Referendum: The 2006 Midterm
GARY C. JACOBSON
Collectively, the American electorate treated the 2006 midterm
congressional elections as a classic referendum on the performance of the
president and his party. Most voters held negative views of both George W.
Bush and his Republican partisans in Congress, and as a consequence, Democrats
won majority control of both chambers for the first time in twelve years.
Table 1 summarizes the results. Democrats picked up thirty seats in the House,
fifteen more than necessary to take over, winning a majority one seat larger than
that held by the Republicans in the previous Congress. They also gained six Senate
seats, all taken fromRepublican incumbents, to win a one-seatmajority in the
upper house. Remarkably, Democrats lost not a single seat in either body, the
first election in U.S. history in which a party retained all of its congressional seats.
According to the political science literature, party fortunes in midterm elections
are broadly shaped by three basic factors: the number of seats the president’s
party already holds, how well the economy is performing, and how the
public views the president’s performance in office.1 Although there is no consensus
on the relative importance of each condition or the way they ultimately
influence voters’ decisions, in combination, they do predict midterm partisan seat
swings with considerable accuracy. This was true in 2006 as well; for example, a
simple model employing standard measures of these variables predicts a twentysix
seat gain for Democrats in the House.2 Because the economy was doing
quite well by customary measures, the model attributes the Republican losses to
the President’s extraordinarily low standing with the public. Bush’s 38 per-
GARY C. JACOBSON is professor of political science at the University of California, San Diego. He
is the author of several books and numerous articles on congressional elections. His most recent book
is A Divider, Not a Uniter: George W. Bush and the American People.
1 For a discussion of this literature and a full set of citations, see Gary C. Jacobson, The Politics of
Congressional Elections, 6th ed. (New York: Longman, 2006), 154–170.
2 The model is in ibid, p. 156, updated to include 2004; with Bush’s approval at 38 percent and real
per capita income up 3.0 percent between the third quarter of 2005 and the third quarter of 2006, and
Political Science Quarterly Volume 122 Number 1 2007 1
cent approval rating in the Gallup Poll taken just before the election was the
lowest for any president since Harry Truman in 1950; had his rating been as
high as it had been in 2002 (63 percent approving), and with the strong economy,
the model predicts that the parties would have broken about even.
All such models need to be taken with a large grain of salt, however, because
their predictions have wide error bands that would include almost any plausible
result (in this model, one standard deviation of the predicted swing covers plus
or minus sixteen seats—that is, for 2006, a Democratic gain of from thirteen to
thirty-five seats). Without question, a strongly pro-Democratic national tide was
running in 2006, but it was by no means certain that it would be sufficient to
overcome the formidable structural advantage the Republicans now enjoy in
congressional elections and deliver control of Congress to the Democrats. The
Republicans’ advantage derives from the fact that their usual voters are distributed
more efficiently across districts and states than are Democratic voters.
As an illustration, consider that the Democrat, Al Gore, won the national popular
vote over George W. Bush by about 540,000 of the 105 million votes cast in
2000, yet the distribution of the 2000 presidential vote across current House
districts yields 240 districts in which Bush won more votes than Gore but only
195 in which Gore outpolled Bush.3 Similarly, Bush won thirty states to Gore’s
twenty despite losing the popular vote nationally.
Membership Changes in the House and Senate in the 2006 Elections
Republicans Democrats Independents
House of Representatives
At the time of the 2006 election 232 202 1
Elected in 2006 202 233
Incumbents reelected 188 191
Incumbents defeated 22 0
Open seats retained 14 12
Open seats lost 8 0
At the time of the 2006 election 55 44 1a
After the 2006 election 49 50 1
Incumbents reelected 8 15
Incumbents defeated 6 0
Open seats retained 1 2 1
Open seats lost 0 0
aBernard Sanders, an independent who caucuses with the Democrats, replaced Senator James Jeffords, a
one-time Republican, who in 2001, became an independent, caucusing with the Democrats, in one of Vermont’s
3 The principal reason for this Republican advantage is demographic: Democrats win a disproportionate
share of minority and other urban voters, who tend to be concentrated in districts with
with Republicans’ seat ‘‘exposure’’ at 2.3 percent, the point prediction is that Democrats pick up
twenty-six seats. Other models predict modest to considerably larger Democratic gains.
2 | POLITICAL SCIENCE QUARTERLY
The Republicans’ structural advantage is nothing new. As Figure 1 shows,
except after the 1964 election, a notably larger proportion of House districts have
leaned Republican than have leaned Democratic (partisan leaning defined here
as having the district vote for the Party’s presidential candidate at least two
percentage points above the national average in the concurrent or most recent
presidential election).4 But changes in electoral behavior over the past thirty
years, particularly the growing partisan consistency of presidential and congressional
voting, has made the advantage far more significant. In past decades,
Democrats were able to win a substantial share of these Republican-leaning
seats, as high as 44 percent in the 1970s (Figure 2). Their ability to win such seats
has fallen sharply since the 1980s, and this is the main reason it was not a
foregone conclusion that Democrats would win the House in 2006. Republicans
lopsided Democratic majorities. But successful Republican gerrymanders in Florida, Michigan, Ohio,
Pennsylvania, and, after 2002, Texas did enhanced the Party’s advantage, increasing the number of
Bush-majority districts by 12, from 228 to 240. See Gary C. Jacobson, ‘‘Polarized Politics and the 2004
Congressional and Presidential Elections,’’ Political Science Quarterly 120 (Summer 2005): 201–203.
4 The substantive point is unchanged if the standard is five rather than two percentage points; entries
are missing for 1962 and 1966 because redistricting altered too many to allow a reliable estimate.
District Partisan Advantage, 1952–2004
51 51 50 51 50
51 51 49 48 50 50 48 45
50 50 50 50 47 47 50 49 51 52 53
12 12 15 15
12 12 18 19 15 15 17
18 18 14 14 17 17 13 13 11 9 8
36 36 35 34
34 34 35 35 35 37
32 32 36 36 36 36 37 37 37 39 39
Republican Advantage Neutral Democratic Advantage
Source: Compiled by author.
THE 2006 CONGRESSIONAL ELECTIONS | 3
have never done particularly well in Democratic territory and remain less
successful than Democrats in this regard; but this is not a problem for them,
because their structural advantage would deliver comfortable Republican House
majorities even if they won only Republican-leaning districts.5 A comparable
analysis shows that similar trends also apply to states as electoral units.6
To understand what Democrats were up against in 2006, it is instructive
to compare their opportunities to those available to the Republicans in 1994, the
year the GOP gained fifty-six House seats and eight Senate seats to win full
control of Congress for the first time in more than four decades. Using the
district-level presidential vote to estimate the underlying district partisanship as
in Figure 2, I calculated that in 1994, Democrats defended sixty-five districts
that leaned Republican (that is, the Republican presidential candidate had run at
least 2 percentage points better than his national average in the district in 1992)
5 Democrats have also done worse in the dwindling number of evenly balanced districts. From the
1950s through the 1980s, they won 66 percent of these districts; since 1992, they have won 53 percent.
6 Gary C. Jacobson, ‘‘Competition in U.S. Congressional Elections’’ in Michael P. McDonald and
John Samples, eds., The Marketplace of Democracy (Washington DC: The Brookings Institution,
Winning against the Partisan Grain, by Decade
1950s 1960s 1970s 1980s 1990s 2002-2004
Democrats Winning Republican-Leaning Districts
Republicans Winning Democratic-Leaning Districts
Note: Leaning districts are defined as those in which the district-level presidential vote was at least two
percentage points higher than the national average for that election.
4 | POLITICAL SCIENCE QUARTERLY
and another forty-nine districts that could be classified as neutral between the
parties (districts where the presidential vote was within two percentage points of
average). Thus, at least in the abstract, Republicans had 114 likely Democratic
targets; they won forty-seven of these seats, comprising 83 percent of their fiftysix
seat gain that year. In 2006, Republicans defended only thirteen seats in
Democratic-leaning districts and another twenty in neutral districts for a total of
only thirty-three ostensibly vulnerable targets. Democrats won twelve of these
seats; the remaining eighteen they gained had to be won on Republican turf.
The configuration of Senate contests also offered more opportunities to Republicans
in 1994 than to Democrats in 2006. In 1994, twenty-two of the thirtyfive
seats to be contested were held by Democrats, and retirements had left
nine of them open; Republicans took six of the open seats and defeated two
incumbents to gain their eight seats. In 2006, Republicans defended fifteen seats,
only one of which was open. Moreover, only three were in states won by the
Democratic presidential candidate in either 2000 or 2004. Democrats won two
of these three ‘‘blue’’ states (Pennsylvania and Rhode Island), but their other
four takeovers occurred in ‘‘red’’ states, and all six were achieved by defeating
sitting Republicans. A national political tide was surging at least as strongly for
the Democrats in 2006 as it had been for the Republicans in 1994, but the Republicans
were defending higher ground, and so the total damage they suffered
was not as extensive.
SOURCES OF THE DEMOCRATIC TIDE
The primary source of the pro-Democratic tide in 2006 was public unhappiness
with the Iraq War and its originator, George W. Bush. Figure 3 displays the
trends in support for the war and approval of Bush’s job performance through
October 2006.7 The two trends track one another closely and move in similar
ways in response to events in Iraq. Cross-sectional analyses also find a close
relationship between opinions on Bush and the war; respondents give consistent
responses—approve of Bush and support the war, or disapprove of Bush
and oppose the war—an average of 84 percent of the time in polls spanning this
period.8 These opinions are far more tightly linked than they were for Bush’s
predecessors in comparable situations—Harry Truman with the Korean War
7 Bush job approval is the Lowess-smoothed summary of responses to 771 polls conducted by the
Gallup, CBS News/New York Times, Los Angeles Times, NBC/Wall Street Journal, ABC/Washington
Post, Quinnipiac, Newsweek, Time, CNN, Bloomberg, Associated Press, and Pew Center for the
People and Press polls reported at http://www.pollingreport.com. Support for the Iraq War is the
Lowess-smoothed summary of 579 polls asking a wide variety of questions (more than forty different
wordings in all) assessing opinions on the war from the same survey sources as well as Fox News,
Knowledge Networks, and Zogby polls.
8 In secondary analysis of 107 of these polls, mean consistency was 83.8 percent with a standard
deviation of 2.9 percent.
THE 2006 CONGRESSIONAL ELECTIONS | 5
(60 percent consistent), and Lyndon Johnson with the VietnamWar (64 percent
consistent).9 In any single survey, the direction of causality is ambiguous—prior
attitudes toward Bush shape reactions to the war, assessments of the war shape
evaluations of Bush—but there seems little doubt that growing disillusionment
with the war has dragged down Bush’s approval ratings over time.
As Figure 3 indicates, public disillusionment with the war was more gradual
than precipitous over the two years following Bush’s reelection, falling
about 10 percentage points over this period. Responses to other questions concerning
the war and its justifications also reveal a gradual erosion of optimism
and support, but with a more pronounced downturn just before the 2006 election.
Figure 4, which displays the trend in beliefs that the war was going well,
serves as an example. Belief that the war in Iraq has made Americans safer
from terrorist attacks shows a similar trajectory and by election day had fallen
below 40 percent. Perhaps more ominous for Republicans, growing pessimism
about the war and its consequences was shared by their own partisans and
Approval of George W. Bush and Support for the War in Iraq
Support the Iraq War Approve of Bush's Job Performance
Saddam Hussein captured
Abu Ghraib scandal
Source: See footnote 7.
9 See Gary C. Jacobson, ‘‘Public Opinion and the War in Iraq’’ (paper presented at the annual
meeting of the American Political Science Association, Philadelphia, PA, 30 August–3 September
6 | POLITICAL SCIENCE QUARTERLY
political independents as well as Democrats (Figure 5), although the partisan
gap on these questions remains very large, with Republicans typically on the
order of 45 percentage points more optimistic than Democrats.10
The distribution of responses to virtually all the diverse questions regarding
the Iraq War places self-identified independents considerably closer, on
average, to Democrats than to Republicans.11 The same holds for evaluations
of the President, with important consequences for voting behavior in 2006,
discussed below. As Figure 6 makes clear, partisan divisions on evaluations of
Bush’s performance continued to be huge during his second term; indeed, by
this measure, Bush is by a wide margin the most divisive president since modern
opinion surveys began asking the approval question more than seventy years
10 See Gary C. Jacobson, A Divider, Not a Uniter: George W. Bush and the American People (New
York: Pearson Longman, 2007), 224–232.
11 The independent category includes those who, when asked, say they lean toward one of the
parties, because data on partisan leaners are not consistently available in the surveys analyzed here.
How Well is the War in Iraq Going?
Very well/somewhat well (CBS/NY Times) Very well/fairly well (Pew) Very well/moderately well (Gallup)
Sources: CBS News/New York Times, Pew Research Center for the People and the Press, and Gallup
polls, accessed at http://pollingreport.com/iraq.htm, 16 November 2006.
THE 2006 CONGRESSIONAL ELECTIONS | 7
ago.12 Although he lost points among ordinary Republicans between his
reelection in 2004 and the midterm, four of five still approved of his performance,
and he was on an upswing with his partisan base as the election
approached. Meanwhile, approval among Democrats had fallen to single digits,
with trends among both sets of partisans, then, pointing to another election
featuring very high levels of party-line voting. The most consequential trend,
however, appears among independents, whose approval ratings of the President
fell about 15 points between the 2004 and 2006 elections.
The sharp partisan divisions provoked by the President made it harder for
Republican campaigns to exploit what was, by the usual measures, a robust
economy. To an extraordinary degree, partisanship now colors Americans’
perceptions of almost anything that can be associated with Bush, including
national economic conditions. Republicans reacted to improving economic in-
Evaluations of How Well the War in Iraq is Going, by Party
Republicans - very/somewhat well (CBS/NY Times) Republicans - very/fairly well (Pew)
Democrats - very/somewhat well (CBS/NY Times) Democrats - very/fairly well (Pew)
Independents - very/somewhat well (CBS/NY Times) Independents - very/fairly well (Pew)
12 Jacobson, Divider, 6–7.
8 | POLITICAL SCIENCE QUARTERLY
dicators by upgrading their assessments, but Democrats and independents have
been comparatively slow to acknowledge the good news (Figure 7). Part of the
reason may be that the benefits of recent economic growth have gone disproportionately
to the affluent and have yet to be felt by ordinary working
people, but whatever the reasons, the data suggest that extolling the economy
was not a particularly effective Republican message for attracting even those
Democrats and independents who thought the economy was a more important
electoral concern than Iraq.
A second major source of the pro-Democratic national tide was public
discontent with the Republican-controlled Congress. In surveys taken during
the month preceding the election, Congress’ approval ratings averaged 27 percent,
lower than Bush’s and only a few points above where they had been when
Republicans had swept the Democrats out in 1994.13 Various scandals, some
involving members’ financial relations with convicted lobbyist Jack Abramoff,
others regarding improper personal behavior, not only gave the Democrats
13 See http://www.pollingreport.com/CongJob.htm, accessed 6 November 2006.
Approval of George W. Bush’s Job Performance, 2001–2006, by Party Identification
Republicans Democrats Independents
Sources: 323 CBS News/New York Times and Gallup polls.
THE 2006 CONGRESSIONAL ELECTIONS | 9
one of their main campaign themes—vote to reject the Republicans’ ‘‘culture of
corruption’’—but also gave them a chance to pick up seats that would not have
been available absent badly tainted Republican incumbents.14 The public also
took a dim view of the Congress’ legislative productivity.15 Negative views of
the Congress, the President, and the war left two-thirds of the electorate dissatisfied
with the way things were going in the United States and believing
that the country had gotten off on the wrong track.16 If the Democrats could
not achieve a national victory under these conditions, it is hard to imagine when
they ever would.
14 These include the seats formerly held by Tom Delay, Bob Ney, and Mark Foley, all of whom had
resigned under a cloud, as well as those defended by Don Sherwood, CurtWeldon, and, in the Senate,
by Conrad Burns.
15 ‘‘Public Disillusionment with Congress at Record Levels,’’ survey report, Pew Research Center
for the People and the Press, 20 April 2006, accessed at http://people-press.org/reports/display.
php3?ReportID=275, 17 November 2006.
16 See the numerous poll results on these questions, accessed at http://www.pollingreport.com/
right.htm, 6 November 2006.
Rating of the Economy, 2001–2006
Percent "Fairly Good" or "Very Good"
Republicans Democrats Independents
Source: Sixty-five CBS News/New York Times polls.
10 | POLITICAL SCIENCE QUARTERLY
STRATEGIC POLITICIANS IN 2006
Favorable national conditions do not, however, automatically deliver victories
to the favored party. Elections are still fought at the local level, and the quality
of candidates and the vigor of their campaigns are crucial determinants of
outcomes. Voters rarely toss out incumbents unless they are offered a qualified
replacement; national issues need effective local sponsors to have their full
electoral impact. Thus, the effects of national political forces are mediated by
the strategic decisions of potential candidates and by the people who control
campaign resources. Unless the favored party’s leaders and activists anticipate
a helpful national tide and prepare to take advantage of it, its effects will be
The Democrats certainly had reason to believe early on that 2006 would be
a good year for them, allowing plenty of time to mobilize candidates and
resources. Responses to the generic House vote question—asking respondents
which party’s candidate they would vote for if the election were held today,
without specifying any candidate’s name—typically gave them a wide lead well
before the election year, a good five points ahead of where they had been at
comparable periods heading into other recent elections (Figure 8). Even discounting
the fact that generic polls always tend to exaggerate the Democrats’
support, the lead was large enough by the end of 2005 to conclude that majority
status was within reach.
This atmosphere helped with recruitment of candidates and fund raising,
and Rahm Emanuel, the chair of the Democratic Congressional Campaign
Committee, and his counterpart in the Senate, Charles Schumer, did an effective
job on both fronts. The proportion of Democratic House challengers with
experience in elective office—one measure of candidate quality—was only average
for recent decades (17 percent), but, as always, experienced candidates were
much more likely to be found in open seats and potentially competitive districts.
Emanuel’s main innovation was to encourage and support candidacies of moderate
to conservative Democrats in districts where mainstream national Democrats
would have faced poor prospects.
More important, though, was that the campaigns of a large proportion
of Democratic candidates showing any promise at all were amply financed
through some combination of contributions and independent spending by party
or outside organizations. Final campaign finance data are not yet available, but
preliminary reports show that the campaigns of at least forty-seven of the fiftyfive
Democrats pursuing Republican-held House seats who ended up with at
least 46 percent of the major-party vote were backed by more than $1 million in
contributions and party spending. Of the remaining eight, two won anyway. For
17 Gary C. Jacobson and Samuel Kernell, Strategy and Choice in Congressional Elections (New
Haven, CT: Yale University Press, 1981), 19–34; Gary C. Jacobson, ‘‘Strategic Politicians and the Dynamics
of House Elections, 1946-86,’’ American Political Science Review 83 (September 1989): 773–793.
THE 2006 CONGRESSIONAL ELECTIONS | 11
only four or five races do the financial data suggest that party officials overlooked
a promising candidate who might have won with a more generous
infusion of cash, an impressive record, considering that the number of seats
estimated to be in play by election handicappers kept growing as the election
drew nearer.18 Republicans also invested massively to defend their vulnerable
House seats—preliminary data indicate that the losing Republican incumbents
on average outspent their Democratic challengers by more than 50 percent19—
and it is conceivable that their efforts saved some incumbents (eleven Democrats
won more than 49 percent of the major-party vote but fell short of a
majority). In general, however, the outcomes of campaigns abundantly funded
by both sides are not determined by who spends more, but by voters’ responses
to the candidates and messages the money is spent to promote.20
Generic House Vote Intention, 1999–2006 (Registered/Likely Voters)
Percent Democratic of Two-Party Vote
Actual vote, 2000: 49.8%
Share of seats: 49.0%
Actual vote, 2002: 47.8%
Share of seats: 47.2%
Actual vote, 2004: 48.5%
Share of seats: 46.5%
Actual vote, 2006: 53.1%
Share of Seats: 53.3%
Source: Compiled by author, largely from data reported at http://www.pollingreport.com, various dates.
18 The five were Larry Kissel (NC 8, 49.9%), VictoriaWulsin (OH 2, 49.4%), Sharon Reiner (MI 7,
48.4%), Nancy Ann Skinner (MI 9, 47.3%), and Larry Grant (ID 1, 47.2%). By Charlie Cook’s calculations,
the number of Republican seats in play (defined as toss-up or leaning Democratic) grew
from 11 in May to 42 in October; see The Cook Political Report accessed at http://www.cookpolitical.
com/races/house/default.php, 6 November 2006.
19 ‘‘House Winners Raised a Record Average of $1.1 Million,’’ press release, Campaign Finance
Institute, 8 November 2006 accessed at http://www.cfinst.org/pr/110806b.html, 27 November 2006.
20 Jacobson, Politics of Congressional Elections, 41–47.
12 | POLITICAL SCIENCE QUARTERLY
Democrats also fielded strong challenges to Republican Senators wherever
prospects looked at all promising,21 and the campaign of every Democrat (and
Republican) in any of the Senate races where the outcome was at all in doubt
was lavishly funded through a combination of contributions, party spending, and
independent spending campaigns. Preliminary data show that the amounts put
into the campaigns of competitive Democrats ranged from more than $8 million
in the low-population states of Montana and Rhode Island to more than $20
million (Missouri); their Republican opponents were at least as well funded. No
Senate candidate in even the most marginally competitive race could reasonably
complain about a shortage of campaign resources.
In short, anticipating a favorable national tide, Democratic operatives, candidates,
and contributors positioned themselves to exploit it, an essential condition
for actually realizing the anticipated gains.
The national climate of opinion gave the Democrats their main campaign
strategies: Attack Republicans for loyally supporting the President and his
misconceived war and for sharing a ‘‘culture of corruption’’ in Congress,
emphasizing the latter especially in states and districts where the incumbent’s
personal record gave the charge local resonance. Frame the choice in national
terms, urging voters to use their franchise to express their unhappiness with the
Republican regime and, most particularly, its leader. Aside from criticizing the
war, emphasize making health care more accessible, raising the minimum wage,
and protecting Social Security—issue domains in which majorities consistently
trust Democrats more than Republicans.
Republican candidates faced a more complicated set of options. One,
standard for the circumstances, was to try to distance themselves from their
party and President (for example, by criticizing aspects of the war or calling for
Defense Secretary Donald Rumsfeld’s resignation), instead emphasizing their
independence, devotion to local interests, and record of delivering valued
projects and services to constituents. The public mood, as well as their Democratic
opponents, worked against this strategy, and it was no more successful
than it had been for Democratic incumbents facing the public’s wrath in 1994.22
An alternative strategy, promoted by the Bush administration and the
President’s political advisor, Karl Rove, was to replace the war in Iraq with
terrorism and homeland security as the dominant electoral focus. The President
obviously had a huge stake in the election, considering its implications for his
21 The list of Democratic challengers in the seven tightest races includes three candidates who had
held statewide offices (governor, attorney general, auditor), two U.S. Representatives, a president of
the state senate, and a former secretary of the navy. The Democrats who retained open seats for the
party were also experienced candidates: the attorney general for Minnesota’s largest county and two
U.S. Representatives (counting Bernie Sanders, an independent who caucuses with the Democrats).
22 Jacobson, Politics of Congressional Elections, 176–182.
THE 2006 CONGRESSIONAL ELECTIONS | 13
remaining legislative agenda as well as the prospect that his administration
could spend its final two years fending off hostile congressional probes of its
decisions and actions. Thus, during the summer, the administration orchestrated
a coordinated, no-holds-barred counterattack against Democratic critics of
Bush and the Iraq War. Taking advantage of the public’s attention to events
commemorating the fifth anniversary of the September 11 attacks, Bush and his
allies sought not only to reinforce the idea that the war in Iraq and the war on
terrorism were one and the same, but also to elevate the conflict to the
equivalent of World War II and the Cold War. The President drew cautionary
parallels between Osama bin Laden’s anti-U.S. fulminations and Lenin’s What
Is To Be Done and Hitler’s Mein Kampf,23 claiming the mantle of Franklin
Roosevelt and Harry Truman as presidents who stood fast in the face of global
threats.24 In addresses delivered around the same time, Vice President Dick
Cheney and Secretary Rumsfeld also variously quoted Roosevelt, took the fight
against Nazism and fascism as precedent, and, more to the political point,
charged that critics of the administration’s policies believed that ‘‘vicious
extremists can be appeased’’25 and that ‘‘retreat from Iraq would satisfy the
appetite of the terrorists and get them to leave us alone’’26
The central message was that Islamic jihadists were as profound a threat
to the existence of the United States as the Axis powers and the Soviet Union
had once been. Democrats, by questioning the wisdom of the Iraq War or the
administration’s conduct of the fight against terrorism more generally, revealed
themselves as appeasers who were blind to the terrorist threat and, if given
control of Congress, would put the security of the United States at grave risk.
As Ken Mehlman, chairman of the Republican National Committee put it,
‘‘The president’s effort to keep Americans safe will grind to a halt with
Democrats in control….’’27 Later in the campaign, Bush told a Republican rally
that ‘‘however they put it, the Democrat approach in Iraq comes down to
this: the terrorists win and American loses.’’28 The campaign was, in short, a
23 GeorgeW. Bush, ‘‘President Discusses the Global War on Terror,’’ transcript of speech delivered
6 September 2006, accessed at http://www.whitehouse.gov/news/releases/2006/09/print/20060905-4.
html, 10 November 2006.
24 George W. Bush, ‘‘President’s Address to the Nation,’’ transcript of speech delivered 9 September
2006, accessed at http://www.whitehouse.gov/news/releases/2006/09/20060911-3.html, 10November 2006.
25 Donald Rumsfeld, (address to the 88th annual American Legion national convention, Salt Lake
City, Utah, 29 August 2006), accessed at http://www.defenselink.mil/Speeches/Speech.aspx?SpeechID=1033,
10 November 2006.
26 Richard B. Cheney, ‘‘Vice President’s Remarks at the Veterans of Foreign Wars National
Convention,’’ (speech, Veterans of ForeignWars national convention, Reno, Nevada, 28 August 2006),
accessed at http://www.whitehouse.gov/news/releases/2006/08/20060828-4.html, 10 November 2006.
27 Quoted in Dan Froomkin, ‘‘Off Message,’’ accessed at http://www.washingtonpost.com/wp-dyn/
content/linkset/2005/04/11/LI2005041100879.html, 18 September 2006.
28 George W. Bush, ‘‘Remarks by the President at Georgia Victory 2006 Rally,’’ Statesboro, GA,
30 October 2006, accessed at http://www.whitehouse.gov/news/releases/2006/10/20061030-4.html, 20 November
14 | POLITICAL SCIENCE QUARTERLY
concerted effort to use the specter of jihadist terrorism to frighten enough
voters into voting Republican in 2006 to keep the Party in control of the House
and Senate, replicating at the congressional level Bush’s successful strategy
against John Kerry in 2004.29
The administration’s campaign included a tactical component in the form
of legislative proposals, submitted to Congress in September, that would put
terrorist suspects outside the protection of the Geneva Conventions, legalize
(unspecified) coercive methods of interrogation, authorize military tribunals to
try terrorist suspects without the usual constitutional protections (including
habeas corpus), and permit the interception of telephone and e-mail messages
between American citizens and contacts abroad without a court order. This
agenda was designed to put Democrats on the spot; any sensitivity to the human
rights or civil liberties issues raised by these proposals, indeed, any opposition
to giving the President unchecked powers to deal with anyone suspected of
links to terrorists as he saw fit, invited the charge of being soft on terrorists.
Those Democrats who balked at this extraordinary delegation of power to the
executive were attacked for being opposed to ‘‘giving President Bush the tools
he needs to protect our country’’ and unwilling to ‘‘bring justice before the eyes
of the children and widows of September 11.’’30 Democrats countered that the
abrogation of fundamental rights for persons accused of terrorism was un-
American, invited international condemnation, and was not needed to combat
and prosecute terrorists. The Republicans had succeeded in cornering them,
however, and fear of being labeled soft on terrorism or willing to ‘‘coddle’’ the
perpetrators of September 11 was enough to induce thirty-four House and twelve
Senate Democrats to support the President’s position on final passage of the bill
authorizing the tribunals and procedures for handling suspected terrorists.
If anyone doubted the terrorism agenda’s electoral motivation, Republican
leaders’ comments after passage laid them to rest. Senate Majority Leader Bill
Frist: ‘‘Do [voters] want to be voting for a party that does unabashedly say, FWe’re
going to have victory in this war on terrorism,_ or a party that says, FWe’ve got to
surrender?_ House Speaker Dennis Hastert: Democrats ‘‘were so bent on protecting
criminals … they’re not allowing us to prosecute these people. The 130
most treacherous people probably in the world, and they want to…release them
out into the public eventually.’’31 (The Democrats who advocated surrender to
terrorists or releasing them from custody naturally went unnamed, as they were
As an attempt to change the subject and to refocus public attention from
what was happening in Iraq to the domestic terrorist threat, the administration’s
29 Jacobson, Divider, 196–197.
30 The first is a quote from John Boehner, Republican majority leader; the second is from James
Sensenbrenner, Jr., chair of the Judiciary Committee; see Charles Babington, ‘‘House Approves Bill
on Detainees,’’ The Washington Post, 28 September 2006.
31 Doyle McManus, ‘‘Detainee Bill Boosts the GOP,’’ Los Angeles Times, 30 September 2006.
THE 2006 CONGRESSIONAL ELECTIONS | 15
campaign largely failed. Although polling data suggested a small upturn in
Republican prospects in late September, whatever traction the campaign had
achieved was soon lost in damaging revelations about the conduct and consequences
of the Iraq War 32 and a revival of the scandal issue with the exposure
and resignation of a Republican congressman whose inappropriate
attentions to male House pages had been known to Republican leaders for
months if not years.33 Moreover, surveys turned up little evidence that ordinary
citizens were swayed by the administration’s rhetoric; it appears, at most, to have
simply reinforced existing (highly partisan) views. In doing so, it may nonetheless
have helped the Republican cause by reviving support among Republican voters
who had been showing signs of disillusionment with the war and thus the
President (see Figure 6). Democrats and independents were a much harder sell;
most of them had long since stopped believing what Bush and his allies were
saying about the war and its justifications.34 The greatest obstacle to the
administration’s attempt to change the subject, however, was the steady stream
of bad news coming out of Iraq; during October, American battle deaths exceeded
100 for the first time since January 2005, and sectarian violence produced
the highest Iraqi monthly civilian death toll recorded to that date.35
The Republicans’ other notable tack, taking a hard-line stance on immigration
through an enforcement-only policy and voting to build a 700-mile fence
along the Mexican border, was also ineffective in attracting voters beyond the
Party’s conservative base. It cost the Party’s candidates support among Latino
voters, the fastest growing segment of the electorate, and was not all that popular
with other groups, either.36 Tellingly, two Republican House candidates in
Arizona, one a six-term incumbent, the other seeking a Republican-held open
seat, who made sealing the border the centerpiece of their campaigns ended up
losing. Republicans also revived their traditional charges that Democrats in
power would raise taxes, overspend, and stunt economic growth, but tax and
spending issues were not high on the list of public concerns in 2006.
As the election approached, the Republicans’ last hope was to outmobilize
the Democrats, as they had in 2002 and 2004, with a carefully prepared
32A classified report from April summarizing the consensus of sixteen U.S. intelligence-gathering
agencies, leaked in late September, concluded that the terrorist threat was growing rather than
shrinking and that the Iraq War was increasing rather than diminishing the number of terrorists. See
‘‘Declassified Key Judgments of the National Intelligence Estimate. Trends in Global Terrorism:
Implications for the United States,’’ April 2006, accessed at http://media.washingtonpost.com/wp-srv/
nation/documents/Declassified_NIE_Key_Judgments_092606.pdf, 29 September 2006. Shortly thereafter,
Bob Woodward published State of Denial (New York: Simon & Schuster, 2006), the title
declaring the book’s thesis.
33 Jackie Calmes, ‘‘Scandal May Further Alienate Republican Base,’’ The Wall Street Journal,
3 October 2006.
34 Jacobson, ‘‘Public Opinion and the War in Iraq,’’ 24–25.
35 Nancy Trejos, ‘‘UN: Iraqi Civilian Deaths at New High,’’ The Washington Post, 22 November 2004.
36 See the collection of survey data at http://www.pollingreport.com/immigration.htm, accessed
18 November 2006.
16 | POLITICAL SCIENCE QUARTERLY
get-out-the-vote operation micro-targeting conservative Republicans and other
voters whose tastes or other characteristics supposedly opened them to Republican
appeals. The experience of the earlier elections, and Karl Rove’s confident
prediction that Republicans would retain their majorities, based, he said, on his
own internal analyses that ignored the polls,37 kept Democrats nervous until the
end, but the pro-Democratic national tide proved too strong to be contained by
any of the Republicans’ organizational countermeasures.
THE NATIONALIZED ELECTION
The Democrats’ efforts to nationalize the election and to make it a referendum
on President Bush and the Republican Congress largely succeeded. As Figure 9
shows, more than a third of the electorate said that their vote for Congress was a
vote against Bush, a noticeably larger proportion than for any of his three
predecessors at midterm, including Bill Clinton in 1994. The reversal from 2002,
when an unusually high proportion of voters said their vote would be an
expression of support for President Bush, is especially striking. The proportion
of voters who said control of Congress would be a factor in their vote was also
considerably higher than usual (Figure 10). Democrats were especially inclined
to take this view (more than 70 percent did so); they were also more enthusiastic
about voting than Republicans.38 Negative opinions of presidential performance
tend to motivate voters more strongly than positive opinions,39 and
among Democrats, strongly negative views of Bush were the norm in 2006. The
extent to which hostility to the President and the Iraq War animated ordinary
Democrats was underlined by the fate of Senator Joseph Lieberman, Al Gore’s
running mate in 2000 but a staunch Bush ally in the Iraq War, who lost the
Connecticut Democratic primary to an anti-war candidate and had to depend
on Republican votes to win reelection as an independent (he continues to
caucus with the Democrats).40
37 ‘‘Karl Rove on Why He Believes the RepublicansWill Keep the House and Senate Despite Polls
to the Contrary,’’ interview broadcast by NPR’s All Things Considered, 24 October 2006, accessed at
http://www.npr.org/about/press/061024_rove.html, 24 November 2006.
38 ‘‘Republicans Cut Democratic Lead in Campaign’s Final Days,’’ research report, Pew Research
Center for the People and Press, 6 November 2006, accessed at http://people-press.org/reports/
display.php3?ReportID=295, 27 November 2006.
39 Samuel Kernell, ‘‘Presidential Popularity and Negative Voting: An Alternative Explanation of
the Midterm Congressional Decline of the President’s Party,’’ American Political Science Review 71
(March 1977): 44–66.
40 According to the exit poll, Lieberman won the votes of 70 percent of the Republicans, 33 percent
of the Democrats, and 54 percent of the independents; only 21 percent of Republicans voted for the
Republican candidate, suggesting an extraordinarily high level of strategic voting on their part.
Connecticut exit poll results are available at http://www.cnn.com/ELECTION/2006/pages/results/
states/CT/S/01/epolls.0.html, accessed 28 November 2006.
THE 2006 CONGRESSIONAL ELECTIONS | 17
VOTING BEHAVIOR IN 2006
Results of the major academic election surveys are not yet available for analysis,
but pre-election surveys and the election day exit poll provide a consistent
account of the broad shifts in voting behavior that contributed to the Democrats’
victory in 2006. As in other recent elections, partisans were very loyal to
their House candidates (Table 2). But whereas in 2002 and 2004, Republicans
were a few points more loyal than were Democrats, the opposite was true in
2006. Moreover, according to the exit polls, the partisan composition of the
electorate was slightly more favorable to the Democrats in 2006 than it had
been in 2004. However, the largest single contribution to the Democrats’ gains,
according to these data, came from independents.41 Voters classifying themselves
as independents had favored Republican candidates in 2002 and had
given the Democrats a modest edge in 2004; in 2006, they broke decisively for
41 Independents who lean toward one of the parties—and who are usually as loyal as weak
partisans—have to be treated as independents in this analysis, because these surveys do not consistently
distinguish partisan leaners from pure independents.
Is Your Vote For Congress a Vote For or Against the President?
0 5 10 15 20 25 30 35 40
Reagan 1982 (1)
Reagan 1986 (2)
GHW Bush 1990 (1)
Clinton 1994 (3)
Clinton 1998 (7)
GW Bush 2002 (4)
GW Bush 2006 (10)
Against the President For the President
Note: The number of polls averaged for each year is in parentheses.
Source: Pew Research Center for the People and the Press, ‘‘October 2006 Survey on Electoral
Competition: Final Topline,’’ 17–22 October 2006, accessed at http://people-press.org/reports/
questionnaires/293.pdf, 15 November 2006.
18 | POLITICAL SCIENCE QUARTERLY
the Democrats. Calculations based on 2004 and 2006 exit poll data indicate that
nearly half the total vote swing to Democratic House candidates between these
elections was supplied by independent voters, although they comprise only
about a quarter of the electorate.
Statewide exit polls indicate that independent voters were also the key
to Democratic victories over Republican Senate incumbents in several of the
‘‘red’’ states. The distribution of partisans and the incidence of party-line voting
reported in Montana, Missouri, and Virginia suggest that the Democrat would
not have won without a clear majority of independent voters (Table 3). It is
noteworthy that Harold Ford, who won but a bare majority of the independent
vote in Tennessee, was the only Democrat in a hotly contested Senate race who
came up short. Ohio evidently would have gone to the Democrat even without
the lopsided support of independents. The same is true of Pennsylvania, but the
Democratic victory in the other ‘‘blue’’ state, Rhode Island, was a product of
the Party’s two-to-one advantage in party identifiers and their desertion of
incumbent Republican Lincoln Chafee; he won 46 percent of the Democrats’
votes in 2000 but only 15 percent in 2006. A similar, if less-pronounced partisan
advantage, ensured that a Democrat retained the only Democratic Senate seat
considered in play on election day (held by Robert Menendez in New Jersey).
Is the Issue of Which Party Controls Congress a Factor in Your Vote?
45 46 44 43
1998 (5) 2000 (2) 2002 (5) 2004 (1) 2006 (6)
Note: The number of polls averaged for each year is in parentheses.
Source: See Figure 9.
THE 2006 CONGRESSIONAL ELECTIONS | 19
A compelling explanation for the Democrats’ lead among independents
in 2006 is provided by the data in Figure 6 showing that the distribution of independents’
opinions on Bush’s performance had grown much more similar
to that of Democrats than of Republicans. In surveys taken during the month
before the election, Bush’s average approval rating among independents of
29 percent was 50 points below his average rating among Republicans (79 percent)
and only 20 points above his rating among Democrats (9 percent). Similarly,
independents’ average level of support for the Iraq War during this
period, at 36 percent, was more than twice as far below that of Republicans
(73 percent) as it was above that of Democrats (19 percent).42 The relationship
between these opinions and voters’ preferences in 2006 is illustrated in
Figure 11. Views on Bush and the Iraq War affected the preferences of people
in all three categories but made a much larger difference for independents
than for partisans. Thus, the predominantly negative opinions on Bush and
the war among independent voters produced a decisive Democratic advantage
in this segment of the electorate.
AGGREGATE VOTING PATTERNS
Although a strong pro-Democratic national tide was running in 2006, it was
not by itself sufficient to deliver control of Congress to the Democrats. The
Democrats’ share of the total House vote nationally increased by nearly five
percentage points over 2004, and the vote swing to Democrats in districts
contested in both 2004 and 2006 averaged a little over five points. But only five
42 See Gary C. Jacobson, A Divider Not a Uniter: GeorgeW. Bush and the American People, updated
with a postscript for 2006 (New York: Pearson Longman, forthcoming, 2008).
Partisan and Independent Voters in U.S. House Elections, 2002–2006 (Percentages)
Loyal Republicans Loyal Democrats Independents Voting Democratic
2002 Pre-election polls (5)a 95 94 46
2004 Pre-election polls (3) 96 94 55
2004 Exit poll 93 91 52
(37)b (37) (24)
2006 Pre-election polls (5) 94 97 61
2006 Exit poll 92 93 59
(36) (38) (26)
Sources: Pre-election polls from ABC News/Washington Post, CBS News/New York Times, Gallup,
Newsweek, and Pew Research Center for the People and Press polls taken in late October and early November
of the election year; ABC and Pew polls did not ask the House vote question in 2004. Exit poll data are from the
National Election Poll exit polls; data include major-party voters only.
aNumber of polls averaged.
bPercent of respondents in partisan category in the exit polls.
20 | POLITICAL SCIENCE QUARTERLY
of the thirty Democratic pickups would have been achieved with a five-point
increase in the Democratic vote over 2004; nineteen required swings of ten or
more points to put the Democrat above 50 percent. The actual swing in the
districts Democrats took from Republicans averaged fourteen points, exceeding
ten points in twenty-three of them. These results underline the crucial
contribution of the Democrats’ strategic deployment of campaign resources—
candidates, money, personnel—to their success in turning a favorable partisan
tide into the victories that produced their House majority. The tide by itself was
not enough; in most districts, it had to be effectively exploited by Democratic
candidates at the district level to have an impact.43 Nonetheless, a couple of lowspending,
relatively obscure Democrats did win unexpected victories, not so
much because local voters had changed their opinions of the Republican
incumbent but because so many former supporters thought it more important
this time to vote their opposition to the Republican regime and its leader.44
The Democrats’ average share of the total vote cast for senator was also up
a little more than five points over 2000, the last time the same set of seats was
contested. Their Senate majority depended on local Democratic swings larger
than the national average in two states (Ohio and Rhode Island), but their
other takeovers came in states where the Republican’s vote margin in 2000
43 Jacobson, Politics of Congressional Elections, 199–200.
44 This category would include Carol Shea-Porter, who defeated Joseph Bradley III in New
Hampshire, and David Loebsack, who defeated James Leach in Iowa.
Partisan and Independent Voting in Hotly Contested Senate Elections, 2006 (Percentages)
Loyal Republicans Loyal Democrats Independents Voting Democratic
Montana 89 91 63
Burns (R) vs. Tester (D) (39)a (32) (29)
Missouri 91 92 54
Talent (R) vs. McCaskill (D) (39) (37) (25)
Ohio 86 91 65
DeWine (R) vs. Brown (D) (37) (40) (23)
Virginia 93 94 56
Allen (R) vs. Webb (D) (39) (36) (26)
Tennessee 94 93 51
Corker (R) vs. Ford, Jr. (D) (38) (34) (28)
Pennsylvania 86 93 72
Santorum (R) vs. Casey (D) (38) (43) (19)
Rhode Island 94 84 45
Chaffee (R) vs. Whitehouse (D) (18) (38) (44)
New Jersey 90 92 48
Kean, Jr. (R) vs. Menendez (D) (28) (41) (31)
Source: VNS exit polls available at http://www.cnn.com/ELECTION/2006/, accessed 10 November 2006;
data include major-party voters only.
aPercent of respondents in partisan category.
THE 2006 CONGRESSIONAL ELECTIONS | 21
had been narrow enough to be overcome with no more than a five-point swing.
Four of the defeated Republicans had themselves entered the Senate by
defeating incumbents, underlining the basic competitiveness of these states. The
Senate results repeat the pattern evident in several other elections (notably
1980, 1982, 1986, and 1994) in which one party swept the lion’s share of the
hotly contested races.45
Historical parallels to the 2006 midterm congressional elections can be found
in the elections of 1950 and 1966 as well as 1994. In all three of these elections,
the economy was in decent shape but voters were unhappy with the president,
in the first two cases, in part because of increasingly unpopular wars (Korea and
Vietnam, respectively), and the president’s party suffered.46 None of these
elections, however, produced a durable change in the party balance in the
electorate, and, according to the data now available, there is little reason to
believe 2006 will be any different. The new Democratic majorities are far from
secure. Democrats did pick up nine seats in Democratic-leaning districts and
Opinions on George W. Bush, the Iraq War, and the Candidate Preference in 2006
80% 16% 32% 65% 9% 87% 74% 19% 29% 63% 16% 78%
Approve Disapprove Right Thing Wrong Thing
Republicans Independents Democrats Republicans Independents Democrats
Bush's Job Performance Rating Judgment on Going to War in Iraq
Percent in Category :
Source: SRBI/Time Magazine Poll #2006-3970, 1–3 November 2006.
45 Jacobson, Politics of Congressional Elections, 200–204.
46 Truman’s Democrats lost a net twenty-eight House seats and five Senate seats in 1950; Johnson’s
Democrats lost forty-seven House seats and four Senate seats in 1966.
22 | POLITICAL SCIENCE QUARTERLY
four in neutral districts (as defined for Figure 1), but eighteen were won in
Republican-leaning territory, including ten districts that had given Bush at least
55 percent of the vote in 2000 and 2004. Even with the advantages of incumbency,
retaining their House majority will be no simple task, for the Republicans’
formidable structural advantage remains securely in place.
The election did disprove one familiar canard: that partisan gerrymandering
has virtually eliminated competitive House districts. At sixty, the number of
House seats won with less than 55 percent of the major-party vote in 2006
was the highest since 1948. The election also confirmed that incumbents’
electoral safety is contingent rather than automatic; fourteen of the twenty-two
losing Republican incumbents had coasted in with more than 60 percent of the
vote in 2004.47 But it is also true that it required a powerful partisan tide to
produce such results. Competition should also be comparatively widespread and
intense in 2008, particularly if, following the usual pattern, the pro-Democratic
tide recedes and Republicans see a chance to win back the Congress.
The Democrats’ prospects for keeping their majorities in 2008 will of course
depend to a considerable extent on what these majorities do in the 110th Congress.
As in 1994, the election was far more a rejection of the ins than an
endorsement of the outs (the mythology regarding the contribution of their
‘‘Contract with America’’ to the Republican triumph in 1994 notwithstanding48).
If the election conveyed any mandate, it was for a change of direction in Iraq,
greater honesty in Congress, and, perhaps, more congressional attention to
matters affecting the lives of ordinary people, such as, for example, access to
medical care. None of these will be easy to achieve. Although most voters are
unhappy with the course of the Iraq War, most also oppose immediate withdrawal,
and congressional Democrats have not found it easy to articulate a
plausible strategy for ending American involvement without leaving behind a
bloody civil war and a vast humanitarian disaster. President Bush will continue to
have the final say in the war’s conduct, and having declared Iraq the main front in
the war on terrorism, it is difficult to imagine that he will disengage short of
something that could be portrayed as victory, for doing so would be to concede
failure on his presidency’s defining mission. Ethics reform in Congress is never
easy, and it is unlikely that Democrats will swear off the earmarking that has
come to symbolize the problem, for it has become part of the pork-barrel culture
that is encoded in the DNA of the institution.49 A Democratic Congress
may be able to boost the minimum wage, but making health care universal and
affordable or shoring up Social Security and Medicare for future retirees are far
47 Thomas Mann, Unsafe at Any Margin: Interpreting Congressional Elections (Washington DC:
American Enterprise Institute, 1980); Jacobson, Politics of Congressional Elections, 48–49.
48 Gary C. Jacobson, ‘‘The 1994 House Elections in Perspective,’’ Political Science Quarterly 111
(Summer 1996): 209.
49 Diana Evans, Greasing the Wheels: Using Pork Barrel Projects to Build Majority Coalitions in
Congress (New York: Cambridge University Press, 2004).
THE 2006 CONGRESSIONAL ELECTIONS | 23
more difficult tasks. And any legislation that cannot win significant Republican
backing will be subject to filibuster in the Senate and Bush’s vetoes.
Republican losses occurred disproportionately among their few remaining
moderates, moving the Party’s center of gravity to the right for the 110th Congress.
The incoming Democrats are ideologically diverse and will most likely
increase the Party’s overall ideological dispersion in both chambers, underlining
the challenge leaders will face trying to unify the Party behind a common
legislative agenda. The consensus of postelection punditry that Democrats will
succeed politically only if they pursue popular centrist policies, resisting any
impulse toward a hard left turn, seems well founded. Among the mass public,
self-described conservatives outnumber liberals by a widemargin, and moderates
outnumber both. Elementary arithmetic makes it plain that Democrats will have
to retain the support of the centrist, swing voters who were essential to their
victory in 2006 if they are to have any chance of repeating it in 2008