Do Redistricting Principles and Practices Affect State Legislative Electoral Competition? Richard Forgette Andrew Garner John Winkle University of Mississippi Abstract Critics of recent congressional and state legislative redistricting have argued that gerrymandering severely undermines electoral competitiveness to the point of violating constitutional equal protection standards. In this paper, we evaluate whether redistricting principles and processes have any measurable consequence on state legislative electoral competition. Besides their substantive importance, state legislative elections also provide greater variance than congressional data for empirically assessing theoretical propositions of redistricting structures. We find that electoral competitiveness in state legislative races declined throughout the 1990s even after term limit reforms were implemented. The proportion of uncontested state legislative seats has doubled since the 1970s and there has also been a slight increase in margin of victory. Our results show that political principles and some traditional, “politically neutral” redistricting principles sig- nificantly decrease the probability of uncontested state legislative elections. We conclude with a discussion of how our findings relate to the redistricting reform debate.
36
Embed
Do Redistricting Principles and Practices Affect US State Legislative Electoral Competition?
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Do Redistricting Principles and Practices AffectState Legislative Electoral Competition?
Richard Forgette
Andrew Garner
John Winkle
University of Mississippi
Abstract
Critics of recent congressional and state legislative redistricting have arguedthat gerrymandering severely undermines electoral competitiveness to the pointof violating constitutional equal protection standards. In this paper, we evaluatewhether redistricting principles and processes have any measurable consequenceon state legislative electoral competition. Besides their substantive importance,state legislative elections also provide greater variance than congressional datafor empirically assessing theoretical propositions of redistricting structures. Wefind that electoral competitiveness in state legislative races declined throughoutthe 1990s even after term limit reforms were implemented. The proportion ofuncontested state legislative seats has doubled since the 1970s and there hasalso been a slight increase in margin of victory. Our results show that politicalprinciples and some traditional, “politically neutral” redistricting principles sig-nificantly decrease the probability of uncontested state legislative elections. Weconclude with a discussion of how our findings relate to the redistricting reformdebate.
1 Introduction
The parties’ electoral parity and the rise of elite partisanship both at the federal and
state levels have heightened sensitivities to political gerrymandering during recent re-
districting cycles. Gerrymandering - drawing legislative lines to advantage a group
or individual - is often viewed as a pervasive and inescapable reality of the tradi-
tionally politicized American redistricting process. This paper examines the effects
of redistricting structures on state legislative electoral competitiveness during the last
two redistricting cycles. Particularly, do statutorily-defined redistricting principles and
processes affect the level of electoral competition in state legislative races? Our anal-
ysis is couched within the apparent gap between social science and legal accounts of
political redistricting.
Social science studies of political redistricting generally present a complex picture
underscoring the muted, variable effects and inherent trade-offs of redistricting princi-
ples and processes (Butler and Cain 1991; Morrill 1981). Early and seminal empirical
studies on redistricting effects discounted them as a major factor on partisan seat
balance and incumbency advantage (Erikson 1972; Ferejohn 1977; Abramowitz 1983;
Born 1985). Rush (2000), among others, concludes that the electoral consequences
of redistricting are ultimately unpredictable for mapmakers given the independence of
American voters. In short, social science accounts generally present a more nuanced
picture of redistricting effects in which intentions often fall short of reality.
In contrast, popular and legal accounts of redistricting - particularly, accounts of
the 2000 cycle - generally have emphasized the partisan and incumbency excesses of
redistricting and more often advocated political reform and stricter judicial enforcement
(Garrow 2002; Toobin 2003; Hirsch 2003) In the wake of recent redistricting, these
popular and legal critics argue that the practice of gerrymandering has progressed
to a point that it now results in severe electoral biases (Issacharoff 2002; Garrow
1
2002; Issacharoff and Pildes 1998). Additionally, they claim that “excessive” partisan
gerrymandering violates the equal protection clause as it consigns electoral majorities
to minority status based on their political views.1 Political gerrymandering claims
should be closely scrutinized by the Courts when electoral districts are drawn by a
nonpartisan body, as occurs in some states.
In this paper, we empirically investigate a legal and popular criticism of contem-
porary redistricting. Specifically, do districting principles specified by a state either
statutorily or constitutionally affect change in legislative electoral marginality and
challenger-entry? Do political principles - such as prohibitions of incumbency protec-
tion or requirements for electorally competitive districts - affect electoral marginality
or candidate entry? We also assess the impact of districting processes on incumbency
protection. Do states with politicized districting processes - ones in which the state
legislature is legally charged with drafting and affirming the districting plan - draw
less competitive districts than states where elected officials are less determinative of
districting plans?
The paper is organized as follows. In the next section, we present a theoretical
explanation for the effect of statutorily-defined, redistricting principles and processes
on state legislative competition. We also report how these redistricting principles vary
across states and present figures showing the variation over time of electoral competi-
tion for state legislative races. The analysis section presents empirical results examin-
ing the effects of districting principles on two measures of electoral competitiveness-the
probability that an election will be uncontested and the margin of victory for contested
races. Finally, we conclude with a discussion of whether reforms in redistricting prin-
ciples or processes enhance political competition in state legislative elections.
2
2 Districting Principles and Electoral Competition
Political scientists have long recognized that legislatures do not have carte blanche con-
trol over redistricting. Cain, MacDonald, and McDonald (2005) describe the current
growth of redistricting process reform and related litigation at the state-level. Cox
(2004) classifies these state redistricting reforms as institution-selecting regulations,
process-based regulations, and outcome-based regulations. Institution-selecting regu-
lations relate to the rules establishing who proposes and approves redistricting plans -
legislatures or commissions. Process-based regulations refer to the formal redistricting
principles and rules directing those mapmakers. Outcome-based regulations, finally,
relate to any explicit standard for evaluating redistricting outcomes - such as minority
vote dilution or partisan bias.
Our interest focuses on process-based (principles) and institution-selecting regula-
tions. Nearly all states formally declare traditional “politically neutral” redistricting
principles in their redistricting laws or constitutions. These principles are intended to
guide state mapmakers and inhibit their ability to manipulate district lines (Barabas
and Jerit 2004; Forgette and Platt 2005). There is significant variation between states’
redistricting principles and within states regarding which principles govern state legisla-
tive and congressional districting plans. Table One shows which states require specific
principles under their state redistricting law. The table demonstrates the substantial
variation across states and regions.2 With the exception of the “cores of prior districts”
principle, states are more likely to adopt formal principles governing redistricting for
their own legislative mapmaking than for congressional mapmaking.
Redistricting principles may be classified into three types: form-based, population-
based, and politically-based criteria. Form-based redistricting criteria - compactness
and contiguity - relate to the prescribed physical shapes of legislative districts. Con-
tiguity simply refers to whether all parts of the district are connected; it is the most
3
widely accepted redistricting criterion across states. Compactness suggests that dis-
tricts should be composed of tightly defined areas. The presumption is that representa-
tives and constituents may find it easier to interact and communicate with each other
with a compactly shaped district. While there are multiple measures of compactness,
the Supreme Court and state supreme courts have not adopted a single one.3
Population-based principles imply some intrinsic democratic value for including
certain voting populations within a district. We include in this category principles for
respecting political subdivisions, preserving communities of interest, and maintaining
district cores. The political subdivisions principle suggests that states should follow
presumably politically neutral and pre-existing political boundaries like county and
municipal lines when drawing legislative and congressional districts. The democratic
value of overlapping legislative lines with other political jurisdictions apparently rests
in increasing voters’ awareness of their elected officials (by conforming lines with the
public’s understanding of boundaries) and clarifying intergovernmental relations be-
tween local, state, and national elected officials. A principle to preserve communities
of interest is less objectively defined than the political subdivision value. That is, states
declaring a “communities of interest” principle typically require citizen meetings held
in locations across a jurisdiction in order to determine where communities of interest
exist. These communities of interest may correspond to existing political or geographic
boundaries; however, the principle has also been used to justify lines inclusive to racial
and ethnic communities that are not necessarily compact (McDonald 2007). Finally, a
district core principle declares a value for maintaining the population center or “core”
within the existing district.
A final class of principles includes explicit political criteria inhibiting or promoting
the likelihood of electing a particular candidate. These political principles include a
value of incumbency protection - a districting criteria allowing or requiring considera-
4
tion of incumbency status. At the other extreme, some states compel incumbent-blind
redistricting, explicitly prohibiting mapmakers from considering incumbency status. In
these cases, states may disallow mapmakers from knowing the location of incumbents’
homes or may limit the use of election information (e.g., party registration data) in the
course of redistricting. For example, Iowa’s redistricting statute expressly forbids the
use of political affiliation, previous election results, the addresses of incumbents, or any
demographic information other than population in creating their redistricting plans.
A nonpartisan Legislative Services Bureau drafts alternative redistricting plans follow-
ing only principles of population equality, contiguity, respect for political subdivisions,
and compactness. Two other states - Washington and Arizona - explicitly require map-
makers to draw lines that promote electorally competitive districts while also satisfying
other stated districting principles. Because incumbency tends to decrease the competi-
tiveness of legislative elections, we expect lower levels of electoral competition in states
where incumbency protection is allowed or even required. In states prohibiting these
practices, we expect legislative races to be relatively more competitive.
How competitive are state legislative elections? Figures 1 and 2 show the trend
over the past 36 years, 1968 to 2004, and demonstrate a substantial decline in electoral
competitiveness with regard to candidate entry. Our two measures of competition
include, first, the percentage of legislative seats that were uncontested each year and,
second, the margin of victory for seats that were contested by both major parties.
As the figure shows, the percentage of uncontested seats has increased steadily from
a low of 20 percent in 1968 to a high of 39 percent in 2002. Thus, the degree to
which state legislators run unopposed in their districts has nearly doubled over the
last several decades, resulting in a dramatic decline in electoral competitiveness at the
state level. In contrast, the margin of victory for contested elections remained relatively
stable and showed only a slight increase through the 1990s. Despite this stability over
5
time, the absolute value for margin of victory remains high. On average, candidates in
state legislative elections win by approximately 25 percent. Figure 2 provides further
evidence of declining competitiveness by charting the percent of electorally marginal
state legislative elections over time. The percent of marginal seats (whether defined
as a 60-40 or a 55-45 split in the district) declined considerably since 1968, although
there was a significant rise in the number of marginal races during the late 1980’s and
early 1990’s, followed by further decline following the 1992 election.
We test whether there is an empirical link between the statutorily-defined redistrict-
ing practices defined in this section and the level of state legislative electoral competi-
tion. Political scientists generally have investigated redistricting effects when modeling
congressional partisan and incumbency gerrymandering.4 The relationship between
redistricting practices and electoral competitiveness has largely gone unexplored at
the state legislative level, however. McDonald (2007) recently notes “the paucity of
research that investigates the political effects of state criteria” (677). The theoretical
rationale for this empirical link is that formal principles lessen the range of possible re-
districting plans available to elected partisans for drawing new lines. Districting plans
in states that formally declare these principles may be more vulnerable to legal chal-
lenges. Thus, mapmakers may engage in less incumbent or partisan gerrymandering.
Compared to congressional redistricting, state legislative maps may even afford
stronger political opportunities. The greater number and small population of state
legislative districts (compared to congressional districts) may allow mapmakers greater
capacity to manipulate district lines for political ends. Additionally, legislative district-
ing permits greater legal population deviation compared to congressional mapmaking
(ten percent compared to one percent). For these reasons, we theorize that formal
principles and institution-selecting regulations promote state legislative electoral com-
petitiveness.
6
Still, there is also reason to suspect that redistricting regulations might have little
impact. Increasingly advanced and sophisticated technology could allow mapmakers to
realize political goals while accommodating legally prescribed districting practices. The
2000 redistricting cycle particularly saw a significant advancement in GIS software and
block-level data available to more mapmakers (Monmonier 2001). The intended effect
of selecting processes (e.g. independent boards) and formal principles may be muted
over time between the 1990 and 2000 redistricting cycles. In the next section, we discuss
the data and methods of our research design and present the results relating different
redistricting practices to cross-sectional variation in state electoral competition.
3 Data and Methods
We examine district-level election data for years following the implementation of each
state’s redistricting plan during the 1990 and 2000 redistricting cycles using the State
Legislative Election Returns, 1967-2003 dataset collected by the Inter-university Con-
sortium for Political and Social Research (ICPSR). The previous section presented
longitudinal trends for uncontested races and margin of victory for contested elections,
which represent our two measures of electoral competition. In the cross-sectional anal-
ysis presented below, these two measures become the dependent variables in our re-
gression models. Margin of victory is measured as the absolute difference between the
vote shares of the two major parties while uncontested races are coded 1 for elections
with only one major party candidate and 0 when both parties ran candidates.
Although the number of multimember districts has declined since the 1970’s, several
states continue to use multimember positions and free-for-all districts for legislative
elections. Multimember position districts require that candidates run for a particular
post and are therefore coded identically to single-member districts. Multimember free-
7
for-all districts, however, have multiple candidates from each of the two major parties
competing for seats in the same district and must be treated differently. Consequently,
we use the method created by Niemi, Jackman, and Winksy (1991) for measuring the
margin of victory in these districts. In addition, following Jewell and Breaux (1991),
multimember free-for-all districts are coded as contested when the total number of
candidates is greater than the total number of seats in the district and are coded as
uncontested when the number of candidates is equal to the number of seats being
contested.
Our primary interest is examining whether and to what extent state laws and redis-
tricting practices affect levels of legislative competition after controlling for district-level
factors such as incumbency and district characteristics. Because legislative districts are
clustered within particular states, we cannot assume that the errors in our regression
models are independently and identically distributed (i.i.d.). Failing to account for the
multilevel nature of the data can lead to serious estimation problems, most notably
smaller standard errors that can make state-level coefficients appear more significant
than they should be (Primo, Jacobsmeier, and Milyo 2007). We therefore use clustered
sample errors to account for the multilevel structure of the dataset.5
To examine the effect of state laws and practices on legislative competition, we
supplemented the ICPSR dataset with state-level data on the statutory and consti-
tutionally stated redistricting principles and practices used during the 1990 and 2000
redistricting cycles. Information about the redistricting principles required by states
was collected from the National Conference of State Legislatures (NCSL) publication
Redistricting Law 2000 (Table 5). Data for the 2000 redistricting cycle was collected
by examining the state constitutions and statutes to determine which states (such as
Arizona) altered their requirements between the two redistricting cycles.
Our first set of state-level factors include formal redistricting principles - compact-
8
ness as well as the protection of incumbents, cores of prior districts, communities of
interest, and existing political subdivisions - required by states when redrawing legisla-
tive district lines. 6 In previous work on congressional redistricting, we have conceptu-
alized redistricting principles based on whether they govern the physical shape of the
district (compactness and contiguity), the people who live in the district (protecting
communities of interest, existing political subdivisions, and cores of prior districts), or
expressly political purposes such as protecting incumbents (Forgette and Platt 2005).
Our argument is that there is a cumulative effect of these principles that limit the
ability of mapmakers to shift political neighborhoods across district lines for parti-
san gains or incumbency protection, especially for the population-based principles. We
therefore include an additive scale indicating the number of population-based principles
a state requires that ranges from zero for states that require none of these principles
to three for states that require all of them. One drawback to this approach is that
it does not allow us to examine which principles, if any, have the strongest effect on
state legislative competition. To more accurately pinpoint which specific principles
have the strongest effect, we also present an alternative specification where each of
the three population-based principles are included in the analysis as individual dummy
variables.7
The other redistricting principles are coded as dummy variables. Compactness is
coded one if states require that district lines be drawn in a compact manner and zero
otherwise. 8 For incumbency protection, some states expressly forbid the protection of
incumbents while other states actually mandate the practice. Prohibiting incumbency
protection is coded one if states forbid mapmakers from considering the residence of
an incumbent or to otherwise protect incumbents when redrawing lines, and is coded
zero otherwise. A separate dummy variable is included that is coded one if states
require incumbency protection and zero otherwise, leaving states that neither require
9
nor prohibit incumbency protection as the baseline.
Turning to the practices and processes used for legislative redistricting, we include
several variables that indicate which actors were responsible for the redrawing of dis-
trict lines. First, states that use a commission or redistricting board to draw district
lines are coded one while those that do not are coded zero.9 We relied on the NCSL
CITATION for data on state redistricting boards and commissions. Second, even when
the legislature or redistricting board oversees the drawing of the maps, legal challenges
sometimes result in the courts imposing their own redistricting plans on the states
(Weber 1995, 2005). States that had court-imposed plans are coded one while those
that did not have court-imposed plans are coded zero.10
Separate dummy variables were also added indicating whether there was divided
partisan control over the legislature (Weber 1995, 2005) and whether the Governor
has a veto over the legislative plan.11 Our final process-related variable measures the
degree to which a state is covered under the preclearance requirement of Section 5 of
the Voting Rights Act. States that have no preclearance requirement are coded zero.
States where only a few townships are required to be precleared by the Department
of Justice (e.g., Hawaii, Nevada, Oregon, Tennessee, and Wyoming) are coded one.
States that require whole counties to be precleared (e.g., California, Florida, North
Carolina, New York, and South Dakota) are coded two while states that must submit
their entire state redistricting plan for approval are coded three. The baseline in the
model therefore represents states with legislative-drawn plans that have unified control,
no gubernatorial veto, no preclearance under the Voting Rights Act, and whose plans
survived court challenges.
Our control variables include a lagged variable for previous margin of victory or
previous uncontested race to control for district effects, whether the state has imposed
term limits on state legislators or not, the presence of an incumbent in the race, and
10
whether the district is a senate or a house seat (coded one for senate and zero for
house). We also control for the level of professionalism of the legislature.12 Finally,
the nature of multimember free-for-all districts could foster closer competition than
single-member or post districts. We therefore add a dummy variable coded one if the
district is free-for-all and zero if it is a single-member or multimember position district.
4 The Effect of Principles and Practices on Legisla-
tive Competition
Table 2 shows the results for the two specifications of the uncontested model. Model
1 uses an additive scale to assess the cumulative effect of population-based principles
while Model 2 allows each population-based principle to enter as a separate dummy
variable. For each model, the first column lists the logit regression coefficients and
statistical significance levels for each variable, the second column shows the clustered
standard errors for these coefficients, and the third column includes the estimated effect
of a discrete change in each independent variable on the probability that the district
will be uncontested. For the dummy variables, the effect represents the difference in the
predicted probability of an uncontested seat for districts that have that characteristic
and those that do not. For the other variables, the effect is the difference in probability
between the lowest observed value of the independent variable and the highest observed
value of the independent variable.
Overall, our results show that population-based principles help make legislative
districts more competitive by reducing the probability of an uncontested race. The
additive scale is negative and statistically significant (p ¿ .03) with a strong substantive
effect. The predicted probability of an uncontested seat is about 17 percent lower for
states that adopt all three population-based principles as compared to those that do
11
not require any of these principles. It appears that the communities of interest and
political subdivisions principles have the strongest effect of the three, indicated by the
negative and significant coefficients in Model 2. Adopting either principle reduced the
chances of an uncontested race by about 9 percent. In contrast, the coefficient for cores
of prior districts in Model 2 was not significant and was actually positive.
The results for the other redistricting principles are mixed. Requiring states to draw
compact districts does not appear to have any effect on contestation. The coefficient for
compactness failed to reach traditional levels of statistical significance in either model
and the substantive effect was rather small. Likewise, the coefficient for prohibiting in-
cumbency protection did not reach traditional levels of statistical significance in either
model. However, states that require mapmakers to consider an incumbent’s residence
or to avoid placing two incumbents in the same district have considerably higher un-
contested rates compared to other states. Depending on the specification of the model,
requiring the protection of incumbents increases the probability of an uncontested race
by 10 to 16 percent. In sum, we find that certain redistricting principles have a strong,
if variable, effect on contestation rates.
In contrast, the process used by states for legislative redistricting does not appear
to have much of an effect on the likelihood of an uncontested race. Only preclearance
under the Voting Rights Act has a consistently strong effect across both models while
the coefficient for independent boards reached traditional levels of statistical signifi-
cance for Model 1, although the significance level for the latter was not overwhelming
(p > .067).13 The probability of an uncontested race was about 16 to 19 percent higher
(depending on the model) when a state must submit its entire plan for preclearance
as compared to states that are not required to submit any portion of their plan for
preclearance. The rest of the process variables failed to reach traditional levels of
statistical significance and had only modest substantive effects on uncontested races.
12
Finally, our control variables performed largely as expected. Term limits made
legislative elections more competitive by lowering the probability of an uncontested
race, although the effect was only significant in one of the models. All of the other
control variables were significant and strong across both models. Notably, multimember
free-for-all districts promoted greater competition by lowering the likelihood of an
uncontested race by about 20 percent. Senate seats and professional legislatures had
more competitive races while incumbents were about 11 percent more likely to run
uncontested than open seat races.
The results using margin of victory as the dependent variable are shown in Ta-
ble 3 and largely confirm the results described above. Population-based redistricting
principles have the strongest effect, with the results largely driven by the communi-
ties of interest and political subdivision variables. The additive scale for redistricting
principles is strongly significant, negative, and adopting all three of these principles
lowers the expected margin of victory in a district by about 15 percent. As before,
compactness does not appear to have any significant effect on legislative competition.
Prohibiting incumbency protection also does not appear to have an effect while requir-
ing incumbency protection creates less competitive districts, though the latter results
are only significant for Model 1.14
However, we do find greater support for the process variables in the margin of
victory models. Independent redistricting commissions and court-imposed plans both
lead to less competitive races, indicating by the positive and statistically significant
coefficients across both models. The margin of victory is about 10 percent higher for
states using independent redistricting commissions and about 6 to 7 percent higher
when a court imposes its own redistricting plan on the state. Preclearance under
the Voting Rights Act also led to less competitive races, which is consistent with our
findings in the uncontested model. States that are fully covered under Section 5 had
13
margins of victory that were about 12 percent higher than those that were not covered
at all. As with the uncontested models, however, divided control of the legislature
and gubernatorial veto had no effect. The control variables also performed the same as
they did in the uncontested model with the exception of the professionalism coefficient,
which did not reach traditional levels of statistical significance.
Finally, we argued above that the effect of redistricting principles could be muted
by increasingly sophisticated technology used for legislative redistricting. To test this
hypothesis, we re-estimated the models by including a interaction between the additive
population-based principle scale and a dummy variable for the redistricting cycle.15 Our
expectation is that the effect of adopting so-called neutral redistricting principles will
have a stronger effect during the 1990’s and a weaker effect during the 2000’s as more
sophisticated technology allowed mapmakers to work around the principles to accom-
plish their redistricting goals. The results of our model support this expectation. The
coefficients for the additive population-based principles scale and the require incum-
bency protection variable in the 1990’s were roughly twice the size of the coefficients
during the 2000 redistricting cycle.16 For example, adopting all three population-based
principles in the 1990’s resulted in a decline in the probability of an uncontested rate
of about 25 percent. In contrast, the probability of an uncontested election declined
by only about 14 percent in the 2000 redistricting cycle for states adopting all three
principles. Similarly, adopting all three principles resulted in a 23 percent decline in
margin of victory during the 1990’s but only lowered margin of victory by about 10
percent during the 2000 round of redistricting.
14
5 Discussion
This paper has investigated the relationship between redistricting principles and elec-
toral competitiveness across state legislative districts. We began by reporting marked
increases in the percent of uncontested races and slight rises in the margins of victory.
We then reported results from models of state legislative marginality and challenger
entry gauging the effects of the 1990 and 2000 redistricting principles and processes.
Critics reasonably wonder if the decline in electoral competitiveness grossly inhibits
the extent to which state legislative bodies function as a basis for democratic respon-
siveness. If it is, can anything be done to stem the electoral tide?
Our study aims for a more nuanced understanding of the effects of declarative
process-based redistricting principles and selecting processes on state legislative elec-
toral competition. Our findings indicate that these elements have marginal but signif-
icant effects. Aside from the primary standards set by federal courts (one person, one
vote; avoidance of retrogression of minority voting rights in covered jurisdictions under
the Voting Rights Act; avoidance of vote dilution by creation of majority-minority dis-
tricts; and, prohibition on racial gerrymandering), state lawmakers or commissions will
choose from available secondary competing principles to guide mapmaking. And those
choices underscore the inescapable political nature of redistricting. Our results show
that two population-based criteria, respect for political subdivisions and the preserva-
tion of communities of interest, lessen the margins of victory as well as the probability
of unconstested seats. The cumulative effect of these principles (as indicated by the
additive scale measure), moreover, produces an even greater consequence.
This state legislative analysis parallels recent studies in congressional redistricting.
Forgette and Platt (2005) find that redistricting principles and processes affected rates
in congressional incumbency protection during the recent redistricting cycle. In the
1992 elections, after the previous redistricting period, 84 U.S. House races were de-
15
cided by a 10 percent margin or less; in the 2002 election, though, 38 races had a
comparable margin. Correspondingly, the number of electorally safe incumbent seats
grew over the decade. In the 1992 election, 65 percent of congressional incumbents
running for reelection won with a 20 percent or greater margin; in the 2002 reelection,
though, 83 percent of incumbents won by this wide margin. These changes in con-
gressional electoral competitiveness occurred among both Republican and Democratic
incumbents.
In other ways, though, the decline in state legislative electoral competitiveness is
different from the decline in congressional races. State legislative races have strikingly
higher rates of uncontested races compared to congressional contests. The effects of
term limits and legislative professionalism also make state legislative contests different
from congressional elections. Finally, our interest in the effect of redistricting principles
presents another difference between state and congressional contests. As noted earlier,
states impose a substantially greater number of redistricting constraints for their state
legislative plans than for their congressional districting plans. Additionally, federal
courts do not require the degree of population equality for state legislative districts
compared to the exactness for congressional districts mandated under federal law.17
Two U.S. Supreme Court rulings from 1973 held (1) that states may deviate 10% from
absolute equality without the need to defend in court, and (2) that even a deviation of
17.4% may be justified .18.
Our findings parallel and extend arguments advanced more than a decade ago by
Butler and Cain (1992), namely, principles of redistricting present inherent trade-offs.
In fact, political redistricting principles - separate from and in addition to the trade-offs
between presumably neutral principles - present tradeoffs for redistricting reformers and
mapmakers. As the summary matrix in Figure 1 [check this] illustrates, political prin-
ciples may be neither wholly compatible with one another nor mutually attainable.
16
The figure arrays alternative political principles on each dimension while describing
their potential inconsistencies. Political principles include values to minimize partisan
bias (state seat share minus vote share for each party), maximize district electoral
or maximize political insulation (restricting mapmakers from examining electoral or
party registration data). It is implausible to assume, for example, that in efforts to
reconfigure legislative districts state decision-makers will be able to maximize electoral
competition and advance minority representation, or to maximize political insulation
and minimize partisan bias. In some instances, pursuing one objective clearly un-
dermines another. In others, relationships between principles become operationally
inconsistent. Unavoidable conflicts of principle occur, in other words, both in theory
and in practice.
What then does this suggest for the ongoing reform debate over legislative redis-
tricting? Is the public disenchanted with partisan redistricting maneuvers, and does
social science research provide guidance? Voters in California and Ohio in 2005 re-
jected ballot initiatives calling for another round of state redistricting within the cur-
rent decade. For the most part Democrats in California and Republicans in Ohio
led popular charges against repetitive redistricting, preferring instead the natural and
once-a-decade model. The common explanation for these actions, observers suggest,
lies in the prevailing perception by the respective electorates that current state party
leaders or prospective ones were engaging in veiled partisan power plays to enhance
legislative seat advantages (Hirsch and Mann 2005). Beyond the issue of timing, and
more importantly, each of the ballot initiatives in California and Ohio would have also
removed from the legislature the task of redrawing district lines. Instead the propos-
als would have authorized special neutral commissions, composed of retired judges in
California and citizens in Ohio, to draft redistricting plans. Voters in Ohio rejected
17
Issue 4 by a decisive 70 percent to 30 percent margin, while counterparts in California
followed suit against Proposition 77 with a solid 59 percent majority.19 Neither of these
ballot initiatives on institution-selecting regulations, however, should be regarded as
routine. Competing political interest groups generated huge contributions to influence
election outcomes. Groups in Ohio poured in more than $10.2 million dollars, while
organizations in California invested more than double that figure, an astounding sum
of $29.5 million.20 Simplifying the debate over the design of redistricting institutions
to a forced choice between legislatures or independent commissions, however, may too
narrowly frame the matter and thereby disserve democratic interests (Cox 2006). Re-
gardless of the designated mapmaker, however, the vexing choices of standards still
remain (Murphy 2005).
Observers in recent years have looked to the U.S. Supreme Court for authoritative
guidance in the matter of partisan gerrymandering. While disposing of the disputes at
hand, two rulings, one in 2004 and another in 2006, however, failed to provide clear
judicial standards. In Vieth v. Jubilerer, 541 U.S. 267 (2004), the Court by a 5-4 vote
affirmed the lower court ruling and dismissed the claim of partisan gerrymandering
raised against the legislative plan to redraw Pennsylvania congressional districts. A
plurality of four Justices led by Antonin Scalia concluded that partisan gerrymander-
ing represented a nonjusticiable issue because no judicial standards could be fashioned
to address it. The remaining members of the Court disagreed, with one unwilling to
foreclose the possibility of discoverable criteria and the other four offering three dis-
tinct standards for adjudicating such claims. The unsatisfying ambiguity perpetuated
by Vieth anticipated a more conclusive action by the Court two years later in the con-
troversial Texas case, League of United Latin American Citizens v. Perry, 548 U.S.
399 (2006). The Court, 5-4, rejected the broad challenge to a congressional redistrict-
ing plan but the analysis advanced by Anthony Kennedy did not itself command a
18
majority. Among the fractured set of opinions, the only ruling that commanded a
majority of Justices required the redrawing of majority-minority districts for Latinos
in the southwest portion of the state. The issue of standards for judicial determination
of impermissible partisanship, however, remained unresolved, for now.
Our analysis indicates that states’ statutory restrictions can promote electoral com-
petition. However, we do not argue or indirectly imply that redistricting is the only
or even the primary cause of declining electoral competitiveness. There are multiple
factors affecting the decline in state legislative electoral competitiveness, including a
complex interaction of geography and politics. Persisting, and in many cases growing,
racial and economic stratification within metropolitan areas has resulted indirectly in a
geographic sorting along party lines (Orfield 2002). The electoral consequences of this
demographic, partisan stratification have been strengthened with more sophisticated
legislative mapmaking tools. Geographic information systems (GIS) allow political
cartographers to anticipate the partisan implications of ever-smaller unit changes in
district boundaries (Monmonier 2001). A political redistricting process limited by
few if any traditional districting principles only heightens the effects of these social
and technological forces. In the end, the demographic and social forces contribut-
ing to declining legislative and congressional electoral competitiveness may be largely
long-term and uncontrollable. Redistricting may only intensify rather than generate
partisan stratification. Legislative electoral competitiveness can be promoted but not
mandated.
Furthermore, even if it could be mandated, our results do not address the connec-
tion between electoral and institutional partisanship. Pundits and political scientists
have often expressed schizophrenic perspectives on the role electoral competitiveness
plays in legislative representation. For instance, while lower competitiveness may sug-
gest a decline of electoral accountability, political scientists have often emphasized the
19
excesses of electoral vote-seeking (Mayhew 1974; Fiorina 1989). Additionally, Brunell
(2005) finds that voters are more likely to approve of legislative performance when they
voted for the winning candidate. In other words, more competitive districts may com-
pel incumbents to engage in more electioneering (and less governing) while lowering
public approval of legislative performance. Furthermore, we wonder whether declining
electoral competitiveness corresponds with stronger public policy preference agreement
within district constituencies. If so, does this add to levels of substantive representa-
tion or policy congruence between elected officials and their constituents? In short,
the effects of declining legislative and electoral competitiveness are multifaceted and
unresolved.
20
Figure 1: Longitudinal Trends in Uncontested Rates and Margin of Victory, 1968-2002
21
Figure 2: Longitudinal Trends in Percent of Marginal Seats, 1968-2002
22
Compact Cores/Prior Comm.’s of Political ProtectDistricts Interest Subdiv.’s Incumbents
Required Required Required Required Cannot ProtectAK NV AL LA AK NE AK ND DE MTAL NJ AR MD AL NV AL NE HI NEAZ NM GA SC CO OK AR NH IA ORCO NY HI OR CA NJ ID UTHI ND Not Required ID SC CO NM KS WYID OH AK ND KS UT CT NV NeitherIL OK AZ NE MD VA GA NY AK NCIA PA CA NH MN VT HI OH AL NDKS RI CO NJ MT WY IA OR AZ NHLA SC CT NM ID PA CA NJME SD DE NV Not Required KS SC CO NMMD UT FL NY AR MS KY SD CT NVMI VT HI OH AZ NC LA TN FL NY
MN VA IA OK CA ND MA TX IL OHMS WA ID OR CT NH MD UT IN OKMO WV IL PA DE NJ ME VA KY PAMT WI IN RI FL MN MI VT LA RINE WY KS SD GA NY MN WA MA SD
KY TN IA OH MS WI ME TNNot Required MA TX IL PA MT WV MI TXAR KY ME UT IN RI NC WY MN WACA MA MI VA KY SD MO WICT NC MN VT LA TN Not Required MS WVDE NH MO WA MA TX AZ IN Must ProtectFL OR MS WI ME WA DE MO AR SCGA TN MT WV MI WI FL OK GA VAIN TX NC WY MO WV IL RI MD VT
Source: National Conference of State Legislatures Reapportionment Task Force (1989)
Table 1: Redistricting Principles Employed for State Legislative Redistricting: 1990’s
23
Model 1 Model 2Coefficient SE Effect Coefficient SE Effect