In “Drawing the Line on District Competition,” Michael McDonald (2006) challenges our conclusion that redistricting
is not responsible for declining competition in House elections. McDonald
claims that this conclusion is incorrect largely because our measure of
district partisanship in 1992 is flawed. According to McDonald, using the
normalized 1992 major party presidential vote as a measure of House district
partisanship in 1992 seriously overstates the competitiveness of House
districts in that year because of the impact of the vote for Ross Perot.
McDonald recommends instead using the 1988 presidential vote to measure
district partisanship in 1992, or using a procedure developed by Gelman and
King (1994a) to estimate the partisan bias of
House districts. With either of these methods, McDonald claims that a
significant proportion of the decline in competitive districts between 1990
and 2002 was caused by redistricting. In addition, contrary to our finding
that nonpartisan redistricting commissions have failed to produce increased
competition in House elections, McDonald claims that nonpartisan
redistricting in Arizona provides a positive example of what can be
accomplished by nonpartisan redistricting commissions elsewhere.
We believe that all of these claims are erroneous. In fact, the evidence
presented by McDonald based on the Gelman-King procedure for estimating the
partisan bias of House districts actually supports our conclusion that
redistricting was not a significant factor in the decline of competitive
House districts between 1990 and 2002.
Before reviewing the evidence, however, we want to emphasize that our
conclusions are supported by other studies that have examined the
consequences of partisan redistricting for competition. While McDonald
implies that previous research in this area has consistently found negative
effects of redistricting on competition, this is not the case.
Studies of both the 1971–1972 and 1981–1982 rounds of redistricting concluded
that partisan redistricting had either neutral or positive effects on
competition. Based on their examination of the 1971–1972 redistricting
cycle, Glazer, Grofman, and Robbins (1987) found
that redistricting produced “minimal change” in competition. Incumbents did
not benefit at the expense of challengers and the main effect of
redistricting was to increase competition when incumbents were forced to run
against each other. Gopoian and West (1984) found
evidence of increased competition as a result of the 1981–1982 round of
redistricting because “leaders appeared to trade security for seats in their
redistricting decisions” (1080).
Redistricting often leads to increased competition because partisans drawing
district lines face a fundamental tension between incumbent protection and
maximizing their party's electoral potential. More often than not, the only
way to shift marginal districts toward the majority party is to cut the
safety margins of majority party incumbents by moving reliable partisans out
of their districts. As Gelman and King (1994b)
explain, “incumbents are often forced to give up votes (hence electoral
safety) in order to increase the number of legislative seats their party is
likely to capture” (541). They conclude that, “far from being a scourge on
the political system in need of major reforms, legislative redistricting has
invigorated American representative democracy” (554).
The Evidence
1. Our conclusions are based on evidence from three different
redistricting cycles, not just the 1991–1992 cycle. Our finding that
there was no significant decrease in the competitiveness of House
districts between 1990 and 1992 is consistent with our findings for the
1981–1982 redistricting cycle and the 2001–2002 redistricting cycle. In
fact, the 2001–2002 redistricting cycle has been the subject of the most
commentary about the impact of redistricting on competition because of
the increased use of computerized databases and mapmaking programs by
state legislatures. However, we find no significant change in the
numbers of safe or competitive districts between 2000 and 2002.
2. McDonald's claim that Ross Perot's candidacy reduced Bill Clinton's
margin of victory over George H. W. Bush in 1992 “by as much as seven
percentage points” (91) is highly questionable. This conclusion is based
on one study of the impact of Perot's candidacy (Lacy
and Burden 1999). It is not supported by any other research
on the 1992 presidential election. In fact, the only other published
study of the impact of Perot's candidacy found that he took more votes
from Bush than from Clinton (Alvarez and Nagler
1995).
Based on evidence from both the American National Election Studies and
national exit polls, it appears highly unlikely that Perot's candidacy
significantly reduced Clinton's margin of victory over Bush. According
to the 1992 National Exit Poll, when Perot voters were asked to name
their second choice candidate, those who expressed a preference divided
almost evenly between Clinton and Bush. Based on these results, Perot's
candidacy had no impact on Clinton's margin over Bush. Nor is there any
evidence that Perot's candidacy caused a disproportionate surge in
Republican turnout. In fact, according to NES data, the difference in
turnout between Republican and Democratic identifiers was smaller in
1992 than in any presidential election since 1952. In addition, national
exit poll data show no significant shift in the partisan composition of
the electorate between 1988 and 1992.
Using the normalized 1988 presidential vote rather than the normalized
1992 presidential vote to measure the competitiveness of House districts
does produce a slightly lower estimate of the number of marginal House
districts in 1992: 148 rather than 164. However, when we use the
normalized 1988 presidential vote to measure the change in the
competitiveness of House districts between 1990 and 1992, as McDonald
recommends, we again find no significant effect of redistricting: the
number of districts classified as marginal actually increases from 146
in 1990 to 148 in 1992.
It is not surprising that using the normalized 1988 presidential vote
produces results that are almost identical to those obtained by using
the normalized 1992 presidential vote. The correlation (Pearson's r)
between these two variables is .95. Moreover, the .70 correlation
between the 1992 Democratic presidential vote and the 1992 Democratic
House vote is identical to the .70 correlation between the 1988
Democratic presidential vote and the 1992 Democratic House vote. There
is no evidence of any attenuation due to the Perot vote in 1992.
3. Most importantly, McDonald's own estimates of the number of
competitive House districts based on the Gelman-King procedure do not
support his conclusion that redistricting accounted for a significant
share of the decline in competitive districts between 1990 and 2002.
According to McDonald's Table 1, using the 45–55% standard for
competitive districts, there was an average decline of 16.0 competitive
districts per non-redistricting election cycle between 1990 and 2002
compared with an average decline of 18.5 competitive districts during
the two redistricting cycles. Similarly, using the 48–52% standard for
competitive districts, there was an average decline of 10.8 competitive
districts per non-redistricting election cycle compared with an average
decline of 13.5 competitive districts during the two redistricting
cycles.
McDonald attributes the entire decline in competitive districts during a
redistricting cycle to the effects of redistricting. But this assumes
that no decline would have occurred in the absence of redistricting,
which is clearly unrealistic. According to McDonald's data, the number
of competitive districts declined by almost as much during
non-redistricting cycles as it did during redistricting cycles.
Regardless of whether one uses the 45–55% standard or the 48–52%
standard for identifying competitive districts, the decline in
competitive districts between 1990 and 2002 that can be attributed to
redistricting was approximately five districts. According to the
Gelman-King estimates presented by McDonald, between 1990 and 2002 there
was a decline of 101 districts in the 45–55% range and 70 districts in
the 48–52% range. Therefore, based on McDonald's data, redistricting
was responsible for only 5% of the overall decline in districts in
the 45–55% range and 7% of the overall decline in districts in the
48–52% range. Over 90% of the decline in competitive districts
between 1990 and 2002 was caused by underlying changes in the
geographic distribution of partisans rather than
redistricting.
4. McDonald's claim that Arizona provides a model of what can be
accomplished by a truly nonpartisan redistricting commission is not
supported by the results of the 2001–2002 round of redistricting or by
recent elections in that state. Before redistricting, three of Arizona's
six House districts were classified as marginal based on the results of
the 2000 presidential election—the difference between George W. Bush and
Al Gore was less than 10 percentage points. After redistricting, only
two of Arizona's eight House districts were classified as marginal. The
average difference between the winning and losing presidential candidate
in Arizona House districts increased from 12.2 percentage points in 2000
to 14.8 percentage points in 2002. By this measure, Arizona's House
districts were less competitive after redistricting than before.
Moreover, in the 2002 and 2004 elections, 15 of Arizona's 16 House
contests were decided by a margin of more than 20 percentage points. Not
one incumbent was defeated or even seriously challenged. The experience
of Arizona should give pause to anyone who believes that shifting
control of redistricting from state legislatures to nonpartisan
redistricting commissions will result in increased competition (Hill 2005).
Conclusion
Much of the evidence presented by McDonald concerning the detrimental
effects of partisan redistricting and the potential benefits of
redistricting reform consists of anecdotes based on his personal
experiences as a consultant to various pro-reform groups. It is
difficult to evaluate the internal or external validity of such
anecdotal evidence. However, we believe that the statistical evidence
presented in our study, based on thousands of House elections over 13
election cycles and three separate redistricting cycles, clearly
demonstrates that redistricting has had little or no impact on the
overall competitiveness of House elections. Moreover, we believe that
the statistical evidence presented by McDonald supports the same
conclusion.
Our purpose here is not to defend partisan control of redistricting.
There is no question that partisan gerrymandering can produce a
short-term electoral advantage for the party that draws the lines (Abramowitz 1983; Niemi and
Abramowitz 1994; Gelman and King
1994b). This clearly happened in several states during the
2001–2002 round of redistricting with Republicans being the main
beneficiaries (Jacobson 2005). We wish Michael
McDonald success in his efforts to undo the results of partisan
gerrymandering in Ohio and other states. However, given the increasing
partisan polarization of the American electorate and the increasing
financial advantage of incumbents in congressional elections, we do not
believe that shifting responsibility for redistricting from state
legislatures to nonpartisan commissions will result in a significant
increase in competition in House elections.
Biography
Alan Abramowitz is Alben W. Barkley Professor of Political Science at
Emory University in Atlanta, Georgia. He is the author of Voice of
the People: Elections and Voting in the United States
(McGraw-Hill, 2004) and numerous articles on voting behavior in U.S.
presidential and congressional elections.
Brad Alexander is a Ph.D. candidate in the department of political
science at Emory University.
Matthew Gunning is a Ph.D. candidate in the department of political
science at Emory University.
