Redistricting Principles and Racial Representation

Jason Barabas, Harvard University and Southern IllinoisUniversityJennifer Jerit, Southertj Illinois University

ABSTRACT

How do traditional redistricting principles—contiguity, communities of interest,political subdivisions, incumbent protection. Section 5 of the Voting Rights Act,preservation of district core, and compactness—affect racial minority representa-tion in congressional districts? Using data from the 2001-02 redistricting process,we Hnd that compactness is the only principle that significantly affects minorityrepresentation, both in terms of majority-minority districts and minority influencedistricts, but these effects are contingent on the size of the minority community andextent of racial segregation in a state. Two other principles. Section 5 pre-clearanceand protecting political subdivisions, improve minority representation in a morelimited way. Thus, race-neutral redistricting criteria like the compactness principle,can dramatically aftect the racial composition of the resulting districts and, thereby,affect minority representation.

THE PRACTICE OF CREATING majority minority electoral districts in recentdecades has led to a fierce debate over whether this policy enhances the politi-cal influence of racial minorities. Some scholars have found that packingAfrican Americans into a small number of legislative districts increases theconservatism of surrounding areas (Bullock 1995; Cameron, Epstein, andO'Halloran 1996; Lublin 1997;Swain 1995).Asaresult, some argue that Afri-can Americans would benefit more from the creation of "minority influencedistricts" (Lublin 1997, 121), where blacks constitute the swing vote that iscrucial to a candidate's victory. Other scholars either downplay the unintendedconsequences of majority-minority districts (Engstrom 1995; Grofman andHandley 1998; Petrocik and Desposato 1998) or champion the descriptiverepresentation that typically results from them (Davidson 1992).

Obscured in this debate is the fact that those who draw political districtsare constrained by a number of rules and principles that govern the redistrict-

State Politics and Policy Quarterly, Vol. 4, No. 4 (Winter 2004}: pp. 415-135

4l6 BARABAS AND JEKIT

ing process in their respective states. Indeed, recent court decisions {Shaw v.Reno 1993; Miller v. fohnson 1995) have emphasized traditional districtingstandards, such as compactness, contiguity, and the preservation of commu-nities of interest, while questioning the constitutionality of using race as thepredominant standard in drawing districts. Even as tlie United States SupremeCourt's position on racial redistricting continues to evolve,' traditional stan-dards are respected, if only because they permit an ostensibly objective—andpresumably race-neutral—assessment of district boundaries {Engstrotii andAlford 2002; Pildes and Niemi 1993). For example, McDonald (1996) arguesthat compactness and contiguity standards are a valuable safeguard againstpolitical shenanigans in a system where the district drawers are governmentofficials (also see Polsby and Popper 1991; Stern 1974; Wells 1982).

But are these traditional districting standards race neutral? Using com-puter simulations, Altman (1998a) shows that the racial impact of the com-pactness standard depends on the distribution of minority voters, whereracial minorities become a majority of a district's voters only if they areboth numerous and concentrated (see also Lowenstein and Steinberg 1985).When minorities are not .sufficiently concentrated, their numbers in anygiven district will be low under a compactness standard. Thus, the effect ofcompactne.ss and other districting standards on the political influence ofminorities is an empirical question, the answer to which likely depends onthe size of the minority community, patterns of racial segregation, and thecombination of redistricting requirements a state adopts.

We investigate this question using demographic, political, and institu-tional data for United States congressional districts after the 2001-02 redis-tricting cycle. We find support for Altman's (1998a) assertion that populationgeography matters. Compactness rules are associated with reduced minorityinfluence, except when they occur in combination with a large and .segregatedminority population, in which case they are associated with greater influence.More generally, we find that traditional redistricting principles have varyingimpacts on minority representation. Some, such as provisions regardingcontiguity and conserving the district core, matter little. Others, such asSection 5 of the Voting Rights Act and rules regarding political subdivisions,enhance minority representation, but in a nuanced way.

POLITICAL REDISTRICTING IN THE UNITED STATES

Because responsibility for electoral redistricting has been left to the states(Butler and Cain 1992), the precise mix of principles guiding the processvaries widely in the United States. Observers of congressional redistricting

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 417

in the states have found seven principles to be particularly important in atleast several states (Wattson 1999).-

• The principle of contiguity holds that a district be a single geographicpiece. A district is considered contiguous if all of its parts touch oneanother at more than one point.

• Redistricters may strive to preserve communities of interest in districts,that is, groups of people united by common social, political, economic,or ethnic characteristics.

• District lines may be drawn to adhere to existing political suhdivisiom,such as city or county boundaries or other local, state, and federal elec-tion districts.

• Incumbent protection refers to the practice of drawing districts so thatmost incumbents are relatively safe in them. Some states explicitlycondone the practice; others prohibit it.

• Section 5 of the Voting Rights Act of 1965 requires that governmentofficials in certain areas of the country ("covered jurisdictions") obtainpre-clearance from the United States Attorney General or the Unit-ed States District Court for the District of Columbia for any policychanges regarding voting, including redistricting plans (http://www.usdoj.gov/crt/votiiig/sec_5/about.htm). The lustice Department or thecourt reviews tliese plans to prevent the dilution of minority votingstrength/

• Redistricters may strive to preserve the district core, drawing new dis-tricts based on the old districts to the extent possible. This practice isthought to maintain voter continuity and minimize voter confusion.

• Compactness refers to the extent to which a district's geography is dis-persed around its center. A district with branches that extend far fromits center may be the result of racial or partisan gerrymandering.

Table I lists the principles that govern congressional redistricting in the43 states that underwent the process after the 2000 United States Census.'^There is considerable variation among the states in their adoption of theseguidelines. No state is governed officially by all seven redistricting standards,although three (Alabama, Arizona, and South Carolina) are governed by six.The overwhelming majority of states that redistricted in 2001-02 adheredto at least one of these seven standards (32), although a sizeable number(11) did not have any formal guidelines. No obvious single characteristicis shared by the states in this latter category, as it includes states both big(Pennsylvania) and small (Connecticut), states with regular (Colorado) andirregular (Maryland) boundaries, and states in all four major regions of the

I

I.I

cDu

•B

sb£>a.

aoIS;

>

X i3

o oU

X X XX X X X X X

X XX

X

X XX

XX X X

XX X

X XX X X XXXX XX X

X X XX X X X X

X XX XXX XXXXX XXXXX

Q uj — O 2 Ul u

c Z3 W

ECo oU

Iou

X X X XXXX

X XXX

X

X X

X

X

X

X

X

XX XX X XX X X

XXX X X X

XXXX X X XXXX

>. X ̂ CC < _̂2 OOO £2

420 DARABAS AND JERIT

country. Scholars have not attempted to ex|ilain the patterns of adoption ofthese standards, but it is an area that is ripe for study.

The state-level variation in Table I provides an ideal opportunity toaddress the question, What impact do these redistricting principles haveon minority representation? For example, with the exception of the slightunder representation of the northeast, there is a good deal o( variation amongthe 19 states that adhere to the compactness principle. The list includesstates tliat lean Democratic {Minnesota and Washington) as well as thosethat typically elect Republicans (Utah and Idaho). Most important, thereare significant differences in the number and distribution of racial minori-ties living in these 19 states; compare, for example, Alabama and Nevada orSouth Carolina and Iowa on these characteristics. In short, the combinationof real world variation in redistricting principle.s and racial demographicsin the states provides a natural experiment that permits us to build uponAltman's (199aa) simulation studies by examining the effect of redistrictingrules empirically.

DATA AND METHODS

We examine the effect of redistricting rules on racial minority political rep-resentation, assessing whether these standards are, in fact, race neutral. Ourtask is complicated by the fact that the reapportionment process adds orsubtracts congressional districts in some states. While it might seem straight-forward to assess minority political representation with, say, the percentageof African Americans that reside in each of a state's congressional districtsbefore and after redistricting, nearly half of the states that redistricted in2001-02 gained or lost seats. Due to this practical problem, we conduct ouranalysis at the state, rather than the district, level. Fortunately, the existingliterature provides guidance on how to conceptualize minority representa-tion at the state level (Brace et al. 1988; Grofman and Handley 1989; Lublin1997; Swain 1995).

Some argue that racial minorities can influence electoral politics even Indistricts where theyare not in the majority (Lublin 1997; Swain 1995). Othershave noted that the presence of Latinos in a district reduces the percentageof blacks needed to assure the election of a Democrat (Grofman, Handley,and Lublin 2001; Lublin 1997,1999). Thus, we can conceive ofthree types ofdistricts in this regard: districts where minority voters (i.e., African Ameri-cans and Hispanics) determine the winner, districts where they serve as theswing vote, and districts where they do not influence election outcomesbecause they constitute too small a percentage of the district. Blacks and

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 421

Latinos can exert influence and gain representation in the first two typesof districts, but probably not the third.^ We operationalize racial minorityinfluence in two ways: the number of congressional districts in a state thathave a minority population of greater than 50 percent (hereafter referredto as "majority-minority districts"), and the number of districts that have aminority population of 33 to 50 percent (hereafter referred to as "minorityinfluence districts").'' Both measures refer to seat counts in the immediateelectoral aftermath of the 2001-02 redistricting cycle.'

For independent variables in our model, we used state constitutions,statutes, and the written guidelines of state redistricting commissions tocreate a dichotomous measure for each of the redistricting principles listedin Table 1 (where 1 = principle required). This series of variables permits usto test the hypothesis that redistricting principles affect the political influ-ence of racial minorities. To assess whether the impact of compactness iscontingent upon the distribution of minorities, we also created a measureof racial segregation for each state. Altman (199aa) argues that compactnessenhances minority representation only in those states that have a large andsegregated minority community. Using county-level population data archivedat the University of Michigan's Population Studies Center, we calculated theIndex of Dissimilarity (ID) for each state (see Appendix).^ The theoreticalrange of the ID is 0.00-1.00, with higher values indicating a higher level ofracial segregation. In our data, the ID ranges from 0.33 (South Carolina) to0.93 (Nebraska).

Because the incidence of majority-minority and minority influence dis-tricts is a direct function of the presence of minority voters in a state, wecontrol for the percentage of African American and Hispanic residents in astate using United States Census data. We also control for the partisan com-position of a state, which we expect to be related to the number of major-ity-minority and minority influence districts. More precisely, we anticipatethat states with unified Democratic control of the state legislature and thegovernor's office would have more of both of these kinds of districts, owingto the fact that racial minorities tend to vote Democratic (Flanigan andZingale 1998; but see McDonald 2004 on how majority-minority districtsmay benefit Republicans)."*

EMPIRICAL RESULTS

Did the use of traditional redistricting principles affect the political represen-tation of racial minorities in states that engaged in congressional redistrictingfollowing the 2000 census? Our first measure of representation is a count of

422 BARABAS AND (EKIT

the number of majority-minority congressional districts in a state, whereAfrican Americans and Hispanics constitute 50 percent or more of a district'spopulation. Our second measure is a count of the number of minority influ-ence districts in a state, where African Americans and Hispanics constitute35-50 percent of a district's population. Since both of our dependent vari-ables are counts bounded by zero at the low end, we use Poisson regressionto estimate our models of effects on them (King 1989).'"

The small number of cases in our analysis (N — 43) presents certainstatistical challenges. Although we want to examine the effect of all sevenredistricting principals on minority representation, the loss of degrees offreedom and the high correlation among some of the principles in our data-set are a cause for concern." Building on a core model that included ourmeasures of minority population, racial segregation, and political control ofthe redistricting process as independent variables, we used likelihood-ratio(LR) tests to determine which of the seven redistricting rules to include inthe model. Any variable that improved the fit of the model at even a veryweak level of statistical significance (p < .20) was retained.All variables pass-ing this threshold were included in the model and, once again, tested with aseries of LR tests. In the end, we were left with three redistricting standardsin our model—compactness. Section 5 of the Voting Rights Act, and protec-tion of political subdivisions—in addition to the political control, minoritypopulation, and racial segregation variables.

Estimated Poisson regression coefficients, robust standard errors, andmarginal effects for our model with both dependent variables are in Table2. As the first model indicates, the compactness standard has a negative andstatistically significant effect on the number of majority-minority congres-sional districts in a state. States that adopt a compactness standard havefewer majority-minority seats, typically just over half of one seat fewer (amarginal effect of-.45). On the other hand, states that in whole or part mustcomply with Section 5of the Voting Rights Act (VRA) are estimated to haveat least one additional majority-minority seat. The terms for the politicalsubdivisions principle and unified Democratic control are not statisticallysignificant in the majority-minority district count model, after controlling forminority population and racial segregation. States with a greater percentageof minorities and more racial segregation also had more majority-minoritycongressional seats, even after controlling for the other factors.

The second set of estimates in Table 2 shows that compactness also had astatistically significant effect on the number of minority influence congres-sional districts in a state, although the overall size of this negative effect wasonly about two-thirds of this effect in the majority-minority district model.

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 42=

Table 2. The Effects of Redistricting Principles and Demographic Factors on MinorityRepresentation in Congressional Districts following the 2000 Census: Poisson RegressionEstimates

txtm pact nessVotingRights Act §5Political subdivisionsUnified dcniocratit controlMinority populationRacial segregation indexConslant

Log-LikelihoodN

Majority-Minority Districts

Coc£f.

-0.80*1.45*

-0.440.456.41*4.41"

-1.78*

StandardError

0.360.460.290.361.49]M

1.30

-31.5443

MarginalEffect

-0.4S1.14

-0.250.303.712.55

—

Minority Influence Districts

Coeff.

-1.51*0.061.27*0.14

12.54*2.35

-5.63*

StandardError

0.570.470.390.351.691.511.42

-44.7043

MarginalEffect

-0.300.010.280.032.520.47

—

Note: Thf dependent variables in these models arc ihc numlwr of majoriiy-minoriiy and minority indurnce districts,rcspetlivtly, in a slate. Thu marginal fffccis arc the partial diTivalivcs with respett lo a givt-n independent variableevaluated at the mean I Long and Freese 1997).

•p •; .05(two-lailedl

On the other hand, Section 5 of the VRA had no discernible impact on thenumber of minority influence districts. This differential effect across thetwo models is consistent with academic interpretations of the VRA. Follow-ing the 1982 amendments to the VRA, officials in the United States JusticeDepartment and many state legislators interpreted the law as a mandateto create majority-minority districts (Canon 1999, 70-3). It should comeas little surprise, then, that Section 5 coverage is related to the number ofmajority-minority congressional districts in a state, but not to the numberof minority influence districts. Far more important to the creation of minor-ity influence districts was whether a state follows the political subdivisionsprinciple. While preserving political subdivisions did not affect the countof majority-minority districts in our analysis, it increased the probabilityof having additional minority influence districts to a statistically significantdegree. Political control, on the other hand, had no effect. Consistent withthe first set of estimates, a state's minority composition added appreciablyto its number of minority influence districts. The racial segregation term, animportant predictor of the number of majority-minority districts in a state,was only marginally statistically significant (p = .11) in the second model.

To understand better the effects of these rules and demographic factors.Figure 1 displays the predicted probability of having a particular numberof majority-minority and minority influence districts based on the modelsin Table 2.'- Panel A of Figure 1 demonstrates that having a compactness

424 BARABAS ANP (ERIT

rule increases the probability of having no majority-minority districts. Moreprecivsely, the model predicts that a state without a compactness rule has a.45 probability of having no majority-minority seats (standard error - .10);with a compactness rule, the probability of having no majority-minority .seatsjumps to .69 (standard error = .07). Contrast this with Panel B. which showsthat states operating under Section 5 of the VRA are more likely to have mul-tiple majority-minority districts than states that are not covered. In fact, theprobability of having none of these districts drops from .70 to .23 (standarderror of the difference - .12) when Section 5 of the VRA is a considerationin a state's redistricting, even controlling for the racial composition of thestate. Panels C and D show, as expected, that the total number of majority-minority districts is predicted to rise along with the proportion of a state'spopulation that is black or Hispanic and its level of racial segregation.

Figure 2 presents similar predicted probabilities, this time using the esti-mates from the Poisson model in Table 2 for minority influence districts.As with majority-minority districts, having a compactness rule decreasestheprobability of having districts in a state in which Hispanics and AfricanAmericans constitute 35-50 percent of the population (Panel A). But protect-ing political subdivisions (Panel B) does the opposite, making it somewhatmore likeiy that a state will adopt at least one minority influence district.Panel C shows that our model predicts that the size of a state's minoritypopulation has a similar effect on the number of minority influence districtsas it had on the number of majority-minority districts. However, the impactof segregation is different on these two measures of minority representation.High segregation virtually guarantees that a state will have at least one major-ity-minority district (Panel D, Figure 1). but there is only a .33 probabilitythat a state will have one or more minority influence districts when racialsegregation is high (Panel D, Figure 2).

A more subtle, but very important, question is how compactness, the sizeof a state's minority population, and racial segregation work together to affectminority representation." Consistent with the prediction of Altman's (1998a)computer simulations, we find that the effect of a compactness standardon minority representation is contingent on the proportion of minoritiesin a state and their geographic distribution. Panel A of Figure 3 shows thepredicted probabilities of having majority-minority districts in states withboth a high proportion of racial minorities and a compactness standardunder varying levels of racial segregation, based on the Poisson model esti-mates in Table 2.'"' In a very segregated state, the probability of having nomajority-minority districts is less than .02, meaning that a segregated stateis virtually guaranteed at least one majority-minority district if it has a com-

Figure I. The Effects of Redistricting Principles and Demographics on the Numberof Majority-Minority Districts in a State

Panel A:Kulv

Pani'l B:VRASvi-tionS

1t "•« j

1 Z•2

^ O.I)

1J I I [L n(t 1 : I I 5 6 7 8

Mdinhcr iir MuJnrlly-MlniiHly DbilricLs

1.0

I 0.81 0.6

% l i . l

£t).i iJjj.

n I ; 1 1 5 6 7 8

Number of MtOurfly-MJmirity l)htrk-is

INo Co C'tnnpacincsx rule

Panel C:Stair Mliiorily Comptisitinn

a J J 1 1 ^ ^ ^0 I 2 .1 -I 5 C 7 8

Number urMitirit)-Min

Figure 2. The Effects of Redistricting Principles and Demographics on the Number

of Minority Influence Districts in a State

1.0

O.R

0.6

0.4

0.2

O.ll

A:

Compactness Rule

a

I) I 2 < I •; I

Number or Minority infUivnt-v

• No compactnest rule

Panil It:

Piilitical Subdivisions Rule1.0 -

0.8

(If.

0.4

U.2

0.0

(1 1 2 3 4 5 6 7

NumtH-r of Mimirlly Innucnt-v

I Nu i>ubdtvi!iiun» luk • SuUiivL'aons rule

.30£

1.0

0.8

0.6

0.4

0.2 -

0.0

Panel CStMte Minority Composition

. J 10 1 2

IIK 4

Niimbn- nt Minority In

1 1 1 ^6 7 8

\cv ntstricts

£•

.2eL

•DU

wVu

£

0.8

KM-

( 1 1

0.0

PanrI D:

miiHinlitis • Maximum itimo

0 1 2 3 4

NumhiT iif Minority

• Lo

Note: Prcilicieil pmbiihilliics were calailaicd using C7(in/v

Figure 3. The EfTect of Racial Segregation in States with a High Minority Population

and 3 Compactness Rule

Panel A:Majority-Minority Districts

0.81

"8 0.40.4 I

ft —H«».

2 3 4 5 6

Number of Majoril>-Minority Districts

Panel B:Minority InliaeDcc Dlslrtcti

2 3 4 5 6 7Number or.Miooiit> iDfluenct Districts

• Low segregation •High segregation

Nou- PredicieJ prababitMe* were cakuUied uung Clarify (KJDg, TOOB. aoA WiRmbers 2000) btied on The model

atiiratn in Table 2. For ihcie predicted piohabiliiy calctibtiam, $1HIC miiionty composiiian m s tct to ibc umpte

naximum and the compactness rule vonabtc wu ict lo 1. The bare show the predicted piobabilitica for the nutnber uf

congrrssionat seals u ihc Irvel ornci«l segrcgxtioa varies fmta the un^\K Tntnimum to imxununi on the Index of

Otlilmilanty (lee Appendix).

428 BARABAS AND TBRIT

pactness rule. The modal value in this scenario is four majority-minoritydistricts (probability = .12; standard error = .07). By contrast, a state witha high percentage of minoritie.s. a compactness standard, and a low level ofsegregation has a .64 predicted probability of having no majority-minoritydistricts (standard error = .21).

Panel B of Figure 3 shows a similar relationship for the predicted numberof minority influence districts. The modal value for a state with a compact-ness rule and a large and segregated minority population is three minorityinfluence districts (probability = .17; standard error = .06). A compactnessrule in the absence of high racial segregation yields a modal prediction ofzero minority influence districts.

To test the robustness of our models, we explored the impact of severalother factors on our dependent variables. Using data from the United StatesCensus, we examined whether changes in a state's African American andHispanic population had an impact on their representation in the 2001-02redistricting cycle.'' While an increase in a state's Hispanic population wasassociated with more minority influence districts, our main substantiveconclusions do not change when we control for changes in the percentageof African Americans or Hispanics in a state.'" We also included regionaldummy variables to control for spatial autocorrelation, but the only patternwe found was the tendency for .southern states to have fewer majority-minor-ity districts.

Finally, our findings should be considered with some caution for tworea.sons. First, only 43 states engage in redistricting at the congressional level,giving us precious few degrees of freedom to disentangle the direct and inter-active effects of these rules. While studying congressional redistricting overtime might help this, reapportionment complicates matters since some stateslose, while others gain, congressional seats each decade. Future research onthe effects of redistricting principles might gain leverage on this problem byexamining state legislative redistricting where the number of states engagingin it is greater and the number of districts in each state is more stable overtime.

A second concern with our research design is that states likely imple-ment these redistricting standards in different ways. For example, accordingto Niemi ami his colleagues {1990), there are over 20 unique measures ofcompactness, and more measures continue to be developed (Engstrom andAiford 2002). Our analysis only considers the legal requirements, not themanner in which or even whether they were implemented (but see Hirsch2002 for anecdotal evidence that these rules are in fact implemented). Ifboth parties agree not to follow a required standard, and if the plan is not

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 429

subsequently challenged in the courts, it is possible to evade. While this sortof chicanery is unlikely to occur in obtaining pre-clearance with Section 5of the VRA, we have no sense of how many states fail to comply with theirother formal standards. Future research with a measure of the redistrictingstandards that actually are followed in each state would likely arrive at a moreprecise estimate of the effects of these standards.''

CONCLUSION

Scholars and political practitioners debate whether explicit racial redistrictingenhances the political influence of racial and ethnic minorities (Cameron,Epstein, and O'Halloran 1996; Engstrom 1993; Grofman and Handley 1998;Lublin 1997; Petrocik and Desposato 1998; Shotts 2003; Swain 1995). Wehave examined whether traditional redistricting principles might also havean effect on minority representation, however unintentional. This is an issueof vital political importance. Although these principles are the subject ofmuch legal and media commentary and are widely applied, there has beenlittle empirical examination of their impact. Our analysis of congressionalredistricting in 2001-02 suggests that both these principles and demographicfactors can affect minority representation, but that these effects are quitenuanced. The compactness standard reduces minority representation, leadingto fewer majority-minority districts and fewer minority influence districts.On the other hand, the mandate to obtain pre-clearance under Section 5 ofthe Voting Rights Act increases the number of majority-minority districts,and the requirement to preserve political subdivisions increases the numberof minority influence districts. Other redistricting principles had no discern-ible effect on minority representation in our data.

Two findings deserve special note. First, we find that the impact of redis-tricting rules is conditional, so that we need to consider these standards inlight of the size and distribution of the racial minority population in a state.In particular, highly segregated states with a compactness standard and a largeminority population are more likely to create majority-minority and minor-ity influence districts. Second, we found that these districting principles haddifferential effects across the two models. While compactness had a greaterimpact on the number of majority-minority districts than it did on minorityinfluence districts, the reverse was true for the political subdivisions rule.Thus, how we operationalize minority influence has some bearing on oursubstantive conclusions.

Recent court cases make this last point especially important. In Page v.Bartels 2001, a three-judge federal court ruled in favor of a plan that dis-

430 BARABAS AND lERIT

mantled three majority-minority districts and created four minority influ-ence districts (Hirsch 2002). If future redistricting plans emphasize effectivemajorities rather than strict numerical majorities (Grofman, Handley, andLublin 2001), our analysis indicates that some factors, such as Section 5 ofthe Voting Rights Act, should become less important in enhancing minorityrepresentation, while others, such as the use of a political subdivisions rule,should become more important. But regardless of how minority politicalinfluence is defined, our results indicate that at least some traditional, andostensibly race-neutral, redistricting principles can have significant effectson minority representation.

APPENDIX: MEASURING RACIAL SEGREGATION WITH THE INDEX OF

DISSIMILARITY

The Index of Dissimilarity (ID) is the suiumaiion of vertical deviations between theLorenz curve and the line of perfect equality (Slack and Rodrigue 2004). The IP rangesfrom 0 to I, and ihecloserit i.s to I. the more di.ssiniilarthc distribution of the measuredcharacteristic is tTom the line ol' perfect equality. The formula for creating Uie ID is:

Consider the following example using the ID to measure racial segregation in NewHampshire:

County

BclknapCarrollCheshireCoosGraftonHillsboroughMerrimackRockinghainSirafTordSullivan

TolatPopulati (V)

O02

0.010,030.000.050.54OOS

0 180,080.01

EqualDistribution ( \ )

0,100 100,100,100.100 100 100 100,100,10

IX-YI

OOS

0,090,070.100,050,440.020.080.020.09

1,03

l p d » or Dissimilarity = (.5 * 1.03) = .52

In this example there is an unequal distribution of African Americans in New Hampshire,with Hillsborough County accounting for 54 percent of tlie state's blacks. As a result, the10 for racial segregation in New Hampshire is a moderately high .52.

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 431

We were concerned that the variation in the number of counties in each state {mean= 67; s.d. = 48) might compromise the comparability of state-level IDs. However, whenwe regressed the racial segregation ID on state population and the number of countiesin a state, neither coefhcient was statistically significant (p = .95 and .25, respectively).

ENDNOTES

The authors are listed in alphabetical order. We thank lanct Box-Steffensmeier, DavidDarmofal, Brian Gaines. Tobin Grant, John Hamman, lohn Iceland, Kris Kanlhak, ScottMcCIurg, Al Melone, lean-Paul Rodrigue, Ed Schatz, Richard Winters, and three anony-mous reviewers for their comments and helpful advice. We also thank Paul Gottemollerfor his research assistance. Timothy Weeks, Mary GalUgan, Stewart Wills, Tom Duffy,Pat Hopkins, and Karen Mau provided demographic data for some of the stales. Earlierversions of this article were delivered at the 2003 Annual State Politics Conference andthe 2003 Annual Meeting of the Midwest Political Science Association. Replication dataare available from the authors' Web sites at: vAvw.siu.edu/departments/coIa/polysci/web-sitc/faculty.htnil.

l.In Easley v. Oomartic 200\. the Supreme Court ruled that political, but not racial,considerations may be used in the creation of legislative districts.

2. None of these principles are mentioned explicitly in the United States Constitution(Canon 1999, chapter 2; Altman 1998b),although some, such as contiguity and compact-ness, were once mandated by federal law {e.g., the Apportionment Act of 1901).

3. Some entire states are subject to pre-clcarancc, while elsewhere, only certain sectionsof the state must obtain pre-clearance. Section 5 of the Voting Rights Act remains in effectthrough 2007, at which point it will sunset, if not renewed. Redistricting plans also havebeen challenged under Section 2 of the Voting Rights Act, which prohibits any state fromabridging the right to vote on the basis of race.

4.Seven states {Alaska, Delaware, Montana, North Dakota, South Dakota. Vermont,and Wyoming) have a single, ai-large congressional district, so they are excluded fromour analysis. We compiled Table 1 using information from state constitutions, statutes,and the written guidelines of state rcdistricting commissions (see lerit, Gottemoller, andBarabas 2003 for details).

5. Failure to incorporate the size of the Latino population in our measure of racialminority influence would lead us to overestimate the proportion of blacks needed in adistrict to elect a black-preferred candidate, especially a Democrat. Of course. Latinosdo not vote as reliably for Democratic candidates as do blacks (Abramson, Aldrich, andRohde 2002; Flanigan and Zingale 1998). Cuban Americans, in particular, tend to voteRepublican.

6. We obtained demographic data on congressional districts from the Center for Votingand Democracy (http://www.tairvote.org/) and the National Committee for an EfifectiveCongress (http://www.ncec.org/index.phtml). For the six states for which these organiza-tions did not have demographic data, we obtained it from the United States Census, statelibrarians, or state officials involved in the redistricting process. These data arc identical(after rounding) to ligures reported in Politics in America 2004 (Hawkings and Nutting2004).

432 BARAUAS AND JERIT

7.Maine redistricted in 2003. Also in 2003, the political parties in some states (e.g.,Texas, Colorado, Pennsylvania, and Georgia) tried to redraw their 2001 -02 congressionaldistrict lines. Lawsuits are pending in several of these states.

8. According to Harrison and Weinberg (1992), the ID is the most widely used measureof spatial distribution. See the Appendix for more information on calculating the ID. Datafor these calculations were obtained from http://www.psc.isr.umich.edu/.

9. Although some states use bipartisan redistricting commissions, in nearly 80 percentof the states, the legislature plays the primary role in the congressional redistrictingprocess (Barone, Cohen, and Cook 2001; McDonald 2004). In addition, the governor in84 percent of the states has some type of veto power in the process (Barone, Cohen, andCook 2001). In most states, then, state-level elected officials have both the incentive andthe institutional means to act on their political preferences.

10. In our dataset, both dependent variables range from 0 to 13, with a modeof 0. Wetested related techniques, such as negative binomial regression and zero-inflated ver-sions of Poisson and negative binomial regression (Long 1997),but we rejected them asunnecessary since tbe dispersion parameter and terms in the inflation models were notstatistically significant.

11. For example, tbe contiguity and communities of interest standards have a Pearson'sr correlation statistic of .59 {p

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 433

Altman, Micah. 1998a. "Modeling the Effect of Mandatory District Compactness onPartisan Gerrymanders." Political Geography 17:989-1012.

Altman, Micah. 1998b. "Districting Principles and Democratic Representation." Ph.D.diss. California Institute of Technology.

Barone, Michael, Richard E. Cohen, and Charles E. Cook, Jr. 2001. Tlie Almanac of Ameri-can Politics 2002. Washington, L">C: National Journal.

Brace, Kimball, Bernard Grofman, Lisa Handley. and Richard Niemi. 1988. "MinorityVoting Equality: The 65 Percent Rule in Theory and Practice." Law and Policy 10:43-62.

Bullock, Charles S., III. 1995. "Affirmative Action Districts: In Whose Faces Will TheyBlow Up?" Campaigns and Elections I6{4):22-3.

Butler. David, and Bruce Cain. 1992. Congressional Redistricting: Comparative Theoriesand Perspectives. New York: Macmillan.

Canon, David T. 1999. Race, Redistricting, and Representation. Chicago: University ofChicago Press.

Cameron, Charles, David Epstein, and Sharyn O'Halloran. 1996. "Do Majority-MinorityDistricts Maximize Substantive Black Representation in Congress?" American PoliticalScience Rt-viVw 90:794-812.

Cohen, lacob. 1988. Statistical Power Atialysis for the Social Sciences. Hillsdale, NJ: Law-rence Erlbaum.

Davidson, Chandler. 1992. "The Voting Rights Act: A Brief History." In Controversies inMinority Voting, eds. Bernard Grofman and Chandler Davidson. Washington, DC:Brookings Institution.

Easleyv. Cromartie. 2001. 532 U.S. 234.Engstrom, Richard L. 1995. "Voting Rights Districts: Debunking the Myths." Campaigns

andEkctions 16{4):24,46.Engstrom, Kichard N., and John R. Alford. 2002. "A New Approach to Measuring Dis-

trict Compactness." Presented at the Annual Meeting of the Midwest Political ScienceAssociation, Chicago.

Flanigan, William H., and Nancy H. Zingale. 1998. Political Behavior of the AmericanElectorate. 9"' ed. Washington, DC: CQ Press.

Grofhian, Bernard, and Lisa Handley. 1989. "Minority Population and Black and His-panic Congressional Success in the 1970s and 1980s." American Politics Quarterly 17:436^5.

Groftnan, Bernard, and Lisa Handley. 1998. "Estimating the Impact of Voting-Rights-Related Distriaing on Democratic Strength in the U.S. House of Representatives." InRace and Redistricting in the 1990$, cd. Bernard Groftnan. New York: Agathon Press.

Grofinan, Bernard, Lisa Handley, and David Lublin. 2001. "Drawing Effective MinorityDistricts: A Conceptual Framework and Some Empirical Evidence." North CarolinaLaw Review 79:\i8i^iO.

Harrison, Roderick |., and Daniel H. Weinberg. 1992. "Racial and Ethnic ResidentialSegregation in 1990." Presented at the Annual Meeting of the Population Associationof America, Denver, CO.

Hawkings, David, and Brian Nutting. 2004. Politics in America 2004: The 108"' Congress.Washington, DC: Congressional Quarterly Press.

Hirsch, Sam. 2002. "Unpacking Page v. Bartels: A Fresh Redistricting Paradigm Emergesin New Jersey." Election Law Joumai 1:7-24.

434 BARABAS AND JERIT

Jerit, Jennifer, Paul Gottemoller, and Jason Barabas. 2003. "U.S. Redistricting Standards:A Protocol for Data Collection." Southern Illinois University at Carbondale. Unpub-lished manuscript.

King, Gary. 1989. Unifying Political Methodology: The Likelihood Theory of StatisticalInference. Ann Arbor, MI: University of Michigan Press.

King, Gary, Michael Tomz, and Jason Wittenberg. 2000. "Making the Mo.st of StatisticalAnalyses: Improving Interpretation and Presentation." American Journal of PoliticalScience 44:347-6].

Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent Variables.London: Sage Publications.

Long. J. Scott, and Jeremy Freese. 2001. Regression Models for Categorical Dependent Vari-ables Using Stata. College Station, TX: .Stata.

Lowenstein, Daniel H., and Jonathan Steinberg. 1985. "The Quest for Legislative District-ing in the Public Interest: Elusive or Illusory?" UCIA Liw Review 33:1-75.

Lublin, David. 1997. The Paradox of Representation: Racial Gerrymandering and MinorityInterests in Congress. Princeton, NJ: Princeton University Press.

Lublin, David. 1999. "Racial Redistricting and African American Representation: A Cri-tique of 'Do Majority Districts Maximize Substantive Black Representation in Con-gress?'." American Political Science Review 93:183-96.

McDonald, Michael D. 1996. "The Why and How of Compact Districts." American Reviewof Politics \7:3Z9^i.

McDonald, Michael P. 2004. "A Comparative Analysis of Redistricting Institutions in theUnited States, 2001-02." State Politics and Policy Quarterly 4:371-95.

Miller v. Johnson. 1995. 515 U.S. 900.Niemi, Richard C , Bernard Grofman, Carl Cariucci, and Thomas Hofeller. 1990. "Mea-

suring Compactness and the Role of a Compactness Standard in a Test for Partisanand Racial Gerrymandering." yo»rMrt/(j/"Po/(fji5 52:1155-81.

Pagev. Batets. 2001. 248 F3d 175.Petrocik, John R., and Scott W. Desposato. 1998. "The Partisan Consequences of Majority

Minority Rcdistricting in the South, 1992 and 1994." Journal of Politics 60:613-33.Pildes. Richard H., and Richard G. Niemi. 1993. "Expressive Harm.s, 'Bizarre Districts,' and

Voting Rights: Evaluating Election-District Appearances after Shaw v. Reno." MichiganLaw Review 92:483-587.

Polsby, Daniel D., and Robert D. Popper. 1991. "Ugly: An Inquiry into the Problemof Racial Gerrymandering under the Voting Rights Act." Michigan Law Review 92:588-651.

Shaw V. Reno. 1993. 509 U.S. 630.Shotts, Kenneth W. 2003. "Does Racial Redistricting Cause Conservative Policy Outcomes?

Policy Preferences of Southern Representatives in the 1980s and 1990s." Journal ofPolitics bS:2\6'23.

Slack, Brian, and Jean-Paul Rodrigue. 2004. "GIni Coefficient." In Transport Geographyon the Web, eds. Jean-Paul Rodrigue, Brian Slack, and Claude Comtois. Hofstra Uni-versity, Department of Economics & Geography, http://people.hofstra.edu/geotrans(accessed April 21, 2004).

Stern, Robert. 1974. "Political Gerrymandering: A Statutory Compactness Standard as anAntidote for Judicial Impotence." University of Chicago Law Review 4\:39i-4l6.

WINTER 2004 / STATE POLITICS AND POLICY QUARTERLY 435

Swain, Carol M. 1995. Black Faces, Black Interests: The Representation of African Americansin Congress. Cambridge, MA: Harvard University Press.

Wattson, Peter. 1999. Redisiricting Law 2000. http://www.senate.leg.state.mn.us/depart-ments/scr/redist/red2000/red-tc.htm (accessed April 26, 2004).

Wells, David. 1982. "Against Affirmative Gerrymandering." In Race and Redistricting inthe 1990s, ed. Bernard Grofnian. New York: Agathon Press.