Political Conditions and the Electoral Effects of Redistricting

Abstract

Redistricting can have considerable electoral consequences because it undermines the incumbency advantage. Numerous voters are drawn into districts with a different incumbent seeking reelection. With regard to vote choice, these redrawn constituents rely more on their partisanship and prevailing political conditions because they lack familiarity with their new representative. Macropartisanship, the aggregate party identification of the electorate, is an excellent barometer of the political climate and hence the partisan direction guiding voters. Because redrawn constituents have at best a tenuous bond with their new incumbent, partisan tides have more influence on their vote choice. Analyses of the 1992 and 2002 U.S. House elections show that higher district percentages of redrawn constituents significantly reduced the vote shares of southern Democratic representatives in 1992 and Democratic incumbents regardless of region in 2002. Given the stated behavioral implications associated with redistricting, these findings speak to the political conditions occurring at the time of these respective elections: a Republican realignment picking up steam in the South in 1992 and a short-term national GOP tide in the first post-9/11 midterm.

Full text

http://apr.sagepub.com/
American Politics Research
http://apr.sagepub.com/content/41/4/623
The online version of this article can be found at:
DOI: 10.1177/1532673X12464545
2012
2013 41: 623 originally published online 5 NovemberAmerican Politics Research Seth C. McKee
Political Conditions and the Electoral Effects of Redistricting
Published by:
http://www.sagepublications.com
can be found at:American Politics ResearchAdditional services and information for
http://apr.sagepub.com/cgi/alertsEmail Alerts:
http://apr.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://apr.sagepub.com/content/41/4/623.refs.htmlCitations:
What is This?
- Nov 5, 2012OnlineFirst Version of Record
- Jun 28, 2013Version of Record >>
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

American Politics Research
41(4) 623 –650
© The Author(s) 2013
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1532673X12464545
apr.sagepub.com
464545APR41410.1177/1532673X124
64545American Politics ResearchMcKee
© The Author(s) 2011
Reprints and permission: http://www.
sagepub.com/journalsPermissions.nav
1University of South Florida, St. Petersburg, FL, USA
Corresponding Author:
Seth C. McKee, University of South Florida, 140 7th Ave S, St. Petersburg, FL 33701, USA.
Email: scmckee@mail.usf.edu
Political Conditions and
the Electoral Effects of
Redistricting
Seth C. McKee1
Abstract
Redistricting can have considerable electoral consequences because it
undermines the incumbency advantage. Numerous voters are drawn into
districts with a different incumbent seeking reelection. With regard to vote
choice, these redrawn constituents rely more on their partisanship and
prevailing political conditions because they lack familiarity with their new
representative. Macropartisanship, the aggregate party identification of the
electorate, is an excellent barometer of the political climate and hence the
partisan direction guiding voters. Because redrawn constituents have at best
a tenuous bond with their new incumbent, partisan tides have more influence
on their vote choice. Analyses of the 1992 and 2002 U.S. House elections
show that higher district percentages of redrawn constituents significantly
reduced the vote shares of southern Democratic representatives in 1992
and Democratic incumbents regardless of region in 2002. Given the stated
behavioral implications associated with redistricting, these findings speak to
the political conditions occurring at the time of these respective elections: a
Republican realignment picking up steam in the South in 1992 and a short-
term national GOP tide in the first post-9/11 midterm.
Keywords
redistricting, political conditions, U.S. House elections, redrawn voters,
macropartisanship
Article
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

624 American Politics Research 41(4)
Redistricting is the spatial redistribution of voters. The simple act of relocating
a district boundary alters the representational relationship for numerous voters
and this can have considerable electoral consequences. Redrawn constituents,
those who reside in a district with a new incumbent because of redistricting, are
naturally less familiar with their representative and this bears directly on their
voting behavior. The incumbency advantage that accrues to those who cultivate
a personal vote with their constituents is severely discounted among redrawn
residents because the representative cannot appeal to these new constituents on
nonpartisan grounds since they have not yet had the opportunity to work for
them (Desposato & Petrocik, 2003; but see Boatright, 2004) and thus nurture a
home style (Ansolabehere, Snyder, & Stewart, 2000). Lacking familiarity with
their new incumbent, these redrawn constituents are more reliant on partisan
cues and the prevailing political conditions when they cast a vote.
Redistricting fosters a high degree of electoral uncertainty in the minds of
representatives (see Fenno, 1978) and it is tied directly to the presence of large
numbers of new constituents (Cain, 1984; Cain & Campagna, 1987). What can
make matters even more perilous for those who seek reelection in a redistrict-
ing cycle is the partisan tenor of prevailing political conditions. The political
climate is a complex mixture of long- and short-term forces. The immediacy of
recent events and how they impinge on presidential approval and the state of
the economy shape the electorate’s macropartisanship, which has a direct bear-
ing on election outcomes (MacKuen, Erikson, & Stimson, 1989). Likewise,
secular changes set in motion by issue evolutions occurring long ago can be
tempered by contemporary political events or they can be accelerated by short-
term conditions that reinforce a partisan realignment.
The latter scenario is exemplified by the political transformation of the
American South.1 This region of the United States finally broke in favor of the
Republican Party in the late 1960s in presidential contests, but it was not until
the 1990s that GOP support in congressional elections reached commensurate
levels. The nearly three decades-old split-level alignment, in which southern-
ers voted Republican in presidential elections and Democratic in House races
was swiftly undone by the dislocating effects of redistricting.2 For southern
Democrats, the incumbency advantage served to thwart strong Republican
challengers and maintain district-level voting majorities (E. Black &
M. Black, 2002). But in the 1990s this electoral status quo was shattered by
the presence of constituents who were poised to vote Republican when drawn
into districts with a different, and most likely, unfamiliar Democratic incum-
bent (McKee, 2010; Petrocik & Desposato, 1998).
At the same time that southern Whites were trending in favor of the
Republican Party, nationally, voters were re-sorting themselves to better fit
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 625
their partisanship with the positions advocated by the major parties
(Levendusky, 2009). Voters recognize the growing policy divide among par-
tisan elites (Ansolabehere & Jones, 2010; Hetherington, 2001), and this
enables them to pick sides. As Bartels (2000) has shown, voters have
responded by moving away from political independence and their growing
partisanship has strengthened the relationship between political affiliation
and vote choice in presidential and congressional elections. The increase in
mass partisanship since the 1970s has weakened the electoral pull of incum-
bency (Fiorina, 2005) and in turn strengthened the role of national forces in
House elections (Jacobson, 2009). Whereas in the 1970s and even in the
1980s, incumbency could considerably mollify the effects of what appeared
to be a harmful redistricting (Born, 1985; Rush, 1992, 1993), by the 1990s
and early 2000s redistricting exhibits a different electoral dynamic.
The political climate is a fundamental factor impinging on the degree to
which redistricting affects election outcomes. Redrawn voters are the elec-
toral wild card because their relatively greater unfamiliarity with the incum-
bent makes them more susceptible to prevailing political conditions. And if
conditions clearly favor a party, redrawn constituents will exhibit greater sup-
port for the party benefiting from the political climate because they are not
restrained by the personal vote most incumbents have cultivated among those
residents they have represented before redistricting.
In this study the electoral effects of redistricting are captured by evaluating
the influence of redrawn constituencies on House vote shares in 1992 and
2002. In 1992 and 2002, political conditions tilted in favor of the Republican
Party, but in the former election the effects of redistricting were only substantial
and significant in the South, where the White electorate continued to realign in
favor of the GOP. By comparison, in 2002, partisan re-sorting had all but run its
course and a more divided electorate behaved in a more predictable manner
when placed in new congressional districts. Even so, the first post-9/11 election
proved a net winner for the Republican Party because terrorism materialized as
an issue voters perceived as being owned by the GOP (Jacobson, 2003). With
this issue casting a massive shadow over the 2002 midterm, redrawn constitu-
ents broke in favor of Republican House candidates, but this had little overall
effect on partisan outcomes because so many districts were successfully drawn
to protect incumbents of both political parties (Hirsch, 2003).
This article proceeds in the following order. First, I introduce some of the
theoretical and conceptual components conditioning the relationship between
redistricting and incumbent support. Then I provide an overview of the 1992
and 2002 redistrictings and discuss how redrawn constituents are expected to
influence electoral outcomes given what we know about the state of political
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

626 American Politics Research 41(4)
conditions and macropartisanship during these elections. Next, I present the data
and methods, followed by the results of the analyses. Finally, I conclude with a
discussion of the electoral effects of redistricting that goes beyond the two elec-
tions studied here to explain how different political conditions have affected
contests under previous and more recent boundary changes.
Redistricting and Incumbent Support
Recent studies of the incumbency advantage and how it interacts with redis-
tricting have greatly improved our understanding of this relationship.
Specifically, the work of Scott W. Desposato and John R. Petrocik (Desposato
& Petrocik, 2003, 2005; Petrocik & Desposato, 1998, 2004) demonstrates that
the incumbency advantage is highly sensitive to the presence of a redistricting.
The key feature of any redistricting is the role of redrawn voters. Will these
voters, who now have a new representative because of a boundary change,
exhibit the same level of support for their new incumbent? Ceteris paribus, the
answer is no and the reason why is because of the informational deficit result-
ing from the lack of a shared history between redrawn voters and their new
incumbent. Redrawn constituents are considerably less likely to be familiar
with their representative vis-à-vis constituents who retain the same incumbent
following a redistricting (Hayes & McKee, 2009; McKee, 2008b) and this
reduces the likelihood of voting for the incumbent (Hood & McKee, 2010).
Simply put, the incumbency advantage does not apply to the vast majority
of redrawn constituents. Without the requisite time to establish a personal
vote with their new residents, it follows that a redrawn voter’s calculus relies
more heavily on partisanship and the extant political conditions shaping the
election. Indeed, as Petrocik and Desposato (2004, p. 366) claim, redrawn
voters resemble open-seat voters because “the absence of personal vote cues
that anchor a predisposition for the incumbent” makes both types of voters
more susceptible to short-term political forces. Petrocik and Desposato refer
to their conceptualization of the incumbency advantage as the “anchor
model” because incumbency serves to reduce the pull of short-term factors
that may favor the party of the challenger. Indeed, lacking the anchoring
effect of incumbency, redrawn voters should drift with the prevailing partisan
tide. There is empirical support for this theory of redrawn voter behavior in
the American South (Petrocik & Desposato, 1998), California (Desposato &
Petrocik, 2003), and national U.S. House elections (McKee, 2008a).
It is of course true that political conditions may not significantly favor one
party over another and thus the electoral impact of a redistricting—in the form
of redrawn voting behavior, should exhibit no significant partisan bias. Instead,
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 627
partisanship and challenger viability should be the most relevant factors for
redrawn voters during a politically neutral environment. And since redistrict-
ing (at least in the short term) eliminates the representational relationship for
numerous voters, quality candidates are more likely to mount a challenge in
a redistricting year (Hetherington, Larson, & Globetti, 2003). Historically,
smoothing over the vicissitudes of specific elections, we see that incumbents
perform worse among their redrawn constituents (Ansolabehere et al., 2000);
no doubt a consequence of their new residents relying more on their partisan
inclinations and greater susceptibility to the appeals of strong challengers,
who are more likely to emerge at those times when incumbents experience a
boundary change.
Thus, controlling for other factors, we can posit two general claims regard-
ing the effects of redistricting on incumbent vote shares: (a) under politically
neutral conditions (no party is clearly advantaged at the time of the election)
incumbents of both parties should receive less support from redrawn voters
and (b) when the political climate favors one party, redrawn voters will only
serve to reduce the vote shares of incumbents who are affiliated with the
disadvantaged party. With respect to the first claim, because redrawn voters
lack the personal vote cue (the anchoring effect of incumbency) they will be
less supportive of their new incumbent regardless of that representative’s
party affiliation.3 As for the second claim, redrawn voters will be much less
supportive of incumbents who represent the party harmed by political tides
because conditions push these voters toward candidates of the favored party.
Conversely, redrawn voter support for incumbents of the advantaged party
may be no different from the support given by same-incumbent voters (a null
finding) because the partisan tide will push redrawn voters toward these
incumbents at a level of support exceeding their partisanship—a boost that is
roughly tantamount to the personal vote. The next section of the article
explains why there are reasons to expect that the second claim is borne out in
both the 1992 and 2002 U.S. House elections.
Redistricting in 1992 and 2002
There is evidence that both the 1992 and 2002 House elections occurred when
short-term political conditions favored the Republican Party. The GOP netted
10 seats in 1992 and 6 in 2002 (Jacobson, 2009). The 2002 elections were only
the third midterm since the Civil War in which the president’s party gained
seats. Many scholars view the 1992 House elections as breaking in favor of the
GOP (see Abramson, Aldrich, & Rohde, 1994; McKee, 2008a; the American
South in the case of Petrocik & Desposato, 1998). The anti-incumbent mood
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

628 American Politics Research 41(4)
disadvantaged Democrats (see Abramson et al., 1994; Gaddie & Bullock,
2000) since they were the majority party in Congress.4 The House Bank scan-
dal proved more detrimental to Democratic incumbents because they were
more likely to have bounced checks and, for this reason, Republican Newt
Gingrich made it a campaign issue (Jacobson, 2009). Finally, the large num-
ber of newly drawn majority-minority districts concentrated the most loyal
Democratic voters in a smaller number of districts and this concomitantly
increased the number of Republican voters in adjacent districts (E. Black &
M. Black, 2002; Lublin, 1997). However, since most new majority Black dis-
tricts were located in southern states (Epstein & O’Halloran, 2000) and all but
one of the 10 Republican seat gains were in the South, whatever GOP redis-
tricting advantage there was, may have been confined to this region.
Despite a large number of incumbent protection plans (Forgette & Platt,
2005), overall, the 2002 redistricting aided the GOP (see Hirsch, 2003; Jacobson,
2009, pp. 8-11; La Raja, 2009; McCarty, Poole, & Rosenthal, 2009; Schweers,
2003). As Jacobson points out, the states that gained seats through reapportion-
ment were heavily Republican while the states that lost districts favored the
Democratic Party. This is not a surprise considering the ongoing population
movement and growth away from the Democratic Frost Belt to the Republican
Sun Belt. What really allowed Republicans to capitalize on reapportionment
was their success in winning districts in those states where the GOP controlled
redistricting and to a lesser degree in those states with divided partisan control
(see Jacobson, 2009, p. 9, Table 2-1). In addition to a favorable redistricting,
short-term forces swung in favor of the GOP. The president’s high approval rat-
ing, Republican candidates’ success in their efforts to “own” the terrorism issue
(Jacobson, 2003), and the fact that a sluggish economy failed to gain electoral
traction (Jacobson, 2009) presaged a Republican edge in the 2002 midterms.
Unlike the nation at large, in the South its ongoing realignment hastened
with the 1992 congressional elections. Whereas 9 of the 11 ex-Confederate
states had Democratic plans in 1992, only four southern states advanced
Democratic objectives in 2002 (see McDonald, 2004). The prevalence of
southern Democratic control of the redistricting process in 1992 belies the
party’s declining status, which is made evident by two important and elector-
ally consequential realities. First, with the exception of Arkansas and
Tennessee, southern states are covered by the Voting Rights Act and of par-
ticular importance, Section 5, which dictates federal preclearance of pro-
posed boundary changes. Second, 1992 denotes the initial point of Republican
ascendancy in southern U.S. House elections (McKee, 2010). This was the
first election in a string of three in which southern Republicans made extraor-
dinary gains in House contests.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 629
As the agency in charge of reviewing district plans, the Department of
Justice (DOJ) leaned hard on several southern states to greatly increase,
wherever geographically possible, their number of majority-minority dis-
tricts (Bullock, 2010; Butler, 2002; Cunningham, 2001). To comply with the
DOJ and hopefully limit the electoral damage to White southern Democrats
representing districts bordering newly created majority Black districts, south-
ern Democratic-controlled legislatures drew extremely convoluted maps,
shocking in their complexity to the naked eye (see the examples provided by
Monmonier, 2001).
Table 1. Percentage of Redrawn Constituents for U.S. House Incumbents Seeking
Reelection in 1992 and 2002.
Median
redrawn (%)
Mean
redrawn (%)
Max
redrawn (%) N
1992 Elections
All incumbents 19 24 309
Democrats 19 23 95 192
Republicans 21 26 100 117
Southern incumbents 18 26 82
Democrats 15 20 69 57
Republicans 37 39 100 25
Nonsouthern incumbents 19 24 227
Democrats 21 25 95 135
Republicans 18 23 100 92
2002 Elections
All incumbents 20 23 296
Democrats 19 21 76 149
Republicans 23 24 97 147
Southern incumbents 17 22 65
Democrats 12 16 54 30
Republicans 25 27 84 35
Nonsouthern incumbents 20 23 231
Democrats 20 23 76 119
Republicans 22 23 97 112
Note: These data include only contested (Democrat vs. Republican) districts with incumbents
running for reelection in 1992 and 2002. Redrawn is the district percentage of constituents
an incumbent inherited after redistricting. The variable was calculated based on data from the
following links: http://mcdc2.missouri.edu/websas/geocorr90.shtml; http://mcdc2.missouri.edu/
websas/geocorr2k.html.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

630 American Politics Research 41(4)
Recognizing the danger of saddling their own incumbents with a large per-
centage of new constituents, southern Democratic mapmakers made sure that
Republican incumbents would receive a much greater percentage of redrawn
residents.5 Table 1 documents the percentage of redrawn constituents for
incumbents in 1992 and 2002, at the national level and regional level (South
and non-South) and according to each party (Democrat or Republican). The
notable partisan disparity in the percentage of redrawn constituents is found in
the South. In 1992 and 2002, southern Republicans represented districts whose
constituent populations were much more likely to be altered vis-à-vis their
Democratic counterparts. Whether it is the median, mean, or maximum, the
percentage of redrawn constituents was markedly higher for southern
Republicans and this was especially true in 1992. Based on measures of cen-
tral tendency, differences in the percentage of redrawn constituents outside the
South according to an incumbent’s party are hardly detectable in 1992 and
2002. In both 1992 and 2002, the higher national percentage of redrawn con-
stituents in Republican districts is being driven by the greater number of new
constituents placed in districts represented by southern Republicans.
Table 2. Macropartisanship of Voters in U.S. House Elections.
Year 1990 1992 Δ(1992-1990) 2000 2002 Δ(2002-2000)
All voters (%)
Democrats 50 51 +1 50 47 −3
Republicans 35 40 +5 43 49 +6
Independents 15 9 −6 8 4 −4
N(1,014) (1,694) (1,173) (951)
Dem minus Rep +15 +11 −4 +7 −2−9
Southern voters (%)
Democrats 54 54 0 47 39 −8
Republicans 28 37 +9 45 57 +12
Independents 18 9 −9 8 5 −3
N(370) (458) (406) (266)
Dem minus Rep +26 +17 −9 +2 −18 −20
Nonsouthern voters (%)
Democrats 48 50 +2 52 50 −2
Republicans 39 41 +2 41 46 +5
Independents 13 9 −4 7 4 −3
N(644) (1,236) (767) (685)
Dem minus Rep +9 +9 0 +11 +4 −7
Note: These data were calculated by the author from the American National Election Studies (ANES)
cumulative file. Independent leaners are classified as partisans. Data include only those respondents who
claimed they voted in the election. Similar results were obtained when independent leaners are classified
as independents. Likewise, exit poll data for these years also yield similar results (the three-category
classification for political affiliation in the exit poll data is Democrat, Republican, and Independent).
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 631
To limit the electoral effects of the less predictable behavior of redrawn
voters, southern Democrats redistricted the lion’s share of these constituents
into Republican-held districts. But they did not go far enough. Democratic line
drawers were hamstrung by the DOJ because its insistence on having southern
states maximize its number of majority-minority districts assured that many
White Democratic incumbents would still be faced with a substantial number
of redrawn voters. Nonetheless, the redistribution of constituents into districts
with different incumbents would have altered only a handful of election out-
comes if these voters’ partisan preferences were static. They were not.
In an insightful analysis, Petrocik and Desposato (1998) demonstrate with
American National Election Studies (ANES) data on southerners’ House vote
choices, that even a substantial reduction in the Black percentage of the dis-
trict population would not be enough to defeat the typical Democratic incum-
bent, if the political environment was in fact neutral. But Petrocik and
Desposato contend that in the South, in 1992 and 1994, partisan tides were
moving decidedly in a Republican direction and under these conditions,
redrawn constituents were at the vanguard of shifting their preferences in
favor of Republican House candidates. The greater exposure of southern
Republican incumbents to the presence of new voters was not electorally det-
rimental because in this region of the country, the ongoing Republican realign-
ment was bolstered by the presence of short-term GOP tides in 1992 and 2002.
Macropartisanship in the 1992 and 2002 House
Elections
The most direct empirical support for the potential influence of political condi-
tions at the time of the 1992 and 2002 elections is illustrated by data on the
macropartisanship of House voters presented in Table 2. To provide evidence
of change in the macropartisanship of the voting electorate, data are presented
for 1990 and 1992, and also for 2000 and 2002. ANES data on the party iden-
tification of respondents who claimed to have voted in these elections are
presented for all voters, southern voters, and nonsouthern voters. For the three
categories of Democrat, Republican, and Independent, those independents who
lean toward a party are coded as identifiers of that party. The results are sub-
stantively the same if these independents are classified as such but since their
voting preferences are more in line with weak identifiers (Keith et al., 1992),
they are designated as partisans. The last row for each set of voters displays the
difference in the percentage of partisans—Democrats minus Republicans.
Finally, two columns show the difference in each row of partisan category
between the consecutive election years (1990 vs. 1992 and 2000 vs. 2002).
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

632 American Politics Research 41(4)
The data in Table 2 once again indicate a very different set of political
conditions in 1992 and 2002. In 1992 the shift in the macropartisanship of the
voting electorate since 1990 is confined to voters in the South, whereas out-
side the region neither party is advantaged by the changing distribution of
identifiers. In the South the formidable advantage in Democratic identifica-
tion is constant between 1990 and 1992: 54% of southern voters call them-
selves Democrats. Starting with a much lower base in 1990, however, there is
a considerable increase in the percentage of southern voters who identify as
Republicans between 1990 and 1992, going from 28% to 37% of the voting
electorate. This large increase in Republican identifiers serves to reduce the
overall Democratic identification advantage in the South by 9 points, narrow-
ing the Democratic surplus from 26 to 17 percentage points in 1992. By com-
parison, for nonsouthern voters, the percentage of Democrats and Republicans
both increase 2 percentage points. In fact, the only commonality in the pat-
tern of macropartisanship between voters in and outside the South, is the
decline in the share of Independents—a pattern that persists through the 2002
elections.
According to the macropartisanship data for 2000 and 2002, the changing
distribution of partisanship favors the Republican Party in and outside the
South although the GOP advantage is much more pronounced in the latter
region. Among all voters, Democratic identification declines 3 percentage
points between 2000 and 2002 and increases 6 points for Republicans, who
now comprise a slightly larger share of the voting electorate (49% Republicans
vs. 47% Democrats). A 7-point Democratic identification advantage in 1990
becomes a 2-point deficit by 2002.
For nonsouthern voters the Republican shift is evident as the percentage of
voters identifying with the GOP increases 5 points, whereas Democratic
identifiers decline 2 points between 2000 and 2002. In the case of southern
voters, we see a mature stage of the Republican partisan realignment in 2002.
The percentage of Democratic voters drops 8 points in 2002 compared to
2000, and a remarkable 15 points since 1992. Even more dramatic is the
increase in the share of Republican voters, who jump from 45% in 2000 to
57% by 2002, more than doubling the total percentage of Republicans since
1990. What was a slim Democratic identification advantage of 2 points in
2000 is rendered an 18-point Republican identification advantage in 2002—
an incredible 20-point Republican shift in the span of 2 years. Finally, irre-
spective of region, the decline in the share of Independents is evident,
demonstrating partisan re-sorting all across the country.
Figure 1 further supports the evidence showing that only in the South did
voters move toward the GOP in their party affiliation in 1992, whereas voters
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 633
in and outside the region both shifted toward the Republican Party in 2002.
The figure graphically displays the percentage difference between Democratic
and Republican identifiers based on exit poll data for the 1990, 1992, 2000,
and 2002 elections. A downward slope between consecutive elections (e.g.,
1990 vs. 1992) indicates a trend in favor of the GOP. Between 1990 and 1992
the decline in Democratic voters versus Republican voters in the non-South
was a mere 0.4 percentage points whereas in the South the decline was a more
notable 2.7 points in the Republican direction. By contrast, between 2000 and
2002, gains in Republican voters are obvious among southern and nonsouth-
ern voters. In the non-South the partisan balance moves 3.5 percentage points
in favor of the GOP and in the South a 3.1 percentage point Democratic mar-
gin in 2000 becomes a 9-point deficit favoring the Republican Party in 2002.
The next section turns to an examination of the effects of redrawn con-
stituents on House elections in 1992 and 2002. Multivariate analysis allows
Figure 1. Macropartisanship by region: 1990 versus 1992, and 2000 versus 2002.
Note: These data are from the national exit polls for 1990 and 1992 (Voter Research
and Surveys), and 2000 and 2002 (Voter News Service). Data are weighted and show the
difference in the percentage of Democratic versus Republican identifiers. The percentage of
Democrats and Republicans in the South, respectively, for 1990, 1992, 2000, and 2002 was as
follows: Democrats = 41.3, 41.7, 41.8, and 36.9; Republicans = 33.2, 36.3, 38.7, and 45.9. The
percentage of Democrats and Republicans in the Non-South, respectively, for 1990, 1992,
2000, and 2002 was as follows: Democrats = 37.5, 38.7, 39.8, and 40.1; Republicans = 34.5,
36.1, 35.3, and 39.1. In the South the corresponding partisan disparities were 8.1, 5.4, 3.1, and
–9.0. In the Non-South the corresponding partisan disparities were 3.0, 2.6, 4.5, and 1.0.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

634 American Politics Research 41(4)
us to determine whether the presence of redrawn residents exhibited an elec-
toral effect on vote shares in these election cycles.
Data, Method, and Results
The research design is set up to capture the effects of redistricting at the time
of the 1992 and 2002 House contests. Because redrawn constituents are no
longer tied to a familiar incumbent, they are in effect barometers of the
political environment at the time of the election. Hence, if political condi-
tions favor a party, then redrawn constituents should reflect this in their vot-
ing behavior as their preferences shift with the partisan tide.6 The underlying
theory is tied directly to the effect of macropartisanship on voting behavior.
To be clear, a shift in macropartisanship that decidedly benefits one of the
major parties will produce an across-the-board swing in voter preferences
toward the favored party,7 but redrawn voters (other factors constant) will be
the most susceptible to voting in accordance with the short-term partisan tide
because they are not constrained by the personal vote. In other words, it is
their lack of a relationship with their new incumbent that makes redrawn
voters the most likely to vote for the candidate affiliated with the party ben-
efiting from favorable short-term political conditions.
After controlling for several relevant factors that affect House vote share, the
key relationship is between the district percentage of redrawn constituents and
the party affiliation of the incumbent (Democrat or Republican). By controlling
for other variables, if there is an independent effect of redrawn constituents on the
House vote according to an incumbent’s party, it is due to the prevailing political
climate at the time of the 1992 and 2002 elections. If the percentage of redrawn
constituents only reduces the support given to Democratic incumbents then this
suggests that conditions in these elections benefited the Republican Party.
Because it is argued that 1992 and 2002 were elections with political conditions
benefiting the GOP, redrawn voters should only depress the votes of Democratic
incumbents. This hypothesis is tested with an interactive model so that
Republican House Vote = β0 + β1 Redrawn + β2 Democratic Incumbent
+ β3 Redrawn × Democratic Incumbent + βΧ + ε.
Redrawn = the district percentage of constituents who are new to the incum-
bent as a result of redistricting. Democratic Incumbent = 1 if the respondent’s
district has a Democratic incumbent seeking reelection and 0 if a Republican
representative runs for another term. Χ is a vector containing the remaining
control variables and ε is the error term. Finally, Redrawn × Democratic
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 635
Incumbent is the main variable of interest. It is expected that the interaction
will be positive and statistically significant. This means that an increase in
redrawn constituents for Democratic incumbents increases the Republican
share of the two-party House vote. Thus, given a political climate that favors
the Republican Party, lacking the anchoring effect of incumbency, redrawn
constituents should decrease the votes of Democratic incumbents.
The district-level data are from three sources: (a) The Almanac of American
Politics (Barone & Cohen, 2003; Barone & Ujifusa, 1993), (b) Congressional
Quarterly (Congressional Districts in the 1990s: A Portrait of America;
Congressional Districts in the 2000s: A Portrait of America; CQ’s House
Ratings), and (c) The Missouri Census Data Center. The Missouri Census
Data Center has a transposition of population tool that allows one to deter-
mine the percentage of redrawn constituents assigned to each House incum-
bent seeking reelection in the 1992 and 2002 elections.8
Ordinary least squares (OLS) models are used to evaluate the district-level
effect of redistricting on Republican vote shares. There are a total of nine
multiple regressions. The first three regressions present national results for
1992, 2002, and the two elections pooled. Because the results of the national
models may in fact mask an effect limited to the South, the next six regres-
sions display the results for the South and the non-South separately. As dis-
cussed, the variable of interest is the interactive term Redrawn × Democratic
Incumbent. The coefficient is expected to be positive and significant so that
an increase in the percentage of redrawn residents for Democratic representa-
tives increases the Republican House vote.
All analyses are limited to contested races (Democrat vs. Republican)9 with
an incumbent10 seeking reelection, and include the following controls: the
Republican two-party percentage (1992 or 2000) of the presidential vote, per-
centage Black voting age population, Congressional Quarterly’s measure of
district competitiveness scaled from safe Democrat to safe Republican (0 = safe
Democrat, 1 = favors Democrat, 2 = leans Democrat, 3 = no clear favorite, 4 =
leans Republican, 5 = favors Republican, 6 = safe Republican), the total change
in the number of House seats for each state, median household income (in thou-
sands), indicators for the political party in control of a state’s congressional
redistricting11 (dummies for Republican plans and Democratic plans, with the
base category consisting of bipartisan or neutral plans), a dummy for South (1 =
11 former Confederate states, 0 = otherwise) in the national models, and a year
dummy (1 = 1992, 0 = 2002) is included for all of the pooled models.12
Table 3 presents the estimates for the national models in 1992, 2002, and the
elections pooled. As anticipated, the interaction term Redrawn × Democratic
Incumbent is positive and significant, providing evidence that Democratic
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

636 American Politics Research 41(4)
incumbents were placed at greater electoral risk as a consequence of represent-
ing redrawn constituents. The effect is largest in 2002 and the region dummy
(“South”) is not significant for this year, suggesting there was a national
Republican tide in this midterm. By comparison, in 1992 the South dummy is
positive and significant, which indicates that the Republican House vote was
notably higher in this region. The findings from Table 3 warrant further investi-
gation by segmenting the analysis on the basis of region: South and non-South.
Table 4 presents the results for the South and the non-South, respectively,
in 1992, 2002, and both contests pooled. By parsing the analysis on the basis
of region we see that in 1992 it was only in the South where Democratic
incumbents were negatively affected by the presence of redrawn constituents.
The interaction is easily significant in the South in 1992, but in the non-South
there is no significant effect. Apparently, in the 1992 redistricting, political
conditions favoring the GOP only manifested themselves in southern House
elections. In 2002, and in the pooled models, regardless of region, we see that
an increase in redrawn constituents for incumbent Democrats increased the
Republican House vote.
Table 3. The Effect of Redistricting on House Vote Shares in 1992, 2002, and Pooled
Contests.
1992 2002 Pooled
Variables of interest
Redrawn constituents (%) −.014 (.029) −.024 (.021) −.022 (.018)
Democratic incumbent −.075 (.019)*** −.113 (.054)* −.085 (.020)***
Redrawn constituents × Democratic incumbent .092 (.036)** .115 (.035)*** .105 (.025)***
Controls
Republican presidential vote (%) .279 (.056)*** .424 (.047)*** .356 (.036)***
Black voting age population (%) −.175 (.035)*** −.037 (.037) −.108 (.025)***
Competitiveness (safe D to safe R) .037 (.003)*** .033 (.009)*** .037 (.003)***
ΔTotal seats in state −.003 (.001)** .0004 (.0049) −.002 (.001)*
Median household income (thousands) .0004 (.0005) .0007 (.0004)* .0005 (.0003)
Republican plan −.034 (.013)** −.014 (.011) −.021 (.008)**
Democratic plan −.0002 (.0089) −.007 (.010) −.001 (.007)
South .022 (.010)* .002 (.014) .016 (.007)*
1992 election -- -- −.018 (.007)**
Constant .265 (.035)*** .214 (.058)*** .239 (.027)***
R2.86 .92 .89
N309 296 605
Note: OLS coefficients with robust standard errors (clustered on the district in the pooled
models) in parentheses. The dependent variable is the Republican percentage of the two-party
U.S. House vote. These data include only contested races (Democrat vs. Republican) with an
incumbent seeking reelection.
*p < .05. **p < .01. ***p < .001, one-tailed tests.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

637
Table 4. The Effect of Redistricting on House Vote Shares, by Region, in 1992, 2002, and Pooled Contests.
Southern districts Nonsouthern districts
1992 2002 Pooled 1992 2002 Pooled
Variables of interest
Redrawn constituents (%) −.049 (.054) .008 (.049) −.039 (.037) .000 (.035) −.040 (.021)* −.020 (.020)
Democratic incumbent −.089 (.026)*** −.265 (.123)* −.138 (.046)** −.065 (.025)** −.032 (.030) −.062 (.019)***
Redrawn constituents × Democratic incumbent .153 (.059)** .206 (.122)* .168 (.048)*** .055 (.044) .105 (.033)** .081 (.028)**
Controls
Republican presidential vote (%) .174 (.106) .435 (.086)*** .257 (.075)*** .351 (.071)*** .440 (.057)*** .415 (.043)***
Black voting age population (%) −.140 (.067)* .091 (.078) −.040 (.057) −.144 (.042)*** −.066 (.035)* −.097 (.027)***
Competitiveness (safe D to safe R) .037 (.004)*** .010 (.019) .031 (.007)*** .036 (.004)*** .047 (.006)*** .039 (.003)***
Δ Total seats in state .009 (.007) .006 (.011) .010 (.005)* −.004 (.001)*** −.004 (.005) −.004 (.001)**
Median household income (thousands) .0018 (.0009)* .0014 (.0007)* .0014 (.0005)** .0002 (.0005) .0005 (.0004) .0002 (.0003)
Republican plan -- .002 (.022) .005 (.023) −.038 (.014)** −.033 (.012)** −.035 (.009)***
Democratic plan .013 (.022) −.015 (.019) .008 (.013) −.007 (.012) −.000 (.012) −.005 (.008)
1992 election -- -- −.003 (.015) -- -- −.024 (.007)***
Constant .287 (.085)*** .287 (.145)* .280 (.077)*** .232 (.043)*** .146 (.032)*** .207 (.026)***
R2.88 .91 .88 .86 .93 .90
N82 65 147 227 231 458
Note: OLS coefficients with robust standard errors (clustered on the district in the pooled models) in parentheses. The dependent variable is the Republican percentage
of the two-party U.S. House vote. These data include only contested races (Democrat vs. Republican) with an incumbent seeking reelection.
*p < .05. **p < .01. ***p < .001, one-tailed tests.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

638 American Politics Research 41(4)
One way to demonstrate the size of the effect of the interaction term is to
calculate the Democratic share of the House vote for Democratic incumbents
given various increments of the district percentage of redrawn constituents.
Because interpretation of interaction terms is not intuitive (see Brambor,
Clark, & Golder, 2006), assigning values for the constituent parts of the vari-
ables of interest and determining the corresponding two-party vote provides a
clearer sense of the substantive effects of redrawn constituents on Democratic
incumbent vote shares. All of these calculations are based on simulations pro-
duced by CLARIFY (Tomz, Wittenberg, & King, 2003).13 Table 5 displays the
two-party share (%) of the Democratic incumbent vote, given select percent-
ages of redrawn constituents (0% to 100% in 20-point intervals; the difference
in the vote based on one standard deviation, the median redrawn, and the
actual maximum percentage of redrawn constituents based on the data shown
in Table 1).
Setting the other variables at their means, in the South in 1992 the
Democratic House vote for Democratic incumbents is 57% with no redrawn
constituents, and 47% Democratic with 100% redrawn constituents. In 2002
in the non-South the Democratic House vote for Democratic incumbents is
54% with no redrawn constituents, and 47% Democratic with 100% redrawn
Table 5. The Two-Party Vote for Democratic Incumbents Based on the Percentage
of Redrawn Constituents.
Percentage of redrawn constituents
Democratic vote 0% 20% 40% 60% 80% 100% Max Δ± 1 SD Median True Max
Nation 1992 57 56 54 53 51 50 7 1.6 56 50
Nation 2002 57 55 53 52 50 48 9 1.6 55 50
Non-South 2002 54 52 51 50 48 47 7 1.2 52 49
South 1992 57 55 53 51 49 47 10 2.4 55 50
South 2002 63 58 54 50 45 41 22 3.9 60 51
Note: Predicted probabilities were calculated from the models in Tables 3 and 4 using CLARIFY
(Tomz, Wittenberg, & King, 2003). All entries were produced at the 95% confidence level
with the remaining variables set at their means. Based on the entry for “South 2002,” moving
from a district with no redrawn constituents (0%) to an entirely redrawn district (100%), the
Democratic incumbent’s vote goes from 63% to 41%—a maximum difference in the two-
party vote of 22 percentage points. In addition, a one standard deviation in the percentage of
redrawn constituents (19% redrawn) moves the Democratic incumbent’s vote by just under
4 percentage points (3.9). Finally, given the median redrawn percentage of constituents (12%
redrawn), the Democratic incumbent vote is 60% and in the case of the Democratic district
with the actual highest percentage of redrawn constituents (54% redrawn), the Democratic
incumbent’s share of the two-party vote is 51%.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 639
constituents. Finally, in the South in 2002 the Democratic House vote for
Democratic incumbents is 63% with no redrawn constituents, and it plum-
mets to 41% Democratic with 100% redrawn constituents.14
Figure 2 provides a more comprehensive look at the effect of redrawn
constituents on the Democratic incumbent vote in the South in 1992 and in
the South and non-South in 2002. The y-axis plots the Democratic share of
the House vote for Democratic incumbents according to the percentage of
redrawn constituents plotted on the x-axis. It is clear from all three plotted
lines that Democratic incumbents lose vote shares as their district percentage
of redrawn constituents increases. The largest effect is found among southern
Democratic incumbents in 2002. Nonetheless, redistricting proved less of an
impediment to reelection in 2002 because incumbents were more electorally
insulated. For example, compared to 1992, in 2002 the median share of the
two-party House vote for Democratic incumbents seeking reelection was
higher and their vote share standard deviations were lower.15
Discussion and Conclusion
This research shows that redrawing congressional boundaries creates a dynamic
electoral environment due to the responsiveness of redrawn voters. Depending
on political conditions, representatives may find their new residents to be either
Figure 2. The effects of redistricting on democratic incumbent vote shares.
Note: These results were calculated from the models in Table 4 using CLARIFY (Tomz,
Wittenberg, & King, 2003).
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

640 American Politics Research 41(4)
friends or foes. The analysis of district-level data show asymmetric support for
U.S. House incumbents during the last two decennial redistrictings. In the South
in 1992 and nationwide (South and non-South) in 2002, redrawn constituents
depressed the votes of Democratic incumbents because the political climate
favored the Republican Party. Without the anchoring effect of incumbency,
redrawn voters shifted in the direction of partisan tides—making them less sup-
portive of Democrats (Petrocik & Desposato, 2004).
Previous research has not explicitly addressed or made clear the regional
distinction between the electoral effects of congressional redistricting in the
1992 and 2002 contests. The 1992 redistricting triggered a swift Republican
rise in southern House elections after three consecutive election cycles in
which the GOP was stuck holding one third of the region’s seats (McKee,
2010). As shown in this study, partisanship data indicate that only in the South
does the voting electorate shift toward affiliating with the GOP in the 1992
House elections (see Table 2 and Figure 1), and this evidence of changing
macropartisanship is the grounds for explaining why only in the South does a
higher percentage of redrawn constituents reduce the votes of Democratic
incumbents.
Furthermore, the shift in the South is not a short-term phenomenon, rather
the Republican tides running through Dixie in 1992 and 1994 (see Petrocik &
Desposato, 1998) were reinforced by the secular Republican realignment of
southern Whites underway since the 1960s and carried forth primarily
through generational replacement (Green, Palmquist, & Schickler, 2002). In
short, southerners primed to vote Republican in 1992 were part of an endur-
ing movement in favor of the Republican Party and short-term political con-
ditions favoring the GOP ensured that redistricting would accentuate this
voting behavior.
Compared to 1992, however, the 2002 elections occurred at a time when
national political conditions swung in favor of the Republican Party. There is
no question that the 2002 redistricting generally protected incumbents from
serious competition, as evidenced by the average vote shares of representa-
tives who ran for reelection, the small number of incumbent defeats and
retirements, and the remarkably small number of districts rated competitive
by Congressional Quarterly (see Jacobson, 2009). Nonetheless, despite the
greater protection afforded both party’s incumbents in 2002, redrawn con-
stituents were much less supportive of Democrats. And contrary to popular
commentary that suggests redistricting makes incumbents immune to
defeat—under the right political conditions no gerrymander is foolproof. For
the sake of comparison, given the prevailing political climate during the 2006
elections, if a national redistricting had been implemented for this midterm
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 641
then redrawn constituents would have severely punished Republican incum-
bents and likewise, Democratic losses in 2010 would have been even more
acute under altered district boundaries.
The 2002 congressional elections highlight a short-term political environ-
ment favoring the Republican Party. To be sure, since 1992 White southern-
ers have been moving to the GOP in House races and partisan re-sorting in
and outside the region was steadily progressing, but only in the South are the
2002 House races occurring against the backdrop of a long-term trend in
favor of the Republican Party. In the non-South, 2002 merely constitutes a set
of short-term political conditions (i.e., the high approval rating of President
Bush, Republican control of the terrorism issue) benefiting Republicans, and
the political behavior of redrawn constituents responds accordingly by turn-
ing against Democratic representatives. In the absence of a long-term shift
toward one of the major political parties, outside the South short-term politi-
cal conditions will increase electoral volatility in a redistricting cycle. With
rough parity in the aggregate in terms of Democratic and Republican identi-
fiers, coupled with a recently growing number of political independents,16
redistricting will exacerbate wins and losses according to which party is the
beneficiary of ephemeral partisan tides (like those favoring Democrats in
2006 and 2008 and Republicans in 2010).17
Redistricting has always been controversial because it has the potential to
skew electoral outcomes in a direction that comports with the intentions of
mapmakers. But intentions and actual results can be two very different things
(on this point see Hill, 1995; Petrocik & Desposato, 1998; Rush, 1992, 1993,
2000). Indeed, before gerrymander entered the American lexicon thanks to
the shenanigans in Massachusetts in 1812 (see Cox & Katz, 2002), “enemies
of the Federal Constitution in Virginia” in 1788 used redistricting “to prevent
the election of James Madison to the first Congress, and fortunately it was
unsuccessful” (Fiske, 1890, pp. 216-217). Fast forward to 1986 in the case of
Davis v. Bandemer, in which the Supreme Court ruled partisan gerrymander-
ing justiciable in congressional elections, and we find ironically enough that
the Indiana state legislative districts that prompted the ruling were intended
to yield large Republican gains, but Indiana Democrats actually netted more
seats (Grofman, 1990, p. 39).18
As a last example of intentions failing to match results, once again con-
sider congressional redistricting plans enacted by Democratic-controlled leg-
islatures in the South for the 1992 elections (e.g., Georgia and North Carolina).
Remarking on how well these boundary changes benefited Republicans,
Grofman and Brunell (2005, p. 184) coined the term dummymander: “a ger-
rymander by one party that, over the course of the decade, benefits the other
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

642 American Politics Research 41(4)
party, and actually looks as if it was designed by that party rather than the
party in power.” The one thing all of these failed gerrymanders have in com-
mon is that political conditions countered partisan intentions.
With respect to electoral objectives, whether one advocates for redistrict-
ing plans that foster competition (e.g., Mann & Cain, 2005; McDonald, 2006)
or instead protect incumbents (see Brunell, 2006, 2008)—it is successful ger-
rymanders that alarm critics of this institutional device. For instance, the
2003 Texas congressional redistricting has become the modern exemplar of
how boundary changes can manufacture substantial partisan seat gains. But
whether successful or not, what should raise concern is the increasing will-
ingness of partisan elites to consider mid-decade redistricting as a means to
gain seats (Levitt & McDonald, 2007). In addition to being costly, a time
consuming diversion from substantive legislative business, and a hyperpoliti-
cal action (Bullock, 2010), mid-decade redistricting further disrupts the rep-
resentational linkage between voters and officeholders. And this is
demonstrated most clearly when partisan gerrymanders actually work, that is,
when political conditions jibe with political intent.
Acknowledgments
The author thanks the editor, anonymous reviewers, Ray Block, Gary Jacobson, Chad
Murphy, and Antoine Yoshinaka for numerous suggestions that have greatly improved
this article.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publica-
tion of this article.
Notes
1. The South refers to the 11 states that seceded from the Union: Alabama, Arkan-
sas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina,
Tennessee, Texas, and Virginia.
2. To be clear, an ongoing Republican realignment was accelerated by the favorable
electoral effects of redistricting.
3. Since partisanship is a more important factor in the vote calculus of redrawn vot-
ers, if redrawn voters are more likely to share the party affiliation of their new
representative as compared to same-incumbent voters, then their support for the
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 643
incumbent could certainly be higher under politically neutral conditions. Partisan
gerrymanders have their greatest likelihood of meeting expectations when short-
term forces are neutral and the partisan loyalties of voters are identifiable because
the political behavior of redrawn voters is more predictable under these condi-
tions.
4. At the time, public opinion of the Democratic-controlled Congress in 1992
reached its nadir for the entire time series measuring approval of this institution
(see Stimson, 2004). Furthermore, Stimson’s quarterly data on the public mood—
a massive aggregation of public opinion data tracking movement of the American
public along a liberal to conservative policy continuum shows that in the fourth
quarter of 1992, the public mood was moving in a conservative direction. In 1992
the percent liberal for the public policy mood dropped from 66.7 in the third quar-
ter to 63.1 in the fourth quarter (data are available on Stimson’s website: www.
unc.edu/~jstimson/index.html).
5. Yoshinaka and Murphy (2009) provide evidence in the 2002 House elections that
under partisan gerrymanders one of the primary means of weakening opposing-
party incumbents is by placing a much higher percentage of redrawn constituents
in their districts (see also McKee, 2010; McKee & Shaw, 2005). Unfortunately,
for southern Democrats, the presence of redrawn constituents did not prove elec-
torally detrimental to Republican incumbents and thus under Republican ger-
rymanders (i.e., Florida in 2002) it was also possible to significantly increase
the percentage of redrawn voters in Republican-held districts in furtherance of
expanding the overall number Republican seats.
6. Let me be clear that the expected relationship between redistricting and voting
behavior is modeled with district-level data and therefore the results cannot speak
directly to the behavior of individuals (the ecological inference fallacy)—rather
the theoretical expectation of individual-level behavior in the context of a redis-
tricting serves as the basis for setting up the district-level models.
7. Of course this shift in voting behavior will not be detectable for the vast majority
of voters; they are nudged in the direction of the partisan tide, but retain a level of
partisanship that overpowers the political shift created by short-term conditions.
8. The redrawn district percentage was calculated from the following websites in
1992 and 2002, respectively: http://mcdc2.missouri.edu/websas/geocorr90.shtml;
http://mcdc2.missouri.edu/websas/geocorr2k.html. Determination of the redrawn
district percentage is not based on congressional district numbers, but rather the
portion of an incumbent’s constituency that is new as a direct consequence of
boundary changes. For instance, in 2002 Republican incumbent Pete Sessions
vacated Texas District 5 in favor of the newly drawn District 32. Only 15.7% of
District 5 constituents (residents whom Sessions represented prior to redistricting)
were redrawn into the new District 32; thus in 2002 Sessions ran for reelection in
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

644 American Politics Research 41(4)
a district where 84.3% of his constituents were redrawn. This calculation of the
redrawn district percentage has been used in several other studies (i.e., Crespin,
2005; McKee, 2008a; Petrocik & Desposato, 1998).
9. In the 1992 House elections all of the district-level analyses exclude the five cases
where two incumbents faced off in the general election: (a) Iowa 2nd—Demo-
crat David Nagle vs. Republican Jim Nussle (b) Louisiana 5th—Democrat Jerry
Huckaby vs. Republican Jim McCrery (c) Louisiana 6th—Republican Richard
Baker vs. Republican Clyde Holloway (Louisiana’s general election takes the form
of an open primary contest) (d) Maryland 1st—Democrat Thomas McMillen vs.
Republican Wayne Gilchrest, and (e) Montana At-Large—Democrat Pat Williams
vs. Republican Ron Marlenee (Montana had two U.S. House districts before reap-
portionment). In the 2002 House elections all of the district-level analyses exclude
the four cases where two incumbents squared off in the general election: (a) Con-
necticut 5th—Democrat Jim Maloney vs. Republican Nancy Johnson (b) Illinois
19th—Democrat David Phelps vs. Republican John Shimkus (c) Mississippi 3rd—
Democrat Ronnie Shows vs. Republican Chip Pickering, and (d) Pennsylvania
17th—Democrat Tim Holden vs. Republican George Gekas.
10. Analysis of open-seat races is beyond the scope of this paper. But it should be
noted that many open-seat elections are in fact the result of redistricting. An unfa-
vorable redistricting can have the effect of prompting an incumbent to retire and
thus the influence of redistricting is felt prior to a single vote being cast. It is well
known that, incumbent retirements, the overwhelming reason for open-seat races,
are mainly done for strategic motives and this is especially true with respect to
redistricting (Cox & Katz, 2002; Friedman & Holden, 2009; McCarty et al., 2009).
Infusing districts with large percentages of redrawn voters is a leading cause of
incumbent retirements and line drawers will target out-party incumbents for defeat
by giving them a disproportionately higher share of new voters (see Cain, 1985;
Murphy & Yoshinaka, 2009; Yoshinaka & Murphy, 2011). Second, the analysis
in this paper is concerned with assessing the relationship between redrawn con-
stituents and incumbent vote shares, and by definition, the concept of redrawn
constituents applies only to incumbents (see Desposato & Petrocik, 2003). Finally,
the analysis does not account for all of the effects of redistricting—those which
influence the decision to retire and other notable effects related to such things
as challenger emergence, the ability to raise campaign donations, strategic voter
mobilization, etc.
11. Michael P. McDonald provided the data indicating state control of congressional
redistricting (Republican plan, Democratic plan, bipartisan, or neutral [typically
a court-drawn map]). McDonald’s data have been used for similar purposes by
other scholars, such as Yoshinaka and Murphy (2011).
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 645
12. The control variables account for several factors that usually affect House out-
comes, like partisanship (presidential vote as a proxy), race, income, and region.
Finally, district competitiveness is included as a more holistic estimate of a candi-
date’s chances of winning election. Though district ratings are a fairly blunt mea-
sure, they are good for gauging the competitiveness of each contest and have the
benefit of not being post hoc. Due to issues of collinearity and the fact that includ-
ing candidate spending measures (i.e., challenger spending, incumbent spending,
and challenger’s portion of total spending by both candidates) does not change
the substantive findings, these variables are omitted. Similarly, the inclusion of a
dummy for challenger quality also raises the endogeneity issue (on this point see
Petrocik & Desposato, 2004), because quality challengers are much more likely
to run against vulnerable incumbents, i.e., those who have a higher percentage of
redrawn constituents (McKee, 2010; Murphy & Yoshinaka, 2009). Models that
include these variables do not change the substantive findings and will be made
available by the author upon request.
13. Specifically, the value for the Democratic incumbent share of the two-party vote
is the mean under a 95% confidence interval with all of the control variables
set to their means. For instance, to determine the two-party vote for southern
Democratic incumbents in 1992, the minimum is calculated so that redrawn con-
stituents = 0, Democratic incumbent = 1, and the interaction = 0 (all other vari-
ables set at their means with a 95% level of confidence). Hence the maximum is
calculated so that redrawn constituents = 1, Democratic incumbent = 1, and the
interaction = 1 (all other variables set at their means with a 95% level of confi-
dence). The values generated for Figure 2 were calculated in the same manner.
14. In 1992 the highest percent redrawn constituents for an incumbent southern Dem-
ocrat was 69% (Richard Ray in GA 3). In 2002 the highest percent redrawn con-
stituents for a nonsouthern Democratic incumbent was 76% (Leonard Boswell in
IA 3), and in 2002 the highest percent redrawn constituents for a southern Demo-
cratic incumbent was 54% (Karen Thurman in FL 5). Boswell won reelection but
the aforementioned southern Democrats both lost.
15. The data are as follows. Southern Democrats in 1992: median vote = 61%, stan-
dard deviation = 9%, N = 57; southern Democrats in 2002: median vote = 66%,
standard deviation = 8%, N = 30; nonsouthern Democrats in 1992: median vote
= 65%, standard deviation = 11%, N = 135; nonsouthern Democrats in 2002:
median vote = 69%, standard deviation = 9%, N = 119.
16. ANES data show that since 2002 the percentage of pure independents is growing.
17. In the non-South, the absence of a partisan realignment means the electoral pen-
dulum will continue swinging. This point can be demonstrated with data from
Pennsylvania House contests. Pennsylvania aptly reflects the volatility created
by redistricting when short-term conditions shift in favor of a political party but
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

646 American Politics Research 41(4)
there is no underlying partisan realignment. The Keystone State was the subject
of the lawsuit advanced by Democrats in Vieth v. Jubelirer (2004), following
the Republican gerrymander implemented for the 2002 House elections. In Vieth
v. Jubelirer (2004), the Supreme Court denied the claim that Pennsylvania Repub-
licans’ 2002 congressional map was an unconstitutional partisan gerrymander.
With hindsight, this was the appropriate ruling because political conditions have
exerted a sizable effect on the partisan makeup of the Pennsylvania House del-
egation. Under the 2002 Republican-drawn plan, the Pennsylvania delegation
went from 10 Democrats and 11 Republicans in 2000 to just 7 Democrats and
12 Republicans after the 2002 elections (reapportionment reduced the number
of districts to 19). But in 2008, when political conditions favored Democrats, the
partisan composition of the delegation flipped to 12 Democrats and 7 Republi-
cans. Finally, in 2010, the most favorable election for the GOP since 1938, the
delegation flipped back to 12 Republicans and 7 Democrats. Seabrook (2010) has
it right when he states that
[r]edistricting does not insulate a party’s electoral majority in the face of
popular sentiment, and the more seats a party attempts to gain through ger-
rymandering, the more it is likely to lose seats in subsequent elections where
the popular vote shifts in the opposite direction. (p. 11)
18. The Supreme Court did not find the Indiana state house and state senate plans
unconstitutional.
References
Abramson, P. R., Aldrich, J. H., & Rohde, D. W. (1994). Change and continuity in the
1992 elections. Washington, DC: CQ Press.
Ansolabehere, S., & Jones, P. E. (2010). Constituent’s responses to congressional roll-
call voting. American Journal of Political Science, 54, 583-597.
Ansolabehere, S., Snyder, J. M., Jr., & Stewart, C. S., III. (2000). Old voters, new vot-
ers, and the personal vote: Using redistricting to measure the incumbency advan-
tage. American Journal of Political Science, 44(1), 17-34.
Barone, M., & Cohen, R. E. (2003). Almanac of American politics 2004. Washington,
DC: National Journal.
Barone, M., & Ujifusa, G. (1993). Almanac of American politics 1994. Washington,
DC: National Journal.
Bartels, L. M. (2000). Partisanship and voting behavior, 1952-1996. American Jour-
nal of Political Science, 44(1), 35-50.
Black, E., & Black, M. (2002). The rise of southern Republicans. Cambridge, MA:
Harvard University Press.
Boatright, R. G. (2004). Static ambition in a changing world: Legislators’ preparations
for, and responses to, redistricting. State Politics & Policy Quarterly, 4, 436-454.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 647
Born, R. (1985). Partisan intentions and election day realities in the congressional
redistricting process. American Political Science Review, 79, 305-319.
Brambor, T., Clark, W. R., & Golder, M. (2006). Understanding interaction models:
Improving empirical analyses. Political Analysis, 14(1), 63-82.
Brunell, T. L. (2006). Rethinking redistricting: How drawing uncompetitive districts
eliminates gerrymanders, enhances representation, and improves attitudes toward
Congress. PS: Political Science & Politics, 39(1), 77-85.
Brunell, T. L. (2008). Redistricting and representation: Why competitive elections are
bad for America. New York, NY: Routledge.
Bullock, C. S., III. (2010). Redistricting: The most political activity in America. Lan-
ham, MD: Rowman & Littlefield.
Butler, K. I. (2002). Redistricting in a post-Shaw era: A small treatise accompanied by
districting guidelines for legislators, litigants, and courts. University of Richmond
Law Review, 36, 137-270.
Cain, B. E. (1984). The reapportionment puzzle. Berkeley: University of California Press.
Cain, B. E. (1985). Assessing the partisan effects of redistricting. American Political
Science Review, 79, 320-333.
Cain, B. E., & Campagna, J. C. (1987). Predicting partisan redistricting disputes. Leg-
islative Studies Quarterly, 12, 265-274.
Congressional Quarterly. (1993). Congressional districts in the 1990s: A portrait of
America. Washington, DC: Congressional Quarterly.
Congressional Quarterly. (2003). Congressional districts in the 2000s: A portrait of
America. Washington, DC: CQ Press.
Cox, G. W., & Katz, J. N. (2002). Elbridge Gerry’s salamander: The electoral con-
sequences of the reapportionment revolution. Cambridge, UK: Cambridge Uni-
versity Press.
Crespin, M. (2005). Using geographic information systems to measure district change,
2000-02. Political Analysis, 13, 253-260.
Cunningham, M. T. (2001). Maximization, whatever the cost: Race, redistricting and
the Department of Justice. Westport, CT: Praeger.
Davis v. Bandemer (1986). 478 U.S. 109.
Desposato, S. W., & Petrocik, J. R. (2003). The variable incumbency advantage: New
voters, redistricting, and the personal vote. American Journal of Political Science,
47(1), 18-32.
Desposato, S. W., & Petrocik, J. R. (2005). Redistricting and incumbency: The new
voter effect. In P. F. Galderisi (Ed.), Redistricting in the new millennium (pp. 35-63).
Lanham, MD: Lexington Books.
Epstein, D., & O’Halloran, S. (2000). Majority-minority districts and the new politics
of congressional elections. In D. W. Brady, J. F. Cogan, & M. P. Fiorina (Eds.),
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

648 American Politics Research 41(4)
Continuity and change in House elections (pp. 87-109). Stanford, CA: Stanford
University Press.
Fenno, R. (1978). Homestyle: House members in their districts. Boston, MA: Little,
Brown.
Fiorina, M. P. (2005). Keystone reconsidered. In L. C. Dodd & B. I. Oppenheimer
(Eds.), Congress reconsidered (8th ed., pp. 159-177). Washington, DC: CQ Press.
Fiske, J. (1890). Civil government in the United States: Considered with some refer-
ence to its origins. Boston, MA: Houghton, Mifflin.
Forgette, R., & Platt, G. (2005). Redistricting principles and incumbency protection in
the U.S. Congress. Political Geography, 24, 934-951.
Friedman, J. N., & Holden, R. T. (2009). The rising incumbent reelection rate: What’s
gerrymandering got to do with it? Journal of Politics, 71, 593-611.
Gaddie, R. K., & Bullock, C. S., III. (2000). Elections to open seats in the U.S. House:
Where the action is. Lanham, MD: Rowman & Littlefield.
Green, D., Palmquist, B., & Schickler, E. (2002). Partisan hearts and minds: Political
parties and the social identities of voters. New Haven, CT: Yale University Press.
Grofman, B. (1990). Toward a coherent theory of gerrymandering: Bandemer and
Thornburg. In B. Grofman (Ed.), Political gerrymandering and the courts
(pp. 29-63). New York, NY: Agathon Press.
Grofman, B., & Brunell, T. L. (2005). The art of the dummymander: The impact
of recent redistrictings on the partisan makeup of southern House seats. In
P. F. Galderisi (Ed.), Redistricting in the new millennium (pp. 183-199).
Lanham, MD: Lexington Books.
Hayes, D., & McKee, S. C. (2009). The participatory effects of redistricting. Ameri-
can Journal of Political Science, 53, 1006-1023.
Hetherington, M. J. (2001). Resurgent mass partisanship: The role of elite polariza-
tion. American Political Science Review, 95, 619-631.
Hetherington, M. J., Larson, B., & Globetti, S. (2003). The redistricting cycle and stra-
tegic candidate decisions in U.S. House races. Journal of Politics, 65, 1221-1234.
Hill, K. A. (1995). Does the creation of majority black districts aid Republicans? An
analysis of the 1992 congressional elections in eight southern states. Journal of
Politics, 57, 384-401.
Hirsch, S. (2003). The U.S. House of unrepresentatives: What went wrong in the latest
round of congressional redistricting. Election Law Journal, 2, 179-216.
Hood, M. V., III., & McKee, S. C. (2010). Stranger danger: Redistricting, incumbent
recognition, and vote choice. Social Science Quarterly, 91, 344-358.
Jacobson, G. C. (2003). Terror, terrain, and turnout: Explaining the 2002 midterm
elections. Political Science Quarterly, 118 (1), 1-22.
Jacobson, G. C. (2009). The politics of congressional elections. New York, NY: Pearson
Education.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

McKee 649
Keith, B. E., Magleby, D. B., Nelson, C. J., Orr, E., Westlye, M. C., & Wolfinger, R. E.
(1992). The myth of the independent voter. Berkeley: University of California Press.
La Raja, R. (2009). Redistricting: Reading between the lines. Annual Review of Politi-
cal Science, 12, 203-223.
Levendusky, M. (2009). The partisan sort: How liberals became Democrats and con-
servatives became Republicans. Chicago, IL: University of Chicago Press.
Levitt, J., & McDonald, M. P. (2007). Taking the “re” out of redistricting: State constitu-
tional provisions on redistricting timing. Georgetown Law Review, 95, 1247-1286.
Lublin, D. (1997). The paradox of representation: Racial gerrymandering and minor-
ity interests in Congress. Princeton, NJ: Princeton University Press.
MacKuen, M. B., Erikson, R. S., & Stimson, J. A. (1989). Macropartisanship. Ameri-
can Political Science Review, 83, 1125-1142.
Mann, T. E., & Cain, B. E. (2005). Party lines: Competition, partisanship, and con-
gressional redistricting. Washington, DC: Brookings Institution Press.
McCarty, N., Poole, K. T., & Rosenthal, H. (2009). Does gerrymandering cause polar-
ization? American Journal of Political Science, 53, 666-680.
McDonald, M. P. (2004). A comparative analysis of redistricting institutions in the
United States, 2001-02. State Politics and Policy Quarterly, 4, 371-395.
McDonald, M. P. (2006). Drawing the line on district competition. PS: Political Sci-
ence & Politics, 39(1), 91-94.
McKee, S. C. (2008a). The effects of redistricting on voting behavior in incumbent
U.S. House elections, 1992-1994. Political Research Quarterly, 61(1), 122-133.
McKee, S. C. (2008b). Redistricting and familiarity with U.S. House candidates.
American Politics Research, 36, 962-979.
McKee, S. C. (2010). Republican ascendancy in southern U.S. House elections. Boul-
der, CO: Westview Press.
McKee, S. C., & Shaw, D. R. (2005). Redistricting in Texas: Institutionalizing
Republican ascendancy. In P. F. Galderisi (Ed.), Redistricting in the new millen-
nium (pp. 275-311). Lanham, MD: Lexington Books.
Monmonier, M. (2001). Bushmanders and bullwinkles: How politicians manipulate elec-
tronic maps and census data to win elections. Chicago, IL: University of Chicago
Press.
Murphy, C., & Yoshinaka, A. (2009). Are mapmakers able to target and protect con-
gressional incumbents? The institutional dynamics of electoral competition.
American Politics Research, 37, 955-982.
Petrocik, J. R., & Desposato, S. W. (1998). The partisan consequences of majority-
minority redistricting in the South, 1992 and 1994. Journal of Politics, 60, 613-633.
Petrocik, J. R., & Desposato, S. W. (2004). Incumbency and short-term influences on
voters. Political Research Quarterly, 57, 363-373.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from

650 American Politics Research 41(4)
Rush, M. E. (1992). The variability of partisanship and turnout: Implications for ger-
rymandering analysis and representation theory. American Politics Quarterly,
20(1), 99-122.
Rush, M. E. (1993). Does redistricting make a difference? Partisan representation
and electoral behavior. Baltimore, MD: Johns Hopkins University Press.
Rush, M. E. (2000). Redistricting and partisan fluidity: Do we really know a gerry-
mander when we see one? Political Geography Quarterly, 19, 249-260.
Schweers, M. (2003). U.S. House races: Republican resurgence after eight lean years.
In L. J. Sabato (Ed.), Midterm madness: The elections of 2002 (pp. 67-76). Lanham,
MD: Rowman & Littlefield.
Seabrook, N. R. (2010). The limits of partisan gerrymandering: Looking ahead to the
2010 congressional redistricting cycle. Forum, 8(2), Article 8.
Stimson, J. A. (2004). Tides of consent: How public opinion shapes American politics.
Cambridge, UK: Cambridge University Press.
Tomz, M., Wittenberg, J., & King, G. (2003, January 5). CLARIFY: Software for
interpreting and presenting statistical results. Version 2.1. Stanford University,
University of Wisconsin, and Harvard University. Retrieved from http://gking.
harvard.edu/
Vieth v. Jubelirer (2004). 541 U.S. 267.
Yoshinaka, A., & Murphy, C. (2009). Partisan gerrymandering and population insta-
bility: Completing the redistricting puzzle. Political Geography, 28, 451-462.
Yoshinaka, A., & Murphy, C. (2011). The paradox of redistricting: How partisan map-
makers foster competition but disrupt representation. Political Research Quar-
terly, 64, 435-447.
Author Biography
Seth C. McKee is an associate professor of political science at the University of
South Florida, St. Petersburg. His research centers on voting behavior, southern poli-
tics, political parties, and redistricting. He is the author of Republican Ascendancy in
Southern U.S. Elections (2010) and numerous journal articles.
at UNIV OF SOUTH FLORIDA on July 13, 2013apr.sagepub.comDownloaded from