Research Article
Psychological Science
21(1) 140­–146
© The Author(s) 2010
Reprints and permission: http://www​
.sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797609356421
http://pss.sagepub.com
Wishful Thinking in the 2008
U.S. Presidential Election
Zlatan Krizan, Jeffrey C. Miller, and Omesh Johar
Iowa State University
Abstract
In elections, political preferences are strongly linked to the expectations of the electoral winner—people usually expect their
favorite candidate to win. This link could be driven by wishful thinking (a biasing influence of preferences), driven by a biasing
influence of expectations on one’s wishes, or produced spuriously. To examine these competing possibilities in the 2008 U.S.
presidential election, a longitudinal study assessed uncommitted young voters’ electoral expectations and preferences over
four time points during the month before the election. The findings indicated clear support for wishful thinking: Over time,
people’s preferences shaped their expectations, but the reverse was not the case. Moreover, these relations were larger among
those more strongly identified with their political party and held even when perceptions of general candidate popularity were
taken into account. Finally, changes in electoral expectations were consequential, as they shaped disappointment in the electoral
results even after taking candidate preferences into account.
Keywords
optimism, predictions, election, motivated reasoning, likelihood judgment
Received 3/25/09; Revision accepted 5/21/09
My friends, we’ve got them just where we want them
—John McCain during the 2008 presidential campaign,
Virginia Beach, VA, October 13, 2008 (Mooney, 2008)
Ask a voter during a political campaign who he or she thinks
will win the election, and the answer should not be surprising.
People are usually confident their candidate will win! This
observation was first documented by Hayes (1936) who, in the
1932 U.S. presidential election, observed that 93% of Roose­
velt supporters predicted Roosevelt would win, whereas 73%
of Hoover supporters predicted Hoover would win. Clearly,
voters’ preferences were foretelling of their electoral predic­
tions. Since then, links between preferences and expectations
have been documented with regard to a variety of social and
political events (e.g., Cantril, 1938; Granberg & Brent, 1983;
Olsen, 1997) and constitute one of the most robust findings in
social psychology. Electoral expectations in particular show
strong links with preferences, with correlations usually
exceeding .50 (see Granberg & Brent, 1983).
What can account for this close correspondence of outcome
preferences and relevant expectations? It might be tempting to
conclude that people believe what they want to believe, mean­
ing that their rose-colored glasses prevent them from forming
accurate expectations about relevant outcomes, in this case
electoral ones (e.g., Babad, 1997). This conclusion is all the
more appealing given what we know about motivated reasoning;
there is overwhelming evidence that people’s beliefs and conclu­
sions can be influenced by their motives and goals in various
ways (e.g., Kunda, 1990). Moreover, at least some laboratory
research has directly implicated desirability of outcomes as a
casual factor in shaping outcome expectations (e.g., Irwin, 1953).
Although seductive, the conclusion that wishful thinking is
responsible for preference-expectation links is unwarranted
for two main reasons. First, the experimental support for desir­
ability bias in predictions has been somewhat elusive—it is
confined mostly to outcome predictions in games of chance,
and there is a lack of evidence for wishful thinking in predic­
tions involving more naturalistic settings (see Krizan & Wind­
schitl, 2007, for a review). In short, the experimental evidence
on wishful thinking seems too equivocal for one to infer that
preference-expectation links are necessarily driven by the
biasing influence of wishes or desires.
Second, there is an obvious alternative to the notion that
preference-expectation links are formed by the causal
Corresponding Author:
Zlatan Krizan, Department of Psychology, W112 Lagomarcino Hall,
Iowa State University, Ames, IA 50011
E-mail: [email protected]
141
Wishful Thinking in the 2008 Election
influence of preferences. Namely, the reverse could be true:
People’s expectations could drive their preferences. In the
political arena, there are multiple ways in which this could
occur. Voters might shift their candidate preferences in order
to support the candidate who is perceived most likely to win
(i.e., get on the bandwagon; see Navazio, 1977; Simon, 1954).
Although voters might also shift their preferences to support
the candidate perceived as the underdog (e.g., Ceci & Kain,
1982), such a case would imply a negative relation between
preferences and expectation rather than the positive one usu­
ally observed. Finally, people’s expectations might influence
their preferences outside awareness, as voters cope with the
possibility of their candidate’s defeat by rationalizing in
advance why the unwanted candidate might not be so bad after
all (see Kay, Jimenez, & Jost, 2002). Together, this evidence
indicates there are numerous ways in which expectations
could influence preferences rather than only wishes and pref­
erences influencing expectations.
In sum, by themselves, preference-expectation links are
insufficient to infer that wishful thinking is operating, even
though such an assumption is often made. In fact, many analy­
ses have examined additional factors that could suggest moti­
vated reasoning is at work. For example, Babad (1997)
demonstrated a higher preference-expectation link among
those more strongly committed to a given candidate and lower
links among those who were given monetary incentives for
accurate predictions or were able to accurately report current
polls. Furthermore, Dolan and Holbrook (2001) examined
National Election Study data from 1984 to 1996 and showed
that preferences are less predictive of expectations among
those with more political knowledge, supporting their case that
more accurate knowledge reduces the influence of one’s wishes.
Granberg and Brent (1983) examined similar national data for
the elections of 1952 to 1980 and again found preferenceexpectation links to be highest among those who were highly
invested but poorly informed.
All of these findings are consistent with the wishful-thinking
hypothesis and the general understanding of motivated
cognition—desirability biases should be stronger among those
more invested in the outcome (e.g., Krizan & Windschitl,
2007), motivation to be accurate or accountable should reduce
judgment biases (e.g., Tetlock & Kim, 1987), and constraints
of reality should pose limits on motivated distortion of judg­
ments (e.g., Kunda, 1990). However, the same findings can
again be explained in alternative ways that do not necessitate
a causal influence of preferences. The critical problem involves
the possibility that acquired knowledge about a candidate
drives both preferences and expectations regarding that candi­
date, effectively producing preference-expectation links (i.e.,
a third-variable explanation; see Krizan & Windschitl, 2007).
For example, if you were a young, uncommitted voter who
learned that Barack Obama was the first African American
president of the Harvard Law Review (a very prestigious legal
publication), it would make sense for you both to develop a
more positive attitude toward him and to anticipate his chances
of winning to be higher. For such reasons, studies examining
political knowledge as a moderating factor should also assess
knowledge content (e.g., how biased is it toward one’s candi­
date) in addition to the amount of knowledge (cf. Dolan &
Holbrook, 2001), although this is typically not done (one rea­
son being the complexity involved in such measurement).
Another possibility is that social influence from one’s peers
(e.g., friends and family) could drive both electoral prefer­
ences and inferences about the electorate as a whole (e.g.,
Fischer & Budescu, 1995). In short, given the potential for
reverse causality and third-variable problems, we should
demand additional evidence before concluding that in a given
context, preference-expectation links solely (or even mostly)
reflect wishful thinking.
A Longitudinal Approach
An effective way to examine causality in real-world settings
that at least partially mitigates these problems is to use longi­
tudinal (i.e., cross-lagged panel) designs (see Finkel, 1995;
Locascio, 1982). Such designs increase our confidence in
causal conclusions to the extent we can demonstrate one vari­
able to predict another at a future point in time even after con­
trolling for both variables at initial assessment, and we can do
so over multiple assessment intervals. As mentioned before,
experimental findings do provide some support for wishful
thinking in predictions (Krizan & Windschitl, 2007), but these
findings speak to how variables interrelate in principle, rather
than how they function in real-world settings (see Henshel,
1980). The same can be said for the potential causal role of
expectations—most research supporting bandwagon and
underdog effects has been experimental (e.g., Ceci & Kain,
1982; Navazio, 1977).
Investigators have already used longitudinal approaches to
examine these causal issues. Lazarsfeld, Berelson, and Gaudet
(1948) examined 1940 national election data that included
assessment of preferences (i.e., voting intentions) and expec­
tations (i.e., predictions of the winner) at two different time
points before the election. They concluded that both causal
pathways were operating; there was evidence both for wishful
thinking and bandwagon effects. Granberg and Brent (1983)
reexamined these findings using more appropriate analyses
and reported additional data from the 1980 election that sug­
gested only a causal effect of preferences. However, they also
used only two time points, and barring replication, the rele­
vance of this finding for contemporary political environments
is unclear.
Purpose
To test for presence of wishful and expectation-driven effects
in the 2008 U.S. presidential election, we conducted a longitu­
dinal investigation involving uncommitted student voters in
which preferences and expectations were assessed at four dif­
ferent time points over 1 month before the election. Although
142
these young adults were not necessarily representative of the
general voting population, there were several key advantages
of the approach we used. First, young voters are an increasingly
important demographic in political elections, playing a critical
role in the election of Barack Obama (Tufts University, 2008).
Second, by focusing on uncommitted voters, we were able to
examine dynamics in preferences and expectations that would
be difficult to capture among more typical voters whose politi­
cal preferences tend to be relatively fixed. Third, focusing on
the month before the election enabled capturing those last shifts
in opinions that typify uncommitted voters considered very
important in shaping election outcomes (e.g., Fenwick, Wise­
man, Becker, & Heiman, 1982). Fourth, we assessed both pref­
erences and expectations as continuous variables, enabling us
to examine more fine-grained changes in preferences or expec­
tations not possible with American National Election Studies
data (www.electionstudies.org), where both are measured as
binary variables. Finally, given that Iowa holds the first caucus
in the election season, students in this state were constantly
exposed to information about the election and had multiple
opportunities to learn about the candidates.
We also assessed individuals’ perceptions of poll results
and favorability of candidates’ media coverage at each time
point. To the extent that preferences are found to shape expec­
tations over time even after taking into account perceptions of
candidates’ general popularity, arguments about such variables
driving preference-expectation dynamics over time become
less compelling. In addition, we examined to what extent iden­
tification with a preferred party might moderate wishful think­
ing, that is, the influence of preferences on expectations that
occurs over time. Although this approach is a conceptual par­
allel of examining political investment as a moderating factor
of preference-expectation links mentioned earlier (e.g., Gran­
berg & Brent, 1983), in this case it would allow for a firmer
test of wishful thinking given the longitudinal nature of the
design. Finally, to establish the importance of changes in elec­
toral expectations, we tested whether these expectations
uniquely predicted elation or disappointment with the election
outcome after taking candidate preferences into account.
Method
Participants
One hundred fifty-three students from Iowa State University
served as participants. Students were encouraged to partici­
pate if they were interested in the election and were not fully
committed to a candidate; 57% did so in exchange for $20,
whereas others participated as one means to earn extra credit
for a psychology course. Retention (87% at completion) was
facilitated by making the full reward contingent on continued
participation. All participants were recruited during Septem­
ber 2008.
Participants were e-mailed a link to an initial Web survey
during early October of 2008, which they all completed before
Krizan et al.
Table 1. Characteristics of the Sample
Variable
Gender (n = 148)
Male
Female
Party affiliation (n = 152)
Democrat
Republican
Independent, other
Ideological orientation (n = 152)
Liberal
Moderate
Conservative
Don’t know
Percentage
32.4
67.6
32.2
21.7
46.1
45.3
25.0
23.7
5.8
October 16. To assess basic background information, we asked
participants to indicate their party affiliation (Democrat,
Republican, or Independent/Other) and general political lean­
ings (rated on a scale ranging from 1, extremely liberal, to 7,
extremely conservative). They also indicated social identifica­
tion with the more preferred party on a social identity measure
developed by Greene (1999). This measure requires responses
to 10 statements (e.g., “This group’s successes are my suc­
cesses” or “The limitations associated with this group apply to
me also”), rated on a 4 point scale, ranging from disagree
strongly, 1, to agree strongly, 4 (α = .87). Finally, participants
reported their age, gender, ethnicity, and ACT score.
Participants’ mean age was 21 years (SD = 2.5), and their
mean ACT score was 24.7 (SD = 5.0). As is typical of collegeage populations in the Midwest, most individuals were Cauca­
sian (87%) and expressed liberal leanings (see Table 1).
Importantly, after the election, 83% of participants indicated
they had voted, with 67% of those having voted for Barack
Obama.
Procedure and measures
Critical measures in the initial survey were as follows. Expec­
tations were assessed first with the question, “Who do you
think will win the U.S. presidential election this year?” which
was rated on a 7-point scale ranging from definitely Barack
Obama, 1, to definitely John McCain, 7, with the additional
option of someone else (which no one chose). Preferences
were assessed with the question, “Who will you vote for in the
upcoming presidential election?” rated on a 7-point scale
ranging from I will definitely vote for Obama, 1, through Don’t
know who will I vote for, or if I will vote at all, 4, to I will definitely vote for McCain, 7. Next, to measure perceptions of can­
didates’ general popularity—which served as covariates in the
main analyses—we asked participants, “Which of the presi­
dential candidates do you think is currently leading in the
national polls?” (response options ranged from Obama to
They are tied to McCain) and “According to what you have
143
Wishful Thinking in the 2008 Election
Analytic strategy
We used structural equation modeling by means of Mplus
(Version5.0; Muthén & Muthén, 2007) to test our main hypoth­
eses. Specifically, we fitted a path model to the longitudinal
data where (a) all repeated measures were allowed to correlate
with one another at each time point, (b) all were regressed on
themselves at the previous time point (i.e., stability relations),
and (c) preferences and expectations were regressed on each
other at the previous time point (i.e., cross-lagged relations).
As this model did not quite reach acceptable standards of fit,
comparative fit index (CFI) < .90, root-mean-square error of
approximation (RMSEA) > .10, we examined modification
indexes, which suggested that freeing a single regression path
from preferences at Time 2 to preferences at Time 4 would
lead to considerable improvement in model fit. The model pre­
sented next includes this modification. Note that none of the
substantive results differed between the prespecified and the
modified model.
Results
Mean-level changes
Although not the focus of the present analysis, it is also
informative to consider how the overall level of people’s
preferences and expectations changed over time. As seen in
Figure 1, most people were aware that Obama was more
likely to win and recognized that his chances improved even
further as Election Day drew nearer (which was widely
reported in the media at the time). However, it is also appar­
ent that sympathizers of both parties grew somewhat more
optimistic a couple of weeks before the election, although in
the case of those identified as Republicans this still meant
that Obama was perceived as slightly ahead. In terms of pref­
erences, there did not seem to be any substantial changes
over time (see Fig. 2).
Democrat
Independent
Electoral Expectation
4.5
4.0
Republican
3.5
3.0
2.5
2.0
1.5
1.0
Time 1
Time 2
Time 3
Time 4
Assessment Point
Fig. 1. Democrats’, Republicans’, and Independents’ predictions of the
winner of the 2008 U.S. presidential election as a function of time. Participants
were asked, “Who do you think will win the U.S. presidential election this
year?” They responded using a 7-point scale ranging from definitely Barack
Obama, 1, to definitely John McCain, 7. The value of 4 denotes indifference. The
assessment points were approximately early October (Time 1), October 18
(Time 2), October 25 (Time 3), and November 1 (Time 4).
Preference-expectation dynamics
Figure 3 presents the path model of relations among prefer­
ences, expectations, and perceptions of general candidate
favorability over time. The model fit was acceptable, χ2(77, N =
153) = 144.70, p < .001, CFI = .94, RMSEA = .076, 90% con­
fidence interval (CI) = [.057, .095], standardized root-meansquare residual (SRMR) = .086. Excluding initial assessments,
Democrat
Independent
4.5
Republican
4.0
Voting Intentions
personally experienced, how would you characterize recent
(i.e., over the last week) media coverage of the presidential
candidates?” which participants rated from strongly favoring
Obama, 1, to strongly favoring McCain, 7.
These measures were administered via the Web again over
three time points in roughly weekly intervals, with the vast
majority of the responses being made during the weekends
of October 18 and October 25 and after the weekend of
November 1. We used partial counterbalancing to control for
possible order effects. Half of the respondents first provided
expectations on the first follow-up, with the order switched on
each subsequent assessment relative to the other half of
respondents. Perceptions of polls and media coverage were
always assessed last. Finally, the day after the election, we
assessed whether participants actually voted and asked them
to rate how satisfied they were with the election result on a
scale ranging from very elated, 1, to very disappointed, 7.
3.5
3.0
2.5
2.0
1.5
1.0
Time 1
Time 2
Time 3
Time 4
Assessment Point
Fig. 2. Democrats’, Republicans’, and Independents’ voting intentions for
the 2008 U.S. presidential election as a function of time. Participants were
asked, “Who will you vote for in the upcoming presidential election?” They
responded using a 7-point scale ranging from I will definitely vote for Obama, 1,
to I will definitely vote for McCain, 7. The assessment points were approximately
early October (Time 1), October 18 (Time 2), October 25 (Time 3), and
November 1 (Time 4).
144
Krizan et al.
Time 1
Time 2
.88*
Candidate
Preference
.11
.05
.56*
Media
Perception
.32
.14
.16†
.37
Poll
Perception
.20
Electoral
Expectation
.00
.33*
Time 3
.89*
Candidate
Preference
.05
.18*.19
Media
Perception
.18
.00
.25*
.35
Electoral
Expectation
.09 .07
Media
Perception
.27
Poll
Perception
.12
.01
.25
.22
.52*
.47*
Candidate
Preference
−.04
.27*
.45
Poll
Perception
.32
.28
.03
Electoral
Expectation
Time 4
.50*
.51*
Candidate
Preference
.12
.10 .03
Media
Perception
.40
.19
.26
.88
Poll
Perception
.27
Electoral
Expectation
Fig. 3. Path model showing the relations among candidate preferences, electoral expectations, and media and poll perceptions of general candidate
favorability over time. Standardized path coefficients are shown. Critical paths are shown in black; paths involving covariates are shown in gray. The
assessment points were approximately early October (Time 1), October 18 (Time 2), October 25 (Time 3), and November 1 (Time 4). Significant and
marginally significant paths are marked, †p < .10, *p < .001.
the R2 for preferences and expectations ranged from .20 to .91.
Inspection of paths in Figure 3 reveals the standard finding
that electoral expectations show more fluctuations than do
voting intentions, although both showed substantial stability
over time. Also as would be expected, expectations tended to
show higher relations with poll and media perceptions than did
preferences.
There was clear evidence for wishful thinking—candidate
preferences consistently predicted electoral expectations at
subsequent assessment, even after taking into account previ­
ous expectations and perceptions of general candidate favor­
ability. Furthermore, these relations became stronger as
Election Day approached, despite the fact that electoral expec­
tations became more stable.
To test whether wishful thinking was more robust among
those who more strongly identified with their preferred party,
we split the sample into two subsamples based on whether
their average score on the social identity measure (Greene,
1999) was above the minimum scale value reflecting disagree­
ment regarding identification. Those scoring at this value (2.1)
or higher were considered to identify highly (n = 91), whereas
those scoring lower than this value were considered to identify
low (n = 60). Next, using multiple-groups analyses, we fitted
the same path model described earlier, with the exception of
covariates (given the low sample size, we focused only on the
fundamental part of the model). This model also fit the data
reasonably well, χ2(22, N = 151) = 37.25, p = .022, CFI = .98,
RMSEA = .096, 90% CI = [.036, .15], SRMR = .032. To deter­
mine whether party identification moderated the strength of
wishful thinking, we tested an alternative model where the
cross-lagged wishful-thinking paths were fixed across both
groups, χ2(25, N = 151) = 42.50, p = .016, CFI = .98, RMSEA
= .096, 90% CI = [.042, .14], SRMR = .06, and then examined
difference in chi-squares and strengths of wishful-thinking
paths. Although the inferential test did not reach significance,
Δχ2(3, N = 151) = 5.25, p = .15, there was clear evidence for
moderation. As seen in Figure 4, cross-lagged wishful-thinking
paths were substantially greater among those more identified
with their preferred party across the first (z = 1.90, one-tailed
p = .03) and third intervals (z = 1.90, one-tailed p = .04), with
the exception of the middle interval (z = .22). Taken together
with the results from the general model including covariates,
these findings support the conclusion that voter’s candidate
preferences biased their expectations.
Consequences of electoral expectations
To examine whether electoral expectations had unique utility in
predicting reactions to the election outcome, we simultaneously
regressed postelection disappointment on preferences and
expectations assessed just before the election (at Time 4). As
typically observed, preferences were a strong predictor of post­
election disappointment (β = 0.77, p < .001), such that those
who favored McCain were much more likely to indicate disap­
pointment with the election outcome. There was also a modest
tendency for expectations to exacerbate disappointment (β =
0.11, p = .07), so that those more optimistic about McCain’s
chances tended to experience marginally more disappointment.
Discussion
In the end, how can we account for links between electoral
preferences and expectations? Experimental literature sug­
gests potential for multiple causal influences, but both correla­
tional and experimental research to date is inconclusive
regarding which influences are the most powerful or the most
prevalent. Among young voters that we followed over the
month preceding the 2008 U.S. presidential election, there was
clear support for wishful thinking—electoral preferences con­
sistently influenced expectations across all the three time
145
Wishful Thinking in the 2008 Election
Standardized Path Coefficient (β)
0.5
High (n = 91)
Low (n = 60)
0.4
0.36
0.3
0.2
0.2
0.27
0.24
0.1
0.09
0
−0.1
−0.2
−0.15
−0.3
−0.4
Interval 1
Interval 2
Interval 3
Assessment Interval
Fig. 4. Results of the cross-lagged analysis testing the effect of wishful
thinking over time. The graph shows standardized coefficients representing
the relation between candidate preference at one assessment point and
electoral expectations at the subsequent assessment point. Interval 1 refers
to paths from Time 1 (early October) to Time 2 (October 18), Interval 2
refers to paths from Time 2 to Time 3 (October 25), and Interval 3 refers
to paths from Time 3 to Time 4 (November 1). Results are shown separately
for individuals who reported high and low social identification with their
preferred party. Error bars indicate standard errors.
intervals. On the other hand, there was no evidence that expec­
tations influenced electoral preferences. Although bandwagon,
underdog, and rationalization effects have empirical support
from field and laboratory experiments (e.g., Ceci & Kain,
1982; Kay et al., 2002; Navazio, 1977), there was no evidence
for these tendencies in our sample. Moreover, wishful thinking
over time was obvious even after taking into account voters’
perceptions of how well candidates fared in the current polls
and media coverage. Although these perceptions do not
exhaust all possible exogenous factors that could produce
preference-expectation links, they do reflect important sources
of information for voters. Finally, wishful thinking was espe­
cially likely among those who strongly identified with their
preferred party. Taken together, these findings make for the
most compelling case to date that electoral expectations are
indeed driven by political preferences. These expectations, in
turn, influence reactions to the election, so that above and
beyond the influence of political preferences, optimism about
the losing candidate can also exacerbate disappointment.
Implications for psychological theory
Whereas the notion that people engage in wishful thinking
about the future has been largely accepted as a truism, empiri­
cal evidence to date has been modest. As we have argued,
preference-expectation links in themselves are unsatisfactory
evidence, and laboratory research has been limited and often
confined to artificial settings (Krizan & Windschitl, 2007).
The present investigation produced compelling evidence for
wishful thinking in a consequential, real-world setting and
should thus be encouraging to researchers examining desir­
ability biases in forecasts.
We also need to ask how candidate preferences influenced
electoral expectations. On one hand, preferences might have
influenced voters’ expectations at the time of assessment,
for example, through selective accessibility of preference-­
consistent information or differential evaluation of favorable
and unfavorable information (see Krizan & Windschitl, 2007,
for discussion of mediating mechanisms). Given the brief
nature of laboratory studies, these types of accounts are gener­
ally considered most relevant for experimental research on
desirability bias. On the other hand, in complex dynamic envi­
ronments such as election campaigns, it is likely that candidate
preference influenced how voters chose, attended to, and
remembered relevant information over time. Although it was
initially thought that partisan preferences influenced relevant
beliefs mainly through selective exposure to congenial infor­
mation, empirical evidence suggests that to be a minor factor;
rather, political preferences mainly bias interpretation of infor­
mation (see Taber, 2003). For example, individuals tend to
hold higher standards of proof for arguments they deem unpal­
atable (e.g., Edwards & Smith, 1996). In sum, it is the iterative
nature of information processing in real-world environments
that is likely responsible for why preferences influence expec­
tations over time, as voters perceive relevant information and
filter it through their partisan lenses (as those more identified
with their preferred party seemed to do in the current data). It
will rest on future research to conduct more fine-grained anal­
yses of such processes in real-world environments.
Implications for political behavior
Note that our results reflect the psychology of a younger group
of uncommitted voters from a state (Iowa) very much informed
about political campaigns. In this sense, there are clear uncer­
tainties about the extent to which dynamics observed here
characterize the voting public in general or other young voters
in particular. However, these findings do speak about political
perceptions of a very important demographic (young adults;
see Tufts University, 2008) during a critical time in an elec­
tion, namely that final month (e.g., Fenwick et al., 1982). Note
also that these data mirrored several patterns generally
observed in national polling data—there was a clear realiza­
tion that Obama was more likely to win as Election Day drew
nearer, and partisans generally became more optimistic regard­
ing their candidate regardless of which one they supported (cf.
Gallup Poll, 2009). In this sense, the sample seemed psycho­
logically representative of a typical young voter, also reflected
in the fact that they favored Obama at about 2 to 1.
Also important was the observation that electoral expecta­
tions played a unique role in exacerbating disappointment with
the electoral results among McCain supporters. Although this
effect was modest, it suggests that campaigns’ efforts to boost
146
optimism during the final preelection weeks might have negative
consequences down the line. In other words, rousing overopti­
mism among the candidates’ supporters can exacerbate negative
reactions should the candidate lose. Such reactions could have
negative consequences on civic behavior, such as promoting
unwillingness to support the ultimate winner or increasing disen­
gagement from the electoral process (cf. Tykocinski, 2001).
In sum, the results suggest that wishful thinking during
elections is likely to be a double-edged sword. Although
increased optimism regarding one’s candidate might motivate
behavior that will help get that individual elected (a primary
reason why campaigns generally foster optimism among their
supporters), it also might lead to complacency before the elec­
tion (e.g., “my candidate will win anyway, so there is no need
to vote”) or negative reactions after elections that could lead to
antagonism toward the elected leader (e.g., “he bought the
election”) or the civic process itself.
Acknowledgments
We thank Doug Bonett, Fred Lorenz, Meifen Wei, and Gary Wells for
their helpful comments on the manuscript.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interests with
respect to their authorship and/or the publication of this article.
References
Babad, E. (1997). Wishful thinking among voters: Motivational and
cognitive influences. International Journal of Public Opinion
Research, 9, 105–125.
Cantril, H. (1938). The prediction of social events. Journal of Abnormal and Social Psychology, 33, 364–389.
Ceci, S.J., & Kain, E.L. (1982). Jumping on the bandwagon with the
underdog: The impact of attitude polls on polling behavior. The
Public Opinion Quarterly, 46, 228–242.
Dolan, K.A., & Holbrook, T.M. (2001). Knowing versus caring: The
role of affect and cognition in political perceptions. Political Psychology, 22, 27–44.
Edwards, K., & Smith, E.E. (1996). A disconfirmation bias in evalua­
tion of arguments. Journal of Personality and Social Psychology,
71, 5–24.
Fenwick, I., Wiseman, F., Becker, J.F., & Heiman, J.R. (1982). Clas­
sifying undecided voters in pre-election polls. The Public Opinion Quarterly, 46, 383–391.
Finkel, S.E. (1995). Causal analysis with panel data. Thousand
Oaks, CA: Sage.
Fischer, I., & Budescu, D. (1995). Desirability and hindsight biases
in predicting results of a multiparty elections. In J.-P. Caverni,
M. Bar-Hillel, F.H. Barron, & H. Jungermann (Eds.), Contributions to decision making—I (pp. 193–211). Amsterdam: Elsevier.
Gallup Poll, Inc. (2009). Election 2008. Retrieved March 2, 2009, from
http://www.statmodel.com/www.gallup.com/poll/election2008​
.aspx
Krizan et al.
Granberg, D., & Brent, E. (1983). When prophecy bends: The preference–
expectation link in U.S. presidential elections, 1952–1980. Journal of Personality and Social Psychology, 45, 477–491.
Greene, S. (1999). Understanding party identification: A social iden­
tity approach. Political Psychology, 20, 393–403.
Hayes, S.P., Jr. (1936). The predictive ability of voters. Journal of
Social Psychology, 7, 183–191.
Henshel, R. (1980). The purpose of laboratory experimentation and
the virtues of deliberate artificiality. Journal of Experimental
Social Psychology, 16, 466–478.
Irwin, F.W. (1953). Stated expectations as a function of probability
and desirability of outcomes. Journal of Personality, 21, 329–
335.
Kay, A.C., Jimenez, M.C., & Jost, J.T. (2002). Sour grapes, sweet
lemons, and the anticipatory rationalization of the status quo.
Personality and Social Psychology Bulletin, 28, 1300–1312.
Krizan, Z., & Windschitl, P.D. (2007). The influence of outcome
desirability on optimism. Psychological Bulletin, 133, 95–121.
Kunda, Z. (1990). The case for motivated reasoning. Psychological
Bulletin, 108, 480–498.
Lazarsfeld, P., Berelson, B., & Gaudet, H. (1948). The people’s
choice: How the voter makes up his mind in a presidential campaign. New York: Columbia University Pres.
Locascio, J.L. (1982). The cross-lagged correlation technique:
Reconsideration in terms of exploratory utility, assumption spec­
ification and robustness. Educational and Psychological Measurement, 42, 1023–1036.
Mooney, M. (2008). Obama takes double-digit lead over McCain.
ABC News. Retrieved March 17, 2009, from http://abcnews​.go​
.com
Muthén, L.K., & Muthén, B.O. (2007). Mplus (Version 5.0) [Soft­
ware]. Retrieved from http://www.statmodel.com/
Navazio, R. (1977). An experimental approach to bandwagon
research. The Public Opinion Quarterly, 41, 217–225.
Olsen, R.A. (1997). Desirability bias among professional investment
managers: Some evidence from experts. Journal of Behavioral
Decision Making, 10, 65–72.
Simon, H.A. (1954). Bandwagon and underdog effects and the pos­
sibility of election predictions. The Public Opinion Quarterly, 18,
245–253.
Taber, C.S. (2003). Information processing and public opinion. In
D.O. Sears, L. Huddy, & R. Jarvis (Eds.), Oxford handbook of
political psychology (pp. 433–476). Oxford, England: Oxford
University Press.
Tetlock, P.E., & Kim, J.I. (1987). Accountability and judgment pro­
cesses in a personality prediction task. Journal of Personality and
Social Psychology, 52, 700–709.
Tufts University, Jonathan M. Tisch College of Citizenship and Public
Service. (2008). Youth turnout rate rises to at least 52%. Retrieved
February 28, 2009, from http://www.civicyouth.org
Tykocinski, O.E. (2001). I never had a chance: Using hindsight tac­
tics to mitigate disappointments. Personality and Social Psychology Bulletin, 27, 376–382.