
election, when President Obama was not yet an incumbent, we find that (now
President) Obama would have carried each of these districts in 2008 with about
65% of the vote.25 Thus, there can be no doubt that, if a viable African-American
candidate wins the Democratic congressional primary in districts configured as
shown in these two illustrative plans, then that candidate of the Democratic party
has a realistic opportunity to win election in CD3 due to cohesive voting from within
the African-American community and cross-over voting from non-black Democrats -even if that candidate is not an incumbent.
In 2012, Obama won in the general election with about 67% of the vote in each
district.
25
47
(c3A3) The best evidence we have available to us for determining whether or not we
should expect that an African-American candidate has a realistic opportunity to win
the Democratic party nomination in these reconfigured versions of CD3 is to project
into these districts the 2013 vote share of the African-American candidate, Justin
Fairfax, in his quest for the Democratic partys nomination to be that partys
candidate for statewide office of Attorney General. Mr. Fairfax was not an
incumbent, and his principal opponent was a white candidate with a strong
background who went on to win the Democratic primary, statewide, and to
subsequently be elected Attorney General of the State of Virginia. Thus, evidence
that Mr. Fairfax would have won the 2013 Attorney General Democratic primary
within the boundaries of the illustrative remedial CD3 districts I have drawn for
the Court, despite the fact that these districts are not majority black in voting age
population, provides very strong evidence that a viable black candidate, even if not
an incumbent, who achieves cohesive support from the minority community and
perhaps also some cross-over support from white Democrats, has a realistic
opportunity to win the Democratic primary within these districts. In fact, as Table 1
demonstrates in both these illustrative configurations of a remedial CD3, Mr.
Fairfax wins easily, with between 60.5% and 63% of the vote.
(c3A41) Indeed, Mr. Fairfax receives more votes in the Democratic primary in the
illustrative Modification 14 of the current version of CD3 than he did in the actual
48
current CD3 (63.0% versus 62.4%), despite the fact that the black voting age
population in the current CD3 is 11.0 percentage points higher than in the
illustrative Modification 16 of the current version of CD3.26 Similarly, Mr. Fairfax
does almost as well in the Democratic primary in the NAACP modification 6 version
of CD3 as he did in the current CD3 (60.5% versus 62.4%), despite the fact that the
black voting age population in the current CD3 is 14.0 percentage points higher
than in the NAACP 6 version of CD3.27
(c3A42) When we look at the 2008 Democratic primary results for Barack Obama,
we find that Mr. Obama ran rather well throughout the state but he runs especially
well in all of central and southeastern Virginia, even in areas where AfricanAmerican voting strength was limited, and even though he had strong white
opponents such as Hillary Clinton. In a compiled election projected into district CD3
in either of the illustrative plans, Obama would easily, indeed overwhelmingly, win
the Democratic primary.
(c3A5) In sum, the combination of the evidence we have examined demonstrates

that in a situation where there is no incumbent white Democratic candidate, an
For a theoretical explanation of this seemingly counterintuitive result see

Bernard Grofman, Lisa Handley & David Lublin, Drawing Effective Minority
Districts: A Conceptual Framework and Some Empirical Evidence, 79 N.C. L. Rev.
1383 (2001).
26
27
See previous footnote.
49
African-American candidate who is the candidate of choice of the African-American

community clearly has a realistic opportunity to win district in CD3 either in the
version of that district drawn in NAACP plan Modification 6, or in the version of
that district drawn in Current Congressional Plan Modification 16.
(c3A6) The fact there is presently an African-American incumbent whose home is

located in CD3 in both the NAACP plan Modification 6 configuration and in the
version of that district drawn in Current Congressional Plan Modification 16 adds
even more confidence to my assessment that, beyond any reasonable doubt, the
versions of CD3 shown in these illustrative plans will avoid any retrogression and
provide the African-American community a realistic opportunity to elect a
candidate of their choice in CD3.28 Similarly, the evidence reviewed above, based on
primary contests involving non-incumbents (Obama in 2008; Fairfax in 2013) shows
that that a viable black candidate, even if not an incumbent, who achieves cohesive
support from the minority community and perhaps also some cross-over support
from white Democrats, has a realistic opportunity to win the Democratic primary in
CD3 as reconfigured in the two illustrative remedial plans.
Between 2002 and 2014, in general elections, the African-American incumbent in

CD3 has never received less than 68.7% of the vote, and in five elections during the
decade he received over 94% of the vote in essentially uncontested election. Even if
that incumbent were to retire, the seat would still be open seat. While the black
voting age population in the district he ran in is higher than that in the
reconfigured remedial versions of CD3 in my illustrative maps, even if we were to
cut his expected vote share by a dozen or so percentage points, he would still be
expected to win the general election in the district. And, of course we have
projections from the vote shares of Obama in the district that, in fact, we would
expect him to still win very handily by a comfortable margin.
28
50
(d1) Table 2 below shows key data for CD4 in each of the illustrative plans. The
next-largest portion of the current CD3 becomes the second largest portion of CD4
in each of the two illustrative plans. Once again, simply to facilitate comparisons,
similar data for the current plan, the Defendant-Intervenors plans, and the
Plaintiffs September 18 plan are also shown in Table 2, even though these plans
are not ones which I would recommend to the Court.
51
Table 2
Comparisons of CD4 among Illustrative /Proposed Remedial Plans
Plan(district4)
Current
Plaintiff's
White
VAP
Black
VAP
Hispanic
VAP
2012
Dem.
President
2008
Dem.
President
2013
Dem
A.G.
Primary
J.Fairfax
Split
localities
Polsby
Popper
Reock
Ehrenberg
63.3%
31.3%
4.0%
48.8%
48.0%
66.0%
0.20
0.20
63.1%
30.7%
4.3%
53.1%
52.2%
62.2%
0.19
0.22
63.3%
31.2%
4.2%
49.1%
48.3%
66.0%
0.19
0.20
63.2%
31.2%
4.2%
49.2%
48.3%
66.0%
0.19
0.20
52.0%
42.0%
4.4%
63.2%
62.2%
62.5%
0.22
0.25
53.6%
40.9%
3.9%
60.9%
60.1%
61.1%
0.20
0.26
IntervenorPlan1
IntervenorPlan2
NAACP Plan
modification 6
Modified
Current
Congressional
plan
modification
16
52
(d2A) In what I have called illustrative NAACP Plan modification 6, the

reconfigured CD4 is contiguous in that you can move from one land portion of the
district to all other land portions of the district without the need to rely on
contiguity via water. As demonstrated in Table 2, it is more compact than any of the
submitted plans provided for comparison purposes in Table 2, both with respect to
the Polsby-Popper measure -- based on the area of a circumscribing circle relative to
the area of the district, and the Reock measure -- based on the degree of
irregularity/jaggedness in the district perimeter. The current CD4 scores .20 on
Polsby-Popper and .20 on the Reock measure; the comparable scores for CD4 in
NAACP modification 6 are .22 and .25. This reconfigured CD4 splits one more city
or county than the current map and two more splits than in some alternative
proposals. In part the number of city/county splits in this configuration is related to
issues of incumbency pairing in terms of the split in Suffolk to pick up the
incumbents home. Earlier versions of the same plan that paired two incumbents
had one less city/county split.29
The incumbent in CD4 lives in Chesapeake and the incumbent in CD2 lives in
Virginia Beach, but they live less than 18 miles apart and they live in the
southeastern corner of the state, where redistricting options are limited by
Chesapeake Bay to the east and the North Carolina border to the south. Once
major changes have been made to the configuration of CD3 to create a
constitutional district, and choices have been made about the shape of CD2 and
other districts, if no incumbents are to be paired, there are constraints on
redistricting options for CD4 that will require some splitting of cities or counties to
insure that the home of the present incumbent of CD4 remains within that district
in the reconfigured plans.
29
53
(d2B) In what I have called illustrative modification 16 to the current map, the
reconfigured CD4 is contiguous in that you can move from one land portion of the
district to all other land portions of the district without the need to rely on
contiguity via water. As demonstrated in Table 2 it is as compact on one measure
and more compact then current CD4 on the other, and more compact than the 4th
congressional district in any of the submitted plans provided for comparison
purposes in Table 2. The current CD4 scores .20 on Polsby-Popper and .20 on the
Reock measure; the comparable scores for CD3 in my illustrative modification 14 of
the present map are .20 and .26. The reconfigured CD4 splits the same number of
cities or counties as the current CD4. In part, the number of city/county splits in
this configuration is related to issues of incumbency pairing. Earlier versions of
each plan that paired two incumbents had one less city/county split.
(d3A) As shown in Table 2, the black voting age population in CD 4 in both NAACP
plan Modification 6 and Current Congressional Plan Modification 16 is above 40%,
and the white voting age population in CD3 in each plan is above 50% (52.0% and
53.6%, respectively).
(d3B) As also shown in Table 2, projecting into these illustrative configurations of

CD4 the actual vote share for Barack Obama in his 2008 statewide contest in the
general election, when President Obama was not yet an incumbent, we find that
(now President) Obama would have carried CD4 in the general election in 2008 in
54
each of these illustrative plans with between 60.1% and 62.2% of the vote. This give
rise, in my professional judgment, to a conclusion that, if a viable African-American
candidate wins the Democratic congressional primary in districts configured as
shown in these two illustrative plans, then, ceteris paribus, that candidate of the
Democratic party has a realistic opportunity to win election in CD4 due to cohesive
voting from within the African-American community and cross-over voting from
non-black Democrats. However because there is a white Republican incumbent
whose home is also in the district, this may be a somewhat closely contested general
election, but the previous level of vote success of the African-American candidate of
the Democratic party in the 2012 general election in CD4, is a 42.9% vote share, in
a district where, in 2012, the incumbent President Obama won only 48.8% of the
vote. If you reconfigure the district to not fragment compact and contiguous
minority population, then CD4 has an Obama vote in the general election in 2012 of
between 60.9% and 62.5% of the vote in the two illustrative plans. Since this vote
share is around 12 to 14 percentage points higher than Obamas vote share in the
current CD4, although we cannot directly project Ms. Wards 2012 vote in the
current CD4 into the reconfigured CD4 districts, it seems a very high probability,
posed as a hypothetical, that were she to have run as the Democratic party nominee
in 2012 in a CD4 whose black voting age population share is around ten percentage
points higher than in the current CD4 where she ran and lost, and where the new
55
district has an Obama vote share above 60%, she would have had a realistic
opportunity to win election.30
(d4A) An important piece of evidence that an African-American candidate has a

realistic opportunity to win the Democratic Partys congressional nomination in
these reconfigured versions of CD4 is that, in the most recent election (2012) where
a Democrat contested the current 4th congressional district, an African-American
candidate, Ella Ward, won the Democratic Partys congressional nomination. Since
the current district CD4 has a much lower African-American voting age population
than the proposed versions of CD4 (31.3% versus 42.0%; 31.3% versus 40.9%), it
seems clear a fortiori that there is a realistic chance that an African-American who
is the candidate of choice of the African-American community can be nominated by
the Democratic party from either of the reconfigured versions of CD4, since the
district is gaining, and substantially so, in African-American population.
(d4B.) Further evidence we have available to us for determining whether or not we

should expect that an African-American candidate has a realistic opportunity to win
the nomination in these reconfigured versions of CD4 is to project into these
districts the 2013 vote share of the African-American candidate, Justin Fairfax, in
his quest for the Democratic partys nomination to be that partys candidate for
A realistic opportunity to be elected does not mean a 100% certainty. With

elections nothing is certain until the ballots have been counted and certified (or
until the outcome has successfully sustained court challenge).
30
56
statewide office of Attorney General. As noted earlier, Mr. Fairfax was not an
incumbent, and his principal opponent was a white candidate with a strong
background who went on to win the Democratic primary, statewide, and to
subsequently be elected Attorney General of the State of Virginia. Thus, evidence
that Mr. Fairfax would have won the 2013 Attorney General Democratic primary
within the boundaries of the CD4 districts in the two illustrative districts I have
drawn for the Court, despite the fact that these districts are not majority black in
voting age population, provides strong evidence that a viable black candidate, even
if not an incumbent, who achieves cohesive support from the minority community
and perhaps also some cross-over support from white Democrats, has a realistic
opportunity to win the Democratic primary within these districts. In fact, as Table 2
demonstrates, in both illustrative configurations of a remedial CD4, Mr. Fairfax
wins easily, with over 60% of the vote (62.5% and 61.1%, respectively).
(d4C). In a compiled election projected into CD4 in either of the illustrative plans,
Barack Obama would easily win the Democratic primary in 2008.
(e) As noted above African-American candidate of choice, Ella Reed, won the
Democratic Party nomination in CD4 in 2012, in the first election held under those
new district lines, but she rather lost the general election with a vote share of only
42.9% vote. Ms. Reed chose not to seek the Democratic nomination in CD4 in 2014.
Indeed, no Democrat chose to seek the Democratic Party nomination in CD4 in
57
2014. I conclude from this latter fact, and the fact that President Obama failed to
carry CD4 in 2012, even though he was then already an incumbent president, that
the perception of voters in CD4 who identify with the Democratic party, including
African-American voters, is that CD4, as presently configured after the 2010 census,
given that it has a sitting Republican incumbent, is not a district where a Democrat
can win election to congress and thus, a fortiori, not a congressional district where
an African-American candidate of choice can win election, even though an AfricanAmerican candidate can win the Democratic party nomination.31
When we turn from specific consideration of individual districts, first CD3 and then
CD4, to the overall features of the two illustrative remedial maps in the full set of
five districts that have been modified (CD1, CD2, CD3, CD4, and CD7) we find that
not only do the illustrative maps resolve the constitutional defects in current CD3 in
a narrowly tailored fashion, and pay respect to the demography and geography of
In some circumstances, it may be easier for a minority candidate of choice to win

the Democratic primary than to win the general election (e.g., when there are few
white Democrats relative to the number of African-American Democrats, and the
combined African-American and non-African-American vote for the Democratic
candidate is not large enough to win a general election); while in other
circumstances it may be harder for a minority candidate of choice to win the
Democratic primary than to win the general election (e.g., when there are many
more white Democrats than black Democrats, but the combined African-American
and non-African-American vote for the Democratic candidate is large enough for a
Democrat to win a general election). But, as emphasized earlier, to have a
realistically drawn minority opportunity district it is necessary to have a realistic
chance to win both a party primary and a general election, running in the latter as
the official candidate of that party. For further discussion of this and related issues
see Bernard Grofman, Lisa Handley & David Lublin, Drawing Effective Minority
Districts: A Conceptual Framework and Some Empirical Evidence, 79 N.C. L. Rev.
1383 (2001).
31
58
the state that results in plans that are far superior to the current map in their
treatment of minority population concentrations, but we also find that the two
illustrative maps, which are drawn according to good government criteria, are
clearly overall superior to the current plan on the criteria of city splits and
compactness.
These comparisons are shown in Tables 3 and 4.
59
Table 3. Compactness Comparisons in the Modified Districts (1, 2, 3, 4, 7)
current
DISTRICT map
Polsby
Popper
1
0.18
2
0.21
3
0.08
4
0.2
7
0.13
NAACP
Mod6
Polsby
Popper
0.2
0.2
0.36
0.22
0.17
Mod16
of
Mod16of
current current
NAACP current
map
map
Mod6 map
Polsby
Popper Reock
Reock Reock
0.23
0.21
0.25
0.25
0.22
0.21
0.22
0.21
0.22
0.12
0.36
0.31
0.2
0.2
0.25
0.26
0.2
0.2
0.25
0.26
Table 4. Locality Split Comparisons in the Modified Districts (1, 2, 3, 4, 7)
DISTRICT currentmap
localitysplits
1
5
2
3
3
6
4
4
7
4
Mod16ofcurrent
NAACPMod6 map
localitysplits localitysplits
4
2
2
5
3
60
4
3
4
4
3
IV. FINDINGS AND RECOMMENDATIONS

For reasons elaborated in the text, I cannot recommend to the Court any of the
eight full plans presented to the Court as of September 18, 2015. Six of the eight
plans, by making changes in the current congressional plan that extend into
districts that are not immediately contiguous to current CD3, are making changes
in more districts that we need to modify for the purpose of being able to create a
fully satisfactory remedial version of CD3. These six plans can be eliminated
immediately on grounds of lack of narrow tailoring. Two others (the two plans
offered by Defendant Intervenors), which do limit changes to only five districts, and
which make relatively minimal changes in the current boundaries of CD3, still fail,
in my judgment, to provide a constitutional plan. First, the brief justifying these
plans does not provide satisfactory responses to issues of contiguity, unnecessary
city and county splits, and lack of compactness in CD3. These issues were among
the key issues in the finding that the current CD3 was unconstitutional. Second,
and even more importantly, the plans fail to pay adequate attention to the nature of
a district that would be needed to provide to achieve a non-retrogressive district
that provided minority voters equal opportunity to elect a candidate of choice. The
minority population placed in CD3 in these plans is far in excess of what is needed
for a minority opportunity district. Thus the plan fails the narrow tailoring test.32
Furthermore, by configuring CD3 in the same general form that it is configured

in the present plan, the Defendant-Intervenors plans (as well as the Plaintiffs plan
and some other submitted plans), arguably pack minority voting strength in CD3,
which then has the effect of fragmenting a potential compact minority population
concentration elsewhere that could form the basis of a reconfigured CD4.
32
61
For these reasons, I cannot recommend the Court adopt either of DefendantIntervenors (very similar) plans. Thus, my recommendation is that the Court adopt
a plan of its own that draws on the best elements of plans that have been submitted
to the Court.
As described in Page, one justification offered by some individual Virginia

legislators for the way in which the present CD3 was configured, was that such a
configuration was required because only a district with a 55% black voting age
majority could provide African-American voters with a realistic opportunity to elect
candidates of choice. That assertion is unsupported by any empirical evidence.
Moreover, in my view as a political scientist specialist on redistricting, not only does
talismanic reliance on a figure of 55% black voting age population impose a bright
line test rejected by the Supreme Court in Alabama Legislative Black Caucus et al.
v. Alabama, 135 S. Ct. at 1270,33 but analyses specific to the eastern part of
Virginia demonstrate that the claim that a 55% minority voting age population is
In Alabama Legislative Black Caucus, the Supreme Court discussed the concept
of retrogression and asserted that avoiding retrogression does not require a covered
jurisdiction to maintain a particular numerical minority percentage, but instead
requires the jurisdiction to maintain a minoritys ability to elect a preferred
candidate of choice. 135 S. Ct. at 1272 (2015). The Court specifically rejected
reliance on a mechanically numerical view as to what counts as forbidden
retrogression Id. at 127374. (This language is quoted in the Page majority
opinion).
33
62
needed in a district to assure African-American voters a realistic opportunity to

elect candidates of choice in CD3 is, factually, flatly wrong.
I have reached the empirical conclusion at the end of the paragraph immediately
above by my own independent conceptual analyses of the basic elements of
elections, such as the two-stage nature of partisan contests, and by my own
independent empirical analyses of demographic and electoral data from eastern
Virginia. In these empirical analyses I have drawn heavily on state wide races
involving biracial contests where there are viable African-American candidates and
viable white candidates, but I have also looked at recent election outcomes in the
present CD3 and CD4. In addition, I have also reviewed data analyzed by Dr.
Handley on the voting cohesion of white and black voters in elections involving a
black candidate, and on estimated turnout levels of minority and non-minority
voters in primaries and generals. I found her work fully supportive of the conclusion
stated in the paragraph above, and her modes of analysis complementary to my
own. Indeed, either her analyses, standing alone, or my analyses, standing alone,
fully support the conclusion that, in parts of central and southeastern Virginia in
the area of the present CD3, an African American voting age population percentage
considerably less than 50% could be the basis for a minority opportunity to elect
district.
Given the demography and geography of the State of Virginia, it is my professional
judgment that the most appropriate way to remedy the constitutional violation
63
identified in CD3 is to redraw CD3 as a Newport News-Hampton-PortsmouthNorfolk based district that is contiguous, highly compact, and with the few city or
county splits that are found in the reconfigured district motivated by population
equality concerns, and/or a desire to assure contiguity by land bridge, or tunnel. As
discussed in detail earlier in this Report, such a remedial district can be drawn with
a substantial minority population that is sufficient to provide minority voters an
equal opportunity to elect candidates of choice. Doing so does not require that the
district have a black voting age majority. Rather it requires that voting be such
that, when the African-American community votes in a cohesive fashion, a
candidate of choice of the minority community can be expected to have a realistic
opportunity to win both a primary and a general election -- with success in the
general election occurring because the minority candidate of choice wins the support
of white voters who share that candidates partisan preferences (i.e., the minority
candidate of choice receives some white cross-over voting support). As
demonstrated by my empirical analyses, and as confirmed by Dr. Hadleys analyses
using in some cases, a different methodology, the minority vote age percentages
shown in the illustrative CD3 districts discussed in this Report realistically offer
African voters an equal opportunity to elect candidates of choice, i.e., create a
minority opportunity district, as part of an map that is, overall, constitutionally
drawn, and does not in any way have race as predominant motive, and where the
remedial district is narrowly tailored to achieve constitutional objectives. Moreover
in my view they are non-retrogressive.
64
It is my professional judgment that the present configuration of CD4 makes it

essentially impossible for the African-American population in central Virginia and
eastward (the Richmond to Chesapeake area) to have a realistic opportunity to elect
a candidate of choice, because that minority population has been fragmented by the
current congressional map. In contrast, the configuration of CD4 in the two
illustrative maps (NAACP 6 and Modified Current Map 16) each provide the
African-American community that would be located in a reconfigured CD4 a
minority opportunity district.
I took as my goal the drawing of a non-retrogressive compact contiguous minority

opportunity to elect district based on good government principles that was
designed to fix the constitutional infirmities in the current CD3 infirmities that
include the packing of minority voting strength into a single district in a fashion not
required to avoid retrogression or to create a minority opportunity to elect district.
The current configurations of CD3 and CD4 reflect a combination of packing of
minority voting strength in central and southeastern Virginia, in CD3, and
fragmentation of minority voting strength in central and southeastern Virginia, in
CD4, that can and should be remedied. Packing and cracking, i.e., fragmentation,
are two sides of the same coin. Once I drew CD3 to address the constitutional
violations in the present CD3 and located it in the Newport News-HamptonPortsmouth-Norfolk area, I found that the geography and demography of the
remaining portions of central and southeastern Virginia were such that the
65
fragmentation of minority voting strength in the southeastern and central part of

the state existing in the current CD4 was remedied simply by including whole in
the reconfigured CD4 the cities of Richmond and Petersburg that previously had
been contained in whole or in part in current CD3, and then naturally extending
CD4 south and east for population purposes until it reached the border of the newly
reconfigured CD3.
While the two plans I submitted to the Court are illustrative, and there may well be
ways of improving them further, it is my professional judgment that each provides
an appropriate and narrowly tailored means of remedying the constitutional
infirmities in the present CD3 in the context of a good government map that does
not have race as a predominant motive, and that is attentive to the legal issues in
this case to which the Court has called attention.
Encs. DLS Reports on NAACP plan Modification 6 and Current Congressional Plan
Modification 14, along with maps of these two plans showing boundaries and
locations of the homes of present congressional incumbents.
66

VA Redistricting Report of Special Master (Nov. 2015)

Uploaded by

Copyright:

Available Formats

VA Redistricting Report of Special Master (Nov. 2015)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

VA Redistricting Report of Special Master (Nov. 2015)

Uploaded by

Copyright:

Available Formats

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 1 of 66 PageID# 6054

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 2 of 66 PageID# 6055

REPORT OF THE SPECIAL MASTER

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 3 of 66 PageID# 6056

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 4 of 66 PageID# 6057

Both illustrative plans fully remedy identified constitutional infirmities in the

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 5 of 66 PageID# 6058

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 6 of 66 PageID# 6059

Full information on each of these two illustrative plans is attached as an Appendix

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 7 of 66 PageID# 6060

FINAL REPORT OF THE SPECIAL MASTER

1. Pursuant to my responsibilities as a special master in Personhuballah v. Alcorn

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 8 of 66 PageID# 6061

(d) obtained (pursuant to an Order of the Court) technical assistance in map

(e) created a large number of illustrative congressional maps in an iterative fashion

2. There are a number of different criteria that can be used to evaluate a

(a) conformity to a standard of one person, one vote;

(b) avoiding either fragmentation or packing of geographically concentrated

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 9 of 66 PageID# 6062

diluting the voting strength of constitutionally protected minorities, and/or lead to

(c) avoiding use of race as a predominant criterion for redistricting;

See below for further elaboration.

In Reports prepared by the State of Virginias Division of Legislative Services,

See below for further elaboration.

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 10 of 66 PageID# 6063

3. In situation such as that applying in Personhuballah v. Alcorn, where a court is

4. Considerations that enter into line-drawing:

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 11 of 66 PageID# 6064

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 12 of 66 PageID# 6065

(b1) In seeking to reach professional judgments as a political scientist in

(b2) Neither of these areas of concentrated minority population concentration

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 13 of 66 PageID# 6066

(b3a) As shown in the plan submitted by Mr. Rapoport, it is possible to draw an

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 14 of 66 PageID# 6067

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 15 of 66 PageID# 6068

(c1A) In drawing illustrative maps for a federal court to remedy constitutional

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 16 of 66 PageID# 6069

wish the Court to consider seriously arises virtually automatically. In particular, in

(c2) It is my professional judgment that there are alternative ways to reconfigure

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 17 of 66 PageID# 6070

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 18 of 66 PageID# 6071

average population size of the type of districts that need to be drawn.

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 19 of 66 PageID# 6072

(g1A) In the context of a plan that is being drawn to remedy a constitutional

See, e.g., Upham v. Seamon, 456 U.S. 37, 43 (1982).

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 20 of 66 PageID# 6073

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 21 of 66 PageID# 6074

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 22 of 66 PageID# 6075

(g3B) Of course, when the configuration of CD3 is so substantially changed, with

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 23 of 66 PageID# 6076

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 24 of 66 PageID# 6077

(g5) A fifth approach to least change was offered by Defendant-Intervenors in

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 25 of 66 PageID# 6078

a constitutional plan. As indicated above, I have addressed the incumbency issue

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 26 of 66 PageID# 6079

absolutely no evidence presented that it would achieve political fairness between

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 27 of 66 PageID# 6080

(b) The plans introduced by Defendant-Intervenors, which make only minimal

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 28 of 66 PageID# 6081

Case 3:13-cv-00678-REP-LO-AD Document 272 Filed 11/16/15 Page 29 of 66 PageID# 6082