Party identification in the 2008 presidential election.
Winneg, Kenneth ; Jamieson, Kathleen Hall
Since Belknap and Campbell tied the concept of party identification
to political behavior in 1951, party affiliation has played a central
role in explanations of individual political behavior in the United
States. Indeed, as Nie, Verba, and Petrocik noted in 1979, in the 1950s
and 1960s, party identification "was the central thread running
through interpretations of American politics," where it was
considered "a stable characteristic of the individual: it was
likely to be inherited, it was likely to remain steady throughout the
citizen's political life, and it was likely to grow in strength
during that lifetime" (47). However, political scientists also
argue that party identification is continually updated to reflect
economic performance and presidential approval (Erikson, MacKuen, and
Stimson 1998; Fiorina 1981; MacKuen, Erikson, and Stimson 1989). Efforts
by Green, Palmquist, and Schickler (1998, 2002) to replicate the
macropartisanship effect found less influence for economic sentiment and
evaluation than MacKuen, Erickson, and Stimson (1989) identified and
elicited questions about the proper form of the time-series model and
measurement (Johnston 2006). Johnston (2006) concluded that the impact
of partisan predisposition on opinion and values, on perceptions of
performance and of candidates, and on the vote itself is strong, but
what is unclear is the "resistance of PID to external forces."
Party identification is important, in part, because of its power to
predict presidential vote choice (Barrels 2000). Recent evidence
suggests that it has become more important over time. Analyzing data
from the American National Election Study (ANES) from 1952 to 2004,
Bafumi and Shapiro (2009) showed that party identification achieved its
highest predictive value in 1996 and 2004. Consistent with that
analysis, the 2008 National Exit Poll revealed that 9 out of 10
self-identified Republicans reported casting their ballot for Arizona
senator John McCain, and 89% of self-styled Democrats sided with the
nominee of their party (CNN 2008). Moreover, Gallup tracking data
revealed that in 2008, "the rank-ordering of the states on the
Democratic-to-Republican continuum generally follows the election
results quite closely--Obama won 22 of the 23 most Democratic states
(West Virginia being the only exception), and McCain won the 17 most
Republican states" (Jones 2009a). Indeed, strong party attachment
predicts straight-ticket voting (Lewis-Beck et al. 2008; Schaffner,
Streb, and Wright 2001).
The drop in the number of citizens reporting a strong tie to a
party, a drop that occurred between the mid-1960s and mid-1970s and
leveled off in the late 1970s and early 1980s, did not change the fact
that most continued to identify with a party. But the change in partisan
affiliation did focus scholars on its change from one election to
another.
In 2004, both the National Annenberg Election Survey (NAES) and
surveys conducted by the Pew Research Center for the People and the
Press showed that the gap in self-identified party affiliation between
Democrats and Republicans was closing. However, because Republicans
turned out in higher numbers than Democrats in 2004, this difference did
not produce a net Democratic advantage on election day when, for the
first time since modern exit polling began in 1976, the number of
self-identified Republicans who voted equaled the number of people
saying they were Democrats (37% each) (CNN 2004; Pew Research Center
2005). As Kerry-Edwards pollster Mark Mellman observed at the 2004
Annenberg election debriefing, "In the '70s and '80s,
Democrats on Election Day had 15-point margins. By the time we got to
the '80s, those were 2 and 3 and 4-point margins. It is right to
say that today this was the first election where the exit polls showed
parity" (Jamieson 2005).
In 2000, the NAES and ANES observed a narrowing of the gap
(Johnston, Hagen, and Jamieson 2004, 42-43). Since 2004, the gap has
instead widened. The parity that Mellman referred to has now moved back
to an advantage for the Democratic Party. What we saw in 2008 was a
reversal in turnout, with Democrats voting in higher proportions than
Republicans (American University 2008). At the same time, Democratic
identification grew. In his analysis of the 2006 congressional elections
and the 2008 presidential and congressional elections, Jacobson (2009)
attributes the change to voters seeing the election more as a referendum
on Bush and the Republican Party. As such, his argument is more in line
with MacKuen, Erikson, and Stimson (1989) and Erikson, MacKuen, and
Stimson (1998). However, it is unclear whether the shift in 2008 is
sufficient to signal a party realignment in favor of the Democrats
(Caraley 2009).
Research Questions and Methodology
This essay will address the following questions: How does 2008
party identification compare with 2004? If there is a widening gap
between Democrats and Republicans, how does that change the demographic
and regional makeup of the parties? Also, how stable was party
affiliation during the course of the 2008 presidential campaign? Was
Democratic self-identification consistently rising as the campaign
progressed, or was there a noticeable level of variability across the
campaign? Barack Obama won the 2008 election with some support from
voters who cast their ballots for George W. Bush in 2004. Was there a
significant shift in party self-identification for those Bush voters?
Also, how did their rating of Bush as president compare with those 2004
Bush voters who did not vote for Obama in 2008?
Our research builds on the work we did in analyzing party
identification in the 2000 and 2004 elections and moves beyond much of
the extant research, which has focused on party changes from campaign to
campaign over time. Rather, the NAES captures changes occurring within
the course of an election. (1)
We employed two distinct methodologies in the 2008 NAES survey: a
rolling cross-sectional telephone survey and a five-wave online panel.
Using the rolling cross-sectional design (Romer et al. 2004, 2006), the
2008 National Annenberg Election Survey interviewed 55,590 adults in the
United States by telephone from December 17, 2007, to November 3, 2008.
The 2004 NAES survey was conducted by telephone from October 7, 2003,
through November 16, 2004, among 77,993 U.S. adults. The 2008 NAES
Internet panel survey completed 95,464 interviews across the five waves
beginning in October 2007 and concluding on January 31, 2009. There was
no NAES Internet panel in 2004.
The telephone survey employed the party identification question:
"Generally speaking, do you usually think of yourself as a
Republican, a Democrat, an Independent or something else?"
Interviewers recorded verbatim responses from those who said
"something else." The question wording used in the NAES
telephone survey was the same as that used in the ANES. The question
that the online survey employed for party identification was slightly
different from the telephone version: "Generally speaking, do you
think of yourself as a Republican, Democrat, Independent, Another party,
please specify, or no preference?" In both instruments, the NAES
interviewers asked the party identification question prior to the vote
intention question, with several questions in between. However, research
suggests that question order has no significant impact on party
identification (McAllister and Wattenberg 1995). In our analysis for
this article, we focus on party identification among adults
self-identified as U.S. citizens for the entire field period of the 2004
NAES and the 2008 NAES. (2)
Results
The results that we present here do not include
"leaners"--that is, independents or "others" who are
asked whether they lean more toward the Democratic or Republican party.
We present only the root question in our findings because the inclusion
of leaners tends to contribute to greater volatility in party
identification (Converse 1966).
Compared with the 2004 NAES telephone survey, aggregate Democratic
Party identification in 2008 increased significantly (i.e., by nearly 3
percentage points) in the telephone survey, from 33.5 % to 36.1%. This
finding is consistent with that of Gallup, which found that Democratic
identification increased 2 points, from 34% to 36%, over that period.
The proportion of self-identified independents increased by a nearly
identical proportion, from 26.1% in 2004 to 28.8% in 2008. Republican
self-identification declined by about half of the amount of the combined
increase in Democrats and independents, from 29.9% to 27.4% (-2.5
percentage points). By contrast, Gallup found a drop from 34% to 28%
(Saad 2008).
One might have expected the Republican decline in our survey to be
larger, but in 2004, more respondents declared themselves to be
something other than a Democrat, Republican, or independent than did
those in 2008. Nevertheless, the increase in the proportion of
self-identified Democrats and independents and the decrease in
self-identified Republicans were all significant. As a result of the
gains in Democratic self-identification and the decline in Republican
self-identification, the gap between the two parties has increased to 9
points (see Table 1). The margin of sampling error for both the 2008 and
2004 findings is less than 1% in either direction. (3)
These findings are very similar to the Pew Research Center's
aggregate survey numbers on party identification from both 2008 and
2004. Pew showed a 36% to 25% Democratic Party edge in 2008 and a 33% to
30% edge in 2004 (Pew Research Center 2005, 2009).
Individual Shift in Party Self-Identification
Nearly all of the respondents who participated in Wave 1 and Wave 5
of our Internet panels maintained their initial party
self-identification in the final wave. A nearly equal proportion of Wave
1 self-identified Democrats (1.7%) and Republicans (1.8%) switched to
the other party. A greater proportion switched their identification to
independent (4.2% of Wave 1 Democrats and 6.7% of Wave 1 Republicans).
The greatest shift was seen among Wave 1 independents. Nearly twice as
many Wave 1 independents identified themselves as Democrats in Wave 5 as
identified themselves as Republicans in that wave (16.5% and 9.8%,
respectively) (see Table 2). (4)
Demographic Breakdowns of Party Identification--Telephone Survey
The breakdown by demographic and geographic variables is based on
all adult citizens, age 18 or older, for the entire NAES telephone field
periods of 2004 and 2008. We demonstrated in 2004 that Republicans had
made gains among registered voters since the 2000 election across many
key demographic variables (Winneg and Jamieson 2005). Comparing all
adults in 2004 and 2008, any gains found in 2004 were either erased or
reversed in 2008. In every demographic category we measured in 2008,
self-identified Republicans declined in number or were not significantly
different in their proportions compared to 2004. The greatest decrease
was found among Hispanics, a group that McCain lost badly to Obama in
the 2008 presidential election. In 2004, 24% of Hispanics and Latinos
said that they were Republicans. In 2008, that number declined by more
than 6 points to 17.7%. While the proportion of Republican Hispanics and
Latinos declined, the share of self-identified Hispanic and Latino
Democrats increased from 40.7% in 2004 to 47.3% in 2008. The shift in
Hispanic and Latino party identification went almost entirely to
Democrats, as there was no change in the proportion of independent
Hispanics and Latinos from 2004 to 2008 (see Table 3).
Additionally, the results reveal increases in Democratic
self-identification among African Americans. Perhaps because of the
uniqueness of the 2008 election, with the first African American
presidential candidate, there was a greater sense of enthusiasm and
eagerness to participate among African Americans. The proportion of
African Americans who called themselves Democrats increased more than 5
points, from 65.2% in 2004 to 70.9% in 2005. Whites declined slightly
among the ranks of Republicans (34.9% to 32.7%), and increased even more
slightly among those calling themselves Democrats, from 28.0% in 2004 to
29.8% in 2008.
Self-identified Democrats increased among Asians as well. The
proportion of Asian American Democrats jumped nearly 5 points, from
35.6% in 2004 to 40.2% in 2008. In short, the growth in the Democratic
Party identification advantage seen in 2008 was largely attributable to
the shifts among minorities, African Americans and Hispanics, rather
than to gains among whites (see Table 3).
Republican declines with commensurate Democratic gains were evident
among those from lower-middle-income households, middle-aged adults
between the ages of 30 and 44, and adults residing in the Northeast
United States (see Table 3).
Furthermore, there was a shift in party identification by marital
status. In 2004, the NAES showed gains in Republican Party
self-identification among those who were married. However, in 2008, the
proportion of married Republicans declined slightly by 3 points, from
34.5% to 31.2%. The proportion of single (unmarried) Democrats grew by 4
points, from 38.7% to 42.7% (see Table 3).
A gender gap continues to exist for party self-identification, and
it grew slightly in 2008. Women remain proportionately more Democratic;
men are only slightly more likely to self-identify as Republicans. In
2008, 41.2% of women in our survey identified as Democrats, compared
with 30.4% of men. Democratic women increased by 3 points, while
Democratic men increased by 2 points. Among men, 28.6% said that they
were Republican, compared with 26.4% of women. These numbers are smaller
than in the 2004 NAES, when 31.7 % of men and 28.3 % of women said that
they were Republicans.
White Evangelical Protestants became less identified with
Republicans, but a plurality continue to call themselves Republican
(48.9% in 2004 versus 45.3% in 2008). (5) The increase among Democrats
identified as white Evangelical Protestants increased slightly, from
21.4% in 2004 to 23.6% in 2008. White Evangelical Protestants
identifying themselves as independents increased by 4 points, from 21.5%
in 2004 to 25.6% in 2008. This suggests that white Evangelical
Protestants, who did not identify with Republicans, Democrats, or
independents in 2004, chose to align with one of the parties in 2008.
Thus, the results show a larger increase in Democrats and independents
than a decrease in Republicans (see Table 4).
Time-Series Analysis of 2008 Party Identification
One of our key assertions in this article is that party
identification was nor static across the course of the 2008 presidential
campaign. We show this statistically by conducting a time-series
analysis to illustrate the variability across the campaign. Figure 1
uses time series to show the variability in party identification during
the 2008 campaign.
[FIGURE 1 OMITTED]
Figure 1 shows changes in party identification from December 17,
2007, to November 3, 2008, based on a regression analysis of the daily
means for each identification--Republican, Democrat, and independent
("other" and "don't know" are excluded). The
line and curves are the predicted regression lines from each analysis.
The regression shows a slight but steady linear decline of Republicans
over the course of the primary and general election campaigns. However,
this movement is not statistically significant, and therefore can be
judged as remaining flat throughout the campaign.
The curvilinear lines for Democrats and independents mirror each
other and show the variability that occurred during the campaign. Rather
than being constant, there is an ebb and flow for Democrats and
independents as the impact of events and campaigns is felt. Although the
changes across time are not dramatic, as we concluded in 2004 (Winneg
and Jamieson 2005), the change that we observed in both of these
elections lends support to the view that party identification should not
be used as a variable in weighting or sample balancing. This further
provides some evidence that party identification is not a constant,
especially in light of the changes from 2004 and the instability that we
see here.
The gap between Republican and Democratic self-identification
widened between 2004 and 2008. While the Republican line remains
essentially flat during the election season, the Democratic line
increases steadily during the Democratic primary campaign and begins to
decline at about the time the primary season ends, close to when Senator
Hillary Clinton suspended her campaign on June 7. This decline continues
throughout the summer and fall. During the same period of the decline in
Democratic Party self-identification, the line representing independents
increases. The increase continued to election day on November 4.
We propose a number of explanations for the rise in independents
along with the decline in Democratic self-identification after peaking
at the end of the primaries. Beginning in mid-July, criticism of Obama
stepped up on two fronts. First, the media began to debate whether
Barack Obama was receiving disproportionately positive coverage
(Maurice, Taney, and Unruh 2008). Second, Obama embarked on an overseas
trip to Europe, where he spoke to massive crowds, especially in Berlin,
Germany. The McCain campaign responded by saying that Obama was more
concerned with earning the acceptance of Europeans than of Americans.
From this criticism sprung the McCain "celebrity" ad. An
August 3 Gallup poll showed the race tightening (Gallup, August 3,
2008).
During the final month of the campaign, we also see an increase in
the number of independents who said that Obama was more likely to raise
taxes, while also thinking that McCain would either lower taxes or leave
them alone. In a regression performed on data from October 8 through
November 3, controlling for candidate favorability and demographic
variables, (6) people who believed that Obama was going to raise taxes
and McCain was going to either lower them or leave them alone were more
likely to switch to an independent identification (b = -.099 [S.E. =
.021], DV Dem = 0 and independent = 0) (see Table 5). We ran a similar
analysis on the 2004 NAES data for Kerry and Bush and found that
independents were more likely than Democrats to believe that Kerry would
raise taxes and that Bush would lower them or keep them at the same
level (b = -.130 [S.E. = .032]) (see Table 5). Still, we cannot show
with certainty the effect these events had on self-identification. (7)
Shift in 2008 Party Identification among 2004 Bush Voters
It is difficult to pinpoint when the increase in Democratic
self-identification occurred. Certainly, the 2006 congressional
elections reflected a shift. One can speculate about the events leading
up to the change, such as the response to Hurricane Katrina, the
continuing war in Iraq, ethical issues with Republican members of
Congress, and the worsening economy. While the NAES data cannot
determine the cause, we can begin to examine the extent of the shift
from 2004 to 2008 by showing how those who said that they voted for
George W. Bush in 2004 identified their party affiliation and voted in
2008. The NAES Internet panel data show that among voters who voted for
Bush in 2004 but cast their ballots for Barack Obama in 2008, 35.8%
identified themselves in 2008 as Democrats, 19.5 % said that they were
Republicans, and 29.1% said that they were independents (see Table 6).
These defectors held significant disaffection with the candidate
for whom they voted four years earlier. This is reflected in the regard
in which they reported holding the George W. Bush administration in
2008. When asked in Wave 5 of the online panel, nearly three in four of
those who voted for Bush and then Obama rated Bush unfavorably, while
two-thirds of those who said they voted for Bush in 2004 but did not
vote for Obama in 2008 rated Bush favorably (see Table 7).
Conclusion
Hershey (2009) suggests that people fall within three categories of
party identification: social identity, perceptual screen, and
changeable. Those who fall within the social identity category are
typically the most loyal, and therefore the most stable. Party
identification is typically passed down to them from their parents. The
second category, party as a perceptual screen, is composed of people who
use party as a way to see the world, filtering out those things that
conflict with their worldview. This group also is likely to be very
stable. The third category consists of people who shift their party
identity based on party stances, the performance of party leaders and
elected officials, and events. Those who do not move from their party
identification, making up the vast majority of people, are those who see
party either as a social identity or as a perceptual screen. The
changeable group, subject to external events and evaluations, is the one
on which we are most focused here and are trying to understand.
Consistent with past research, most people in the 2008 NAES data
did not change party affiliation. Data collected from December 17, 2007
to November 3, 2008 show that party affiliation fluctuated. The pattern
was different in the two elections. While in 2004, the gap between
Democratic and Republican Party identification narrowed as the election
season progressed, the 2008 NAES shows a widening of the gap.
In 2008, there were differences in the amount of movement in each
party. As the campaign progressed, both Democratic and independent
self-identification shifted. Republican Party self-identification showed
no noticeable change. We assume that Republican self-identification was
the lowest of the three, as those who remained in the party were
hard-core party loyalists.
Democratic Party self-identification was significantly greater than
Republican and independent self-identification throughout the campaign.
By the end of the campaign, however, our data showed a decline in
Democratic self-identification and an increase in independent
self-identification. Nevertheless, in aggregate, the Democratic Party
retained its advantage. One explanation for the shift is that in the
NAES, we found a narrowing of the gap in favorability by McCain during a
two-week period between September 15 and September 28, so one might
surmise that a shift toward independents would follow. However, as our
essay notes, the change is very slight. Still, the cause remains an open
question.
There was some change in the demographic makeup of the parties
between 2004 and 2008, with the biggest changes observed among
Hispanics. The Democrats made major gains among this important voting
bloc. The Republican Party made slight gains among men and women.
However, the Democratic Party retained its edge among women. We observed
larger increases in Republican Party identification in those with lower
educational levels.
The 2008 NAES survey data are consistent with the notion that those
who are shifting identification are doing so because they are updating
their evaluation of the economy and the incumbent (Fiorina 1981;
MacKuen, Erikson, and Stimson 1989). As much of the general election was
about the economy, and clearly those former Bush voters who switched to
Obama had a strong negative evaluation of the forry-third president,
there is some evidence supporting macropartisanship. However, here, too,
we want to caution that the changes were slight. Consistent with the
macropartisanship view, as we go to press, there seems to be a narrowing
in the gap between self-identified Democrats and Republicans (Jones
2009b). This closely follows a period in which President Obama's
approval ratings have declined and news reports have been filled with
attacks and counterattacks over health insurance reform. At the same
time, unemployment and the deficit remained high. The result appears to
be a reenergizing of the Republican base and disaffection among
independents for Obama's policies, the current state of the
economy, or both.
Appendix: Methodology
The National Annenberg Election Survey (NAES) is conducted during
each presidential election by the Annenberg Public Policy Center at the
University of Pennsylvania. The 2008 NAES is based on a telephone and
online survey. Telephone interviews began on December 17, 2007, and
concluded on November 3, 2008. The online survey was conducted as a
five-wave panel survey from October 7, 2007, and concluded on January
31, 2008.
The 2004 NAES is based on telephone interviews that began on
October 7, 2003, and concluded on November 16, 2004.
In both the 2004 and 2008 NAES, the sample of telephone exchanges
called was randomly selected by a computer from a complete list of
thousands of active residential exchanges across the country. Within
each exchange, random digits were added to form a complete telephone
number, thus permitting access to both listed and unlisted numbers.
Within each household, one adult was designated by a random procedure to
be the respondent for the survey. The interviewing was conducted by
Schulman, Ronca, Bucuvalas, Inc. The results have been weighted to take
account of household size and number of telephone lines into the
residence and to adjust for variation in the sample relating to
geographic region, sex, race, age, and education.
The data for the NAES Internet component were collected by
Knowledge Networks. Participants were selected from Knowledge
Networks' pool of respondents, who were originally recruited
through random-digit dialing (RDD), and therefore a national probability
sample. (8) Knowledge Networks provides respondents with WebTV units and
Internet access for participation in their surveys. Because a random
subset from the Knowledge Networks pool was drawn for this study, the
NAES utilizes a random sample of the U.S. adult population. In
comparison to telephone RDD surveys, Knowledge Networks samples have
been shown to be equally representative of the U.S. population (e.g.,
see Berrens et al. 2003; Chiang and Krosnick 2009).
To reduce the effects of any nonresponse and noncoverage bias in
panel estimates, a poststratification raking adjustment is applied using
demographic distributions from the most recent data from the Current
Population Survey (CPS). The distributions for Internet access, however,
were obtained from Knowledge Panel recruitment data, as this measurement
is not collected as part of the CPS.
Comparable distributions were calculated using all completed cases
from the field data. Because study sample sizes are typically too small
to accommodate a complete cross-tabulation of all the survey variables
with the benchmark variables, an iterative proportional fitting is used
for the poststratification weighting adjustment. This procedure adjusts
the sample data back to the selected benchmark proportions. Through an
iterative convergence process, the weighted sample data are optimally
fitted to the marginal distributions.
After this final poststratification adjustment, the distribution of
the calculated weights are examined to identify and, if necessary, trim
outliers at the extreme upper and lower tails of the weight
distribution. The poststratified and trimmed weights are scaled to the
sum of the total sample size of all eligible respondents.
The 2004 NAES was conducted from October 7, 2003, through November
16, 2004, reaching 77,993 adults in the United States. The 2008 NAES
telephone survey interviewing reached 55,590 adults. (9) The 2008 NAES
Internet component included 95,464 completed interviews across five
waves from October 1, 2007, through January 31, 2009. (10)
In theory, in 19 out of 20 cases, results for all adults in 2004
will differ by just over one-third of 1 percentage point (+0.35%), up or
down, from what would have been obtained by interviewing all American
adults. For smaller subgroups, the margin of sampling error would be
higher. For the 2008 telephone survey, the margin of sampling error
would be less than one-half of 1 percentage point ([+ or -] 0.41%), up
or down. For smaller groups, the margin of sampling error would be
higher.
In addition to sampling error, the practical difficulties of
conducting any survey of public opinion may introduce other sources of
error into the poll. Variations in the wording and order of questions,
for example, may lead to somewhat different results.
References
American University. 2008. "Much-Hyped Turnout Record Fails to
Materialize, Convenience Voting Fails to Boost Balloting: Report from
the Center for the Study of the American Electorate." News release,
November 6. htrp://wwwl.media.american.edu/electionexperts/
election_rurnout_08.pdf (accessed January 13, 2010).
Bafumi, Joseph, and Robert Y. Shapiro. 2009. "A New Partisan
Voter." Journal of Politics 71 (January): 1-24.
Barrels, Larry. 2000. "Partisanship and Voting Behavior,
1952-1996." American Journal of Political Science 44 (January):
35-50.
Belknap, George, and Angus Campbell. 1951. "Political Party
Identification and Attitudes Toward Foreign Policy." Public Opinion
Quarterly 15 (Winter): 601-23.
Berrens, Robert P., Alok K. Bohara, Hank Jenkins-Smith, Carol
Silva, and David L. Weimer. 2003. "The Advent of Internet Surveys
for Political Research: A Comparison of Telephone and Internet
Samples." Political Analysis 11 (Winter): 1-22.
Caraley, Demetrios James. 2009. "Three Trends over Eight
Presidential Elections, 1980-2008: Toward the Emergence of a Democratic
Majority Realignment?" Political Science Quarterly 124 (Fall):
423-42.
Chiang, Linchiat, and Jon A. Krosnick. 2009. "National Surveys
via RDD Telephone Interviewing versus the Internet: Comparing Sample
Representativeness and Response Quality." Public Opinion Quarterly
73 (Winter): 641-78.
CNN. 2004. 2004. "2004 Election Exit Poll Results."
http://www.cnn.com/ELECTION/2004/pages/
results/states/US/P/00/epolls.0.html (accessed January 13, 2010).
--. 2008. "2008 Election Exit Poll Results."
http://www.cnn.com/ ELECTION/2008/results/ polls/#USP001 (accessed
January 13, 2010).
Converse, Philip E. 1966. "The Concept of a Normal Vote."
In Elections and the Political Order, eds. Angus Campbell, Philip E.
Converse, Warren E. Miller, and Donald E. Stokes. New York: Wiley, 9-39.
Erikson, Robert S., Michael B. Mackuen, and James A. Stimson. 1998.
"What Moves Macropartisanship? A Response to Green, Palmquist, and
Schickler." American Political Science Review 92 (December):
901-12.
Fiorina, Morris P. 1981. Retrospective Voting in American National
Elections. New Haven, CT: Yale University Press.
Gallup. 2008. "Gallup Daily: Election 2008."
http://www.gallup.com/ poll/107674/Gallup-Daily-Election-2008.aspx
(accessed January 13, 2010).
Green, Donald, Bradley Palmquist, and Eric Schickler. 1998.
"Macropartisanship: A Replication and Critique." American
Political Science Review 92 (December): 883-99.
--. 2002. Partisan Hearts and Minds: Political Parties and the
Social Identities of Voters. New Haven,
CT: Yale University Press.
Hershey, Marjorie R. 2009. Party Politics in America. 13th ed. New
York: Pearson Longman.
Jacobson, Gary C. 2009. "The 2008 Presidential and
Congressional Elections: Anti-Bush Referendum and Prospects for the
Democratic Majority." Political Science Quarterly 124 (Spring):
1-30.
Jamieson, Kathleen H., ed. 2005. Electing the President, 2004: The
Insiders' View. Philadelphia: University of Pennsylvania Press.
Johnston, Richard. 2006. "Party Identification: Unmoved Mover
or Sum or Preferences?" American Review of Political Science 9
(August): 329-51.
Johnston, Richard, Michael Hagen, and Kathleen H. Jamieson. 2004.
The 2000 Election and the Foundations of Party Politics. Cambridge, UK:
Cambridge University Press.
Jones, Jeffrey M. 2009a. "Democratic Advantage in Party
Affiliation Shrinks." September 2.
http://www.gallup.com/poll/122693/
Democratic-Advantage-Party-Affiliation-Shrinks.aspx (accessed January
13, 2010).
--. 2009b. "State of the States: Political Party
Affiliation." January 28. http://www.gallup.com/
poll/114016/state-states-political-party-affiliation, aspx (accessed
January 13, 2010).
Lewis-Beck, Michael S., William G. Jacoby, Helmut Norpoth, and
Herbert Weisberg. 2008. The American Voter Revisited. Ann Arbor:
University of Michigan Press.
Maurice, Alexandre, Ryan Taney, and Kristin Unruh. 2008. Campaign
for President: The Managers Look at 2008--Timeline. Presented at the
John R Kennedy School of Government at Harvard University, December
11-12.
MacKuen, Michael B., Robert S. Erikson, and James A. Stimson. 1989.
"Macropartisanship." American Political Science Review 83
(December): 1125-42.
McAllister, Ian, and Martin P. Wattenberg. 1995. "Measuring
Levels of Party Identification: Does Question Order Matter?" Public
Opinion Quarterly 59 (Summer): 259-68.
Nie, Norman H., Sidney Verba, and John R. Petrocik. 1979. The
Changing American Voter. Cambridge, MA: Harvard University Press.
Pew Research Center for the People and the Press. 2005.
"National Security More Linked with Partisan Affiliation: Politics
and Values in a 51%-48% Nation." http://people-press.org/reports/
display.php3?ReportID=236 (accessed January 13, 2010).
--. 2009. "Independents Take Center Stage in Obama Era: Trends
in Political Values and Core Attitudes: 1987-2009."
http://people-press.org/report/517/ political-values-and-core-attitudes
(accessed January 13, 2010).
Romer, Daniel, Kate Kenski, Paul Waldman, Chris Adasiewicz, and
Kathleen H. Jamieson. 2004. Capturing Campaign Dynamics: The National
Annenberg Election Survey. Oxford, UK: Oxford University Press.
--. 2006. Capturing Campaign Dynamics 2000 and 2004: The National
Annenberg Election Survey. Philadelphia: University of Pennsylvania
Press.
Saad, Lydia. 2008. "As Independents Shrink, Democrats
Gain." July 2. http://www.gallup.com/poll/
108619/Independents-Shrink-Democrats-Gain.aspx (accessed January 13,
2010).
Schaffner, Brian E, Matthew J. Streb, and Gerald Wright. 2001.
"Teams without Uniforms: The Nonpartisan Ballot in State and Local
Elections." Political Research Quarterly 54 (March): 7-30.
Winneg, Kenneth, and Kathleen H. Jamieson. 2005. "Party
Identification in the 2004 Election." Presidential Studies
Quarterly 35 (September): 576-89.
(1.) In Winneg and Jamieson (2005), we presented data among
registered voters. In this article, we present data among all adults
regardless of registration, as those who were not registered at the time
of the interview may eventually have registered, so we would be missing
party preference.
(2.) The methodological structure of the rolling cross-sectional
telephone design prevents us from looking at individual shifts in party
self-identification across time. Rather, with the telephone survey data,
we can observe aggregate shifts across the campaign. In the 2008 NAES
Internet panel study, however, we are able to note individual shifts
from the initial wave (Wave 1) and the final postelection wave (Wave 5).
When we discuss the telephone survey, we are focusing on aggregate
changes. When we discuss the Internet panel, we are focusing on
individual shifts in party self-identification.
(3.) The margin of error for the 2004 data is [+ or -] 0.035%. The
margin of error for the 2008 telephone data is [+ or -] 0.042%.
(4.) We compared changes in attitudes toward Obama with changes in
party identification to see whether candidate preferences produced a
change in party identification or whether changes in party
identification stimulated a change in candidate preferences. We found no
effect or relationship.
(5.) The analysis of party identification by religion is based on
data collected from July 7, 2004, through November 16, 2004, and from
July 7, 2008, through November 3, 2008. The 2008 NAES did not insert a
question measuring Evangelical and born-again Christians until July 7,
2008.
(6.) The model controlled for age, education, race, gender, and
favorability toward Obama (feeling thermometer).
(7.) Finally, strong partisans tend to pay closer attention to
political campaigns than weaker partisans. To investigate shifts in
party identification over the course of the campaign, we also examined
whether close attention to the campaign was also associated with Bush
voters voting for Obama, the belief that the economy was worse than it
was in the prior year (both national and personal), and whether the
country was headed in the right direction or off on the wrong track. Our
analysis showed that campaign attention was not significantly associated
with any of those behaviors or beliefs.
(8.) The household recruitment rate is the percentage of
respondents recruited using RDD telephone methods who agreed to be in
the Knowledge Networks sample; for this analysis, the household
recruitment rate is 23.7% for Wave 1 and 21.0% for Wave 5. Of those in
the Knowledge Networks sample who were asked to complete the survey,
67.4% completed Waves 1-5.
(9.) The response rate for the 2008 telephone survey was 20%. The
response rate for the 2004 NAES telephone survey was 21%. All response
rates were calculated based on guidelines set forth by the American
Association for Public Opinion research.
(10.) By wave, 19,190 interviews were completed in Wave 1; 17,747
in Wave 2; 20,052 in Wave 3; 19,241 in Wave 4; and 19,234 in Wave 5.
KENNETH WINNEG AND KATHLEEN HALL JAMIESON
Annenberg Public Policy Center, University of Pennsylvania
Kenneth Winneg is the managing director of the National Annenberg
Election Survey. His research interests are in political communication,
the electoral process, and the Internet and political participation.
Kathleen Hall Jamieson is Elizabeth Ware Packard Professor of
Communication at the Annenberg School for Communication and Walter and
Leonore Annenberg Director of the Annenberg Public Policy Center at the
University of Pennsylvania. She codirected the 2008 National Annenberg
Election Survey.
AUTHORS' NOTE: Dan Romer, associate director of the Annenberg
Public Policy Center, also contributed to this article.
TABLE 1
National Self-Reported Party Identification--NAES Telephone Survey
2004 NAES * 2008 NAES ** Net Change
Republican 29.9% 27.4% -2.5
Democrat 33.5% 36.1% +2.6
Independent *** 26.1% 28.8% +2.7
* Asked of 77,993 U.S. adults between October 7, 2003, and
November 16, 2004.
** Asked of 55.590 U.S. adults from December 17, 2007, through
November 3, 2008.
*** Independents are respondents who indicate they are
"independents." This number excludes those who give a response of
"something else," "don't know," or refuse to answer.
TABLE 2
NAES Online Panel Party Identification Shifts from Wave 1 to Wave 5
Republican Democrat
Wave 1 Wave 1
Wave 5 Party Identification (n = 3,691) (n = 4,011)
Republican (W5) 87.8% 1.7%
Democrat (W5) 1.8% 91.3%
Independent (W5) 6.7% 4.2%
Other party/no preference (W5) 3.8% 2.6%
Independent Wave 1
Wave 5 Party Identification (n = 2,829)
Republican (W5) 9.8%
Democrat (W5) 16.5%
Independent (W5) 66.9%
Other party/no preference (W5) 6.8%
TABLE 3
Party Identification by Key Demographic Groups-2008 versus 2004
(NAES Telephone Survey) among Adults
Generally speaking, Republican
do you usually think
of yourself as a ... 2004 * 2008 **
Total 29.9% 27.4%
Men 31.7% 28.6%
Women 28.3% 26.4%
High school or less 27.1% 24.2%
Some college 31.9% 29.5%
College degree or more 32.8% 29.8%
Men high school or less 28.5% 24.4%
Women high school or less 25.9% 24.1%
Men some college 32.6% 30.9%
Women some college 31.2% 28.3%
Men college degree or more 35.9% 32.1%
Women college degree or 29.6% 27.5%
more
18-29 years old 25.1% 24.4%
30-44 years old 32.5% 27.9%
45-64 years old 29.4% 27.3%
65 and over 31.8% 29.8%
Men 18-29 27.3% 30.6%
Women 18-29 23.0% 22.7%
Men 30-44 34.6% 29.1%
Women 30-44 30.5% 26.8%
Men 45-64 31.2% 28.3%
Women 45-64 27.8% 26.4%
Men 65 and over 32.7% 31.0%
Women 65 and over 31.1% 28.9%
Household income below $35,000 22.8% 19.8%
$35,000 to less than $75,000 31.8% 27.6%
$75,000 and over 37.6% 33.9%
Men below $35,000 24.2% 20.4%
Women below $35,000 21.8% 19.4%
Men $35,000-$75,000 33.1% 29.0%
Women $35,000-$75,000 30.6% 26.4%
Men $75,000 and over 38.7% 35.1%
Women $75,000 and over 36.1% 32.4%
Whites 34.9% 32.7%
African Americans 5.4% 4.8%
Hispanic/Latinos 24.0% 17.7%
Asians 23.3% 19.4%
Native Americans 20.5% 22.2%
White men 36.5% 33.4%
White women 33.6% 32.0%
African American men 6.7% 4.9%
African American women 4.4% 4.6%
Hispanic/Latino men 27.8% 19.0%
Hispanic/Latino women 20.8% 16.5%
Union Household 22.9% 21.5%
Union men 24.5% 22.3%
Union women 21.3% 20.7%
Married 34.5% 31.2%
Single 22.2% 20.3%
Married men 36.0% 31.7%
Married women 33.0% 30.8%
Single men 23.5% 22.2%
Single women 21.1% 18.8%
Northeast 25.9% 22.5%
Midwest 28.6% 26.6%
South 32.4% 30.1%
West 30.8% 28.9%
Northeast men 29.2% 23.7%
Northeast women 23.2% 21.4%
Midwest men 30.4% 27.6%
Midwest women 26.9% 25.7%
South men 33.1% 31.3%
South women 31.7% 29.1%
West men 32.7% 30.2%
West women 28.9% 27.6%
Urban 24.2% 22.3%
Suburban 32.2% 29.5%
Rural 31.9% 29.8%
Urban men 26.0% 23.8%
Urban women 22.5% 21.0%
Suburban men 34.0% 30.8%
Suburban women 30.6% 28.2%
Rural men 33.4% 29.6%
Rural women 30.5% 29.9%
Generally speaking, Democrat
do you usually think
of yourself as a ... 2004 * 2008 **
Total 33.5% 36.1%
Men 28.4% 30.4%
Women 38.1% 41.2%
High school or less 35.7% 38.4%
Some college 30.7% 34.3%
College degree or more 32.6% 34.7%
Men high school or less 31.8% 34.2%
Women high school or less 39.1% 42.3%
Men some college 25.8% 28.3%
Women some college 35.0% 39.3%
Men college degree or more 25.4% 27.2%
Women college degree or 39.8% 41.9%
more
18-29 years old 31.6% 36.3%
30-44 years old 30.6% 34.4%
45-64 years old 34.7% 36.1%
65 and over 38.7% 38.6%
Men 18-29 26.5% 30.6%
Women 18-29 36.5% 41.9%
Men 30-44 25.4% 28.1%
Women 30-44 35.5% 40.4%
Men 45-64 29.7% 30.8%
Women 45-64 39.3% 40.8%
Men 65 and over 34.3% 33.0%
Women 65 and over 42.0% 42.8%
Household income below $35,000 38.8% 43.9%
$35,000 to less than $75,000 32.3% 35.8%
$75,000 and over 29.0% 31.7%
Men below $35,000 33.7% 38.6%
Women below $35,000 42.3% 47.3%
Men $35,000-$75,000 27.7% 30.1%
Women $35,000-$75,000 36.9% 41.3%
Men $75,000 and over 24.5% 26.2%
Women $75,000 and over 34.5% 38.2%
Whites 28.0% 29.8%
African Americans 65.2% 70.9%
Hispanic/Latinos 40.7% 47.3%
Asians 35.6% 40.2%
Native Americans 35.1% 30.6%
White men 23.7% 24.9%
White women 31.9% 34.4%
African American men 57.5% 66.2%
African American women 71.2% 74.2%
Hispanic/Latino men 35.7% 43.3%
Hispanic/Latino women 44.8% 50.5%
Union Household 42.8% 42.6%
Union men 38.3% 37.1%
Union women 47.5% 48.6%
Married 30.4% 32.7%
Single 38.7% 42.7%
Married men 26.4% 28.0%
Married women 34.4% 37.2%
Single men 32.3% 35.6%
Single women 43.9% 48.2%
Northeast 36.0% 39.7%
Midwest 32.3% 34.8%
South 33.7% 35.6%
West 32.4% 34.8%
Northeast men 29.6% 34.2%
Northeast women 41.3% 44.6%
Midwest men 27.6% 29.0%
Midwest women 36.7% 40.0%
South men 28.8% 29.8%
South women 38.0% 40.4%
West men 27.8% 28.9%
West women 36.9% 40.5%
Urban 40.2% 43.4%
Suburban 31.0% 33.3%
Rural 30.8% 32.2%
Urban men 34.3% 37.8%
Urban women 45.5% 48.2%
Suburban men 25.8% 27.3%
Suburban women 35.8% 38.9%
Rural men 27.2% 27.6%
Rural women 34.1% 36.3%
Generally speaking, Independent ***
do you usually think
of yourself as a ... 2004 * 2008 **
Total 26.1% 28.8%
Men 29.2% 32.7%
Women 23.3% 25.2%
High school or less 24.9% 28.1%
Some college 26.9% 28.6%
College degree or more 27.5% 30.0%
Men high school or less 27.4% 31.8%
Women high school or less 22.8% 24.6%
Men some college 30.5% 32.3%
Women some college 23.8% 25.4%
Men college degree or more 31.2% 34.3%
Women college degree or 23.8% 25.9%
more
18-29 years old 28.7% 28.0%
30-44 years old 25.9% 29.2%
45-64 years old 26.8% 29.9%
65 and over 22.3% 26.2%
Men 18-29 31.2% 31.4%
Women 18-29 26.3% 24.6%
Men 30-44 28.7% 33.7%
Women 30-44 23.2% 25.0%
Men 45-64 30.0% 33.5%
Women 45-64 23.9% 26.7%
Men 65 and over 26.2% 30.4%
Women 65 and over 19.4% 23.0%
Household income below $35,000 25.7% 27.3%
$35,000 to less than $75,000 26.6% 29.3%
$75,000 and over 26.4% 28.9%
Men below $35,000 29.0% 31.1%
Women below $35,000 23.5% 24.8%
Men $35,000-$75,000 29.6% 32.8%
Women $35,000-$75,000 23.6% 24.8%
Men $75,000 and over 29.2% 32.5%
Women $75,000 and over 22.9% 25.9%
Whites 27.2% 30.3%
African Americans 18.7% 18.1%
Hispanic/Latinos 23.7% 23.7%
Asians 29.9% 32.7%
Native Americans 28.3% 34.5%
White men 29.8% 33.9%
White women 24.8% 27.0%
African American men 24.7% 22.4%
African American women 14.2% 15.2%
Hispanic/Latino men 25.2% 26.5%
Hispanic/Latino women 22.4% 21.5%
Union Household 24.8% 29.0%
Union men 27.6% 33.3%
Union women 21.9% 24.2%
Married 25.5% 29.0%
Single 27.2% 28.3%
Married men 28.2% 32.7%
Married women 22.9% 25.4%
Single men 31.2% 32.7%
Single women 24.1% 24.8%
Northeast 28.8% 30.6%
Midwest 28.1% 30.5%
South 23.9% 27.1%
West 25.3% 27.6%
Northeast men 31.7% 34.6%
Northeast women 26.3% 27.1%
Midwest men 31.3% 34.6%
Midwest women 25.2% 26.8%
South men 27.4% 31.0%
South women 20.7% 23.8%
West men 27.8% 31.4%
West women 22.8% 23.8%
Urban 25.2% 26.5%
Suburban 26.8% 29.8%
Rural 26.0% 23.9%
Urban men 28.9% 30.2%
Urban women 22.0% 23.4%
Suburban men 30.1% 33.7%
Suburban women 23.7% 26.1%
Rural men 27.9% 33.6%
Rural women 24.2% 25.8%
* Asked of 77,993 U.S. adults between October 7, 2003, and
November 16, 2004.
** Asked of 55,590 U.S. adults from December 17, 2007, through
November 3, 2008.
*** Independents are respondents who indicate they are
"independents." This number excludes those who give a response of
"something else," "don't know," or refuse to answer.
TABLE 4
Party Identification among White Evangelical Protestants (from
July 6, 2008, when the Evangelical question was asked, compared
to July 6, 2004)
Republican Democrat
2004 * 2008 ** 2004 * 2008 **
White Evangelical Protestants 48.9% 45.3% 21.4% 23.6%
Catholics 28.0% 26.1% 37.2% 38.6%
Independent ***
2004 * 2008 **
White Evangelical Protestants 21.5% 25.6%
Catholics 26.5% 27.9%
* Asked of 9,381 U.S. adults classified as white Evangelical
Protestants and 8,677 U.S. adults classified as Catholics between
July 6, 2004, and November 16, 2004.
** Asked of 4,635 U.S. adults classified as white Evangelical
Protestants and 5,163 U.S. adults classified as Catholics between
July 6, 2008, and November 3, 2008.
*** Independents are respondents who indicate they are
"independents." This number excludes those who give a response of
"something else," "don't know," or refuse to answer.
TABLE 5
Effect of Belief that Obama (2008)/Kerry (2004) Will Raise Taxes
on Switching from Democrat to Independent, 2008 versus 2004
Unstandardized Unstandardized
B (SE) B (SE)
(10/8-11/03/2008) (10/13-11/01/2004)
Constant .399 *** (.061) .253 *** (.085)
Race (white/nonwhite) -.102 *** (.018) -.007 (.066)
Education -.015 *** (.003) -.011 * (.005)
Gender (male/female) -.082 *** (.015) .086 *** (.024)
Age .001 (.000) .000 (.001)
Fear of reverse discrimination .003 (.003) n.a.
Favorable to Obama (2008)/ .061 *** (.003) .065 *** (.004)
Favorable to Kerry (2004)
Obama will raise taxes/McCain -.099 *** (.021) -.130 *** (.032)
will keep same or lower
(2008)
Kerry will Raise Taxes/Bush
Keep Same or Lower (2004)
[R.sup.2] 0.191 0.226
* p < .05; ** p < .01; *** p < .001.
N = 3,853 for 2008; 1,370 for 2004.
TABLE 6
Self-Identified Party among 2004 Bush Voters Who Voted for
Obama in 2008
2004 Bush Voters12008
Obama Voters (Internet Panel
Postelection Wave)
(n = 1,025)
Republican 19.5%
Democrat 35.8%
Independent 29.1%
TABLE 7
Bush Favorability (Feeling Thermometer) among 2004 Bush Voters
2004 Bush Voters/2008
2004 Bush Voters/2008 DID NOT VOTE FOR
Obama Voters Obama Voters
(Internet Panel (Internet Panel
Postelection Wave) Postelection Wave)
(n = 1,022) (n = 5,527)
Unfavorable (0-49) 73.2% 20.9%
Neither Favorable Nor 11.5% 12.7%
Unfavorable (50)
Favorable (51-100) 15.3916 66.496
