Content uploaded by Stella M. Rouse
Author content
All content in this area was uploaded by Stella M. Rouse on Feb 05, 2020
Content may be subject to copyright.
The Impact of Electoral Arrangements on Minority Representation:
District Magnitude and the Election of African American State Legislators*
Paul S. Herrnson
Department of Political Science
University of Connecticut
paul.herrnson@uconn.edu
Stella M. Rouse
Center for Democracy and Civic Engagement
Department of Government and Politics
University of Maryland
srouse@umd.edu
Jeffrey A. Taylor
Westat
jeffreytaylor@westat.com
Abstract
Previous research shows that multimember districts (MMDs) disadvantage African American
candidates. However, these studies focus on only a few aspects of the electoral process and they
may be time bound. Using a new dataset, we examine the impact of district magnitude (the
number of candidates elected from a single constituency) on the emergence, nomination, and
general election of African Americans to state legislatures. Using data from recent elections to
the Maryland state legislature, we find no evidence that district magnitude dims the electoral
prospects of African American candidates. Our findings suggest that biases attributed to MMDs
may have resulted from laws, partisan practices, customs, and political attitudes. The
implementation of the Voting Rights Act, broad societal changes, and strategic adjustments by
Black candidates and voters may have mitigated the effects of previous biases resulting in the
election of more African Americans in MMDs and other districts.
2
*Non-copyright version
African Americans historically have faced many obstacles when seeking public office.
Included among these are gerrymandering, literacy tests, and other discriminatory practices
rooted in racial prejudice. The conventional wisdom is that multimember districts (MMDs),
including at-large districts, are among the institutional arrangements that deter the political
candidacies of African Americans, hinder their success, and minimize the descriptive
representation of Blacks in government. However, the conventional wisdom relies on
experiences from the nineteenth through mid-twentieth centuries. This was a time when wily
politicians and party committees manipulated election laws, adopted discriminatory party rules,
and used even more base tactics to harm the electoral opportunities of Blacks. It is difficult to
know with certainty the degree to which MMDs harmed the prospects of African American
politicians during this period because they were part of a much larger systematic effort to
undermine Blacks’ political representation. This raises questions about whether MMDs, in and of
themselves, continue to disadvantage African American candidates.
African Americans and other minorities have made dramatic gains in the electoral arena
since the mid-twentieth century, and some of these gains have been in MMD elections. The
increased numbers of Blacks who run for and are elected to federal, state, and local offices,
including Barack Obama, the nation’s first African American president, strongly suggest that
some aspects of the political environment that previously hindered the candidacies of
traditionally underrepresented groups have changed or are of less consequence. Nevertheless, the
Supreme Court decision in Shelby County v. Holder serves as a reminder that racial progress in
the electoral arena is not guaranteed. Given many officeholders continue to be elected in MMDs
3
and the many changes that have occurred in American politics, it seems appropriate to reassess
the impact of district magnitude on African American representation.
We use a new data set consisting of candidate, strategic, and demographic variables for
the state legislative elections in Maryland in 2006 and 2010 to consider the effects of district
magnitude on the three key phases of the electoral process: candidate emergence, nomination
contests, and general elections. Our results raise doubts about the effects of district magnitude on
African American politicians. They suggest the conventional wisdom may be in need of revision.
Is District Magnitude an Impediment to the Election of African American Legislators?
Understanding the impact of district magnitude on minority representation is important.
As recently as the 1960s, a majority of states used MMDs to elect some state legislators and
almost half of all state legislators nationwide were elected in MMDs. Ten states currently elect
state legislators to at least one chamber with MMDs. Roughly 15% of all state legislative seats
are elected from MMDs (Ballotpedia ND). Many counties, cities, towns, school boards, and
other special governments also elect some members in MMDs. Moreover, the states have used
MMDs, single-member districts (SMDs), combinations of MMDs and SMDs, and other electoral
configurations to elect members of the U.S. House of Representatives over the course of the
nation’s history (Martis 1982). It was not until the late 1960s that MMDs were no longer used to
elect U.S. House members.
The principal argument against MMDs is rooted in the concept of vote dilution. It can be
summarized as follows: MMDs, especially those of greater magnitude, thwart African American
representation because they place Black voters into districts where they lack sufficient numbers
to elect a corresponding number of Black candidates, whereas dividing the same geographic area
into two or more SMDs that include at least one SMD with sufficient Black voters to elect a an
4
African American to office provides Blacks with better descriptive representation (Parker 1984;
Niemi, Jackman, and Winsky 1991). A less frequently made argument is based on ballot structure
and voter behavior. It states Black candidates are disadvantaged in MMD elections because
African American voters are less likely than whites to participate in these and other down-ballot
races, and the Blacks who participate in MMD elections select a smaller number of the total
allowable candidates than whites (Brady et al. 2001, but see Herrnson, Taylor, and Curry 2015).
The predominant literature focuses on and supports the vote dilution argument, and it
provides evidence that historically MMDs hindered the election of Black candidates (Bullock
and Gaddie 1993; Gerber, Morton, and Rietz 1998; Richardson and Cooper 2003). It also shows
that when states switched from MMDs to SMDs in the 1960s, the number of African Americans
elected to state and local offices increased (Engstrom and McDonald 1982; Grofman, Migalski,
and Noviello 1986; Lenz and Pritchard 1991; Grofman and Davidson 1994; Gerber, Morton,
Rietz 1998; Orey 2000).
However, the argument against MMDs has several shortcomings. First, it downplays the
historical role of gerrymandering, literacy tests, and the other institutions and processes designed
to thwart the election of racial and ethnic minorities in MMDs and SMDs (Hasen 2004; Winburn
2008; Meier and Rutherford 2014). Second, it does not fully consider the effects of the
concentration of African American voters in urban areas (Moncrief and Thompson 1992;
Trounstine and Valdini 2008), and it underestimates the importance of majority-minority districts
in facilitating minority candidacies in both SMDs and MMDs (Lenz and Pritchard 1991; Lublin
1997; Hadley, Grofman, and Arden 1998; Swers and Rouse 2011). Third, it does not take into
account the importance of how voting is structured. As Brockington et al. (1998) show, MMD
elections that use cumulative and limited voting result in similar levels of minority representation
5
as SMD elections. Fourth, the case against MMDs underestimates the effects of the Voting
Rights Act of 1965 and other reforms that combat malapportionment, voter disenfranchisement,
and other impediments to minority voting. Research on city council and other local elections
demonstrated that Black candidates fared much better in MMDs, including at-large districts,
following the implementation of these reforms (Welch 1990, Herrick and Welch 1992).
Attitudinal and behavioral changes among voters also complicate the argument that
MMDs result in the underrepresentation of African Americans in elective office. There is
evidence of less discrimination against Black candidates among white voters (Highton 2004) and
a greater willingness among whites to support Black officeholders (Hajnal 2001, 2006). Indeed,
whites tend to view Black politicians as displaying more integrity, passion, empathy, having
overcome greater adversity, and better able to address issues of racial equality than white
politicians (Schneider and Bos 2011). Whites also consider Black politicians as equally
confident, charismatic, and determined as their white counterparts (Schneider and Bos 2011). In
an era of political polarization, partisanship may trump race when whites enter the voting booth.
The constituencies and ideas a party represents may have a greater impact on whites’ voting
decisions than the racial attributes of the candidates who appear on the ballot (Juenke and Shah
2016; Swain 1993; Lublin 2004).
The strategies adopted by African American candidates and voters also are unaccounted
for by traditional arguments against MMDs. Many Black candidates, including President Obama,
disseminate messages and images that whites find appealing (Ford, Johnson, and Maxwell 2009).
African American voters have been making use of “single-shot” (“bullet”) voting, which
involves selecting only the African American candidate(s) running in a MMD (and purposefully
selecting fewer than the total allowable candidates) in order to avoid giving votes to the white
6
candidates in the race (Engstrom and McDonald 1986, 1993; Herron and Sekhon 2005; Smith
2008).
Finally, observations from the political arena raise questions about the effects of MMDs,
at-large elections, and elections held in districts of greater magnitude on the candidacies of
minorities. The support of many African American members of Congress for proposals to allow
the states to elect U.S. House members in MMDs (e.g., Snyder and Ueda 2007) suggests these
electoral arrangements may no longer disadvantage contemporary Black politicians and voters.
Data and Methods
What is the impact of district magnitude on the election of African Americans to the state
legislature? As noted above, most of the extant literature concludes MMDs have negative effects
on African Americans’ descriptive representation (Meier and Rutherford 2014). However, many
of these studies are time-bound and unable to disentangle the effects of MMDs from those of
other political institutions, laws, and practices. Most studies also do not account for the
attitudinal and behavioral changes that have occurred among candidates and voters since the
1960s. Moreover, most analyze only one part of the electoral process. We use a comprehensive
approach to investigate the impact of MMDs. We not only examine the election prospects of
African Americans running for the state legislature, but also the emergence and nomination of
these candidates. Our skepticism about the extant literature leads us to hypothesize that race will
not have a significant impact on the three stages of the electoral process:
H1: Candidate emergence: African Americans are just as likely to run in districts with
greater magnitude as districts with smaller magnitude as are other candidates.
H2: Primary elections: African Americans perform as well as non-African American
candidates in primary elections held in districts with greater magnitudes.
7
H3: General elections: African Americans perform as well as non-African American
candidates in general elections held in districts with greater magnitudes.
The optimal conditions for testing these hypotheses are as follows. First, the data would
be collected from elections to one legislative chamber so that relevant election laws, political
processes, and norms are constant and cannot influence the results. Second, there would be
significant variation in the candidates’ race, incumbency, and constituencies. Third, voter turnout,
party competition, and population size would be similar to many other states. Combined these
conditions would provide for strong internal validity and make it possible to generalize beyond
one state.
We test the hypotheses using data from elections to the Maryland House of Delegates in
2006 and 2010. No state or set of elections perfectly fits the conditions for testing our
hypotheses, but state delegate elections in Maryland closely approximate a number of them.
Maryland uses bloc with partial abstention MMD elections; voters may cast as many votes as
there are seats, may cast up to one vote per candidate, and may choose to cast fewer votes than
the number of seats. Maryland elects some house members from MMDs of various magnitudes
and others from SMDs. Maryland’s constitution establishes one senator and three delegates be
elected from each state senate district. As a result, its redistricting process has to accommodate
the competing ambitions of members of two legislative chambers rather than one. The state’s
constitution also requires political boundaries, such as counties, cities, or municipalities, be left
intact whenever possible for state house districts. Both sets of requirements impose extraordinary
constraints on legislators’ abilities to manipulate district boundaries (Taylor, Herrnson, and Curry
2018). Indeed, the Maryland State Supreme Court redrew the state’s post-2000 legislative
districts after the state legislature was unable to accomplish this task.
8
Moreover, the nesting of state house districts within senate districts facilitates
comparisons between different combinations of SMDs and MMDs held in a given geographic
areas. Altogether, 49% of the members of the Maryland House of Delegates are elected in three-
member district elections, 19% in two-member districts, and 32% in single-member districts.
Roughly, 65% of the 468 house candidates who ran in 2006 and 2010 contested three-member
districts with the rest fairly evenly distributed across one- and two-member districts.1
Maryland’s racial composition and history are well suited to testing the hypotheses. The
state has a sizeable Black population (29%) and a substantial number of African American
legislators (24%). It also has a history of racial discrimination: it did not abolished slavery until
1864, and it enacted statutes enforcing segregation after the Civil War (Smith and Willis 2012).
Other states that use a combination of MMDs and SMDs to elect state house members, such as
Vermont, New Hampshire, and West Virginia, lack demographic diversity, African American
representation in the state legislature, and a history of discrimination. In short, while Maryland is
not a traditional southern state (as defined by being a part of the Old Confederacy) it is the
southernmost state that continues to employ MMDs, and its racial composition and racial history
resembles that of many southern states.
Maryland’s demographics and politics also provide other advantages. Its population of
5.3 million is just below the national average and it is somewhat more racially diverse, more
educated, and more affluent. Like many states, Maryland’s politics have been long dominated by
one-party (the Democrats), but reasonable two-party competition is evident as the opposing party
1 Voting in district 36 in Maryland changed in 2010 from a traditional three-member plurality
system to a three-member seat designate electoral system, in which voters within the district may
select one candidate depending on their county of residence. This procedure means that District 36
in 2010 is no longer classifiable as a bloc SMD or a MMD so we exclude it from the analyses.
9
(the Republicans) has recently run competitive races for governor and occupied the governor’s
mansion from 2003 through 2007, 2015 through 2019, and presumably through 2022—when the
current GOP governor’s second term will end. The 2006 and 2010 elections are well-matched to
our hypotheses tests because they are characterized by opposing electoral trends: the former
election occurred in a national political environment that advantaged Democrats; the latter took
place in a political climate favorable to Republicans.
The data used to assess the impact of MMDs on the candidacies of African Americans
came from three sources. The Maryland State Board of Elections compiled election returns
and campaign spending data. Information about the candidates’ racial profiles was collected
from campaign websites and newspaper reports. The U.S. Census provided district
demographic data.
Our initial analyses use descriptive statistics to help determine whether the relationships
between race, district configurations, and electoral success are worthy of a comprehensive
analysis. First, we provide an overview of the relationships between district magnitude and who
runs for the legislature, wins the nomination, and wins the general election.2 We conduct these
analyses for all districts as well as for the subsets of districts in which a majority of voters
identify as African American compared to non-majority African American districts. The unit of
analysis for this and the models that follow is the district/year.
The dependent variable used to test candidate emergence is based on the numbers of
African American and other candidates who run for the lower chamber of the Maryland State
House of Delegates. The primary election and general election hypotheses tests are based on
African American candidates’ success in these elections relative to the successes of other
2 See Appendix Table A-1 for summary statistics for the dependent and independent variables
analyzed.
10
candidates. These analyses rely on four independent variables: whether the candidate wins or
loses the primary election, the candidate’s vote share in the primary election, whether the
candidate wins or loses the general election, and the candidate’s vote share in the general
election. Our analyses compare African American candidates to all others; however, virtually
identical results are generated when the comparison group consists solely of white candidates.3
We limit the analysis to Democratic candidacies because African American candidates and
voters affiliate overwhelmingly with the Democratic Party. Only seven African American
Republicans ran for the House of Delegates during the elections under study and none were
victorious. Most candidates and voters of traditionally underrepresented groups affiliate with
the Democratic Party (e.g., Verba, Schlotzman, Brady 1995; Griffin and Newman 2008).4
Regardless of race, ethnicity, or gender, once politicians align themselves with one party, their
career options are largely limited by their political affiliation.
The dependent variable used to test the candidate emergence hypothesis—the
Percentage of African American candidates—is based on the race of each candidate who
entered a Democratic primary in a given election cycle (the unit of analysis is the district-
election year). We use fractional probit regression to test our candidate emergence hypothesis,
which is an appropriate estimation method for models where the dependent variable is a
proportion bounded by zero and one (Papke and Wooldridge 1996). The key independent
variables are two dummy variables indicating whether the district is a Two-member district or a
Three-member district, with SMDs as the comparison group.5
3 The results are presented in Tables SA-1 through SA-5 of the online supplemental appendix.
4 We did not study Latinos, Asian Americans, or other racial and ethnic groups because they
constitute a small number of candidates or legislators in Maryland.
11
When estimating the model for the candidate emergence hypothesis, we use additional
variables to account for differences in the electoral environment and the negative effects of
crowding (the number of candidates in an election) on a candidate’s decision to run. The variable
% Same-party incumbents equals the number of incumbents of the candidate’s party who contest
a district divided by the number of seats in the district. % Opposing-party incumbents equals the
number of incumbents of the opposite party divided by the number of seats. The racial
composition of the district delegation and the constituency also are expected to influence the
number of African American politicians who compete for a legislative seat and their likelihood of
winning (Moncrief and Thompson 1992; Trounstine and Valdini 2008). % African American
incumbents represents the percentage of district incumbents who were African American prior to
the election. We also expected districts with more African American voters to attract more Black
candidates for office. % African American voters indicates the percentage of the district
population that identifies as African American. Substituting a dummy variable that indicates
whether the percentage of the district population that identifies as African American is above
50% produces similar results. 6
We test the primary elections and general elections hypotheses using data related to each
candidate in each election year. We test the primary election hypothesis with two dependent
variables. Primary winner is coded 1 if the candidate won the primary election and 0 otherwise.
Relative primary vote share, a measure of competition similar to that developed by Hirano and
Snyder (2009), allows for comparisons across districts of different magnitudes. It operationalizes
vote share as the votes earned by each candidate in an election relative to the average number of
5 We use dummy variables as opposed to a single measure of district magnitude because this
approach allows for the possibility that the effect of magnitude is non-linear.
6 See Tables SA-6 through SA-8 of the online supplemental appendix.
12
votes earned by the candidate’s opponents. Values below 1 indicate that a candidate earned fewer
votes than the average; values above one indicate a candidate performed above average. The
final two dependent variables, used to test the general election hypothesis, are similar to those in
the model for primary elections. General election winner equals 1 if the candidate won the
election and 0 otherwise. Relative general election vote share compares the number of general
election votes earned by each candidate to the average votes earned by all candidates in the
district. Our reliance on several dependent variables increases our confidence in the findings.
For the primary and general election hypothesis tests, the main independent variables are
the candidate’s race, coded 1 if the candidate is African American and 0 if otherwise, and two
dummy variables indicating whether the district is a Two-member district or a Three-member
district (SMD is the comparison group). We control for a number of district-level factors that are
expected to influence electoral success. We control for incumbency because it harms challenger
performance in both the primary and the general election (Stone and Maisel 2003; Stone, Maisel,
and Maestas 2004). We expect challengers who run against one or more incumbents of their own
party in the primary and challengers who contest an open primary in anticipation of facing one or
more opposing-party incumbents in the general election to be less successful than candidates
who run for seats where no incumbent is seeking reelection (e.g., Canon 1990). % Incumbents is
simply the number of incumbents who contest an election divided by the number of seats in the
district. To control for the effect of the number of candidates who run in an election on each
candidate’s prospects, we include % Contested seats, which equals the number of candidates
divided by the number of seats. We also include controls for the % African American incumbents
and the % African American voters. As is the case with most recent studies of state legislative
elections, we do not include a measure of challenger quality (Barber 2016; Barber, Butler, and
13
Preece 2016); this is largely because too few state legislative candidates have previous office
holding experience (Moncrief, Squire, and Jewell 2001).
Campaign funding also is important to electoral success. We control for each
candidate’s expenditures relative to the average expenditures of all of the candidates in a race.
This measure is operationalized as:
S pending advantage =
(
Candidate expenditures - A verage expenditures
)
Average expenditures
Where candidate expenditures are the total expenditures made by the candidate prior to the
election, and average expenditures is the average expenditures made by all candidates prior to
the election. For primary candidates, we include only funds spent prior to the primary election.
For general election candidates, we include funds spent during the primary season and in the
general election because the effects of primary spending usually carry over into the general
election. We expect that candidates who spend more will perform better than their opponents.
Finally, we include the dummy variable Year 2010 to control for variations in the political
environment across the two election cycles.
We estimate the effects of district magnitude and race on African American candidates’
likelihood of primary success and winning the general election using logistic regression.7 We
also estimate the effects of MMDs and race on candidates’ vote shares. Because vote shares are
continuous variables, we use ordinary least squares (OLS) regression. We include interactions
for African American X two-member district and African American X three-member district in
7 We also estimated the impact of district magnitude on the percentage of African Americans
who win the primary and the percentage of African Americans who win the general election in
each election district. These results show that MMDs do not disadvantage African American
candidates for the state legislature (see Table SA-9 of the online supplemental appendix).
14
the OLS models because, unlike logistic regression, OLS does not assume interactive effects
among explanatory variables. We use Wald tests to estimate the significance of interaction
terms. We transform the results from the multivariate models into predicted effects (the models
are presented in the Appendix).
Results
Does district magnitude penalize contemporary African American politicians who run
for the state legislature? The evidence from our preliminary hypothesis tests suggests that the
impact of district magnitude varies by the stage of the electoral process (see Table 1). The table
presents the percentages of African Americans in the population, who run for office, secure the
nominations, and win the election. Each row contains data for district demographics, candidate
information, and election outcomes for districts of the same magnitude. The first row contains
statistics for the three SMD elections in state senate seats that each contain three separate house
SMDs and for the one SMD election in state senate seats made up of one house SMD and one
two-member house district. The next row consists of the two-member house districts (in senate
districts that also have one house SMD). The third row consists of three at-large house member
seats nested in one senate seat (i.e., a three-member district). It is worth noting that the
preliminary findings in the third and fourth columns are to some degree contingent on the
findings in the columns that precede them. That is, the percentages of Black candidates who
run for the nomination affects the percentage who run in the general election, which, in turn
influences the percentage elected.
[Table 1 about here]
The results in the first section of the table are for all districts. The first two columns in
the table show that the percentage of African American candidates is roughly proportional to
15
the percentage of African Americans in the population, regardless of district magnitude. Had
significantly fewer Black candidates emerged from districts of greater magnitude, one could
argue the preliminary results suggest that MMDs discourage potential Black candidates from
running for office. The figures in the second and third columns demonstrate the nominations
process takes a toll on Black candidates, regardless of the magnitude of their districts. Given
that more African Americans contest MMDs than SMDs one would expect the percentages of
those who lost in the former primaries to be much greater than those defeated in the latter. This
is clearly not the case for three-member seats and, arguably not the case for two-member seats,
providing preliminary support for our expectations. The results for the general election show
that African Americans performed very well in three-member house districts, and less well in
two-member and single-member districts.
The last two sections of Table 1 divide the districts on the basis of whether African
American voters were in the majority or the minority. As expected, Black candidates in
majority-Black districts are more successful than others. However, district magnitude does not
appear to have an effect on the emergence or success of Black candidates in either majority or
non-majority Black districts. Additional preliminary tests that rely on a modified form of
McGhee’s (2014) efficiency gap measure of partisan bias also raise doubts about the negative
effects of MMDs on the election prospects of African American state legislative candidates (see
also Stephanopoulos and McGhee 2015).8
Our first set of multivariate results show that district magnitude has a significant impact
on the candidacies of African Americans. The predicted values in Figure 1 are derived from a
fractional probit regression that includes appropriate control variables. They show that African
Americans make up a significantly higher percentage of the candidates in three-member
8 See Table SA-10 and the ensuing discussion presented in the online supplemental appendix.
16
districts than in single-member districts. The differences between two-member districts and
SMDs, while suggestive, are not statistically significant. The findings strongly support our
expectation that district magnitude does not deter African Americans from running for the state
legislature.
[Figure 1 about here]
The first multivariate test of the primary elections hypothesis assesses the impact of
district magnitude and race on the likelihood of winning the nomination. Figure 2 presents the
predicted probabilities derived from a logistic regression model that includes the racial
composition of the district, campaign expenditures, and other appropriate control variables. The
top panel shows there are no significant differences in the success rates of Black and non-Black
candidates who run in districts of the same magnitude. The bottom panel is for non-incumbent
candidates, who face the longest odds against winning a nomination. These results also show
there are no significant differences between African American and other primary contestants.
As with previous hypothesis tests, the results demonstrate that African American candidates
fare no worse than other candidates regardless of the magnitude of the district they compete in.
[Figure 2 about here]
The second multivariate test of the primary elections takes into consideration the fact that
winning isn’t everything over the course of a political career; a strong showing can encourage a
candidate to run for office again and eventually lead to electoral success (Maisel and Stone
1997). The predicted effects for vote shares in Figure 3 more than substantiate the results for
primary success. They show that African American candidates who contest a primary in either a
two-member or a three-member district perform significantly better than those who run in
17
SMDs. The results for non-African American candidates stand in marked contrast: those in
MMDs do not garner a larger share of the vote than those in SMDs.
[Figure 3 about here]
The final set of hypothesis tests assess the impact of race and district magnitude on
general election outcomes and vote shares. These results raise even more questions about the
broadly accepted generalization that districts of greater magnitudes are harmful to African
American representation in the legislature. The predicted effects in Figure 4 show there are no
significant differences in the success rates of African American candidates, compared to non-
African American candidates. This finding holds for all candidates and for non-incumbents.
The findings for general election candidates’ vote shares reinforce the findings for electoral
success. Figure 5 shows that African American candidates running in three-member districts
perform significantly better than African Americans competing in SMDs, which is similar to
the pattern for other candidates.
[Figures 4 and 5 about here]
Conclusion
Our findings challenge the longstanding belief that multimember districts continue to
hinder the advancement of African American politicians. We demonstrate that rather than
discourage their candidacies, MMDs slightly increase the willingness of African Americans to
run for the state legislature. MMDs also do not undermine the primary or general election
prospects of African American candidates. Under some circumstances, African American
candidates in MMDs garner slightly more primary and general election votes than their non-
Black counterparts.
There are several explanations for the differences in the findings reported here and in
18
earlier studies. First, we research a larger set of dependent variables than those included in
previous studies. Second, the elections we examine are more recent, suggesting the results
previous studies attributed to MMDs may have been the consequence of other factors, such as
racial gerrymandering and voter intimidation. Third, strategic adaptations, including
deracialized strategies and bullet voting, likely offset some hindrances to the participation of
African American politicians and voters. Fourth, as the election of President Barack Obama and
other African Americans suggests, the increased diversity of the electorate and changing voter
attitudes appear to have lessened the impact of race on the voting decisions of a significant
number of voters. This change is consistent with an increasing willingness on the part of many
voters to elect candidates from other traditionally underrepresented groups, including women;
racial, religious, and ethnic minorities; and individuals who identify themselves as lesbian, gay,
bisexual, transgender, queer, intersex, or asexual. Finally, our study differs from most (e.g.,
Grofman, Migalski, and Noviello 1986) in that it assesses the impact of race and MMDs solely
on Democrats in recognition of fact that the careers of the vast majority of contemporary
African American politicians are rooted in Democratic Party politics.
We recognize that our results are based on a single case study. However, one of the
benefits of the case study approach is strong internal validity. Moreover, Maryland has a history
of race relations and a set of electoral arrangements that have been shown to hamper the
election of African American state legislative candidates in the past. These factors combine
with the state’s present demographics, current political conditions, and our reliance on a variety
of measures of candidate emergence, primary outcomes, and general election success, to
strengthen our results and bolster their generalizability.
Our results have important implications for the election of minority candidates and
racial representation. They reinforce the idea that electoral institutions are complex,
19
interrelated, and never completely neutral; put simply, different rules advantage different
groups of candidates and their supporters at different times. They also suggest that outcomes
attributed to one electoral arrangement may result from the interaction of broader sets of laws,
practices, and political and socioeconomic conditions. Few would question that throughout
much of U.S. history MMDs were part of a political context that impeded the election of
African American politicians and disempowered African American voters. However, this study
and other recent research (e.g., Meier and Rutherford 2014) suggest the underrepresentation of
minorities is the product of other institutional structures, processes, and norms. Hasen (2004)
and Winburn (2008), for example, maintain that partisan gerrymandering has become the
primary obstacle to minority descriptive representation. The adoption of strict voter
identification laws in many states also has depressed turnout among minority groups (Sobel and
Smith 2009).
Our finding that MMDs do not systematically hinder the election of African American
state legislative candidates has implications for minority candidacies for a wide range of
offices, including two-thirds of all contests for school boards, county councils, and other
municipal offices that are filled using MMD elections (Municipal Year Book 1998). Because
most candidates for the U.S. House, U.S. Senate, and statewide offices are drawn from state
legislatures and other state and local offices, the findings also have relevance for minority
representation in higher levels of government. Regardless of these implications, the results
support the thesis that African Americans are just as, if not more, likely to seek and capture a
Democratic nomination and win the general election in a multimember state legislative district
as other candidates.
20
21
References
Ballotpedia, ND. “State Legislative Chambers that Use Multimember Districts.” Available at:
https://ballotpedia.org/State_legislative_chambers_that_use_multi-member_districts,
November 29, 2017.
Barber, Michael. 2016. “Ideological Donors, Contribution Limits, and the Polarization of State
Legislatures.” Journal of Politics 78 (1): 296-310.
Barber, Michael, Daniel Butler, and Jessica Preece. 2016. “Gender Inequalities in Campaign
Finance: A Regression Discontinuity Design.” Quarterly Journal of Political Science 11
(2): 219-248.
Brady, Henry E., Justin Buchler, Matt Jarvis, and John McNulty. 2001. “Counting all the Votes:
The Performance of Voting Technology in the United States.” Report published by the
Survey Research Center and Institute of Government Studies, University of California,
Berkeley.
Brockington, David, Todd Donovan, Shaun Bowler, and Robert Brischetto. 1998. “Minority
Representation under Cumulative and Limited Voting.” Journal of Politics 60(4): 1108-
25.
Bullock, Charles S. III and Ronald Keith Gaddie. 1993. “Changing from Multimember to Single-
Member Districts: Partisan, Racial and Gender Consequences.” State and Local
Government Review 25: 155-163.
Canon, David T. 1990. Actors, Athletes, and Astronauts: Political Amateurs in the United States
Congress. Chicago: University of Chicago Press.
Engstrom, Richard and Michael McDonald. 1982. “The Underrepresentation of Blacks on City
Councils: Comparing the Structural and Socioeconomic Explanations for
South/NonSouth Differences.” Journal of Politics 44: 1088-99.
____________. 1993. “‘Enhancing’ Factors in At-Large Plurality and Majority Systems: A
Reconsideration.” Electoral Systems, 12(4): 385-401.
22
____________. 1986. “The Effect of At-Large Versus District Elections on Racial
Representation in U.S. Municipalities.” In Bernard Grofman Arend Liphart eds. Electoral
Laws and Their Political Consequences. New York: Agathon Press.
Ford, Pearl K., Tekia A. Johnson, and Angie Maxwell. 2009. “‘Yes We Can’ or ‘Yes We Did’?
Perspective and Retrospective Change in the Obama Presidency.” Journal of Black
Studies 40(3): 462-483.
Gerber, Elizabeth R., Rebecca Morton, and Thomas A. Rietz. 1998. “Minority Representation in
Multimember Districts.” American Political Science Review 92: 127-144.
Gonzales, Erica. 2018. “20 Big Firsts and Historic Wins from 2018 Midterm Elections.”
Harper’s Bazaar, November 7.
https://www.harpersbazaar.com/culture/politics/a24781771/who-won-midterm-elections-
historic-wins/.
Griffin, John D. and Brian Newman. 2008. Minority Report: Evaluating Political Equality in
America. Chicago: The University of Chicago Press.
Grofman, Bernard, Michael Migalski, and Nicholas Nociello. 1986. “Effects of Multimember
Districts on Black Representation in State Legislatures.” Review of Black Political
Economy 14: 65-78.
Grofman, Bernard and Chandler Davidson. 1994. “The Effect of Municipal Election Structure on
Black Representation in Eight Southern States.” In Chandler Davidson and Bernard
Grofman eds. Quiet Revolution in the South. Princeton: Princeton University Press.
Hadley, Lisa, Bernard Grofman, and Wayne Arden. 1998. “Electing Minority-Preferred
Candidates to Legislative Office: The Relationship Between Minority Percentages in
Districts and the Election of Minority-Preferred Candidates.” In Bernard Grofman ed.
Race and Redistricting in the 1990s. New York: Agathon Press.
Hanmer, Michael J. and Kerem Ozan Kalkan. 2013. “Behind the Curve: Clarifying the Best
Approach to Calculating Predicted Probabilities and Marginal Effects from Limited
Dependent Variable Models.” American Journal of Political Science 57: 263-277.
23
Hajnal, Zoltan L. 2001. “White Residents, Black Incumbents, and a Declining Racial Divide.”
American Political Science Review 95 (3): 603-617.
_____________. 2006. Changing White Attitudes toward Black Political Leadership. New York:
Cambridge University Press.
Hasen, Richard L. 2004. “Looking for Standards (in All the Wrong Places): Partisan
Gerrymandering Claims after Vieth.” Election Law Journal 3: 626-642.
Herrick Rebekah and Susan Welch. 1992. “The Impact of At-Large Elections on the
Representation of Black and White Women.” National Political Science Review, 3: 62-
77.
Herrnson, Paul S., James M. Curry and Jeffrey A. Taylor. 2015. “The Impact of District
Magnitude on Voter Drop-Off and Roll-Off in American Elections.” Legislative Studies
Quarterly 40: 627-650.
Herron, Michael C. and Jasjeet S. Sekhon. 2005. “Black Candidates and Black Voters: Assessing
the Impact of Candidate Race on Uncounted Vote Rates.” Journal of Politics 67: 154-
177.
Herron, Michael C. 1999. “Postestimation Uncertainty in Limited Dependent Variable Models.”
Political Analysis 8(1): 83–98.
Highton, Benjamin. 2004. “White Voters and African American Candidates for Congress.”
Political Behavior 26 (1): 1-25.
Hirano, Shigeo, and James M. Snyder, Jr. 2009. “Using Multimember District Elections to
Estimate the Sources of the Incumbency Advantage.” American Journal of Political
Science 53: 292-306.
24
Juenke, Eric Gonzalez Juenke and Paru Shah. 2016. Demand and Supply: Racial and Ethnic
Minority Candidates in White Districts. The Journal of Race, Ethnicity, and
Politics. 1(1): 60-90.
Lenz, Timothy O., and Anita Pritchard. 1991. “Florida Legislators' Perceptions of the Change to
Single-Member Districts.” State & Local Government Review, 23(3): 134–138.
Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent Variables.
Thousand Oaks, CA: Sage
Lublin, David. 1997. The Paradox of Representation: Racial Gerrymandering and Minority
Interests in Congress. Princeton: Princeton University Press.
___________. 2004. The Republican South: Democratization and Partisan Change. Princeton,
NJ: Princeton University Press
Maisel, L. Sandy and Walter J. Stone. 1997. “Determinants of Candidate Emergence in U.S.
House Elections: An Exploratory Study.” Legislative Studies Quarterly 22: 79-96.
Martis, Kenneth C. 1982. The Historical Atlas of United States Congressional Districts. 1789-
1983. New York: The Free Press.
McGhee, Eric. 2014. “Measuring Partisan Bias in Single-Member District Electoral Systems.”
Legislative Studies Quarterly 39(1): 55-85.
Meier, Kenneth J. and Amanda Rutherford. 2014. “Partisanship, Structure, and Representation:
The Puzzle of African American Education Politics.” American Political Science Review
108: 265-280.
Moncrief, Gary, and Joel Thompson. 1992 “Electoral Structure and State Legislative
Representation: A Research Note.” Journal of Politics 51 (1): 246–56
Moncrief, Gary, Peverill Squire, and Malcolm Jewell. 2001. Who Runs for the Legislature?
Upper Saddle River: Prentice Hall.
Municipal Year Book. 1998. Washington, DC: International City/County Management
Association.
25
Niemi, Richard G., Simon Jackman, and Laura R. Winsky. 1991. “Candidacies and
Competitiveness in Multimember Districts.” Legislative Studies Quarterly 16: 91-109.
Orey, Byron D’Andra. 2000. “Black Legislative Politics in Mississippi.” Journal of Black
Studies, 30(6): 791-814.
Papke, Leslie. E., and Wooldrige, Jeffrey M. 1996. “Econometric methods for fractional response
variables with an application to 401(k) plan participation rates.” Journal of Applied
Econometrics, 11(6): 619-632.
Parker, Frank R. 1984. “Racial Gerrymandering and Legislative Reapportionment,” in Minority
Vote Dilution. Chandler Davidson, ed. Washington, DC: Howard University Press.
Richardson, Jr., Lilliard E. and Christopher A. Cooper. 2003. “The Mismeasure of MMD:
Reassessing the Impact of Multimember Districts on Descriptive Representation in the
United States.” Unpublished Manuscript, Truman School of Public Affairs, University of
Missouri.
Schneider, Monica C. and Angela L. Bos. 2011. “An Exploration of the Content of Stereotypes
of Black Politicians” Political Psychology, 32 (2011): 205-233.
Shelby County, Ala. v. Holder 2013, 133 S. Ct. 2612, 570 U.S. 2, 186 L. Ed. 2d 651.
Smith, C. Fraser. 2008. Here Lies Jim Crow: Civil Rights in Maryland. Baltimore, MD: Johns
Hopkins University Press.
Smith, Hebert C. and John T. Willis. 2012. Maryland Politics and Government: Democratic
Dominance. Lincoln: University of Nebraska Press.
Snyder, James M. Jr., and Michiko Ueda. 2007. “Do Multimember Districts Lead to Free-
Riding? “Legislative Studies Quarterly 22: 649-679.
Sobel, Richard and Robert Ellis Smith. 2009. “Voter-ID Laws Discourage Participation,
Particularly among Minorities, and Trigger a Constitutional Remedy in Lost
Representation.” PS: Political Science and Politics 42(1): 107-110.
26
Srivastava, Virendra K. and David E. A. Giles. 1987. Seemingly Unrelated Regression Equation
Models: Estimation and Inference. New York: Marcel Dekker.
Stephanopoulos, Nicholas O. and Eric M. McGhee. 2015. “Partisan Gerrymandering and the
Efficiency Gap.” The University of Chicago Law Review 82 (2): 831-900.
Stone, Walter J. and L. Sandy Maisel. 2003. “The Not-So-Simple Calculus of Winning: Potential
U.S. House Candidates’ Nomination and General Election Prospects.” Journal of Politics
65(4): 951-977.
Stone, Walter J., L. Sandy Maisel, and Cherie D. Maestas. 2004. “Quality Counts: Extending the
Strategic Politician Model of Incumbent Deterrence.” American Journal of Political
Science 48: 479-95.
Swain, Carol M. 1993. Black Faces, Black Interests: The Representation of African Americans in
Congress. Cambridge: Harvard University Press.
Swers, Michele L. and Stella M. Rouse. 2011. “Descriptive Representation: Understanding the
Impact of Identity on Substantive Representation of Group Interests.” In Eric Schickler
and Frances Lee eds. The Oxford Handbook of the American. New York: Oxford
University Press.
Taylor, Jeffrey A., Paul S. Herrnson, and James M. Curry. 2018. “The Impact of District
Magnitude on the Legislative Behavior of State Representatives.” Political Research
Quarterly 71: 302–317.
hornburg v. Gingles, 1986. 478 U.S. 30, 106 S. Ct. 2752, 92 L. Ed. 2d 25.
Trounstine, Jessica and Melody E. Valdini. 2008. “The Context Matters: The Effects of Single-
Member versus At-Large Districts on City Council Diversity.” American Journal of
Political Science 52(3): 554-569.
U.S. Census Bureau. 2000. ‘Demographic Data Profile Search.” Census of Population and
Housing. https://www.census.gov/prod/www/decennial.html.
Verba, Sidney, Kay Lehman Schlozman, and Henry Brady. 1995. Voice and Equality: Civic
Voluntarism in American Politics. Cambridge: Harvard University Press.
27
Welch, Susan. 1990. “The Impact of At-Large Elections on the Representation of Blacks and
Hispanics.” Journal of Politics 52(4): 1050-76.
Winburn, Jonathan. 2008. The Realities of Redistricting: Following the Rules and Limiting
Gerrymandering in State Legislative Redistricting. New York: Lexington Books.
Zellner, Arnold. 1962. “An Efficient Method of Estimating Seemingly Unrelated Regressions
and Tests for Aggregation Bias.” Journal of the American Statistical Association 57(298):
348-68.
28
Table 1: The Emergence and Success of African American Candidates
Percentage of Blacks Among
Populatio
n
Primary
Candidates
Primary
Winners
General
Election
Winners
All districts
One-member district 16.60% 16.00% 10.50% 23.50%
Two-member district 19.30% 23.50% 16.30% 19.10%
Three-member district 35.30% 34.00% 29.10% 33.50%
N 5,642,116 501 270 202
Majority Black districts only
One-member district 74.64% 100.00% 100.00% 100.00%
Two-member district 60.04% 38.46% 37.50% 37.50%
Three-member district 74.16% 77.27% 78.33% 78.33%
N2,462,30
6161 70 70
Non Majority Black districts only
One-member district 14.46% 10.63% 5.56% 13.33%
Two-member district 12.93% 14.29% 11.43% 7.69%
Three-member district 21.03% 11.56% 6.20% 7.69%
N3,179,81
0340 200 132
Notes: Candidate and elections data collected by the authors. Population figures are from the
2010 U.S. Census.
29
Figure 1: The Predicted Percentages of African American Candidates in Democratic
Primary Elections
SMD 2MD 3MD
0%
5%
10%
15%
20%
25%
30%
35%
Percentage Black Candidate
Notes: Predicted values obtained using the margins command in Stata based on fractional probit
regression estimates of the percentage of Black candidates competing in the Democratic primary
presented in Appendix Table A-2. Error bars represent the 95% confidence interval around the
estimate.
30
Figure 2: The Impact of District Magnitude and Race on Primary Election Success
Three-member district
Two-member district
Single-member district
0% 20% 40% 60% 80% 100%
Maryland Nonincumbent Democrats
Non-African American African American
Average Probability of Primary Election Success
Three-member district
Two-member district
Single-member district
0% 20% 40% 60% 80% 100%
Maryland, All Democrats
Non-African American African American
Average Probability of Primary Election Success
Notes: The horizontal bars represent average effects. The lines represent 90% confidence
intervals. Results are based on logistic regression estimates presented in Appendix Table A-3.
Calculations were made by manipulating the values for African American and incumbent while
holding all other variables at their observed values (see Hanmer and Kalkan 2013) using
Herron’s (1999) simulation method. Overlapping confidence intervals indicate that there is no
statistically significant difference in the likelihood of primary election success of African
American versus non-African American Democratic candidates.
31
Figure 3. Predicted Primary Vote Shares for African American and non-African American
Candidates
SMD 2MD 3MD
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
Black Nonblack
Relative vote share
Notes: Error bars are 95% confidence intervals calculated by generating 1,000 bootstrapped
replications of the model predictions based on the model OLS estimates presented in Table A-4.
32
Figure 4: The Impact of District Magnitude and Race on General Election Success
Three-member district
Two-member district
Single-member district
0% 20% 40% 60% 80% 100%
Maryland Nonincumbent Democrats
Non-African American African American
Average Probability of General Election Success
Three-member district
Two-member district
Single-member district
0% 20% 40% 60% 80% 100%
Maryland, All Democrats
Non-African American African American
Average Probability of General Election Success
Notes: The horizontal bars represent average effects. The lines represent 90% confidence
intervals. Results are based on logistic regression estimates presented in Appendix Table A-5.
Calculations were made by manipulating the values for African American and incumbent while
holding all other variables at their observed values (see Hanmer and Kalkan 2013) using
Herron’s (1999) simulation method. Overlapping confidence intervals indicate that there is no
statistically significant difference in the likelihood of primary election success of African
American versus non-African American Democratic candidates.
33
Figure 5: Predicted General Election Vote Shares for African American and non-African
American Candidates
SMD 2MD 3MD
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
Black Nonblack
Relative vote share
Notes: Error bars are 95% confidence intervals calculated by generating 1,000 bootstrapped
replications based on the OLS regression estimates presented in Table A-6.
34
Appendix
Table A-1: Summary Statistics for the Variables Used in the Analysis
Candidate
Emergence
(Elections)
Primary
Elections
(Candidates)
General Elections
(Candidates)
Mean SD Mean SD Mean SD
Dependent variables
% African American candidates 0.23 0.33 -- -- -- --
Primary election vote share -- -- 1.00 0.56 -- --
Primary election success -- -- 0.54 0.50 -- --
General election vote share -- -- -- -- 1.09 0.27
General election success -- -- -- -- 0.75 0.43
District magnitude
Single-member district 0.30 0.46 0.10 0.30 0.14 0.35
Two-member district 0.18 0.39 0.14 0.34 0.16 0.37
Three-member district 0.52 0.50 0.76 0.42 0.70 0.46
Candidate race
African American -- -- 0.31 0.46 0.24 0.43
Other -- -- 0.69 0.46 0.76 0.43
Candidate characteristics
Incumbent -- -- 0.35 0.48 0.59 0.49
Nonincumbent -- -- 0.65 0.48 0.41 0.49
Spending advantage (Primary) -- -- 0.00 1.01 -- --
Spending advantage (General) -- -- -- -- 0.31 0.95
Electoral competition / challengers
% Contested seats -- -- 2.21 0.94 1.71 0.40
% Same-party incumbents 0.58 0.43 -- -- -- --
% Opposing-party incumbents 0.29 0.42 -- -- -- --
% Incumbents -- -- 0.35 0.25 0.79 0.27
% African American incumbents 0.19 0.35 -- -- -- --
Racial composition of district
% African American voters 0.21 0.41 0.32 0.47 0.26 0.44
Election year
2006 0.49 0.50 0.53 0.50 0.50 0.50
2010 0.51 0.50 0.47 0.50 0.50 0.50
N 126 501 270
Table A-2: The Impact of District Magnitude on Candidate Emergence
among Democratic Candidates
35
Two-member district 0.425
(0.317)
Three-member district 0.223
(0.230)
% Same-party incumbents -0.0843
(0.392)
% Opposing-party incumbents 0.216
(0.470)
% African American incumbents 1.468***
(0.334)
% African American voters 0.0255***
(0.00551)
Year 2010 0.231
(0.165)
Constant -2.334***
(0.451)
N 126
Notes: The dependent variable is the proportion of candidates in the election who are
African American. Cell entries are fractional probit regression coefficients. Standard
errors in parentheses. **p<.05 *p<.10, two-tailed tests.
36
Table A-3: The Impact of District Magnitude on the Success of
Democratic Primary Candidates
Two-member district 0.736**
(0.322)
Three-member district -0.208
(0.226)
African American -0.306
(0.446)
Incumbent 3.069**
(0.447)
Primary spending advantage 1.773**
(0.262)
% African American voters 0.001
(0.007)
% Incumbents -2.442**
(0.507)
Primary crowdedness -1.509**
(0.144)
Year 2010 0.321**
(0.145)
Constant 3.686**
(0.334)
N 501
Notes: Figures are logistic regression coefficients. Standard errors corrected
for clustering in parentheses. *p<.10 **p<.05, two-tailed tests.
Table A-4: The Impact of District Magnitude on the Relative Vote Shares of
Democratic Primary Candidates
Two-member district 0.007
(0.044)
Three-member district -0.02*
(0.012)
African American 0.038
(0.044)
African American X two-member district 0.088
(0.202)
African American X three-member district 0.107**
(0.046)
Incumbent 0.529***
(0.080)
Primary spending advantage 0.256***
(0.001)
% African American voters -0.002**
(0.001)
% Same-party incumbents -
0.503***
(0.081)
% Primary contested seats 0.01
(0.007)
Year 2010 -0.012
(0.007)
Constant 1.005***
(0.014)
Adjusted R-squared 0.524
N 501
Notes: The dependent variable is relative primary vote share, calculated as the votes
earned by each candidate in an election relative to the average number of votes earned by
all the candidates in the district. Figures are OLS regression coefficients. Standard errors
corrected for clustering in parentheses. *p<.10 **p<.05, two-tailed tests.
Table A-5: The Impact of District Magnitude on the Success of
Democratic General Election Candidates
Two-member district 1.065
(1.206)
Three-member district 2.154**
(0.971)
African American 1.285
(1.660)
Incumbent 4.810**
(1.305)
Overall expenditures 1.805**
(0.722)
% African American Voters 0.150**
(0.044)
% Incumbents -1.819
(1.572)
General election crowdedness -2.194
(1.470)
Year 2010 -1.091
(0.862)
Constant 1.294
(3.381)
N 270
Notes: Figures are logistic regression coefficients. Standard errors corrected
for clustering in parentheses. *p<.10 **p<.05, two-tailed tests.
Table A-6: The Impact of District Magnitude on Relative Vote Shares of
Democratic General Election Candidates
Two-member district 0.043
(0.045)
Three-member district 0.198**
(0.041)
African American 0.128**
(0.058)
African American X two-member district -0.003
(0.104)
African American X three-member district -0.101*
(0.055)
Incumbent 0.195**
(0.037)
Spending advantage 0.066**
(0.016)
% African American Voters 0.004**
(0.001)
% Incumbents -0.135**
(0.052)
% Contested seats 0.338**
(0.052)
Year 2010 -0.058**
(0.028)
Constant 0.231*
(0.122)
Adjusted R-squared 0.521
N 270
Notes: The dependent variable is relative general election vote share, calculated as
the votes earned by each candidate in an election relative to the average number of
votes earned by all the candidates in the district. Figures are OLS regression
coefficients. Standard errors corrected for clustering in parentheses. *p<.10
**p<.05, two-tailed tests.