Some Inconvenient Truths About Redistricting
I go away on vacation and I return to find that an article claiming to
debunk redistricting "myths" has been published in The New Republic. Given
my research on redistricting and competition, the article comes as a bit of
an amusement as one of my major articles is "The Myth of the Vanishing
Voter" which debunked the decline in voter turnout.
The authors, John Freidman and Richard Holden, claim that redistricting
actually increases electoral competition, contrary popular writings and my
own work, which the authors are apparently not aware of. Not surprisingly,
I'll debunk the debunkers, and I've selected the title "Some Inconvenient
Truths About Redistricting" for this e-mail as a play on Al Gore's recent
movie debunking debunkers on the subject of global warming. My critique is
aimed at the academic paper and is technical at times (perhaps I'll write
this up in an op-ed format), so as a synopsis:
1. The authors present no new findings about redistricting that hasn't been
covered in the literature. The literature supports that redistricting
generally temporarily increases electoral competition and decreases
incumbency reelection rates for reasons outlined below.
2. The authors do not provide compelling evidence explaining the decline in
electoral competition following in 2002 and 2004, which should have seen the
usual increase in electoral competition. It is this current low state of
competition that I and others are most concerned about and lay at least some
blame on redistricting.
3. Despite the authors' findings, it is possible to construct redistricting
institutions that will draw competitive districts that increase electoral
competition.
Why now publish this (non-peer-reviewed) research in a left-leaning popular
journal? I have been accused of being a left-winger for my positions on
redistricting reform, despite that I've worked with Democrats and
Republicans on this issue. I care about making legislatures more responsive
to the voters regardless of the party in control. It is very clear that the
most recent round of redistricting Republicans were advantaged by partisan
gerrymanders in key states. Redistricting reform seemed a way for Democrats
to make sure that they would not suffer similarly in 2012.
Now, with Bush's approval rating bringing down Republicans in state races,
it appears that Democrats might pick up governorships and other key
positions that determine redistricting, and thus be advantaged in 2010 when
these positions come up for reelection. Republicans may very well find
themselves on the short-end of redistricting in 2012 in key states of New
York, California, and Illinois, and Democrats might be able to force
redistricting into the courts in places like Ohio and Pennsylvania (and in
Ohio, Democrats might control the state legislative redistricting, creating
a Texas-like congressional re-redistricting situation in 2013). Intoxicated
by their potential gains, Democrats are abandoning reform for power, as most
obviously demonstrated in Democratic Ohio House legislative leaders'
parliamentary maneuverings to keep a redistricting reform referendum off the
ballot this year, even to the point of voting against the same initiative
many Democrats supported last year. The New Republic appears to be jumping
on the bandwagon.
My critique follows:
===============
No new findings
===============
First, the incomplete literature review is a concern, though perhaps
forgivable since the authors are economists and not political scientists. I
do not mean to belittle their contribution by saying this. However, because
they do not know the literature, they make some inaccurate claims in the
manuscript and fail to fully interpret their findings within the context of
the literature.
For example, the authors claim that "Though not specifically focused on the
incumbent reelection rate, Gelman and King (1994a) provide the best evidence
towards the matter at hand." Actually the literature is vaster than this
one article. The authors are essentially replicating previous work on the
electoral effects of redistricting, with a methodological innovation.
Perhaps most relevant to the author's modeling approach is:
Glazer, Amihai, Bernard Grofman and Marc Robbins. 1987. "Partisan and
Incumbency Effects of 1970s Congressional Redistricting." American Journal
of Political Science 31(3): 680-707.
The authors come to similar conclusions as previous literature: that
redistricting generally reduces incumbent reelection rates in the
short-term., though the Glazer et al.'s findings are a bit more complex than
this and the authors need to place their findings in the context of this and
other published research.
===============
Redistricting lowers incumbency advantage in the immediately following
election
===============
This finding seems to be contrary to the claim that redistricting has
resulted in fewer competitive elections, so what gives?
1. Redistricting forces paired incumbents in states where apportionment
reduces the number of seats and all incumbents seek reelection. This
dynamic is not mentioned in the article and presumably is uncontrolled for.
2. Redistricting temporarily reduces incumbency advantage as incumbents find
themselves representing new constituents who they have not built up name
recognition. In addition to the Ansolabehere et al. (2000) cited in the
paper, the authors should read:
Desposato, Scott W. and John R Petrocik. 2003. "The Variable incumbency
Advantage: New Voters, Redistricting, and the Personal Vote." American
Journal of Political Science 47(1): 18-33.
3. Strategic incumbents wait for redistricting to contest the temporarily
vulnerable incumbents. "Cox and Katz (1996) and Cox and Katz (2002) claim
that the cause is the interaction between gerrymandering and challenger
quality. Unfortunately they do not present systematic evidence to support
this intriguing claim." Perhaps the authors should read this, too:
Hetherington, Marc J., Bruce A. Larson, and Suzanne Globetti. 2003. "The
Redistricting Cycle and Strategic Candidate Decisions in U.S. House Races."
Journal of Politics. 65(4):1221-35.
===============
Some inconvenient truths about redistricting
===============
Redistricting generally lowers incumbency advantage and this is generally
evident in Congressional Quarterly's rankings of "competitive races." The
number of competitive races generally increases and then declines as
incumbents become entrenched in their new districts.
For a nice graph since 1970, see:
Jacobson, Gary. 2003. "Terror, Terrain, and Turnout: Explaining the 2002
Midterm Elections." Political Science Quarterly 118(1): 1-22.
As a modeling choice, this suggests a lag-structure to redistricting,
consistent with the authors statistical results. Immediately following
redistricting, incumbents are more vulnerable, but over time, they become
less vulnerable. If redistricting has a long-term effect of lowering
competition, then it might be picked up as a negative time-trend, which is
what the authors find.
But something different happened in 2002: normally when elections become
more competitive, they become less so. This is the concern of those like
myself and others that redistricting has reduced electoral competition.
===============
Why might 2002 be different from previous redistricting?
===============
1. Whatever disagreement I might have with Abramowitz (uncited in the
authors paper, see the January, 2006 issue of PS for our debate on the
effects of redistricting on competitive districts), we both agree that the
number of competitive congressional districts, those with an even balance of
partisan declined during the last round of redistricting.
2. Add to this, Jacobson (2003) also finds that Jacobson finds "three
quarters of the marginal districts were made safer by redistricting" and
that the redistricting musical chairs sat a net of nineteen incumbents with
districts leaning towards the other party in new seats leaning towards their
party.
Thus, I can point to some inconvenient truths in the 2002 redistricting
cycle that reduced the number of competitive races in 2002. In 2004, the
number of competitive races declined even further as the few
incumbent-paired general election races sorted themselves out (despite the
Texas re-redistricting).
What do the authors have to say about 2002? The authors do two analyses,
one that is irrelevant to what happened in 2002 since it estimates
redistricting effects since the 1910s. It is very likely that redistricting
was conducted under much different circumstances then. For example,
congressional districts were not required to be of equal population. There
are clearly some important omitted variables in their analysis, and the
authors even admit as much. The resulting statistical bias is unknown.
To the authors credit, they run separate models with data from 1972-2004 and
decade specific redistricting models. But even 1972 might be different than
2002, for example, the currently closely divided House might have resulted
in a more risk-adverse strategy during redistricting for 2002.
Still, the statistical results for 2002 are haphazard and depend on the
modeling choice. This seems inconsistent with the Congressional Quarterly
(and every other election handicapper) analysis that found fewer competitive
elections in 2002 than in 2000. I rather suspect that the cause is the low
level of competition in 2000 and the currently high incumbency reelection
rates. (Perhaps an inappropriate measurement of incumbent pairings is also
working here.) Given these high levels, it would have been nearly
impossible for redistricting to increase incumbent reelection rates. This
point raises a statistical issue that likely confounds the analysis: there
is insufficient variation in the dependent variable (incumbency reelection
rates) to obtain good estimates.
===============
Conclusions
===============
Where we agree:
1. We agree that redistricting technology has had a limited effect. See:
Micah Altman, Karin Mac Donald, and Michael P. McDonald. 2005. "Pushbutton
Gerrymanders? How computing has changed redistricting." in Party Lines:
Competition, Partisanship and Congressional Redistricting, Bruce Cain and
Thomas Mann, eds. Washington, DC: Brookings Press.
2. We agree that there are long-term declines in competition independent of
redistricting.
However, why we might agree on this point, the causes offered in the
author's conclusion are purely speculative. Money, the Voting Rights Act,
better campaigns? The authors make no statistical test to back up these
claims. In fact, the downward decline in competition seems to simply come
down to one variable: a downward trend variable, which is hardly
explanatory.
Let me add another cause that at least seems to have some evidence backing
it up: the regional electoral realignment as the South and other rural
states complete their transition from the Democratic to Republican parties.
===============
Addendum: Poor measurement of types of redistricting plans, and is it
possible to increase competition through redistricting?
===============
The authors consider all redistricting plans as one of three types: a
bipartisan, partisan, or court-ordered plan.
To score bipartisan or partisan plan the authors identify if a "party
controls all the relevant branches of a state government." Using this
approach, they have clearly miscoded redistricting, for example, Hawaii has
a bipartisan commission and although Democrats had unified control of
California, they adopted a bipartisan compromise with Republicans in order
to avert a Republican sponsored redistricting reform initative.
This scoring also misses a crucial aspect of court ordered plans: my perusal
of Congressional Quarterly synopses of redistricting plans revealed that
courts more often accepted plans proposed during the redistricting process
by party leaders and did not draw their own maps, though some courts did
employ special masters to draw maps. Calling these "court" maps does not
accurately measure the intent of these maps.
Overall, I believe that the intent of the redistricting plan, not the
authority, should be relevant in the scoring of the redistricting "type."
See:
Michael P. McDonald. 2004. "A Comparative Analysis of U.S. State
Redistricting Institutions." State Politics and Policy Quarterly 4(4):
371-396.
A careful review of the redistricting plan types listed in their appendix
reveals that they might have haphazardly taken this approach: the 2002 Texas
court-ordered plan is scored as a bipartisan plan -- which it generally was
-- and not a court ordered plan.
Thus, the redistricting "type" variable is miscoded and any conclusions
drawn from models with these variables should be considered suspect.
While the authors claim that redistricting is not the cause of the decline
in incumbency reelection rates, they cannot state with any authority whether
or not alternative redistricting institutions might increase electoral
competition (note that these two are different questions!). As I explain in
my 2006 PS piece, and expound further in my forthcoming contribution to my
edited book with John Samples on electoral competition, it is possible to
create redistricting institutions that will draw districts that increase
electoral competition.
------------
Dr. Michael P. McDonald
Assistant Professor, George Mason University
Visiting Fellow, Brookings Institution
Mailing address:
(o) 703-993-4191 George Mason University
(f) 703-993-1399 Dept. of Public and International Affairs
mmcdon@gmu.edu 4400 University Drive - 3F4
http://elections.gmu.edu Fairfax, VA 22030-4444