Substantial Political Consequences,
A Practitioner's Perspective on
Redistricting
Clark Bensen
POLIDATA® Political Data Analysis
The very essence of districting is to
produce a different - a more "politically fair" - result than would be
reached with elections at large, in which the winning party would take
100% of the legislative seats. Politics and political considerations are
inseparable from districting and apportionment. The political profile of
a State, its party registration, and voting records are available
precinct by precinct, ward by ward. These subdivisions may not be
identical with census tracts, but, when overlaid on a census map, it
requires no special genius to recognize the political consequences of
drawing a district line along one street rather than another. It is not
only obvious, but absolutely unavoidable, that the location and shape of
districts may well determine the political complexion of the area.
District lines are rarely neutral phenomena. They can well determine
what district will be predominantly Democratic or predominantly
Republican, or make a close race likely. Redistricting may pit
incumbents against one another or make very difficult the election of
the most experienced legislator. The reality is that districting
inevitably has and is intended to have substantial political consequences.1 (emphasis added.)
Justice White summarizes both the reality of drawing district boundaries as well as the limited perspective of many reviewers of the redistricting phase of the apportionment process in the epigraph.
First, the reality is that partisan politics
are an inherent part of the line-drawing exercise for congressional or
state legislative districts. The goal for most partisan entities will be
to at least maintain their status quo, if not to increase their control,
demographics sometimes notwithstanding.
Second, many observers (i.e., those who
review a plan after it has been drawn) view the process as if it were a
two-dimensional jigsaw puzzle in which all defined pieces can easily be
placed in their appropriate location or merely swapped with another.
Practitioners (i.e., those who actually participate in the drawing of
the plan) view the process as a multi-dimensional puzzle more akin to
Rubik's Cube. However, unlike that popular puzzle, there is generally no
easy solution as the interplay of all stakeholders sometimes defies
logical concepts.
The Partisan
Goals of Redistricting
From the perspective of a partisan, the aim
of redistricting is generally to minimize the variation in the potential
partisan control of the delegation or legislative chamber. Partisan
drafters prefer to categorize potential districts into three groups:
those that are clearly Republican, those that are clearly Democrat, and
those that are not clearly in either of the other two groups and in a
general sense are termed as "competitive". This is a perspective quite
different from that of some special interest groups, many of which favor
minimizing, or eliminating, the partisan influence on the line drawing.
As Justice White observed, "The reality is
that districting inevitably has and is intended to have substantial
political consequences." By classifying districts into the three groups,
the partisan hopes to maximize limited resources and focus its campaign
efforts on selected districts. This would, generally, mean expending
efforts on the "competitive" districts as well as a few selected weak
incumbents who are vulnerable or for whom special circumstances have
become evident.
This classification of districts also allows
for some continuity of the caucuses for both parties in the chamber or
delegation. American legislative bodies, especially the U.S. House, are
organizations with infrastructures based upon seniority and experience.
Members who are likely to have easy reelection efforts are more likely
to concentrate on their career as an actor in the process of governing.
This is especially the case in some smaller states with respect to their
congressional delegation. In such situations, it may be the state
government that wants to maintain some continuity in the delegation so
that the state may still have access to federal largesse and have better
influence on legislation.
Nonpartisan
Commissions or Competitive Districts
The most recent move to minimize the
influence of partisan politics in redistricting runs counter to the
goals of partisan drafters. These movements can either be characterized
as being motivated by the well-meaning so-called "good-government"
folks, true independents, or frustrated politicians.
The "good-government" groups are likely to
acknowledge that partisan politics is a legitimate consideration, but
that it should not predominate. The independents are likely to believe
that partisan politics are inappropriate to the process. The frustrated
politicians are likely to endorse the concept of nonpartisan
redistricting only after they realize they are going to be locked-out of
control for the ensuing decade. Regardless of their motivation, these
groups seem to prefer lines that are either drawn in a nonpartisan
manner and/or plans that create "competitive" districts.
Nevertheless, an initial consideration for
any districting reform must redound to an understanding of the basis of
representation. There are several schools of thought on this issue but
the basic issues here are whether districts should distribute the
partisan vote in a manner more "fair" than would occur in at-large
elections or whether districts are intended to represent communities of
interest but happen to use partisan elections to reach that end, or a
bit of both. For example, the rationale for "competitive" districts may
be different in a state in which the vast majority of voters cast their
ballots for one party, versus a state in which the balance of power is
close.
As to drawing competitive districts, the
devil is in the details. What constitutes such an animal? How many
districts should be denominated as "competitive"? What information is to
be used to assess the degree of competition? Which elections, years,
offices are chosen? Is party registration even available for the state?
Is party registration relevant? How close must the district be? How
consistent must the assessment of competition be? How predictable are
the election analyses? Are the analyses based upon previous
district-based offices? Are the analyses based upon a set of objective
statewide offices? Is a complete turnover of the congressional
delegation an acceptable impact?
As to drawing nonpartisan plans, there is
likely to be strong concurrence amongst partisans that plans drawn by
the so-called "nonpartisans" are not very likely to be nonpartisan
districts. The reason for this is that, unfortunately, many
"nonpartisans" would not recognize a partisan districting plan. Just
putting blinders on to ignore partisan information in a database is not
sufficient. Nor is it sufficient to have a nonpartisan procedure with
only "independent" citizens. If, in fact, nonpartisan plans need to be
drawn, I believe the best way to create these plans is to use partisan
line-drafters with tight restrictions.
What type of restrictions could be used to
create plans in which partisan politics doesn't predominate? One
starting point would be to predetermine what the general partisan split
for the state has been for a period of recent elections and then produce
alternative plans that would generally reflect this statewide voting
pattern. Of course, one particular to be addressed in this situation is
what is to be reflected, the votes cast or the seats won. Votes cast in
single-member winner-take-all districts do not share a one-to-one
relationship with seats won. In fact, the party that gets a majority of
the votes cast2 normally receives a bonus in seats won. This
situation is a byproduct of what academics call the "seats-votes" curve.
A concept related to such a discussion is the
notion of partisan symmetry. This is reviewed for each plan through an
analysis of election results for recent statewide, non-districted,
partisan offices. The goal would be to create a plan that was neutral in
its impact vis-ˆ-vis the parties. That is, a plan would be considered to
be "fair' if either party could win the same percentage of the seats
with the same percentage of the votes. For example, a party winning 55%
of the vote may win 62% of the seats. If the other party could also win
62% of the seats when it garnered 55% of the vote, each party would have
a fair chance to control and the plan would be neutral in its partisan
advantage.
Limitations
on Partisan Goals
Redistricting is not a random process. Nearly
every unit of geography is assigned to a district for a discrete reason
or reasons. If it is not obvious why a unit was assigned to a district,
the normal assumption would be that the political perspective of the
line-drafter benefited from the assignment. This may, or may not be, a
valid assumption. There are several restraints on the ability of the
partisan line-drafter to implement the partisan goals.
These areas of concern include:
1. partisan interests: the overall goal for the party agenda;
2. member interests: the preference of incumbents or candidates;
3. legal requirements: constitutional or statutory restrictions;
4. demographics: limitations based upon the shifting of
population;
5. traditional redistricting principles: compactness, contiguity,
preservation of political subdivisions;
6. communities of interest: these may vary state to state;
7. voter interests: concerns voiced by the public;
8. special interests: concerns voiced by organized groups.
Tools of the
Trade
There are several "tools" that line drafters
use to accomplish their ends. These include some very obvious techniques
as well as some less obvious ones that may have non-obvious effects.
Some of the general tools are:
1.
Pairing: assigning more than one incumbent to a district
that can only elect one member. There is some language in court opinions that
this is acceptable so long as it doesn't discriminate against one party.3
This can also be a very effective tool as, regardless of the outcome of the
primary and/or general election, at least one member definitely loses.
2.
Packing: placing as many as possible of the incumbent's
supporters in one district so as to waste the votes. In a broader sense this
means placing as many of the opposing party's supporters in as few districts
as possible creating supermajorities.
3.
Cracking: fragmenting an area of the incumbent's supporters
into two or more districts so as to dilute votes. In a broader sense this
means placing as many of the opposing party's supporters in as many districts
as possible creating permanent minorities.
4.
Dislocation: shifting the core of an incumbent's current
district from one area to another. This may be a small change or a wholesale
change in the makeup of a district. This may also occur merely by carving
the incumbent's residence from the old district, yet leaving the old district
relatively intact.
5.
Overpopulating/underpopulating districts: this can be
quite an effective tool to the drafter as it can unobtrusively transfer a
full seat or more from one party column to another in legislative districting.
This was the technique that caused the Georgia Democrats trouble in their
districting following the 2000 census.4 In that case the most overpopulated
districts favored Republicans and/or were districts that had already experienced
large population growth over the decade. Conversely, the most underpopulated
districts favored Democrats and/or were areas of low population growth. Due
to the zero tolerance for population deviation in congressional plans, its
main impact would be on legislative districting. However, just as in dislocation,
the nature of the population base can still cause some population disparity.
For example, there are huge disparities between the percentage of eligible
voters amongst some population subgroups. Combining some subgroups together
into even a congressional district may significantly affect the number of
voters in a district. Census persons are not necessarily voters. In fact,
based upon the number of total votes cast for president in the 2000 election
(in states with at least 2 districts) the range was from a high of 366,006
in MN-6 to a low of 76,762 in CA-33.
6.
Subdivision splits: in some cases the increase in the
number of units, i.e., counties, or cities, may cause increased campaign effort
for a member. Similarly, drawing a very non-compact, or a barely contiguous
district could cause similar problems. On the other hand, this technique could
also be used to strengthen a district that can't otherwise have a traditional
appearance. A variation on this technique is to use bodies of water for "contiguity".
Plan
Assessment
Even before a plan is subject to judicial
review, the likely impacts of the plan are reviewed by the partisan
drafters and others.
Of course, the first thing for the partisan
to review is the overall party control. This is accomplished by a review
of the previous elections aggregated for the planned districts. It will
generate a range of likely events, based upon what elections are used to
make the assessment. Some drafters use a variety of election averages
for this purpose. For example, a) several collections of races that were
close races statewide; b) some races that were near the top, or near the
bottom, of the winning scale; c) a well-known race, such as a recent
presidential or gubernatorial race; one or more races that were
so-called "base races" which are more likely to reflect generic partisan
strength and not candidate factors.
Incumbents will want to assess their
likelihood of reelection and they will focus on the degree to which the
core of the previous district follows through to the new one. This can
be viewed from two perspectives, how much of the old district follows
the member to the new one as well as how much of the new one is from the
old district. This can be especially a problem in a state that has the
number of congressional members changed due to the congressional
apportionment.
Of increasing importance in many states is a
review of the districts designed to elect members of minority groups,
i.e., generally Blacks/African-Americans or Hispanics/Latinos. In some
states the same review applies to American Indians, other Native
Americans, and other population groups for legislative districting. The
federal Voting Rights Act of 1965 has two basic sections: Section 2,
which covers every state, and Section 5, which covers at least a portion
of 16 states and requires that any changes in election practices must be
pre-cleared by the Department of Justice5 before they can be
enforced.
The key to plan assessment is that the
interests of all stakeholders must be reviewed, and the sooner the
better. If an accommodation can be reached during the process, it is
advisable to consider it then. Defending it later, during legislative
debate (should there be any) or in court could prove very problematic.
Plan Review
In the end, all sorts of persons and groups
will review a plan. Almost every one of these reviewers will find some
fault with the plan. Some may want a modest change, some may want to
start all over again, and some may have a ready alternative plan.
A problem with many alternatives is that they
are not complete plans but only focus on one region of the state. This
goes back to a limitation of resources for the alternative proponent.
First, it is considerably more difficult to draft a plan for the entire
state than a region. Second, many alternative regional plans simply
ignore the interaction with other parts of the state. It may appear that
a swap could be made at the margins of the regional alternative.
However, given the interplay of interests and the tight population
restrictions6 under which all plans are drawn, this is often
a very difficult task. Perhaps this is one reason why last-minute floor
amendments can cause undue grief in the inevitable litigation.
Given the inevitability of litigation and the
concomitant review by persons from different perspectives, line drafters
would be well-advised to bring at least some experts from several
disciplines into the process as early as possible. Of course, resource
limitations can impact the use of expertise. At the bare minimum, legal
counsel with expertise in redistricting needs to be retained before any
serious discussions are held. However, having some experts from the
social science disciplines reviewing plans may assist in both creating a
better plan as well as a better litigation posture. These experts do not
need to be the same ones that will be called at trial.
Regardless, the review of a plan in any
litigation is often comprehensive and complex. Drafters are frequently
called upon as witnesses to explain the instructions they were given by
their partisan leaders and to help explain the trade-offs that were made
to accommodate competing interests. For this reason alone, it is helpful
to have a team of drafters, one who can actually manipulate the GIS7
software package and handle the technical details and one who can work
with the stakeholders to fashion a plan to address multiple interests
and work out compromises.
Preparation for the trial may take months.
Any redistricting trial is likely to be at least a full week of very
intense round-the-clock activity. Then there is, usually, the long
pregnant pause before the decision of the court is announced. Then, in
most cases, the appeals begin. This can drag on for months surely and in
many situations for years.
Preparation
The redistricting phase of the apportionment
process, though intense and very demanding on all resources, can be a
very intellectually stimulating and interesting project. Because of this
intensity and its opportunity to end or create political careers, it can
produce strong friendships as well as strong enmities.
The key to survival of the process is,
however, preparation. Preparation covers the resources in all areas of
concern, but especially in the political and data areas. Many members
elected in the next few years will be the leaders for the 2011-2012
process. Many partisan entities will begin the first phases of their
2010 operations following the 2004 elections.
Election data for the 2004 election cycle
will be the first critical element of any 2010 redistricting
database. However, these data must begin to be processed in 2005, not
2009. Matching precinct-level election results to the geography for the
2010 census is a long process that can not be completed for several
years. However, precinct geography for the current elections must be
collected contemporaneous with the election or one risks having it gone
forever.
To have a good chance of success during the
redistricting process, redistricting must be viewed as a decade-long
project that has varying levels of intensity and resource allocation.
Notes
1. Gaffney v. Cummings, 412 U.S.
735 (1973).
2. This raises the problem of which votes are
used to determine the "votes cast". Using the district-based result for
the chamber or delegation is likely to exacerbate discrepancies caused
by districts with uncontested or weak contests. A way to minimize this
problem is to use statewide results aggregated for the districts. This
is a standard means used by practitioners for assessing districts.
3. See Larios
v. Cox, N.D., GA,
2004, No. 1:03-CV-693-CAP, the three-judge court opinion.
4. See Cox v. Larios, 542 U.S.
____ (2004), affirming the order of a three-judge panel below with only
a concurrence by Justice Stevens and a dissent by Justice Scalia.
5. The VRA
(42 U.S.C. 1973c) covers nine states entirely (AL, AK, AZ, GA, LA, MS,
SC, TX and VA). In seven additional states only selected jurisdictions
are covered (CA, FL, MI, NH, NY, NC, and SD).
6. Generally, the population deviation for
congressional districts should be as near to a zero person deviation as
practicable. For legislative districts there is a bit more leeway. Given
a good-faith nondiscriminatory effort, population differences may be allowed to
a range of 10%, e.g., from a minus 3% to a plus 7%, to reflect rational
and legitimate state interests, e.g., maintaining political subdivision
boundaries. The difference between the two standards is that the
congressional standard is based upon Article 1, sec. 2 of the U.S.
Constitution but the legislative standard is based upon the Fourteenth
Amendment.
7. Geographic
Information Systems (GIS) are the computer software that allow the
integration of geography and demographics from the Bureau of the Census
with state-specific information collected by stakeholders, e.g.,
political election results allocated to each census block. Most GIS
systems are implemented on personal computers and were available for a
few thousand dollars as of 2001. The most widely used systems were
Maptitude and AutoBound.
Clark H.
Bensen (J.D., Vermont Law School) is a consulting data analyst and attorney.
He is the founder of POLIDATA® Political Data Analysis, a demographic
and political research firm located outside Washington, D.C. He is also
the publisher of POLIDATA® Demographic and Political Guides. His email
address is clark@polidata.org.
Table of Contents | | Editor's Introduction | | Special Orders
| News | | Announcements | | Other Issues of Extensions
| HOME | | Contact Us |
Teaching & Research | | Public Outreach | | Congressional Archives | | Graduate Fellowship
Copyright, The Carl Albert Center