Download Report

Cameron Brick
Representative publications
p. 2
Brick, C., & Lewis, G. J. (2014). Unearthing the “green” personality: Core traits predict
environmentally friendly behavior. Environment and Behavior.
p. 27
Binning, K. R., Brick, C., Cohen, G. L., & Sherman, D. K. (2015, forthcoming). Going
along versus getting it right: The role of self-integrity in political conformity. Journal of
Experimental Social Psychology, 56, 73-88.
p. 43
Updegraff, J. A., Brick, C., Emanuel, A. S., Mintzer, R. E., & Sherman, D. K. (2014).
Message framing for oral health: Moderation by perceived susceptibility and motivational
orientation in a diverse sample of Americans. Health Psychology.
Environment
and
Behavior
http://eab.sagepub.com/
Unearthing the ''Green'' Personality: Core Traits Predict
Environmentally Friendly Behavior
Cameron Brick and Gary J. Lewis
Environment and Behavior published online 15 October 2014
DOI: 10.1177/0013916514554695
The online version of this article can be found at:
http://eab.sagepub.com/content/early/2014/10/14/0013916514554695
Published by:
http://www.sagepublications.com
On behalf of:
Environmental Design Research Association
Additional services and information for Environment and Behavior can be found at:
Email Alerts: http://eab.sagepub.com/cgi/alerts
Subscriptions: http://eab.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://eab.sagepub.com/content/early/2014/10/14/0013916514554695.refs.html
>> OnlineFirst Version of Record - Oct 15, 2014
What is This?
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
554695
research-article2014
EABXXX10.1177/0013916514554695Environment and BehaviorBrick and Lewis
Article
Unearthing the “Green”
Personality: Core Traits
Predict Environmentally
Friendly Behavior
Environment and Behavior
1–24
© 2014 SAGE Publications
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0013916514554695
eab.sagepub.com
Cameron Brick1 and Gary J. Lewis2
Abstract
Pro-environmental attitudes and behaviors show substantial individual
differences, and exploring their predictors can help reveal the origins of
pro-environmental behavior. Basic personality traits may provide a partial
explanation, but it is unclear which personality traits are reliably associated
with pro-environmental behaviors. This article uses a specific type of
environmental behavior, reducing greenhouse gas emissions, to clarify
which personality correlates are most robustly associated with behavior,
and to test mediation of those effects through attitudes. A large (N = 345)
sample of United States adults representative in age, gender, and ethnicity
completed the 100-item HEXACO personality inventory, a novel self-report
measure of behaviors that reduce greenhouse gas emissions, and scales
of environmental and political attitudes. Accounting for demographics,
emissions-reducing behaviors were most strongly predicted by Openness,
Conscientiousness, and Extraversion, and these effects of personality were
mediated by attitudes toward the natural environment. These observations
broaden the understanding of the etiology of environmental attitudes and
behavior.
1University
2University
of California, Santa Barbara, CA, USA
of York, York, UK
Corresponding Author:
Cameron Brick, Department of Psychological and Brain Sciences, University of California,
Santa Barbara, CA 93106, USA.
Email: [email protected]
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
2
Environment and Behavior
Keywords
personality, carbon footprint, environmental behavior, environmental
attitudes, climate change
Individual behaviors that negatively affect the environment, including driving cars, energy use, and diet, account for major ecological damage and are a
threat to human society (Intergovernmental Panel on Climate Change [IPCC],
2013). Many psychological factors predict individual environmental behaviors, including attitudes, values, and norms (e.g., Kaiser, Wölfing, & Fuhrer,
1999; Stern, 2000). Despite the established relationship between personality
and behavior across diverse domains (e.g., Paunonen, 2003), basic personality traits (e.g., the Big Five; John & Srivastava, 1999) have been infrequently
used as predictors of individual environmental behaviors, and in particular,
use of the HEXACO model (Lee & Ashton, 2004) is rare. Moreover, in
addressing personality predictors of environmental behavior, none of the
studies have focused on a critical component of environmentalism: individual actions that reduce greenhouse gas emissions (IPCC, 2013). This article
advances our understanding of the personality bases of environmentalism by
examining links between the widely used and psychometrically sound
HEXACO personality framework and self-reported emissions-reducing
behaviors, and tests whether environmental attitudes mediate the predicted
effects.
Personality: A Brief Overview
Individuals differ on stable psychological features (Eysenck & Eysenck,
1985). These differences have been conceptualized at many levels, from
broad temperaments of approach and avoidance motivation (Elliot & Thrash,
2002; Gray, 1981) to various taxonomies of personality traits. A vast literature on personality structure and assessment supports a descriptive theory of
five broad and replicable personality traits, often labeled as follows:
Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism,
and referred to as the “Big Five” (John & Srivastava, 1999). Openness reflects
rich, abstract thinking and an appreciation for variety and unusual experiences. Conscientiousness is indicated by high levels of self-discipline, respect
for duty, and desire for achievement. Extraversion is characterized by an
energetic engagement with the world, sociability, and breadth of activities.
Agreeableness is the tendency to value social harmony and getting along with
others. Finally, Neuroticism is the tendency to experience negative emotions,
such as anger, anxiety, and depression (McCrae & Costa, 1997). A six-factor
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
3
model (HEXACO; Ashton & Lee, 2007; Ashton, Lee, & de Vries, 2014; Lee
& Ashton, 2004) also demonstrates discriminant validity. The HEXACO
model re-labels Neuroticism “Emotionality”—although Emotionality and
Neuroticism are not interchangeable (Lee & Ashton, 2004)—and identifies a
sixth core trait, Honesty-Humility, which taps sincerity, fairness, greed avoidance, and modesty. Honesty-Humility shares variance with Agreeableness
and Conscientiousness as conceptualized in Big Five models, and its inclusion as a separate factor contributes useful and unique personality variance
when predicting attitudes and behavior (Lee & Ashton, 2005; Lee, Ashton,
Ogunfowora, Bourdage, & Shin, 2010). Honesty-Humility (but not
Agreeableness) predicts active cooperation and Agreeableness (but not
Honesty-Humility) predicts non-retaliation (Hilbig, Zettler, Leist, &
Heydasch, 2013). Each of the six HEXACO traits has demonstrated predictive validity through associations with life outcomes and behaviors (e.g.,
Caspi, Roberts, & Shiner, 2005). Personality can also be measured within
traits, by fractionating each trait into facets (e.g., DeYoung, Quilty, &
Peterson, 2007). This level of analysis can reveal the underlying components
responsible for the main effect of traits. Below, we review the literature on
core personality and environmental behavior.
Personality and Environmental Behavior
Core personality traits, such as the Big Five and HEXACO dimensions, are
promising candidates for individual differences predictors of environmental
behavior because they are cross-culturally reliable (McCrae & Costa, 1997),
have excellent internal validity, and may partially determine factors such as
attitudes. Broad models of environmental behavior (Kaiser et al., 1999; Stern,
2000) situate basic personality earlier in the chain of causation than values,
ideology, and attitudes. This causal hierarchy is supported by longitudinal
evidence showing the enduring effects of early personality on outcomes later
in life (e.g., Block & Block, 2006), and in particular by evidence that early
temperament (e.g., in 3-year-olds) predicts later values, attitudes, and behaviors before values and attitudes could have existed for the individual (e.g.,
Caspi & Silva, 1995; Slutske, Moffitt, Poulton, & Caspi, 2012). This path
structure is also consistent with common theorizing within the personality
literature (e.g., McCrae & Costa, 1999). Therefore, empirical papers can seek
to test path models between core personality, intermediate constructs of attitudes, values, and beliefs, and then behavior. Next we summarize findings on
which personality traits predict environmental behavior. Each association is
zero-order and p < .05 unless otherwise stated.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
4
Environment and Behavior
There is a strong argument for how Openness relates to environmentalism.
Openness is characterized by flexible, abstract thinking, exactly what is necessary to imagine long-term and long-distance environmental consequences
such as those associated with climate change. Openness also has a component of counterculture. Since the status quo is damaging the environment,
becoming concerned about the environment means rejecting the is-ought fallacy that the way things are reflects the way they should be. This rejection
requires intellect and alternative thinking. In line with this logic, Openness
has shown the most robust links to environmentalism. In three studies and
five samples (Hilbig, Zettler, Moshagen, & Heydasch, 2012; Hirsh &
Dolderman, 2007; Markowitz, Goldberg, Ashton, & Lee, 2012), Openness
showed moderate associations (rs = .23-.46) with environmental intentions,
goals, or self-reported behavior. The first hypothesis is based on these
findings.
Hypothesis 1 (H1): Of the core personality traits, Openness will show the
strongest unique prediction of emissions-reducing behavior.
The other core personality traits have shown mixed results. One study
reported a significant effect for Conscientiousness, r = .14 (predicting electricity conservation; Milfont & Sibley, 2012, Study 2). This report is atypical
because it contains the only study in which Openness was non-significant as
a predictor of behavior. Two studies reported small or inconsistent effects of
Conscientiousness on environmental concern (Hirsh, 2010) and self-reported
environmental behavior (Markowitz et al., 2012); another report contained
one study with no effect on behavior and a second with a moderate relationship, r = .22 (Hilbig et al., 2012). Despite expectations that duty and selfdiscipline would relate to conservation behaviors, previous work has fallen
short of delivering strong evidence for Conscientiousness. For Extraversion,
two samples using structural equation modeling and accounting for the other
traits showed that Extraversion predicted environmental behavior at r = .20
and in a second sample r = .31 (correlation coefficients between latent factors; Hilbig et al., 2012). In a study using multiple measures of Extraversion
across two samples, Extraversion correlated with environmental behavior at
the zero-order level at rs = .04-.19 (Markowitz et al., 2012; several correlations n.s.). In another study, there was no significant relationship between
Extraversion and environmental goals, r = .09, n.s. (Hirsh & Dolderman,
2007). Overall, there is moderate evidence for a unique contribution for
Extraversion.
Agreeableness is also linked with environmental behavior. The personality framework in each study is noted below since Agreeableness in the Big
Five and HEXACO are not equivalent. Big Five Agreeableness broadly taps
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
5
motivation for social harmony, concern for others, and cooperation, but
HEXACO Agreeableness does not contain elements of empathic concern or
humility/modesty (the latter is tapped by HEXACO Honesty-Humility), and
so the HEXACO arguably provides a more focal measure of pure
Agreeableness than the Big Five. Within the HEXACO framework, HEXACO
Agreeableness can be viewed as capturing common variance among characteristics including forgivingness, gentleness, flexibility, and patience, whereas
HEXACO Honesty-Humility can be viewed as capturing common variance
among characteristics including sincerity, fairness, greed avoidance, and
modesty. One paper found a strong effect of Big Five Agreeableness on environmental goals, r = .31 (Hirsh & Dolderman, 2007), and another set of three
studies using an abbreviated Big Five personality scale observed that
Agreeableness modestly predicted electricity conservation, r = .15 (Milfont
& Sibley, 2012, Study 2). Using the HEXACO, one paper reported r = .30 for
Agreeableness (Hilbig et al., 2012, Study 1). However, that same paper
reported a sample with no effect for HEXACO Agreeableness (Hilbig et al.,
2012, Study 2), and another paper using both HEXACO and Big Five measures found no relationship with Agreeableness across two samples
(Markowitz et al., 2012). This pattern of results may indicate that Big Five
Agreeableness is more robustly associated with environmentalism, although
this account does not fully explain the mixed results to date (e.g., Markowitz
et al., 2012, report a null association for Big Five Agreeableness).
The role of Honesty-Humility is unclear. In one report, it correlated with
environmental behavior r = .41 and r = .42 in two samples and accounted for
more variance than all of the other traits combined (Hilbig et al., 2012); however, in the only other study using the HEXACO, Honesty-Humility showed
no relationship with environmental behavior (Markowitz et al., 2012).
Potential explanations include that the German versus United States populations have meaningful differences; that the behavioral measures between the
studies tapped different types of environmental behavior; and that because
Hilbig et al. (2012) did not control for age or gender in their analyses, but
Markowitz et al. (2012) did, it’s possible that the relationship between
Honesty-Humility and environmental behavior is a cohort effect. Another
report measured the Big Five and also added a Honesty-Humility trait, creating a mix of the two approaches, and found that Honesty-Humility,
Agreeableness, and Openness each predicted environmental concern (Sibley
et al., 2011). The second hypothesis is based on the existing results concerning Agreeableness and Honesty-Humility.
Hypothesis 2 (H2): Agreeableness and Honesty-Humility are expected to
uniquely predict environmental behavior accounting for the other personality traits.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
6
Environment and Behavior
More specifically, we expected Honesty-Humility to relate more strongly
than Agreeableness to pro-environmental behavior because HonestyHumility reflects the lack of exploitation of others. In contrast, HEXACO
Agreeableness most closely reflects the lack of retaliation (Hilbig et al.,
2013). Next, we turn to individual differences beyond the Big Five and
HEXACO frameworks, and their associations with environmental behavior.
Values, Attitudes, and Environmental Behavior
Environmental attitudes are a key individual difference that predict behavior.
As mentioned above, broad models of environmental behavior place attitudes
at an intermediate casual step later than personality and values and earlier
than behavior (e.g., Kaiser et al., 1999; Stern, 2000). There is substantial
empirical evidence for a mediating role of attitudes between personality and
behavior (Conner & Abraham, 2001) and between values and behavior
(Milfont, Duckitt, & Wagner, 2010). The second paper included samples
from three countries, which boosts the external validity of this claim. The
third hypothesis is based on this framework and evidence.
Hypothesis 3 (H3): Environmental attitudes will mediate the relationships between personality traits and environmental behavior.
Two validated self-report attitude scales show consistent associations with
environmental behavior (e.g., Hirsh & Dolderman, 2007; Markowitz et al.,
2012, Study 2). The New Ecological Paradigm (NEP; Dunlap, Van Liere,
Mertig, & Jones, 2000) measures environmental values and concern, and the
Connectedness to Nature Scale (CNS; Mayer & Frantz, 2004) measures emotional connectedness with the natural world, and the CNS is less reflective
and cognitive than the NEP. In studies where both scales are included and
their common effects partialled out, the more affective CNS appears to more
strongly predict environmental behavior than the more reflective and cognitive NEP (Mayer & Frantz, 2004), which reinforces the importance of affective values in the etiology of environmental behavior. Many other
environmental values measures appear in the literature, and the focus in the
current article on these two scales is because previous work suggests they
provide sufficient coverage to explain the relationship between personality
and behavior. In Markowitz et al. (2012, Study 2), the relationship between
Openness and environmental behavior was mediated by the NEP and CNS. In
Hilbig et al. (2012), the relationship between Honesty-Humility and ecological behavior was partially mediated by a related attitudes scale. In addition,
Agreeableness and Openness were the strongest trait predictors of environmental concern in a large sample of German adults (Hirsh, 2010). These
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
7
results show that core personality partially drives environmental attitudes and
beliefs. When investigations of environmental behavior include attitude and
belief measures, these studies can help verify broad models that explore the
proximate and distal causes of environmental behaviors.
Besides personality traits and environmental attitudes, previous research
on pro-environmental behavior has also examined the role of sociodemographic variables such as political orientation (e.g., Costa & Kahn,
2013; Gromet, Kunreuther, & Larrick, 2013). Political conservatism predicts
lower concern for the natural environment, even after accounting for demographic variables related to political orientation such as age and gender (Allen,
Castano, & Allen, 2007; Dunlap, Xiao, & McCright, 2001), and it also predicts less environmental behavior (e.g., Gromet et al., 2013). However, the
unique predictive value of political orientation for behavior has not been demonstrated, because personality, political orientation, environmental attitudes,
and behavior have not yet been modeled together. Personality and environmental attitudes may partially account for the effects of political orientation on
environmental behavior. Next, we explore the limitations of the existing literature on individual actions that affect the natural environment.
Environmental Behavior
Pro-environmental behavior is multidimensional and perhaps too diverse to
measure in a brief self-report scale (Balderjahn, 1988; Diekmann &
Preisendörfer, 1998; Stern, 2000). Environmental behavior spans dissimilar
actions such as using fabric softener during laundry (Hilbig et al., 2012) and
working from home (Markowitz et al., 2012). Covering just private environmental actions is challenging with a brief measure (e.g., Kaiser et al., 1999),
and therefore existing behavioral scales lack comprehensive content coverage, and may be more heterogeneous between papers than they first appear.
The goal in the current study of introducing a novel behavioral scale was to
specify a face-valid type of pro-environmental behavior that is of interest to
public policy. Emissions-reduction is one of the most important aspects of
individual environmentalism (IPCC, 2013). Emissions-reducing behaviors
uniquely span diverse types of environmental actions (transportation, diet,
energy use) while still belonging to a coherent category. The new scale therefore has the potential to increase construct coverage and validity.
Current Study
Previous work provided important insights into which personality traits
might predict individual differences in emissions-related behaviors, but this
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
8
Environment and Behavior
is the first study to directly test the question. A large, geographically diverse
United States sample reported their personality with the 100-item HEXACO
Personality Index-Revised, a psychometrically sound and well-established
instrument (Lee & Ashton, 2004) that allows for facet-level analyses, and
then completed a novel and meaningful measure of environmental actions:
emissions-reducing behavior (see below).
In addition to the three hypotheses, demographics will be used to test for
cohort effects, facet-level results will be compared to previous research, and
we report the first examination in this literature of two-way trait interactions.
These results will clarify the predictive power of each trait for emissionsreducing behavior and inform the mixed literature on the role of
Honesty-Humility.
Material and Methods
Participants and Procedure
In all, 345 U.S. adults completed an online survey through Amazon MTurk.
Student populations have narrow geographic and age range and poor external
validity for many psychological inquiries (Henrich, Heine, & Norenzayan,
2010). MTurk samples can improve external validity through more representative population sampling, and many classic effects are reliable across
MTurk samples (Berinsky, Huber, & Lenz, 2012; Mason & Suri, 2012). The
current sample was diverse compared with most student samples:
M (SD)Age = 36.7 (13.2) years; 53.3% female; 80.0% White, 5.8% Black,
6.7% Asian, 5.2% Latino, and 2.3% Other; modal education = bachelor’s
degree (36.8%); and modal household income = $25,001-$60,000 (42.0%).
Improving on previous studies that relied on a single region, Internet Protocol
addresses revealed the sample spanned 47 U.S. states and no one participated
from a non-U.S. location (MaxMind, 2013). Fifty-five additional participants
did not complete the outcome measure and were excluded. This drop-out rate
is typical for surveys with modest payment. To ensure a large sample, we
aimed for a sample size of 300+ after exclusions, and finalized data collection
before hypothesis testing.
Measures
Personality. Participants first completed the 100-question HEXACO-PI-R
(Lee & Ashton, 2004), which assesses personality across six core traits and
yields 24 sub-trait facets (see Tables 1 and 2 for reliability). Example item
(Conscientiousness): “When working, I often set ambitious goals for myself,”
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
9
Brick and Lewis
Table 1. Means, Standard Deviations, Reliability, and Zero-Order Correlations
Between HEXACO Traits, Political Conservatism, Environmental Attitudes, and
Emissions-Reducing Behavior.
r(345)
H
Em
eX
A
C
O
Pol
NEP
CNS
EB
M (SD)
3.42
(0.72)
.89
3.20
(0.58)
.80
.11
3.16
(0.67)
.88
.04
–.15*
3.04
(0.67)
.88
.29*
–.07
.32*
3.64
(0.55)
.84
.31*
.00
.32*
.13
3.58
(0.68)
.87
.11
.00
.09
.05
.13
3.04
(1.85)
n/a
–.08
–.12
.09
–.02
.10
–.14
3.61
(0.74)
.90
.28*
.26*
–.05
.06
.13
.24*
–.43*
3.54
(0.74)
.90
.28*
.17*
.11
.23*
.17*
.47*
–.22*
.57*
2.85
(0.51)
.73
.22*
.07
.22*
.14*
.25*
.28*
–.08
.36*
.47*
Cronbach’s
H
Em
eX
A
C
O
Pol
NEP
CNS
Note. Alpha = .01 to provide a more stringent threshold in light of multiple comparisons. H = HonestyHumility; Em = Emotionality; eX = Extraversion; A = Agreeableness; C = Conscientiousness; O =
Openness; Pol = political conservatism; NEP = New Ecological Paradigm; CNS = Connectedness to Nature
Scale; EB = emissions-reducing behavior.
*p .01.
rated 1 (strongly disagree) to 5 (strongly agree). Reducing concern about
content overlap between personality and behavior, none of the items mention
conservation or environmentalism, and only one refers to nature (“Sometimes
I like to just watch the wind as it blows through the trees” [Openness–Aesthetic Appreciation]). As noted above, the HEXACO trait items differ from
the Big Five scales, but the constructs overlap considerably and recent work
suggests similar trait relationships with environmental behavior (Markowitz
et al., 2012), with the possible exception of Agreeableness (see above).
Attention check. Twenty-two participants (6.4%) failed a standard attention
check consisting of a question ostensibly about what activities the participant
enjoys. At the end of a long instruction block, participants were asked to
ignore the question, select “Other,” and write a specified word to demonstrate
they were paying attention (Oppenheimer, Meyvis, & Davidenko, 2009,
Study 1). Carelessness in questionnaire response (defined as 5 consecutive
repeated values) was uncorrelated with failing the attention check, r(346) =
.02, p = .78, and there was no difference on visual inspection in the length or
quality of written responses between individuals passing or failing the attention check. Finally, the main results did not substantively change when those
who failed were excluded. The attention check appeared not to yield useful
discrimination, so all participants were retained.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
10
Environment and Behavior
Table 2. Means, Standard Deviations, and Zero-Order Correlations of HEXACO
Facets With Political Conservatism, Environmental Attitudes, and EmissionsReducing Behavior (N = 345).
M (SD)
H: Sincerity
H: Fairness
H: Greed avoidance
H: Modesty
Em: Fearfulness
Em: Anxiety
Em: Dependence
Em: Sentimentality
eX: Social self-esteem
eX: Social boldness
eX: Sociability
eX: Liveliness
A: Forgiveness
A: Gentleness
A: Flexibility
A: Patience
C: Organization
C: Diligence
C: Perfectionism
C: Prudence
O: Aesthetic
appreciation
O: Inquisitiveness
O: Creativity
O: Unconventionality
Cronbach’s
Pol
NEP CNS
EB
.19* .19*
.12
.15*
.27* .26*
.33* .28*
.16* –.01
.21* .05
.12
.12
.23* .31*
.02
.07
–.04
.11
–.07
.11
–.06
.05
–.03
.13
.12
.25*
.10
.25*
.02
.10
.07
.11
.14* .19*
.14* .18*
.03
.01
.30* .51*
.17*
.21*
.16*
.11
.00
–.02
.08
.14
.11
.18*
.24*
.14
.08
.14
.14
.08
.16*
.18*
.25*
.17*
.33*
3.29 (0.90)
3.48 (1.01)
3.20 (0.97)
3.69 (0.79)
3.12 (0.82)
3.45 (0.85)
2.87 (0.81)
3.37 (0.81)
3.72 (0.76)
2.82 (0.89)
2.85 (0.92)
3.25 (0.89)
2.67 (0.86)
3.26 (0.84)
2.95 (0.78)
3.29 (0.91)
3.52 (0.90)
3.83 (0.74)
3.63 (0.65)
3.57 (0.73)
3.55 (0.88)
.79
.85
.80
.75
.65
.71
.71
.69
.71
.76
.81
.81
.80
.78
.66
.79
.78
.78
.58
.70
.69
–.06
.02
–.08
–.17*
–.10
–.11
–.07
–.06
.06
.05
.09
.07
.02
–.04
–.02
–.01
.10
–.01
.09
.09
–.11
3.65 (0.84)
3.59 (0.91)
3.54 (0.76)
.70
.77
.63
–.07
–.10
–.15*
.14*
.17*
.16*
.32*
.36*
.29*
.20*
.25*
.10
Note. Alpha = .01 to provide a more stringent threshold in light of multiple comparisons.
Pol = political conservatism; NEP = New Ecological Paradigm; CNS = Connectedness to
Nature Scale; EB = emissions-reducing behavior.
*p .01.
Environmental attitudes and values. Next, participants reported their attitudes
and values regarding the environment with two scales. The 15-item NEP
(Dunlap et al., 2000) is a widely used attitudes measure, with items such as
“Humans are severely abusing the environment” and “The earth has plenty of
natural resources if we just learn how to develop them” (reversed), 1 (strongly
disagree) to 5 (strongly agree), Cronbach’s alpha = .90. The 14-item CNS
(Mayer & Frantz, 2004) taps affective as well as cognitive content with items
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
11
such as “I often feel a kinship with animals and plants” and “I have a deep
understanding of how my actions affect the natural world,” rated 1 (strongly
disagree) to 5 (strongly agree), Cronbach’s alpha = .90.
Environmental behaviors: Reducing emissions. Fifteen behaviors were adapted
from reports of how individual behaviors impact climate change (Carbon
Footprint, 2012; IPCC, 2013) and from pilot studies, and items measuring
personal frequency of those behaviors were combined into a scale (see also
Brick, Kim, & Sherman, in press). Behaviors were included when they were
repeated and accessible to a wide variety of demographics, for example,
“How often do you walk, bicycle, carpool, or take public transportation
instead of driving a vehicle by yourself?” and “How often do you eat meat?”
(reversed), rated 1 (never) to 5 (always); see Appendix. Actions like buying a
high-efficiency furnace or insulating the home were excluded, because
although these have a significant impact on emissions, they are only applicable to homeowners and are infrequent, limiting their utility for tapping
ongoing psychological processes in a diverse population. The previous literature demonstrates it can be difficult to identify individual differences measures of personality that stably relate to behavior. This novel behavioral
measure targets repeated actions and each data point represents an average
across many individual decisions, features which help to reveal trait effects
(Fleeson, 2004). Behavior frequency ratings ranged from 2.15 to 3.93 with
skew |x| < 1.2, although kurtosis ranged from 1.83 to 3.70. Principal components analysis with oblimin rotation yielded a single coherent factor explaining 24.1% of the variance, and Cronbach’s alpha was .76. Participants also
reported the frequency of performing five non-emissions environmental
behaviors as a pilot for future studies (see Brick, Kim, & Sherman, in press),
and they are not discussed here further.
Uncertainty. Two exploratory questions regarding participants’ feelings of
uncertainty did not show significant relationships with any key variables and
are not discussed further.
Political orientation. Political orientation was measured as in the American
National Election Studies (Center for Political Studies, 2013), and yielded a
continuous rating from 1 (strong Democrat) to 7 (strong Republican).
Demographics, suspicion, and fluency check. Participants reported their age,
gender, education, household income, and ethnicity; guessed at the goal of
the research; reported any technical problems or comments; and wrote 1 to 2
open-response sentences to demonstrate fluency in English. All measures and
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
12
Environment and Behavior
conditions are reported above and no unreported studies or samples were
excluded.
Results
Alpha was set at .01 for all tests to provide a more stringent threshold in light
of multiple comparisons.
Trait Correlations
All analyses below have 345 participants. Descriptive statistics and zeroorder trait correlations are presented for the NEP, CNS, and emissions behavior to allow the full model to be reconstructed (see Table 1). All the HEXACO
traits except Emotionality (r = .07, p = .22) were positively correlated at the
zero-order level with emissions-reducing behavior, rs
.17, ps
.01.
Supporting H1, this association was most strongly observed for Openness,
r = .28, p < .001 (this value was not statistically tested against the other zeroorder correlations).
Facet Correlations
The average correlation between facets within personality traits was r = .35
(range = .14-.52), showing that the facets within each trait were overlapping
but distinct. Next, we report the zero-order facet correlations with the NEP,
CNS, and emissions behavior (see Table 2). At least one facet from each trait
was significantly correlated with behavior. The one previous study with
HEXACO facets reported significant zero-order Openness, Extraversion, and
Conscientiousness facet correlations with emissions-reducing behavior
(Markowitz et al., 2012), and the facet effects reported there partially overlap
with those observed here. The facet most associated with environmental
behavior was Openness–Aesthetic Appreciation in both studies, Markowitz
et al. (2012): r = .42, the current study: r = .33, ps < .001. Aesthetic
Appreciation in the HEXACO Personality Inventory-Revised reflects the
enjoyment of beauty in music, poetry, art galleries, and wind in the trees.
Regression
A hierarchical linear regression explored which personality traits robustly
predicted emissions-reducing behavior (see Table 3). In Step 1, the six
HEXACO traits were entered, and Honesty-Humility, Extraversion,
Conscientiousness, and Openness were the strongest predictors of behavior
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
13
Brick and Lewis
Table 3. Hierarchical Linear Regression Predicting Emissions-Reducing Behavior
(N = 345).
Step 1
Predictor
Step 2
Standardized
Honesty-Humility
Emotionality
eXtraversion
Agreeableness
Conscientiousness
Openness
Age
Female
Education
Income
Political conservatism
Openness × Conscientiousness
Openness × eXtraversion
Conscientiousness ×
eXtraversion
.13 (.06)
.08 (.05)
.16* (.06)
.03 (.05)
.13 (.06)
.23* (.05)
Step 3
(SE)
.08 (.06)
.06 (.06)
.13 (.06)
.05 (.06)
.13 (.06)
.23* (.05)
.08 (.06)
.04 (.06)
.00 (.05)
.07 (.05)
–.06 (.05)
.08 (.06)
.06 (.06)
.14 (.06)
.05 (.06)
.13 (.06)
.24* (.05)
.08 (.06)
.04 (.06)
.00 (.05)
.07 (.05)
–.06 (.05)
–.02 (.06)
.01 (.05)
.00 (.05)
Note. Alpha = .01 to provide a more stringent threshold in light of multiple comparisons.
*p .01.
(βs = .13-.23, ts 2.3, ps .02). Contrary to H2, neither Agreeableness nor
Honesty-Humility uniquely predicted behavior. The significant results are
vulnerable to cohort effects, so in Step 2, age, gender, income, education, and
political conservatism were entered as demographic covariates. These covariates help isolate the effects of personality, and none were significant unique
predictors of behavior when accounting for core traits. Although political
conservatism showed a correlational trend at the zero-order level with fewer
emissions-reducing behaviors, r = –.10, p = .08, in Step 2 political conservatism did not predict behavior, p = .21. However, these covariates did reduce
the effect of Honesty-Humility to non-significance, β = .08, t = 1.29, p = .20.
Further supporting H1, Openness still uniquely predicted emissions-reducing
behavior (β = .23, t = 4.50, p < .001), and Conscientiousness (β = .13, t =
2.39, p = .02) and Extraversion (β = .13, t = 2.34, p = .02) trended toward
significance at alpha = .01. In Step 3, mean-centered traits were combined to
form interaction terms for Openness × Conscientiousness, Openness ×
Extraversion, and Conscientiousness × Extraversion. None of the interactions was significant, ps .66, nor did their inclusion alter the previous
results.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
14
Environment and Behavior
Mediation
In a confirmatory analysis, the NEP and CNS were entered as potential mediators of the relationship between Openness and emissions-reducing behavior,
using nonparametric bootstrapping (Preacher & Hayes, 2008) because it
offers several advantages including no distributional assumptions. All mediation effects are reported with unstandardized coefficients and the point and
interval estimates are interpreted directly avoiding the terms full and partial
as recommended by Hayes (2013). This analysis included the covariates of
age (B = .01, t = 2.66, p < .01), female gender (B = –.01, t = 0.27, p = .79),
ethnicity (White vs. Non-White; B = –.08, t = −1.29, p = .20), political orientation (B = .02, t = 1.48, p = .14), education (B = .02, t = 1.11, p = .27), and
income (B = .03, t = 1.34, p = .18). A total of 10,000 bootstrapped samples
(N = 345) indicated that the effects of Openness on the NEP (B = .21, t = 4.09,
p < .001) and the CNS (B = .48, t = 9.45, p < .001) were both significant. The
effects of the mediators on behavior were also significant, NEP (B = .13, t =
2.80, p < .01) and CNS (B = .26, t = 5.58, p < .001). As hypothesized, the NEP
and CNS mediated the effect of Openness on behavior: the indirect effect of
Openness on behavior was significant (B = .15, 95% CI = [.10, .21]), but the
direct effect accounting for the mediators was not (B = .06, t = 1.44, p = .15,
95% CI = [–.02, .15]; overall model R2 = .28). Participants who were high in
Openness held positive environmental attitudes and performed more behaviors to reduce emissions, and this effect was mediated by environmental attitudes, supporting H3 (see Figure 1).
In an exploratory analysis, we tested a second mediation model using
Conscientiousness for the predictor, again including the covariates of age
(B = .00, t = 2.34, p = .02), female gender (B = –.03, t = .50, p = .62), ethnicity
(White vs. Non-White; B = –.09, t = −1.44, p = .15), political orientation
(B = .01, t = .93, p = .35), education (B = .02, t = 1.12, p = .26), and income
(B = .03, t = 1.14, p = .25). A total of 10,000 bootstrapped samples (N = 345)
indicated that the effects of Conscientiousness on the NEP (B = .19, t = 2.93,
p < .01) and the CNS (B = .22, t = 3.15, p < .01) were both significant. The
effects of the mediators on behavior were also significant, NEP (B = .12, t =
2.52, p = .01) and CNS (B = .28, t = 6.51, p < .001). The NEP and CNS mediated the effect of Conscientiousness on behavior: the indirect effect of
Conscientiousness on behavior was significant (B = .08, 95% CI = [.03, .14]),
and the direct effect accounting for the mediators was still significant (B =
.15, t = 3.11, p < .01, 95% CI = [.05, .24]; overall model R2 = .29). Participants
who were high in Conscientiousness also held positive environmental attitudes and performed more behaviors to reduce emissions, and this effect was
mediated by environmental attitudes, supporting H3 (see Figure 2). To follow-up on the marginal effect of Extraversion, an exploratory mediation was
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
15
Brick and Lewis
NEP
.13 [.04-.22]
.21 [.11-.32]
indirect = .15 [.10-.22]
Openness
direct = .06, n.s. [-.02-.15]
.48 [.38-.58]
Emissionsreducing
behavior
.26 [.17-.35]
CNS
Figure 1. Bootstrapping mediation of Openness on emissions-reducing behavior
through environmental attitudes, controlling for age, gender, political orientation,
education, and income, shown with unstandardized path coefficients and 95%
confidence intervals in brackets (N = 345, R2 = .28).
Note. NEP = New Ecological Paradigm, CNS = Connectedness to Nature Scale.
All paths p < .01 except for the direct effect of Openness on behavior accounting for the
mediators (p = .15).
also run for Extraversion. It revealed a direct effect on behavior, B = .13, p <
.001, and a weak indirect path through the NEP and CNS, B = .04, entirely
accounted for by the CNS, reflecting that about a third of the marginal total
effect of Extraversion on behavior was mediated by the CNS. Please contact
the first author with requests for data.
Discussion
We examined the role of personality in a critical type of environmentalism:
emissions-reducing behavior. Openness and Conscientiousness independently predicted these behaviors, and their effects were mediated by proenvironmental attitudes. The finding for Openness is consistent with previous
reports. However, only one paper reported a unique effect for Conscientiousness
on environmental behaviors (Milfont & Sibley, 2012), and notably it included
three large samples with diverse participants. The lack of other
Conscientiousness findings may reflect the unique properties of emissionsreducing behaviors among other environmental actions, but may also be due
to other samples having been geographically limited with potential restrictions in range on the independent and/or dependent variables. In the current
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
16
Environment and Behavior
NEP
.12 [.03-.21]
.19 [.06-.32]
indirect = .08 [.03-.14]
Conscientiousness
direct = .15 [.05-.24]
.22 [.08-.35]
Emissionsreducing
behavior
.28 [.19-.36]
CNS
Figure 2. Bootstrapping mediation of Conscientiousness on emissions-reducing
behavior through environmental attitudes, controlling for age, gender, political
orientation, education, and income, shown with unstandardized path coefficients
and 95% confidence intervals in brackets (N = 345, R2 = .31).
Note. NEP = New Ecological Paradigm, CNS = Connectedness to Nature Scale.
All paths p < .01.
study, zero-order correlations were observed of Emotionality, Agreeableness,
and Honesty-Humility with emissions-reducing behavior, and in line with
previous work distinguishing active cooperation from non-retaliation, the
correlation of Honesty-Humility with behavior appeared larger than that for
Agreeableness. However, these effects dropped out when controlling for the
other HEXACO traits and key demographics. Consistent with Markowitz et
al. (2012), after controlling for age and gender, Honesty-Humility did not
uniquely predict pro-environmental behavior. Because these results call into
question previous reports of Honesty-Humility’s unique role (i.e., Hilbig et
al., 2012), we requested and were graciously provided access to both studies’
data from Hilbig et al. (2012; data from personal communication, May 13th,
2014). Using linear regression, Honesty-Humility still uniquely predicted
environmental behavior when including age and gender as covariates, indicating that these mixed results for Honesty-Humility were not likely due to
confounding age effects in those samples. Future research can explore other
possibilities for the remaining discrepancy, including the different sample
populations and environmental behavior scales. These results support H1,
that Openness is a key predictor of emissions-reducing behavior, and provide
evidence against H2, that Agreeableness or Honesty-Humility substantially
predict such behaviors.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
17
The current model revealed that interactions between Extraversion,
Conscientiousness, and Openness did not improve predictions of emissionsreducing behavior. Each of the effects was non-significant, but not all possible interactions were tested. Further studies may choose to include trait
interactions. Next, exploratory analyses were performed at the facet level,
and the findings largely mirrored the factor-level zero-order correlation
results, with facets from each domain correlating significantly with emissions-reducing behavior. There was no suggestion of facet-level suppression
on trait-level effects. Brief personality scales that only measure broad factors
appear sufficient for most future work on core personality traits in this area.
The facet associations can also help interpret the main results. The strongest facet predictor of emissions-reducing behaviors was Openness–Aesthetic
Appreciation. In contrast, Openness–Unconventionality was uncorrelated
with emissions-reducing behavior. This suggests that emissions-reducing
behavior was not driven by low rebelliousness or contrariness, as might be
predicted under a model where environmentalism were partly an outlet for
unconventional self-expression, and where such behavior would not reflect
concerns with environmental issues beyond the need for a rebellious identity.
Instead, the facet results revealed that pro-environmental behavior was predicted by appreciation and connection to the aesthetics and the natural environment. This finding, seen here and in Markowitz et al. (2012), reveals new
research questions. We encourage researchers to use more detailed measures
of aesthetic appreciation to further evaluate its relationship with environmental behavior and possibly identify differences between types of aesthetic targets (e.g., music; paintings). Mean shifts in personality traits are observed
across the lifetime (Roberts, Walton, & Viechtbauer, 2006), so perhaps training in aesthetic appreciation such as the study of visual arts, even when thematically unrelated to nature, could increase pro-environmental behavior.
All of the Conscientiousness facets including Organization, Diligence,
Perfectionism, and Prudence were significant predictors of emissionsreducing behavior, unlike in Markowitz et al. (2012). A few explanations are
possible. A recent report shows Conscientiousness ratings for every U.S.
state, and it varies considerably (Rentfrow et al., 2013). The standardized
scores in that report make it difficult to see if Oregon specifically has lower
than average variance in Conscientiousness. Because the Markowitz et al.
sample was entirely from the state of Oregon, that could have obscured variation in Conscientiousness facets through restricted range. However,
Markowitz et al. kindly provided us with HEXACO means and standard
deviations for their sample, and the values for Conscientiousness are closely
overlapping with the current study (Markowitz et al., 2012, data from personal communication, April 24th, 2014). Therefore, restricted range is not the
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
18
Environment and Behavior
explanation. Another possibility is that emissions-reducing behaviors are particularly self-sacrificing and characterized by less consumption, which contrasts with environmental behaviors that could be fun, for example attending
“environmental rallies” (Markowitz et al., 2012, p. 99). These discrepancies
reinforce the argument that covering even just private environmental actions
is too ambitious for a brief measure, and suggest that previous behavioral
scales may have lacked content coverage and are more varied between studies than they appear. The new scale presented here was not designed to integrate all of the previous findings, but rather to provide novel insight into
emissions-reducing behaviors. A limitation of the current design is that measuring environmental attitudes before behavior could potentially have caused
social desirability effects. This could compromise the direct interpretability
of the means of reported behavior. However, universal increases of reported
behavior would not have changed the trait findings.
The effect of Openness on emissions-reducing behavior was mediated by
attitudes to the point where the direct effect became non-significant (this does
not preclude the existence of additional, independent mediators). The relationship between the CNS, NEP, and environmental behavior was previously
known. Incorporating the mediation results of Markowitz et al. (2012) and
the current study suggest a new addition: these attitude scales meaningfully
overlap with the personality construct Openness and its components, especially Aesthetic Appreciation. The effect of Conscientiousness was also
mediated, but the direct effect remained significant, indicating that additional
mediating variables are required to fully explain the effects of
Conscientiousness on environmental behavior. Behaviors have such varied
and complex causes that any effects of personality will likely be mediated
and moderated by intervening individual differences (e.g., values, attitudes)
and context (e.g., the presence of observers).
The lesser mediation of Conscientiousness by the NEP and CNS suggests
these scales may not capture the full diversity of pro-environmental attitudes.
A close examination reveals language that suggests classic, liberal environmentalism (e.g., “Like a tree can be part of a forest, I feel embedded within
the broader natural world,” CNS), and when they mention consequences they
focus on harm and fairness, the hallmarks of liberal morality (Haidt, 2007;
for example, “Plants and animals have as much right as humans to exist,”
NEP). These scales focus less on the moral foundations associated with conservatism, such as purity, which includes concerns about physical and spiritual contamination. We therefore call for new scales of environmental
attitudes that allow for different ways of relating to the environment. For
example, a speculative scale item that may better represent how conservatives relate to pro-environmental attitudes could be “When we pollute the
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
19
earth, we pollute ourselves,” and a potential item that might better mediate
Conscientiousness could read, “It is our duty to protect and conserve wildlife.” New scales could better mediate the effects of conservatism and high
Conscientiousness on environmental behaviors.
Recent studies have shown that congruency between environmental
appeals and political values can increase pro-environmental behavior
(Feinberg & Willer, 2012; Gromet et al., 2013; Kidwell, Farmer, & Hardesty,
2013). The current results showed no unique contribution of political orientation for emissions-reducing behavior. Therefore, messages targeting personality differences could be more effective than those based on political
orientation. It is possible that targeting high Openness and targeting
Democrats could be aiming at the same people, but the modest correlation
between Openness and political conservatism in this sample (r = –.14, p =
.01) suggests they comprise distinct groups. The behaviors in the emissionsreducing scale do not appear to signal environmental identity as strongly as
attending environmental rallies, using pro-environmental stickers, or driving
a hybrid-electric car, all of which appear in previous work. This weaker link
between emissions-reducing behaviors and identity could explain the modest
effect of political conservatism above.
It would be unwise to generalize from these results to all cultures. Although
the structure of five or six personality traits appears universal and crossculturally reliable (McCrae & Costa, 1997), the effects of personality on
behavior may be expressed differently based on cultural and social context.
In an individualistic culture like the United States (Markus & Kitayama,
1991), higher Conscientiousness might predict certain kinds of environmental behavior, but in a collectivistic culture such as Japan, higher
Conscientiousness could promote greater sensitivity to interpersonal harmony, and only predict environmental behavior when social norms reflect
positive attitudes for such behavior. Thus, the effect of personality on environmental behavior could be moderated by social context. Different types of
environmental behavior may even be predicted by different personality traits
across populations, which could explain the mixed results for
Conscientiousness between the current study and Markowitz et al. (2012).
These observations open new avenues for research, and also create difficulty
in interpreting results from geographically limited populations. A strength of
the current study is the geographically diverse sample spanning 47 U.S.
states. Future studies are encouraged to use diverse samples and explore cultural effects through cross-cultural comparisons.
In summary, core personality traits predicted emissions-reducing behavior, in particular Openness and Conscientiousness, and these links were
mediated by environmental attitudes. Political orientation had no unique
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
20
Environment and Behavior
contribution to predicting behavior, which suggests previous work focused
on political orientation may have overlooked the relative importance of personality and environmental attitudes. Stable individual differences and social
influence processes are both important for understanding and changing
behaviors that affect the natural environment (see Fleeson, 2004), and core
personality is a robust, reliable individual difference associated with environmental behavior.
Appendix
Emissions-reducing behaviors (Cronbach’s α = .78), rated from 1 (never) to
5 (always).
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
When you visit the grocery store, how often do you use reusable
bags?
How often do you walk, bicycle, carpool, or take public transportation instead of driving a vehicle by yourself?
How often do you drive slower than 60 mph on the highway?
How often do you go on personal (non-business) air travel? [reversed]
How often do you compost your household food garbage?
How often do you eat meat? [reversed]
How often do you eat dairy products such as milk, cheese, eggs, or
yogurt? [reversed]
How often do you eat organic food?
How often do you eat local food (produced within 100 miles)?
How often do you eat from a home vegetable garden (during the
growing season)?
How often do you turn your personal electronics off or in low-power
mode when not in use?
When you buy light bulbs, how often do you buy high-efficiency
compact fluorescent (CFL) or LED bulbs?
How often do you act to conserve water, when showering, cleaning
clothes, dishes, watering plants, or other uses?
When you are in PUBLIC, how often do you sort trash into the
recycling?
When you are in PRIVATE, how often do you sort trash into the
recycling?
Acknowledgment
We thank David Sherman and his research group, and the reviewers and editor, for
helpful comments.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
21
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article: This research was supported by Grant
DGE-0707430 from the National Science Foundation.
References
Allen, R. S., Castano, E., & Allen, P. D. (2007). Conservatism and concern for the
environment. Quarterly Journal of Ideology, 30, 1-25.
Ashton, M. C., & Lee, K. (2007). Empirical, theoretical, and practical advantages of
the HEXACO model of personality structure. Personality and Social Psychology
Review, 11, 150-166. doi:10.1177/1088868306294907
Ashton, M. C., Lee, K., & de Vries, R. E. (2014). The HEXACO honesty-humility, agreeableness, and emotionality factors: A review of research and theory. Personality
and Social Psychology Review, 18, 139-152. doi:10.1177/1088868314523838
Balderjahn, I. (1988). Personality variables and environmental attitudes as predictors
of ecologically responsible consumption patterns. Journal of Business Research,
17, 51-56. doi:10.1016/0148-2963(88)90022-7
Berinsky, A. J., Huber, G. A., & Lenz, G. S. (2012). Evaluating online labor markets
for experimental research: Amazon.com’s Mechanical Turk. Political Analysis,
20, 351-368. doi:10.1093/pan/mpr057
Block, J., & Block, J. H. (2006). Nursery school personality and political orientation
two decades later. Journal of Research in Personality, 40, 734-749. doi:10.1016/j.
jrp.2005.09.005
Brick, C., Sherman, D. K., & Kim, H. S. (in press). I can’t recycle in public: Visibility
and pro-environmental behaviors.
Carbon Footprint. (2012). Carbon footprint calculator. Retrieved from http://www.
carbonfootprint.com/calculator.aspx
Caspi, A., Roberts, B. W., & Shiner, R. L. (2005). Personality development: Stability
and change. Annual Review of Psychology, 56, 453-484. doi:10.1146/annurev.
psych.55.090902.141913
Caspi, A., & Silva, P. A. (1995). Temperamental qualities at age three predict personality traits in young adulthood: Longitudinal evidence from a birth cohort. Child
Development, 66, 486-498.
Center for Political Studies. (2013). The American National Election Study guide
to public opinion and electoral behavior. Ann Arbor: University of Michigan.
Retrieved from http://electionstudies.org/nesguide/citation.htm
Conner, M., & Abraham, C. (2001). Conscientiousness and the theory of planned
behavior: Toward a more complete model of the antecedents of intentions
and behavior. Personality and Social Psychology Bulletin, 27, 1547-1561.
doi:10.1177/01461672012711014
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
22
Environment and Behavior
Costa, D. L., & Kahn, M. E. (2013). Energy conservation “nudges” and environmentalist ideology: Evidence from a randomized residential electricity field experiment. Journal of the European Economic Association, 11, 680-702. doi:10.1111/
jeea.12011
DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets and domains:
10 aspects of the Big Five. Journal of Personality and Social Psychology, 93,
880-896. doi:10.1037/0022-3514.93.5.880
Diekmann, A., & Preisendörfer, P. (1998). Environmental behavior discrepancies between aspirations and reality. Rationality and Society, 10, 79-102.
doi:10.1177/104346398010001004
Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). Measuring
endorsement of the new ecological paradigm: A revised NEP scale. Journal of
Social Issues, 56, 425-442. doi:10.1111/0022-4537.00176
Dunlap, R. E., Xiao, C., & McCright, A. M. (2001). Politics and environment in
America: Partisan and ideological cleavages in public support for environmentalism. Environmental Politics, 10, 23-48. doi:10.1080/714000580
Elliot, A. J., & Thrash, T. M. (2002). Approach-avoidance motivation in personality:
Approach and avoidance temperaments and goals. Journal of Personality and
Social Psychology, 82, 804-818.
Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual differences.
New York, NY: Plenum Press.
Feinberg, M., & Willer, R. (2012). The moral roots of environmental attitudes.
Psychological Science, 24, 56-62. doi:10.1177/0956797612449177
Fleeson, W. (2004). Moving personality beyond the person-situation debate: The
challenge and the opportunity of within-person variability. Current Directions in
Psychological Science, 13, 83-87. doi:10.1111/j.0963-7214.2004.00280.x
Gray, J. A. (1981). A critique of Eysenck’s theory of personality. In P. H. J. Eysenck
(Ed.), A model for personality (pp. 246-276). Berlin, Germany: Springer.
Gromet, D. M., Kunreuther, H., & Larrick, R. P. (2013). Political ideology affects
energy-efficiency attitudes and choices. Proceedings of the National Academy of
Sciences, 110, 9314-9319. doi:10.1073/pnas.1218453110
Haidt, J. (2007). The new synthesis in moral psychology. Science, 316, 998-1002.
doi:10.1126/science.1137651
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process
analysis: A regression-based approach. New York, NY: Guilford Press.
Henrich, J., Heine, S., & Norenzayan, A. (2010). The weirdest people in the world?
(SSRN Scholarly Paper No. ID 1601785). Rochester, NY: Social Science
Research Network.
Hilbig, B. E., Zettler, I., Leist, F., & Heydasch, T. (2013). It takes two: Honesty–
humility and agreeableness differentially predict active versus reactive cooperation. Personality and Individual Differences, 54, 598-603. doi:10.1016/j.
paid.2012.11.008
Hilbig, B. E., Zettler, I., Moshagen, M., & Heydasch, T. (2012). Tracing the path
from personality—via cooperativeness—to conservation. European Journal of
Personality, 27, 319-327. doi:10.1002/per.1856
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Brick and Lewis
23
Hirsh, J. B. (2010). Personality and environmental concern. Journal of Environmental
Psychology, 30, 245-248. doi:10.1016/j.jenvp.2010.01.004
Hirsh, J. B., & Dolderman, D. (2007). Personality predictors of consumerism and
environmentalism: A preliminary study. Personality and Individual Differences,
43, 1583-1593. doi:10.1016/j.paid.2007.04.015
Intergovernmental Panel on Climate Change. (2013). AR5 synthesis report.
Intergovernmental Panel on Climate Change. Retrieved from http://www.ipcc.
ch/report/ar5/wg1/
John, O., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. In L. Pervin & O. John (Eds.), Handbook of
personality: Theory and research (pp. 102-138). New York, NY: Guilford Press.
Kaiser, F. G., Wölfing, S., & Fuhrer, U. (1999). Environmental attitude and ecological behaviour. Journal of Environmental Psychology, 19, 1-19. doi:10.1006/
jevp.1998.0107
Kidwell, B., Farmer, A., & Hardesty, D. M. (2013). Getting liberals and conservatives to go green: Political ideology and congruent appeals. Journal of Consumer
Research, 40, 350-367. doi:10.1086/670610
Lee, K., & Ashton, M. C. (2004). Psychometric properties of the HEXACO Personality
Inventory. Multivariate Behavioral Research, 39, 329-358.
Lee, K., & Ashton, M. C. (2005). Psychopathy, machiavellianism, and narcissism in the
five-factor model and the HEXACO model of personality structure. Personality
and Individual Differences, 38, 1571-1582. doi:10.1016/j.paid.2004.09.016
Lee, K., Ashton, M. C., Ogunfowora, B., Bourdage, J. S., & Shin, K.-H. (2010).
The personality bases of socio-political attitudes: The role of honesty–humility
and openness to experience. Journal of Research in Personality, 44, 115-119.
doi:10.1016/j.jrp.2009.08.007
Markowitz, E. M., Goldberg, L. R., Ashton, M. C., & Lee, K. (2012). Profiling
the “pro-environmental individual”: A personality perspective. Journal of
Personality, 80, 81-111. doi:10.1111/j.1467-6494.2011.00721.x
Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224-253.
Mason, W., & Suri, S. (2012). Conducting behavioral research on Amazon’s
Mechanical Turk. Behavior Research Methods, 44, 1-23. doi:10.3758/s13428011-0124-6
MaxMind. (2013). GeoLite2 free downloadable databases. Waltham, MA. Retrieved
from http://dev.maxmind.com/geoip/geoip2/geolite2/
Mayer, F. S., & Frantz, C. M. (2004). The connectedness to nature scale: A measure of individuals’ feeling in community with nature. Journal of Environmental
Psychology, 24, 503-515. doi:10.1016/j.jenvp.2004.10.001
McCrae, R. R., & Costa, P. T., Jr. (1997). Personality trait structure as a human universal. American Psychologist, 52, 509-516. doi:10.1037/0003-066X.52.5.509
McCrae, R. R., & Costa, P. T., Jr. (1999). A five-factor theory of personality. In L. A.
Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (2nd
ed., pp. 139-153). New York, NY: Guilford Press.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
24
Environment and Behavior
Milfont, T. L., Duckitt, J., & Wagner, C. (2010). A cross-cultural test of the value–
attitude–behavior hierarchy. Journal of Applied Social Psychology, 40, 27912813. doi:10.1111/j.1559-1816.2010.00681.x
Milfont, T. L., & Sibley, C. G. (2012). The big five personality traits and environmental engagement: Associations at the individual and societal level. Journal of
Environmental Psychology, 32, 187-195. doi:10.1016/j.jenvp.2011.12.006
Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of
Experimental Social Psychology, 45, 867-872. doi:10.1016/j.jesp.2009.03.009
Paunonen, S. V. (2003). Big Five factors of personality and replicated predictions of behavior. Journal of Personality and Social Psychology, 84, 411-422.
doi:10.1037/0022-3514.84.2.411
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for
assessing and comparing indirect effects in multiple mediator models. Behavior
Research Methods, 40, 879-891. doi:10.3758/BRM.40.3.879
Rentfrow, P. J., Gosling, S. D., Jokela, M., Stillwell, D. J., Kosinski, M., & Potter, J.
(2013). Divided we stand: Three psychological regions of the United States and
their political, economic, social, and health correlates. Journal of Personality and
Social Psychology, 105, 996-1012. doi:10.1037/a0034434
Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level
change in personality traits across the life course: A meta-analysis of longitudinal
studies. Psychological Bulletin, 132, 1-25. doi:10.1037/0033-2909.132.1.1
Sibley, C. G., Luyten, N., Purnomo, M., Mobberley, A., Wootton, L. W., Hammond,
M. D., . . .Robertson, A. (2011). The Mini-IPIP6: Validation and extension of a
short measure of the Big-Six factors of personality in New Zealand. New Zealand
Journal of Psychology, 40, 142-159.
Slutske, W. S., Moffitt, T. E., Poulton, R., & Caspi, A. (2012). Undercontrolled
temperament at age 3 predicts disordered gambling at age 32: A longitudinal study of a complete birth cohort. Psychological Science, 23, 510-516.
doi:10.1177/0956797611429708
Stern, P. C. (2000). New environmental theories: Toward a coherent theory of
environmentally significant behavior. Journal of Social Issues, 56, 407-424.
doi:10.1111/0022-4537.00175
Author Biographies
Cameron Brick is a PhD candidate in Psychological and Brain Sciences at the
University of California, Santa Barbara, USA. He studies how individuals think about
and react to environmental problems such as climate change.
Gary J. Lewis, PhD, is a Lecturer in Psychology at the University of York, York, UK.
His research broadly addresses issues concerning the biological and environmental
etiologies of social attitudes and personality.
Downloaded from eab.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on October 16, 2014
Journal of Experimental Social Psychology 56 (2015) 73–88
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology
journal homepage: www.elsevier.com/locate/jesp
Going along versus getting it right: The role of self-integrity in
political conformity
Kevin R. Binning a,⁎, Cameron Brick b, Geoffrey L. Cohen c, David K. Sherman b
a
b
c
University of Pittsburgh, USA
University of California, Santa Barbara, USA
Stanford University, USA
H I G H L I G H T S
•
•
•
•
Participants who afﬁrmed their self-integrity were not swayed by political norms.
Participants who afﬁrmed their self-integrity were swayed by evidentiary data.
The effects persisted over time and transferred to novel political stimuli.
The effects were moderated by participants' identiﬁcation with America.
a r t i c l e
i n f o
Article history:
Received 11 February 2014
Revised 23 July 2014
Available online 06 September 2014
Keywords:
Social inﬂuence
National identity
Conformity
Presidential approval
Opinion polling
a b s t r a c t
People often conform to the opinions of ingroup members, even when available evidence suggests that the group
is misinformed. Following insights from the social identity approach and self-afﬁrmation theory, it was hypothesized that people conform to salient opinions in an effort to maintain global self-integrity. In a series of experiments examining Americans' approval of President Obama and his policies, approval was consistently swayed by
normative information (national polling data) but not by evidentiary information (indicators of national
economic health), except under theory-predicted conditions. When participants had satisﬁed their sense of
self-integrity with a self-afﬁrmation exercise (Democrats in Study 1, Republicans in Study 2), or when they
had low levels of American identiﬁcation and thus were less concerned with national norms (Democrats and Republicans in Study 3), they showed the opposite pattern and were swayed by evidence in spite of contradicting
normative information. The extent to which people are inﬂuenced by norms versus evidence in political judgment is shaped by social identity, one aspect of self-integrity. The results highlight a social psychological
means to attenuate and potentially reverse conformity in the face of contradicting evidence, a ﬁnding with
both practical and theoretical implications.
© 2014 Elsevier Inc. All rights reserved.
We are no longer led by men. We are led around by the polls.
[Edward Bernays (1945)]
Opinion polls reﬂect public opinion and, through processes of social
inﬂuence, can also shape it. The observation that polls can “wag the dog”
and causally affect opinion was made by Edward Bernays (1891–1995),
a pioneer in the ﬁeld of public relations and the science of political spin
(Tye, 1998; see also Ceci & Kain, 1982; Marsh, 1985; Simon, 1954).
⁎ Corresponding author at: Department of Psychology, University of Pittsburgh, 3103
Sennott Square, Pittsburgh, PA 15260, USA. Fax: +1 412 624 9149.
E-mail address: [email protected] (K.R. Binning).
http://dx.doi.org/10.1016/j.jesp.2014.08.008
0022-1031/© 2014 Elsevier Inc. All rights reserved.
Social psychological research has found that descriptive norms – that
is, norms that describe how typical group members think, feel, or
behave (Grube, Morgan, & McGree, 1986; Terry & Hogg, 1996) – can
powerfully affect individual behavior (e.g., Schultz, Nolan, Cialdini,
Goldstein, & Griskevicius, 2007), particularly when norms are regarded
as neutral and authoritative (Petty & Cacioppo, 1986), as polls often are
(Bernays, 1945). People's susceptibility to normative social inﬂuence
has implications for democratic decision-making and speaks to longstanding concerns about psychology of conformity and independence
in judgment (see Asch, 1951; Cohen, 2003; Kahan, Jenkins‐Smith, &
Braman, 2011; Sherif, 1936). People who make evidence-based
decisions that diverge from the group can play an essential role in
preventing destructive group processes such as groupthink (Janis,
1982) and can halt social inertia toward what John Adams called “the
tyranny of the majority” (Adams, 1794, p. 261). This paper explores
74
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
conditions that foster independence in political judgment and resistance from the sway of normative information.
We offer a self-integrity approach (Steele, 1988; see also Sherman &
Cohen, 2006) to understand when people are likely to conform to
salient ingroup norms (normative information, such as opinion polls
about the state of the economy), and when they are likely to
engage in independent judgment based on probative indicators of fact
(evidentiary information, such as concrete economic indicators like
unemployment or housing sales). The division between normative
and evidentiary information has roots in theory suggesting that people
process these two types of information in qualitatively different ways
(see Deustch & Gerard, 1955), and that, depending on how people
construe the psychological environment, either type of information
may carry judgment (e.g., Campbell & Fairey, 1989; Chen, Shechter, &
Chaiken, 1996).
We argue that normative and evidentiary information are likely
to produce different effects on judgment because each type of information serves unique psychological functions (e.g., Katz, 1960;
Snyder & DeBono, 1985). People may use normative information
for collective-level goals (to get along, to ﬁt in), whereas they may
use evidentiary information in pursuit of individual-level goals (accuracy or neutrality in judgment). Our central thesis is that people's
concerns with their self-integrity in the social context will determine
the type of information that they use. Integrating insights from the
social identity approach (Tajfel & Turner, 1986; Terry & Hogg,
1996; Turner & Reynolds, 2011) and self-afﬁrmation theory (Cohen
& Sherman, 2014; Sherman & Cohen, 2006; Steele, 1988), we explain
when and why people are likely to follow the crowd at the expense of
evidence-based decision-making. We suggest that knowledge of the
role of self-integrity in information processing highlights a means to
halt conformity and foster independent evaluations based on salient
evidentiary data.
Normative information and collective identity
People's perceptions of descriptive group norms are powerful predictors of a range of diverse outcomes, such as exercise behavior
(Terry & Hogg, 1996), environmental conservation (Schultz et al.,
2007), judgments of prejudice (Binning & Sherman, 2011), and likelihood of voting (Coleman, 2007). Such conformity is pervasive, in part,
because conformity can be socially and evolutionarily adaptive
(Coultas, 2004). Normative information helps specify how to behave
in ambiguous situations, informing people about what is seen as the
right way to act for “people like us” (e.g., Abrams, Wetherell,
Cochrane, Hogg, & Turner, 1994; Binning, 2007; Hogg & Reid, 2006).
Group norms provide information about how to maintain acceptance
in the group and, by extension, how to avoid becoming a “black
sheep” (Marques, Yzerbyt, & Leyens, 1988). People are especially likely
to conform to groups they ﬁnd attractive (Jackson & Saltzstein, 1958)
and to groups that have a high level of cohesion or interdependence
(Deustch & Gerard, 1955). Notably, people are generally unaware of
the powerful impact of normative information on their own attitudes
and behavior (Cialdini & Goldstein, 2004; Cohen, 2003; Latanè &
Darley, 1970; Nolan, Schultz, Cialdini, Goldstein, & Griskevicius, 2008;
Ross & Ward, 1996).
Social identity theory (Tajfel & Turner, 1979) and the related selfcategorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell,
1987; Turner & Reynolds, 2011) help explain how group norms get
their power. A core assumption of the social identity approach is that
the self-concept is constructed along an individual–social continuum
(Tajfel & Turner, 1986), such that people pursue individual goals and
behaviors (e.g., accuracy and individual performance) when selfcategorized at the individual level, but they pursue group goals and
behaviors (e.g., to maintain a positive social identity) when selfcategorized on the social level. Deﬁnitions of the self shift ﬂuidly along
this continuum, as people deﬁne themselves as an individual in one
context and as a member of a social group in the next, depending on
what aspect of identity is salient in that context (Tajfel & Turner,
1986; Turner & Reynolds, 2011).
Another core assumption of the social identity approach is that
regardless of which aspect of the self-concept is salient, people
have a basic motivation to maintain a positive self-concept. When
collective identity is made salient, people may be compelled to go
along with group norms in order to be a good group member and
maintain their positive standing. When individual identity is made
salient, people may instead ignore group normative information
and strive to maintain a positive individual identity. In a study on
group norms for physical exercise behavior, for example, people's
own exercise behavior varied in line with group norms, but only
among individuals who strongly identiﬁed with the group (Terry &
Hogg, 1996). People with low identiﬁcation, by contrast, were not affected by group norms but rather by perceptions of behavioral control, an individual-level factor. The information people attended to
was determined by the immediate relevance of each type of information to the self-concept.
When norms conﬂict with evidence: a self-integrity approach
In many cases, average group beliefs and norms converge with
available evidentiary information (see Hardin & Higgins, 1996; Insko,
Drenan, Solomon, Smith, & Wade, 1983; Surowiecki, 2004). In political
contexts, when norms align with evidence, normative information
simply provides a reﬂection or barometer of reality, which is the presumptive purpose of most opinion polling. However, a critical question
for both democratic decision-making and the present research is what
happens when the group ignores or disregards emerging facts and evidence (e.g., Fast, Heath, & Wu, 2009). In such cases, resistance to group
norms and attention to evidence could be advantageous — at least from
the standpoint of people's desire to be accurate and independent
decision-makers (Chaiken, Liberman, & Eagly, 1989). Imagine, for example, that the public tends to believe that the economy is in decline
when major indicators suggest that the economy is on the rebound. If
someone is asked for their opinion about the state of the economy,
they would be more accurate if they followed the economic evidence
and ignored the bubble created by popular consensus. However, as
scholars have noted for well over a century (Asch, 1951; Le Bon,
1897), it is often difﬁcult to go against the group. When categorized at
the collective level, going against group norms requires going against
a part of one's self (e.g., Tajfel & Turner, 1986).
Self-afﬁrmation theory (Sherman & Cohen, 2006; Steele, 1988) offers a perspective to understand how people may transcend the pressure of collective identities on judgment. The theory suggests that
although concerns with individual and collective identity may ﬂuctuate
from context to context, an overarching psychological goal is to maintain a global sense of self-integrity: a general feeling of being efﬁcacious,
adequate, and “good enough” (Cohen & Sherman, 2014; Sherman,
2013; Sherman & Cohen, 2006; Steele, 1988). We argue in the present
research that this general goal guides what people attend to in their environment. In some situations, people are concerned more with collective goals and are therefore likely to rely on group norms in judgment.
In other situations, people may feel less attached to the group norms
and instead have an interest in being accurate or independent (e.g., a
neutral judge or referee). However, common to both of these situations
is people's concern with self-integrity. We hypothesize that regardless
of whether people rely on normative or evidentiary information, they
do so in an effort to maintain global self-integrity. Following this reasoning, by manipulating global self-integrity it should be possible to shift
the manner in which people process information. Under certain circumstances, manipulating self-integrity should halt conformity and orient
people toward independent, evidence-based decision-making.
To illustrate how global self-integrity concerns might shape conformity and independence, we use the experimental paradigm developed
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
in self-afﬁrmation research. Typically, to examine the role of selfintegrity motivations in a social psychological process, two experimental conditions are compared (for a review see McQueen & Klein, 2006).
In the control condition, participants are exposed to information that
threatens their sense of self-integrity (e.g., information indicating they
engage in risky health behaviors) and their psychological responses
are observed. People generally respond defensively and resist or downplay the implications of such evidentiary information (Dunning, 2003;
Kunda, 1987; Liberman & Chaiken, 1992). The other experimental condition involves a self-afﬁrmation exercise. Prior to being exposed to the
threatening information, participants write about an important personal value (e.g., their sense of humor, love of family) unrelated to the
threatened domain. They typically express why this value is important
to them and a time in their life when it was particularly important. Because values are central to people's self-deﬁnition (Rokeach, 1973), this
exercise afﬁrms people's self-integrity (Steele, 1988). In this condition,
people tend to be more open to the threatening information, as if the assurance they got from the afﬁrmation exercise allowed them to evaluate
the subsequent information more openly and candidly. The theory
holds that by satisfying global self-integrity, afﬁrmations psychologically equip people to tolerate threats to self-integrity and respond less defensively (Cohen & Sherman, 2014; Sherman & Hartson, 2011).
In political contexts where integrity is maintained by conforming to
group norms and resisting evidentiary data, the act of securing selfintegrity prior to judgment should attenuate conformity. Conversely,
in contexts where self-integrity is maintained by being neutral, accurate, and even-handed (e.g., being a neutral judge or referee), afﬁrmation may attenuate such accuracy motivation (see Cohen et al., 2007).
Thus, self-afﬁrmation should attenuate conformity in contexts where
people conform out of a desire to maintain self-integrity.
Previous work supports the idea that self-afﬁrmation attenuates
identity-defensive processes. For example, people who self-afﬁrmed
prior to group activity were more willing to acknowledge wrong-doing
by an ingroup (Adams, Tormala, & O'Brien, 2006; Čehajić-Clancy,
Effron, Halperin, Liberman, & Ross, 2011); they showed lower partisan
bias in the days prior to a presidential election (Binning, Sherman,
Cohen, & Heitland, 2010); and they displayed fewer group-serving attributions for group success and failure (Sherman & Kim, 2005). When
Americans were presented with a report that was critical of U.S. foreign
policy, afﬁrmed participants were less partisan in their evaluations of
the report (Cohen et al., 2007). By satisfying global self-integrity prior
to engaging in judgment, self-afﬁrmation appeared to relieve pressure
associated with the collective self and salient social identities. Notably,
such research has typically focused on how afﬁrmations can increase receptiveness to threatening information. Research has yet to examine
whether afﬁrming self-integrity can actually increase resistance to normative pressure.
We suggest that afﬁrmations of self-integrity should not only
make group members resistant to normative political information,
they should also simultaneously increase accuracy motivation and
openness to facts and evidence. Prior research supports this argument. For example, in several studies afﬁrmed participants were
more likely to engage deliberative, systematic thinking. Afﬁrmed
participants were more sensitive to the strength (versus weakness)
of arguments when responding to questions about capital punishment (Correll, Spencer, & Zanna, 2004; see also Klein, Harris,
Ferrer, & Zajac, 2011) and were more persuaded by evidence challenging their views toward capital punishment (Cohen, Aronson, &
Steele, 2000). In the health domain, afﬁrmed participants became
better calibrated to their personal risk levels after evaluating threatening health messages, as afﬁrmed individuals saw themselves at
high risk when they were at high risk and at low risk when they
were at low risk (Grifﬁn & Harris, 2011; also see Harris & Epton,
2009, 2010; Harris & Napper, 2005). These ﬁndings suggest that afﬁrmed participants are less likely to over-rely on heuristic cues like
group norms and more likely to be persuaded by factual evidence.
75
In summary, afﬁrmations have been shown to reduce both identity
defense and to increase accuracy motivation. The present research attempts to integrate and test both research outcomes simultaneously.
People's judgment may be guided by either norms or evidence, depending on which type of information is relevant to maintaining selfintegrity. Following the continuum metaphor of the self-concept from
social identity theory, normative and evidentiary information may
each be relevant to different aspects of the self-concept. Normative information should guide judgment when social identity is made salient
because such information informs group members how to maintain
self-integrity by responding like good or typical group members. Evidentiary information should guide judgment, by contrast, when selfintegrity has been afﬁrmed and group members are therefore freed
from the need to maintain the integrity of their social identity. Rather
than being swayed by group norms, afﬁrmed group members may be
more concerned with individual-level goals of accuracy and independence in judgment.
Overview and predictions
Conformity is a ubiquitous social phenomenon, and political contexts provide rich and consequential forums to study it. In the present
research we focus on how Americans go about the task of evaluating a
sitting United States president, Barack Obama. The standards for judging
presidents are ambiguous (e.g., with regard to who or what they are
compared), and research suggests that people actually seek out opinion
polls when faced with difﬁcult or ambiguous political decisions
(Boudreau & McCubbins, 2010). Exposure to such group normative information may directly shape public opinion, a possibility that has led
a number of countries to restrict or ban the reporting of pre-election
opinion polls (Chung, 2012).
The scope of the hypotheses that we test broadens with each of
the three studies presented below. Study 1 is an initial demonstration that self-afﬁrmation can untether group members from the inﬂuence of salient ingroup norms. For this study we draw on actual
news reports that made American identity salient while spinning
the president's poll numbers in opposite ways (favorably or unfavorably), and we examine how control and afﬁrmed participants respond to the polls. Study 2 examines the hypothesis that while
afﬁrmations can increase group members' resistance to salient
norms, afﬁrmations should simultaneously increase reliance on evidentiary information. Thus we examine not only how selfafﬁrmation affects responses to normative information
(i.e., approval of Obama's handling of the economy), but also how
it affects responses to evidentiary information regarding the
president's performance (i.e., actual economic activity during
Obama's tenure as president). Based on ﬁndings that afﬁrmation effects can persist over time (see Cohen & Sherman, 2014 for a review),
Study 2 included a four-month follow-up to test for the effects of the
original afﬁrmation.
Both Study 2 follow-up and Study 3 further tested key theoretical
predictions. If, as suggested by our interpretation of social identity theory, normative and evidentiary information are processed with respect
to different psychological goals, then simultaneously contrasting normative and evidentiary information should yield predictable responses.
We theorized that control participants would be responsive to norms
but not to evidence, whereas afﬁrmed participants would be responsive
to evidence but not to norms. To test this, norms and evidence were
contrasted within a single persuasive message about the president
(e.g., people think President Obama is doing great on the economy,
but the economic data suggest otherwise vs. people think President
Obama is doing poorly on the economy, but the economic data suggest
otherwise).
These predictions are motivated by insights from the social identity
approach and self-afﬁrmation theory. Study 3 seeks to illustrate the theoretical integration of these approaches. First, the social identity
76
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
approach suggests that people who identify with a social category tend
to pursue social identity goals more so than individual-level goals when
that identity is salient (e.g., Terry & Hogg, 1996; see also Cohen et al.,
2007). Study 3 sought to test this idea by examining how individual differences in social identiﬁcation inﬂuence attention to normative and evidentiary information. We predicted that higher identiﬁcation with
America is predictive of conformity to normative but not evidentiary information, but critically, this should not hold among individuals with
low levels of identiﬁcation, as those norms are not relevant to the selfconcept in low-identiﬁed individuals. Instead, these individuals may remain categorized at the individual level and, as a result, be more likely to
pursue individual-level goals of accuracy and independence in judgment (cf. Terry & Hogg, 1996). Self-afﬁrmation should, in turn, attenuate the style of processing people engaged in to maintain self-integrity
in a given situation.
Study 1
This study was conducted in mid-2009, when President Obama had
an approval rating of roughly 56% (Gallup, 2014) and was attempting to
rally support for health care reform legislation. Although overall approval for President Obama was strong, his approval ratings had declined from their initial highs during the early part of his presidency.
As such, it was possible to spin the president's poll numbers in different
ways: by focusing on the objective strength of the poll numbers, a favorable picture could be painted; by focusing on the relative decline in poll
numbers, a more negative picture could be painted. After researching
the current news coverage of the president's poll numbers, we located
two news articles, one framing the president's poll numbers positively
(Anderson, 2009) and another framing them negatively (Schoen &
Rasmussen, 2009). Using actual content from these articles, we then
manipulated popular opinion by giving half the participants the favorable content that Obama was soaring in the polls and the other half
the unfavorable content that Obama was sinking in the polls. Thus, the
articles were intended to manipulate normative information in an ecologically valid fashion (whereas in Study 2 and Study 3 we constructed
ﬁctitious articles to rigorously control for content and strength of
message).
Prior to reading a news article, participants were randomly assigned
to either a control condition or a self-afﬁrmation condition. Our primary
question was whether this self-afﬁrmation manipulation would affect
how participants responded to the articles. In the control condition,
we used the social identity approach as a basis for our predictions.
Based on the idea that people tend to conform to salient ingroup
norms, in the control condition approval should conform to the national
polling data: participants who receive the positive spin should be more
favorable toward Obama than participants who receive the negative
spin.
In the self-afﬁrmation condition, we predicted that bolstering
self-integrity prior to reading the polling articles would untether
the self-concept from the implications of the norms. Thus, we expected that unlike control participants, afﬁrmed participants would
not be swayed by the article spin. To help rule out the possibility
that participants simply responded favorably (or unfavorably) to
all targets when they received positive (or negative) news about
Obama, we examined participants' evaluations of a non-salient
group, Congressional Republicans, whom we predicted would not
be swayed by the polling articles. Finally, as exploratory measures,
we predicted that control (Democratic supporting) participants
would donate more money to the Democratic Party in a hypothetical
allocation task and that their self-feelings would be more positive in
the favorable spin than in the unfavorable spin condition. The allocations and self-feelings of afﬁrmed participants, by contrast, should be
affected less by the article spin in favor or against Obama and his
policies.
Method
Participants and design
One hundred ﬁfteen adults (70% women; 70% White, 19% Asian
or Asian American, 9% African American, 4% Latino/a; MAge =
34.50 years) were recruited from a university-maintained listserv of
US residents and were compensated with a $5 US gift card to an online
retailer. To help limit noise created by variance in political persuasion,
participants were recruited if they indicated an afﬁliation with the Democratic Party during pre-screening and were randomly assigned to one
of four cells in a 2 (Afﬁrmation status: Afﬁrmation vs. Control) × 2 (Normative information: Soaring in polls vs. Sinking in polls) factorial design. Participants were not made aware that they were pre-selected
based on their political party preference.
Procedure
After providing consent, participants completed a standard afﬁrmation manipulation (see McQueen & Klein, 2006). All participants
were presented with a list of 11 nonpolitical values.1 Participants in
the self-afﬁrmation condition were instructed to “pick the one
value that is most important to you” and to type that value in a text
box. Next, participants were instructed to “think about a time
when your #1 value or characteristic was important to you” and to
“write a few sentences about a time when this value was important.”
Consistent with previous research, participants in the control condition received the same list but indicated their least important value
and described why someone else might think the value is important
(see Binning et al., 2010; Cohen et al., 2007). A great deal of research
has employed a similar approach to self-afﬁrmation and supports the
idea that the afﬁrmation condition reduces defensiveness and increases self-security, whereas the control condition produces reactions that are orthogonal to participants' sense of self-integrity
(McQueen & Klein, 2006; Sherman & Hartson, 2011).
All participants were then randomly assigned to receive one of two
news articles, which were formatted to resemble an authentic online article from a high -proﬁle, international newspaper. Each article included
factual content from news articles circulating 1–2 months prior to the
study, but each had a very different political spin (i.e., Anderson, 2009;
Schoen & Rasmussen, 2009). Participants randomly assigned to the
Obama Soaring condition read an article titled, “Obama Soaring in the
Polls,” which described how Obama was performing well in the polls
(e.g., “56% of Americans gave Obama an excellent or good job rating”),
together with a picture of Obama smiling and waving toward the camera. Participants assigned to the Obama Sinking condition read an article
matched for length and level of detail titled, “Obama's Poll Numbers are
Falling to Earth,” which described how the American people were
showing increasing doubts about Obama's presidency (e.g., “33% of
Americans now strongly disapprove of the presidents' performance”).
This article featured a picture of Obama with his hand on his forehead
while looking down at the ground.
After reading their respective news articles, all participants
watched two one-minute YouTube videos. In the ﬁrst video, President Obama described three pillars of his health care reform agenda,
which was being debated by Congress at the time of the study. In the
second video, Republican congressional leaders presented a rebuttal
to the president's position. Immediately after the videos, participants
were asked to complete the measures below.
1
The 11 values listed for the manipulation were Artistic skills/esthetic appreciation,
Sense of humor, Relations with friends/family, Spontaneity/living life in the moment, Social Skills, Athletics, Musical ability/appreciation, Physical attractiveness, Creativity, Business/managerial skills, and Romantic values.
77
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
Approval of Obama
Participants evaluated Obama and his approach to health care using
the following items: “How balanced and objective is Obama's health care
outline?” (1 = Not balanced or objective at all; 9 = Very balanced and
objective), “In general, how knowledgeable about health care issues is
Barack Obama?” (1 = Not knowledgeable at all; 9 = Very knowledgeable).
Two additional items measured general approval using wording from a
recent Gallup Poll (Saad, 2009): “Do you approve or disapprove of the
way Barack Obama is handling his job as president?” (1 = Strongly
disapprove; 9 = Strongly approve), and “How would you rate the job
Barack Obama has been doing as president so far?” (1 = Excellent,
2 = Good, 3 = Just Okay, 4 = Poor, 5 = Terrible). This item was
reverse-scored and transformed to a 9-point scale. The four items loaded
on a single factor and the mean composite of all four items displayed
good reliability (α = .89, M = 7.00, SD = 1.41).
Evaluations of Republicans
Although no polling data about Republicans were presented to participants, we examined if evaluations of Republicans were affected by
the presidential polling data, the afﬁrmation manipulation, or both together. Participants completed four items that paralleled the four
Obama-evaluation items above: “How balanced and objective is the Republicans' health care outline?” “In general, how knowledgeable about
health care issues is the Republican leadership?” “Do you approve or
disapprove of the way Republicans are handling their job in Congress?”
and, “How would you rate the job Republicans in Congress have been
doing?” The composite also showed unidimensionality and good
reliability (α = .85, M = 3.96, SD = 1.59). Three participants did not
complete this measure and their data were treated as missing.
Hypothetical allocation
Having seen the arguments from both sides of the health care debate,
participants were provided a scenario in which they considered having a
ﬁxed sum of $10,000 to distribute between the Democratic and Republican Parties to help work toward health care reform. They were presented
with ﬁve options for allocating the money, with higher scores
representing higher sums of money given to the Democratic Party
relative to the Republican Party (1 = $0 for Democrats and $10,000 for
Republicans; 2 = $2500 for Democrats and $7500 for Republicans; 3 =
$5000 for each party, 4 = $7500 for Democrats and $2500 for Republicans;
5 = $10,000 for Democrats and $0 for Republicans) (M = 3.83, SD =
0.84). One participant did not complete this measure and is treated
as missing.
2
In all studies, the focal outcome variable was a composite of participants' approval for
President Obama, and we measured it in a consistent manner across replications of the key
ﬁndings (i.e., the focal measure in Study 2 is analogous to that used in Study 1, and the focal measure in Study 3 is analogous to that used in the Study 2 follow-up). In addition, we
report analyses of data on participants' self-feelings and assessments of Republicans
(Studies 1 and 2), and we report results from a theoretically relevant allocation task used
in Study 1. In the interest of full transparency (Simmons, Nelson, & Simonsohn, 2011), we
note that data from several exploratory variables are not reported here. These assessed
participants' beliefs about the appropriate role of government in daily life (in Study 1),
their attitudes about Sarah Palin, a measure of Republicans' collective-self-esteem, and
perceptions of American opinion about Obama (in Study 2), their feelings of selfcertainty, life satisfaction, and construal levels (in the Study 2 four-month follow-up),
and a behavioral measure of media preferences (in the Study 2 follow-up and Study 3).
Across all studies, an item that tapped perceptions that President Obama was ideologically
biased was also assessed but not analyzed because it consistently did not load together
with the main approval items. The former measures were exploratory because they were
tangential to the key dependent measure assessing participants' conformity to the normative political information. Additionally, an exploratory manipulation was introduced following participants' completion of the main dependent measure in Study 3. Some
participants were randomly assigned to read an article that attempted to threaten them
in the health domain because we were interested in potential spill-over effects of the afﬁrmation beyond the ﬁrst domain of judgments. Additional data relevant to this hypothesis
and obtained after this manipulation are not reported here but may be the subject of a
forthcoming report.
Self-feelings
Near the end of the study they were asked to respond to a single
item “How are you feeling about yourself right now?” on a nine-point
scale, with end points labeled 1 (Very negatively) and 9 (Very positively).
This one-item measure of self-feelings has been used in prior research to
assess momentary feelings after completing afﬁrmation manipulations
(Cohen et al., 2000; Sherman, Nelson, & Steele, 2000). Two participants
did not complete this measure and their data were treated as missing.
Results and discussion
The composite of evaluations of President Obama were subjected to
a 2 (Afﬁrmation Status: Afﬁrmation vs. Control) × 2 (Normative information: Soaring vs. Sinking) between-subjects analysis of variance
(ANOVA). Results revealed a main effect of normative information,
such that favorability toward Obama was higher when he was soaring
in the polls (M = 7.24, SE = .18) than when he was sinking in the
polls (M =6.73, SE = .19), F(1, 111) = 3.93, p = .050, η2P = .03.
The main effect for afﬁrmation status was not signiﬁcant, F(1, 111) =
0.12, p = .732. However, as hypothesized, there was a signiﬁcant twoway interaction, F(1, 111) =5.80, p = .018, η2P = .05, depicted in Fig. 1.
Simple effects tests revealed that in the control condition, participants
were signiﬁcantly swayed by group norms, as they were more favorable
toward Obama when he was soaring in the polls (M = 7.51, SE = .26)
than when he was sinking the polls (M = 6.37, SE = .28), F(1, 111) =
8.80, p = .004, η2P = .07. Also as hypothesized, afﬁrmed participants
resisted the pressure of group norms, as there was no difference in favorability toward President Obama when he was up in the polls (M = 6.98,
SE = .24) than when he was down in the polls (M =7.09, SE = .26), F(1,
111) = 0.10, p = .752.
Evaluations of Republicans
In contrast to evaluations of President Obama, there were no main
effects, Fs b 1.0, ps N .875, nor a two-way interaction, F(1, 108) = .01,
p = .934, in participants' evaluations of the Republican leadership, consistent with predictions.
Hypothetical allocation
Participants' ingroup allocation scores were subjected to a 2 (Afﬁrmation status) × 2 (Normative information) ANOVA. Results revealed
a marginal main effect for afﬁrmation status, F(1, 110) = 3.30, p =
.072, η2P = .03, suggesting that participants in the self-afﬁrmation condition were marginally less biased in their allocations to the ingroup
(M = 3.71 or $6775 of the $10,000 allocated for Democrats, SE = .10)
than were participants in the control condition (M = 3.98 or $7450
for Democrats, SE = .11). Notably, this main effect suggests that selfafﬁrmation slightly attenuated partisan identity bias, perhaps because
FAVORABILITY TOWARD OBAMA
Measures2
9
8
7
6
Obama soaring
5
Obama sinking
4
3
2
1
Control
Aﬃrma"on
Fig. 1. Democrats' mean favorability toward President Barack Obama (±1 standard error)
as a function of afﬁrmation status and normative data regarding Obama's status in the
current polls in Study 1.
78
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
participants' identity as Democrats may have been active and affecting
their overall responses. The effect was qualiﬁed by a signiﬁcant twoway interaction, F(1, 110) = 12.24, p b .001, η2P = .10. Simple effects
tests showed that in the control condition, participants' allocation decisions followed the national norms, as they allocated marginally more
money to the Democrats when President Obama was soaring in the
polls (M = 4.18 or $7950, SE = .15) than when he was falling in the
polls (M = 3.78 or $6950, SE = .17), F(1, 110) = 3.16, p = .078,
η2P = .03. By contrast, afﬁrmation participants showed the opposite pattern, as they actually gave signiﬁcantly more to the Democrats when
Obama was falling in the polls (M = 4.03 or $7575, SE = .15) than
when he was soaring in the polls (M = 3.38 or $5950, SE = .14), F(1,
110) = 10.61, p = .001, η2P = .09. We return to this ﬁnding in the
discussion.
Self-feelings
Participants' self-feelings were analyzed with a 2 (Afﬁrmation
status) × 2 (Normative information) ANOVA. The analysis revealed a
signiﬁcant main effect of normative information, F(1, 109) = 4.07, p =
.046, η2P = .04, such that the Democratic participants felt signiﬁcantly
more positive about themselves when President Obama was soaring in
the polls (M = 6.80, SE = .20) than when he was falling in the polls
(M = 6.20, SE = .22). There was no main effect for afﬁrmation status
F(1, 109) = 0.18, p = .677. However, the two-way interaction term
approached signiﬁcance, F(1, 109) = 3.89, p = .051, η2P = .03. Simple effects tests revealed that the self-feelings of participants in the control condition were strongly inﬂuenced by the polling data, as the (Democratic)
participants reported signiﬁcantly more positive feelings about themselves when Obama was soaring in the polls (M = 7.03, SE = .30)
than when he was falling in the polls (M = 5.84, SE = .33), F(1, 109) =
7.23, p =.008, η2P = .06. By contrast, the self-feelings of participants in
the afﬁrmation condition exhibited no difference when Obama was soaring (M = 6.57) versus falling in the polls (M = 6.55), F(1, 109) = 0.001,
p = .973. Afﬁrmed participants' self-feelings appeared to be untethered
from their political party's success or failure, as it did not ﬂuctuate in
response to the polling information.
Summary and next steps
Study 1 provided the ﬁrst direct evidence that afﬁrmations of
self-integrity affect responsiveness to normative social inﬂuence.
The attitudes of control participants, all of whom were selfidentiﬁed Democrats, were swayed by the salient American polling
norms. In line with predictions from the social identity approach to
group norms, participants were more favorable toward President
Obama and his approach to health care, they allocated more hypothetical dollars toward Obama's party, and they may have felt better
about themselves when Obama was soaring in the polls than when
he was sinking in the polls. By contrast, afﬁrmation appeared to
free participants from the implications of the group norms, as their
evaluations of Obama were not swayed by the group norms, they actually gave more dollars to Republicans, and their self-feelings were
unmoved by whether Obama was soaring or sinking in the polls. The
national norms did not appear to affect participants' evaluations of
Republicans, a non-salient but relevant social group (cf. Binning
et al., 2010). Also consistent with expectations of the social identity
approach, this latter ﬁnding suggests that participants' evaluations
of Obama were responding to the national norms but not to other,
non-salient partisan norms.
One potential limitation of Study 1 is that all participants were Democrats and therefore shared both their Democratic and national identity
with the president. For instance, from participants' self-feelings, it is unclear if control participants were responding as Americans or as Democrats. Participants felt better when Obama was soaring than when he
was sinking in the polls, which could be expected both among selfidentiﬁed Americans and among self-identiﬁed Democrats. The main
effect on the allocation task suggested that Democratic identity may
have been active. We reasoned that although Obama is a Democrat,
only the national norms were manipulated while Democratic norms
were not. Thus, we expected exposure to national norms to tap
American, not Democratic identity. Following this logic, an equivalent
pattern of results should emerge for Republican participants, a prediction we examine in the next study.
Study 2 aims to replicate the core ﬁnding of Study 1 with a sample of
self-identiﬁed Republicans. If normative pressure sways participants'
evaluations of Obama as a function of salient American norms as
reﬂected in polls, then these salient American norms should exert their
inﬂuence among Republicans as they did with Democrats. Although
counter-intuitive on the surface, this prediction rests on the theoretical
assumption that participants' social identity concerns are driven by
whatever identity is most salient in the social context. Because we aim
to highlight American national norms (and not Republican or Democratic norms), we expected that Republicans would be swayed in the direction of national norms, even when such norms ran counter to their
interests as Republicans (i.e., to see Obama perform poorly). This study
therefore presents a strong test of the idea that it truly is salient national
norms and not partisan norms to which participants are responding
(cf. Bolsen, Druckman, & Cook, 2014; Leeper & Slothuus, 2014).
In addition, Study 2 compares and contrasts how afﬁrmation affected participants' responses to normative (polling) information versus evidentiary (factual) information. Half the participants received polling
information regarding how Americans view President Obama's handling of the economy, and the other half received evidentiary information regarding how the economy was actually doing under Obama's
economic stewardship (e.g., unemployment rate). The distinction
between these two types of information maps onto the distinction
between normative versus informational forms of social inﬂuence
(e.g., Chaiken, Giner-Sorolla, & Chen, 1996; Deustch & Gerard, 1955;
Petty & Cacioppo, 1986). Thus we manipulated afﬁrmation status (Afﬁrmation vs. Control) and orthogonally manipulated both the favorability
of the information (favorable vs. unfavorable to Obama), as in Study 1,
and also the type of information participants received (normative vs.
evidentiary), yielding a 3-factor design. We predicted a three-way interaction: whereas control participants should tend to be swayed by normative but not evidentiary information, afﬁrmation participants
should tend to be swayed by evidentiary but not normative information.
Our dependent measures consisted of participants' approval of President Obama, their approval of Republicans, and their feelings of selfworth in response to the manipulations.
Study 2
When this study was conducted in early 2010, Obama's approval
among the American people hovered around 50% (Gallup, 2014). Although the economic recession had technically ended in 2009, national
unemployment remained high (just under 10%) and was seen as among
the top problems facing the country (Jones, 2010). At the same time, the
U.S. stock market had recorded big gains over the previous year, with
the S&P 500 rising 24%. As such, the economic picture was uncertain,
and the mix of favorable and unfavorable data made it possible to spin
the public's perceptions of the economy under Obama (normative information) as well as the actual state of the economy under Obama (evidentiary information) in either negative or positive directions. Actual
reports at the time of this research indicated that favorable public opinion about Obama had been slipping (see Condon, 2010). Meanwhile, the
Federal Reserve Bank of Philadelphia's (2010) ﬁrst quarter Survey of
Professional Forecasters projected a low but upwardly revised projection of a 2.7% annual rate of growth for the US economy in 2010 after
contracting − 2.8% in 2009 (The World Bank, 2014). As such, both
Obama's popularity and the state of the economy could be plausibly described as weak/declining or strong/growing.
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
Method
Participants and design
Participants were 159 adults (62% female; 88% White, 6% Asian or
Asian American, 4% Latino/a, 2% from other categories; MAge =
35.69 years) recruited from the same listserv used in Study 1, but this
time emails were sent only to self-identiﬁed Republicans. Participants
received a $7 gift card to an online retailer. Participants were randomly
assigned to one of the eight cells of the 2 (Afﬁrmation status: Afﬁrmation vs. Control) × 2 (Direction of data: Favorable toward Obama vs. Unfavorable toward Obama) × 2 (Type of data: Normative vs. Evidentiary)
between-subjects factorial design. Four months after the completion of
the study, participants were invited to take part in another study in exchange for a $5 gift card, and 124 (78%) completed the follow-up study.
Procedure
After providing consent, participants received the same afﬁrmation
manipulation used in Study 1 and then read a news article written
and formatted to appear to be from a prominent global news agency
(a different agency than in Study 1). Half the participants completed a
similar procedure as participants in Study 1. Participants assigned to
this study's normative conditions read one of two articles that featured
a picture of a large crowd. Those in the favorable normative data condition read an article titled, “Poll: More Americans Have Faith in Obama's
Policies,” which reported four pieces of data describing Obama's rise in
the national polls with respect to his handling of the economy
(e.g., approval of Obama's handling of the economy “had steadily improved more than 6%”). Participants in the unfavorable normative
data condition read a parallel article titled, “Poll: More Americans
Doubt Obama's Policies,” which cited four pieces of data on the
president's declining approval with parallel wording (e.g., approval of
Obama's handling of the economy “had steadily declined more than
6%”).
The remaining participants were assigned to one of two evidentiary
conditions that did not contain normative information and instead manipulated factual economic indicators. The evidentiary articles were
matched for length and level of detail with the normative articles and
featured a close-up photo of a stock ticker board with reporting on
four pieces of economic data gathered during Obama's tenure. Participants who received favorable evidentiary data read an article titled, “Experts: Economic Data Lend Support to Obama's Policies,” which
reported on four signals of improving economic strength under Obama's
leadership (e.g., gross domestic spending had “steadily increased by 6%”
during the current year). Participants assigned to receive unfavorable
evidentiary data read an article titled, “Experts: Economic Data Cast
Doubt on Obama's Policies,” which reported four signals of declining
economic strength under Obama's leadership (e.g., domestic spending
had “steadily decreased by 6%”).
Notably, both the normative and evidentiary information articles
were fabricated for the purposes of this research and constructed to
use identical numbers and comparable levels of detail and complexity.
A pilot test3 on a bipartisan sample of 60 participants using the news
materials suggested there was a marginal tendency for participants to
regard information unfavorable to Obama as more “reliable” than favorable information (F[1, 57] = 2.89, p = .096), perhaps reﬂecting the
public's declining opinion of Obama noted above (Condon, 2010). However, there was no difference in the perceived reliability of the normative versus evidentiary information (F[1,57] = 0.08, p = .776), nor
was there a two-way interaction on perceived reliability between
3
Participants in the pilot test were randomly assigned to cells of a 2 (Direction of data:
Favorable toward Obama vs. Unfavorable toward Obama) × 2 (Type of data: Normative vs.
Evidentiary) between-subjects factorial design and asked to indicate “how reliable was
the data presented in the article” with endpoints labeled 1 (Not at all reliable) and 7 (Very
reliable) (Ms = 3.72 and 4.35; SEs = .26 and .27, for favorable and unfavorable data,
respectively).
79
direction and type of data (F[1,57] = 0.05. p = .826). As such, we
went forward with the materials but aimed to follow up with participants four months after the initial manipulation to examine whether
the effects of the manipulations held over time.
Approval of Obama
Participants responded to four items that paralleled the four items
used in Study 1, but which substituted “economic issues” where Study
1 had used “health care issues.” The composite ranged from 1 (unfavorable toward Obama) to 9 (favorable toward Obama; α = .81; M = 3.74,
SD = 1.80).
Evaluations of Republicans
Although participants did not receive any Republican-speciﬁc articles or stimuli, we assessed their perceptions of Republicans with two
items: “How balanced and objective is the Republicans' approach to
economic issues?” (1 = Not balanced or objective at all; 9 = Very
balanced and objective) “In general, how knowledgeable about economic issues is the Republican leadership?” (1 = Not knowledgeable at all;
9 = Very knowledgeable; α = .73; M = 5.68, SD = 1.50).
Feelings of self-worth
Participants completed the feelings of self-worth scale (Brown &
Dutton, 1995), which we used to expand our measure of self-feelings
beyond the single-item measure used in Study 1. Participants indicated
how well each of eight items described how they were feeling “right
now” on a scale from 1 (Not at all) to 7 (Very much). Reﬂecting the
scale's conceptual structure, four items assessed feelings that directly
implicate the self: humiliated*, proud, ashamed*, pleased with myself
(*reverse-scored; α = .62; M = 5.35, SD = 0.97). Four more items
assessed more general mood and were not directly related to the self:
glad, unhappy*, sad*, happy (α = .83; M = 5.20, SD = 1.31).
Results and discussion
Study 2 was designed to examine two potential effects of selfafﬁrmation. First, it tested the Study 1 ﬁndings with Republicans, examining whether self-afﬁrmation can untether people from salient normative group pressure, as previously observed with Democrats; and
second, it examined whether self-afﬁrmation would simultaneously
make people more attuned to evidentiary data. Evaluations of
President Obama were subjected to a 2 (Afﬁrmation status: Afﬁrmation
vs. Control) × 2 (Direction of data: Favorable toward Obama vs. Unfavorable toward Obama) × 2 (Type of data: Normative vs. Evidentiary)
between-subjects ANOVA. The analysis yielded a main effect of direction of data, F(1, 151) = 4.01, p = .047, η2P = .03, indicating that favorability toward Obama was higher when the data were favorable (M =
3.88, SE = .19) than when they were unfavorable (M = 3.35, SE =
.18). However, this effect was moderated by the predicted three-way interaction, F(1, 151) = 3.99, p = .047, η2P = .03 (see Fig. 2, top panel).
We decomposed the three-way interaction by examining the effects
within the normative condition (two way effect: F[1, 151] = 1.93,
p = .167) and evidentiary condition (two way effect: F[1, 151] =
2.02, p = .154). The pattern of responses in the normative conditions
replicated Study 1: participants in the control condition were signiﬁcantly more favorable toward President Obama when he was up in
the polls (M = 4.19, SE = .40) than when he was down (M = 3.08,
SE = .37), F(1, 151) = 4.24, p = .041, η2P = .03. By contrast, as in
Study 1 afﬁrmed participants were untethered from the group norms,
as there was no difference in their approval when Obama was up
(M = 3.67, SE = .37) versus down (M = 3.60, SE = .37) in the polls,
F(1, 151) = 0.02, p = .890.
The pattern further suggested that control participants were unaffected by the (favorable vs. unfavorable) economic evidence. Participants in the control condition were not responsive to the data, as their
favorability toward Obama was low overall and did not differ when
80
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
Immediately A#er the Manipula"on
Norma!ve (Polling) Data
8
7
6
Improving data
5
Declining data
4
3
2
FAVORABILITY TOWARD OBAMA
FAVORABILITY TOWARD OBAMA
9
1
9
Eviden!ary (Economic) Data
8
7
6
5
4
3
2
1
Control
Aﬃrma"on
Control
Aﬃrma"on
Four Months A#er the Manipula"on
Norma!ve (Polling) Data
8
7
6
Improving data
5
Declining data
4
3
2
1
FAVORABILITY TOWARD OBAMA
FAVORABILITY TOWARD OBAMA
9
9
Eviden!ary (Economic) Data
8
7
6
5
4
3
2
1
Control
Aﬃrma"on
Control
Aﬃrma"on
Fig. 2. Republicans' mean favorability toward Obama (±1 standard error) as functions of afﬁrmation status, direction of data, and type of data immediately after the manipulation (top)
and at four-month follow-up (bottom) for Study 2.
the economy was improving (M = 3.60, SE = .39) versus declining
(M = 3.67, SE = .37), F(1, 151) = 0.02, p = .895. By contrast, among
afﬁrmed participants, favorability toward Obama was marginally higher
when the economy was improving (M = 4.07, SE = .40) versus declining (M = 3.05, SE = .38), F(1, 151) =3.53, p = .062, η2P = .02. The overall pattern therefore suggested that afﬁrmation attenuated conformity
to polls but increased openness to economic data.
Evaluations of Republicans
Evaluations of Republicans were tested with a 2 (Afﬁrmation status) ×
2 (Direction of data) × 2 (type of data) ANOVA. The analysis did not yield
any signiﬁcant main effects, Fs b 1.34, ps N .247, or interactions, Fs b 1.0,
ps N .398.
Feelings of self-worth (FOSW)
Separate analyses were conducted on the two facets of the FOSW
scale (self-feelings and general feelings; cf. Brown & Dutton, 1995).
First, a 2 (Afﬁrmation status) × 2 (Direction of data) × 2 (type of data)
ANOVA was conducted on self-feelings. The analysis did not yield any
signiﬁcant main effects, Fs b 1.0, ps N .497. There was, however, a significant Afﬁrmation status × Direction of data (two-way) interaction, F(1,
151) = 4.41, p = .037, η2P = .03. Although the simple effects comparisons were not signiﬁcant, they were in the direction (symmetrical to
Study 1) that control (Republican) participants felt worse about themselves when Obama faced improving data (M = 5.24, SE = .16), than
when he faced declining data (M = 5.53, SE = .15), F(1, 151) = 1.71,
p = .193. Afﬁrmed participants, by contrast, felt marginally better about
themselves when Obama faced improving data (M = 5.46, SE = .16)
than when he faced declining data (M = 5.10, SE = .15), F(1, 151) =
2.77, p = .098, η2P = .02. There were no other two-way interactions,
Fs b 1.77, ps N .185, nor a three-way interaction, F(1, 151) = 0.06, p =
.809. These results are consistent with the idea that the afﬁrmation
changed the way participants felt in response to identity-relevant political information. They are also suggestive of a complex relationship between social identities and responsiveness to political information: it
seems that participants may go along with salient national norms (as
seen in the normative information conditions), even when those
norms threaten the self. Although such inferences are tentative given
the partial and marginal nature of these ﬁndings, the overall pattern
suggests, consistent with Study 1, that the afﬁrmation can reduce and
even reverse the impact of positive and negative identity-relevant information on immediate feelings about the self (see Sherman & Kim,
2005).
A 2 (Afﬁrmation status) × 2 (Direction of data) × 2 (Type of data)
ANOVA on the general mood dimension of the FOSW scale did not reveal any main effects Fs b 2.39, ps N .124, or interactions Fs b 1.75,
ps N .187. Such a ﬁnding is consistent with other research suggesting
that self-afﬁrmations affect the way people feel about themselves but
tend not to change general positive mood (Cohen et al., 2000; Fein &
Spencer, 1997; Schmeichel & Martens, 2005; Sherman & Kim, 2005;
Sherman et al., 2000; for an exception see Koole, Smeets, Van
Knippenberg, & Dijksterhuis, 1999).
Four-month follow-up
Study 2 provided evidence suggesting that self-afﬁrmation can reverse the tendency for people to follow their group at the expense of following evidence. Based on theorizing that afﬁrmation might have
81
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
4
A chi-squared test A chi-squared test revealed that the sample at Time 2 did not differ
from the sample at Time 1 in their assignment to conditions χ2(7) = 1.19, p = .991; nor
were their differences in gender, χ2(1) = 1.06, p = .303, or age, t(275) = 1.08, p = .282.
There was, however, a marginal decline in approval for Obama at Time 2 (M = 3.31,
SE =.16) compared to Time 1 ( M = 3.60, SE = .13), t(281) = 1.84, p = .067. This may
be partly attributable to a decline in Obama approval also seen in the general population
over this time period (Gallup, 2014).
9
FAVORABILITY TOWARD OBAMA
changed participants' motivational goals when processing political information, we sought to examine if this effect held up over time.
The 124 participants who completed the follow-up were given the
identical measure of Obama support at the very beginning of the
follow-up study.4 They were given this measure without reminding
them of their previous condition assignment (or the nature of the
prior study). The critical three-way interaction replicated, F(1,
116) = 6.84, p = .010, η2P = .06, indicating that the original manipulations continued to affect participants' attitudes over time (see Fig. 2 bottom panel).
Next, participants in the four-month follow-up received a new manipulation. Speciﬁcally, we sought to address what would happen
when normative and evidentiary information directly conﬂicted. To
take a recent example, the New York Times recently published an article
with the headline, “Obama's Puzzle: Economy Rarely Better, Approval
Rarely Worse” (Calmes, 2014). That is, sometimes public norms suggest
that the president is unpopular, but the factual data indicate that his
economic policies are effective. How would participants respond to
such conﬂicting information? We predicted that control participants
would be swayed by the normative information contained in the article
(and not by the evidentiary information), whereas afﬁrmation participants would be swayed by the evidentiary information (and not by
the normative information).
To get at this idea, we randomly assigned all participants to read a
new article about President Obama. They either received an article
headlined, “Americans Have Doubts About Obama, But Economy is
Growing” or an article with the opposite headline, “Americans Continue
to Support Obama, But Economy is Weakening.” Each article contained
one paragraph describing how Obama was faring in the polls (either
soaring or sinking) and one conﬂicting paragraph describing how the
economy under Obama was actually doing (either declining or rising).
To control for order effects, the order of presentation of normative and
evidentiary information was counterbalanced for each subject. We
then assessed evaluations of Obama's economic policies with three
new items that used nine-point scales. These questioned the likelihood
that “Obama's economic policies will help the economy,” whether
Obama had “the econom
y on the right track,” and if Obama has “the right leadership qualities to
lead the country out of the recession.” The three items formed a reliable
composite (α = .91; M = 3.12, SD = 1.85).
The hypotheses were supported. Keeping in mind that the new article represented a fourth factor in our experimental design, we used a
statistical model that included all prior manipulated factors, the new
manipulation, and all two- and three-way interactions terms among
these variables. Given the lack of power in the design, we omitted the
four-way interaction term from this model, although the effects reported below hold when it was included. There was a signiﬁcant two-way
interaction (Afﬁrmation status × New article content), F(1, 109) =
5.59, p = .020, η2P = .05 (see Fig. 3), and this effect was not moderated
by the two other factors, Fs b 1.0, ps N .679. Consistent with the ﬁndings
thus far, simple effects tests suggested that evaluations by control participants were swayed by the normative information, as they trended
toward being more favorable toward Obama when the poll information
was favorable (M = 3.41, SE = .45) than when it was unfavorable
(M = 2.48, SE = .31), F(1, 109) = 2.64, p = .107, η2P = .02. By contrast,
evaluations by afﬁrmation participants were swayed by evidentiary information, as they were marginally more favorable toward Obama
when the economy was doing well (M = 3.50, SE = .32) than when
the economy was doing poorly (M = 2.68, SE = .32), F(1, 109) =
8
7
6
Rising polls,
sinking economy
5
Rising economy,
sinking polls
4
3
2
1
Control
Aﬃrma"on
Fig. 3. Republicans' mean favorable evaluations of President Obama's economic policies
(±1 standard error) in response to a news article that contained conﬂicting normative
and evidentiary information at Study 2 four-month follow-up (net of the effects of the
prior manipulated factors).
3.31, p = .072, η2P = .03. In summary, the afﬁrmation had lasting effects
on the type of information that participants were responsive to, decreasing their responsiveness to normative information and increasing
their responsiveness to evidentiary information.5 We return to this experimental design, in which normative and evidentiary information simultaneously conﬂict, in Study 3.
Summary and next steps
Study 2 provides more evidence that adhering to normative and evidentiary information is moderated by concerns about self-integrity.
Participants were exposed to either normative information, evidentiary
information, or (in the Study 2 follow-up) both types of information simultaneously. Participants' American identity was potentially relevant
to both types of information: whereas normative information conveyed
how group members were evaluating Obama's economic policies, evidentiary information conveyed how Obama's economic policies were
actually performing. But consistent with our expectations, participants
in the control condition were only responsive to the normative information. Only the normative information – which was conveyed via (ﬁctitious) polling data on how most Americans felt about Obama and his
handling of the economy – provided information about how to behave
and respond like a typical group member. Evidentiary information, we
argue, was useful not for gleaning how to behave like a typical group
member, but for pursuing individual level goals of accuracy and independence in judgment. If the experimental manipulation successfully
activated collective identity goals of conformity but not individual
level-goals of accuracy, participants in the control condition should be
expected to conform to collective norms but to ignore or resist the implications of evidentiary data. This was the case immediately after exposure to the stimulus and four months later when participants were recontacted and presented with new stimuli.
5
The original three factors (but not the new article) also interacted to affect participants' evaluations following the new article (3-way interaction), F(1, 109) = 6.43,
p = .013, η2P = .06. This pattern of results suggested that the manipulations from fourmonths earlier continued to affect participants' judgments after exposure to the new article in the predicted fashion. That is, independent of the new manipulation, no-afﬁrmation
participants were signiﬁcantly more favorable toward Obama if they had been previously
exposed to favorable normative data (M = 4.24, SE = .77) rather than unfavorable normative data (M = 2.37, SE = .42), F(1, 109) = 4.51, p = .036, η2P = .04. Afﬁrmed participants resisted the sway of normative inﬂuence (Ms = 3.26 and 3.37, SEs = .39 and .45, for
favorable and unfavorable normative data, respectively), F(1, 109) = 0.04, p = .853. The
pattern reversed for evidentiary information: no-afﬁrmation participants were not inﬂuenced by the evidentiary data about Obama they had seen four months earlier
(Ms = 2.42 and 2.75, SEs = .52 and .43, for favorable and unfavorable evidentiary data, respectively), F(1, 109) = 0.24, p = .624. By contrast, afﬁrmation participants were more
favorable toward Obama when they had been exposed to favorable evidentiary data
(M = 3.52, SE = .52) than when they had been exposed to unfavorable evidentiary data
(M = 2.19, SE = .43), F(1, 109) = 3.94, p b .050, η2P = .04.
82
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
The role of self-integrity in driving the above results can be inferred
from participants' responses after self-afﬁrmation. We predicted that afﬁrmations of self-integrity in a domain unrelated to the experimental
context would effectively lead self-evaluation to be less tightly linked
to focal social identities. When relieved from pressures to behave like
a typical group member, participants in the afﬁrmation condition may
be more prone to pursue individual goals of accuracy and independence
in judgment. The results bore this reasoning out. Afﬁrmation appeared
to produce an enduring shift in people's situational goals away from
“going along” and toward “getting it right.” This effect persisted over
time for the domain of evaluating Obama's performance, and it continued to affect participants' evaluations of new information about Obama
and the economy.
Our explanation for these effects rests on several theoretical arguments. In particular, we argue that in the control condition, the experimental context induced participants to self-categorize at the collective
level and to thereby pursue collective identity goals of conformity and
not individual identity goals of accuracy. The above results support
this view, but they do not directly examine the role of social identiﬁcation. For example, following this logic, participants with a low level of
identiﬁcation as American should be less likely to conform than participants with a high level of American identiﬁcation. Such individuals
might not self-categorize as Americans and, as such, would tend not to
conform to the American norms as expressed in the polling data. Lowidentiﬁed individuals may instead be concerned with individual
goals of accuracy (Terry & Hogg, 1996). Research suggests that the
individual-level self tends to be the most primary or default dimension
of self-deﬁnition (Gaertner, Sedikides, Vevea, & Iuzzini, 2002), and thus
when identiﬁcation with a salient identity is low, participants should remain categorized at the individual level. In Study 3, we seek to directly
test our predictions about the role of social identiﬁcation in shaping the
type of information that matters to self-integrity and the tendency for
people to follow norms, even in the face of conﬂicting evidence.
First, we hypothesized that among highly-identiﬁed Americans, selfintegrity concerns should be especially likely to come online in the face
of salient American group norms. As such, in these situations it is among
the highly identiﬁed for which afﬁrmation can untether the self-concept
from social identity and thereby open people up to evidentiary
information.
Second, we predict that in these situations, control participants who
do not identify with America should tend to follow the evidentiary but
not the normative polling information. Since people with low American
identiﬁcation should not be compelled to follow the salient American
group norms, they should be more open to evidentiary data and willing
to neglect the salient normative data. As such, there are two general
conditions under which people may be inclined to follow evidentiary
data and neglect group normative data. The ﬁrst is when people who
identify with the group self-afﬁrm prior to judgment, thereby alleviating the need to protect the integrity of salient social identity. The second
is when low-identiﬁed individuals are exposed to the same information.
Their low level of identiﬁcation means that they are not invested in the
salient group norms in the ﬁrst place. As such, we theorize that afﬁrmation should have contrasting effects among the high and low identiﬁed.
Speciﬁcally, afﬁrmation should attenuate conformity and enhance accuracy in judgment among people who identify strongly with Americans,
but it should have the opposite effect and attenuate accuracy while enhancing conformity among people with low American identiﬁcation.
Thus, afﬁrmed and low-identiﬁed individuals may process information
in similar ways, showing a stronger proclivity toward “getting it right”
than with “going along.” Study 3 tests these possibilities by measuring
participants' American identiﬁcation and testing whether it moderates
the effects of afﬁrmation.
Study 3 also sought to expand the generalizability of the ﬁndings by
recruiting participants from a new subject pool including both Republicans and Democrats. In examining only Democrats (Study 1) or only Republicans (Study 2), both previous studies held partisan identity
constant, which helped to reduce variability in partisan identity. However, doing so also made it difﬁcult to investigate the role of political orientation. Thus, Study 3 sampled participants across the political
continuum to model its effects. We expected that American identiﬁcation would moderate the effects of afﬁrmation, and that this should
occur both prior to and after controlling for the inﬂuence of political
orientation.
Study 3
The ﬁnal study was conducted in the summer of 2012, during the Republican presidential primary season and at the end of President
Obama's ﬁrst term. We modeled the stimulus materials on the Study 2
follow-up, and we sampled participants across the political spectrum
— both Republicans and Democrats. The main purpose of this study
was to test whether the tendency for afﬁrmation to lead to greater receptivity of evidentiary over normative information (the pattern observed in the Study 2 follow-up) would be further moderated by
American identiﬁcation. Because this study sampled participants across
the political spectrum, we also modeled the effects of political identiﬁcation and compared it with the effects of American identiﬁcation
using multiple regression. We predicted that because American norms
but not partisan norms were made salient by the polling stimuli, only
American identiﬁcation should moderate the proposed effects.
Method
Participants, procedure, and design
Two hundred forty-nine Americans were recruited from Amazon
Mechanical Turk and each was paid $0.55 to participate. We restricted
the sample to participants with IP addresses located within the United
States. Twenty-four participants had a signiﬁcant amount of missing
data (including the key dependent measures), and one participant experienced a technical problem. The analyses reported below are based
on the remaining 224 participants (54% women; 73% White, 10%
Asian or Asian-American, 8.4% African-American, 6.2% Latino/a). The
sample had a median age of 30 years (M = 34.76) and a median
education level of college graduate (1 = Some high school; 6 = Graduate
degree; M = 3.58, SD = 1.14). After indicating their American
identiﬁcation and political orientation, participants were randomly
assigned to one of four cells in a 2 (Afﬁrmation status: Afﬁrmation vs.
Control) × 2 (Article type: Favorable normative data and unfavorable evidentiary data vs. Unfavorable normative data and favorable evidentiary
data) factorial design.
American identity and political orientation
Participants ﬁrst completed several scales related to identity. They
indicated their level of identiﬁcation with America with three questions
modeled from previous measures of group identiﬁcation (e.g. Huo,
Binning, & Molina, 2010): “Being an American is an important part of
who I am”, “I am proud to be an American”, and “When someone praises
the accomplishments of the United States, I feel it is a personal compliment to me” (1 = Strongly disagree; 7 = Strongly agree). These three
items were combined into a mean composite that displayed excellent
reliability (α = .93, M = 4.94, SD = 1.59). Participants also reported
their political orientation: “Which political party do you support the
most?” (1 = Green; 2 = Democratic; 3 = Independent; 4 = Republican;
5 = Libertarian; 6 = None [recoded as Independent]), “Would you say
your political beliefs are closer to the Democrats or the Republicans?”
(1 = Much closer to the Democrats; 4 = In the middle; 7 = Much closer
to the Republicans), and “When it comes to politics, do you consider
yourself to be liberal, moderate, or conservative?” (1 = Very liberal;
2 = Liberal; 3 = Somewhat liberal; 4 = Moderate; 5 = Somewhat conservative; 6 = Conservative; 7 = Very conservative). The three items were
all coded on a 7-point scale and then averaged into a mean composite
with higher scores indicating a more right-leaning orientation. The
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
measure displayed good reliability (α = .88, M = 3.34, SD = 1.52). The
composite of political orientation revealed the sample leaned to the left,
as 62% of participants scored below the midpoint of the scale, 10% were
on the midpoint, and 28% scored above the mid-point of the scale. The
liberal skew of the data is a common feature of samples of from Amazon
M-Turk (Berinsky, Huber, & Lenz, 2012). Although liberally skewed data
can be problematic when trying to make general claims across the political spectrum, we believed that there was enough variability in political
orientation to allow us to test our key hypotheses.
News article
The two news articles were very similar to the articles introduced in
the Study 2 follow-up. The article appeared to be from a reputable national newspaper. Half the participants were randomly assigned to an
article reporting that 55% of Americans approved of President Obama's
handling of the economy despite evidence that the economy was ﬂagging (“Americans Continue to Support Obama, But Economy is Weakening”). The other half of participants received the opposite information,
which indicated that 55% of Americans disapproved of Obama's handling of the economy despite evidence that the economy was improving
(“Americans Have Doubts About Obama, But Economy is Growing”).
Approval of Obama
Immediately after reading the article, participants evaluated Obama
with the same three items reported at the end of the Study 2 follow-up
(participants indicated if they thought “the economy on the right track,”
the likelihood that “Obama's economic policies will help the economy,”
and if Obama has “the right leadership qualities to lead the country out
of the recession”). As above, participants answered the questions using
a 9-point scale with end points labeled 1 (Strongly disagree) and
9 (Strongly agree). The composite displayed excellent reliability (α =
.95, M = 4.99, SD = 2.07).6
Results and discussion
Analytic approach
To test the study hypotheses, we used a multiple regression approach with the methods recommended by Aiken and West (1991). Because we did not have full experimental control over all the
independent variables, we controlled for participant age, gender, ethnicity (White = 0; non-White = 1), and educational achievement
(1 = Some high school; 6 = Post-graduate degree), to help isolate the
unique contribution of American identiﬁcation on approval. The correlations among the measured variables used in the regression models
are displayed in Table 1.
We tested our hypotheses using two different multi-step regression
models, the ﬁrst controlling for political orientation and the second not
controlling for political orientation. In the ﬁrst model, we tested all
control variables on Step 1, the three focal main effects on Step 2
(mean-centered American identiﬁcation, afﬁrmation status, and article
type), the three two-way interaction terms on Step 3 (American ID ×
Afﬁrmation status; American ID × Article type; Afﬁrmation status × Article type) and the critical three-way interaction term on Step 4 (American
ID × Afﬁrmation status × Article type). Model 2 included all of these
terms and also the main and interactive effects of political orientation.
That is, the second model also included political orientation on Step 2,
three additional two-way terms on Step 3 (American ID × Political
orientation; Afﬁrmation status × Political orientation; Article type ×
Political orientation), three additional three-way terms on Step 4
6
Participants also answered two questions asked in Study 2, “How balanced and objective is Obama's health care outline?” (1 = Not balanced or objective at all; 9 = Very balanced and objective), and “In general, how knowledgeable about health care issues is
Barack Obama?” (1 = Not knowledgeable at all; 9 = Very knowledgeable).” Including these
two items in the composite did not affect the reliability or signiﬁcance of the focal threeway effect reported below, however, they were excluded from the analyses to keep the
primary measure for this study the same as was used in the Study 2 follow-up.
83
(America ID × Afﬁrmation status × Political orientation; America
ID × Article type × Political orientation; Afﬁrmation status × Article
type × Political orientation), and the four-way interaction term on
Step 5 (American ID × Afﬁrmation status × Article type × Political
orientation). This four-way term did not explain a signiﬁcant
amount of variance and is not discussed further.
Approval of Obama's economic policies
Did American identiﬁcation moderate the impact of the experimental variables on approval of Obama's economic policies? In both models
(with and without political orientation), the central hypotheses
pertained to a three-way effect in Step 4 involving American ID, afﬁrmation status, and article type. In short, we expected the patterns seen in
the prior studies to hold only among those with high American ID. In
both models the three-way effect was signiﬁcant on Step 4 (Model 1:
B = –.76, SE = .37, β = –.25, p = .043; Model 2 :B = –.84, SE = .31,
β = –.27, p = .007) and revealed patterns in line with our
hypotheses. Table 2 displays all regression coefﬁcients from Step 4
from each analysis. Fig. 4 displays the estimated means for each condition at high (+1 SD) and low (–1 SD) American identiﬁcation. Controlling for the inﬂuence of political orientation (Model 2) appeared to help
to isolate the unique contribution of American identiﬁcation on the dependent measure.7
Focusing on Model 2, we decomposed the three-way interaction by
ﬁrst examining the simple two-way interactions between the manipulated variables (afﬁrmation status × article type) at high (+ 1
SD) and low (− 1 SD) American identiﬁcation. Among highly
identiﬁed Americans, the two-way interaction approached signiﬁcance,
B = –1.16, SE = .68, β = –.23, p = .088. As hypothesized, simple effects
tests found that afﬁrmed participants' evaluations of Obama were lower
when the economy was sinking (Estimated M = 5.22, SE = .33) than
when the economy was rising (Estimated M = 6.05, SE = .32), B =
− .82, SE = .46, β = –.18, p = .076. This did not hold among control
participants, as highly identiﬁed control participants were not responsive to the different articles (Estimated Ms = 5.23 and 5.57, SEs = .34
and .36, for sinking and rising economy, respectively), B = .34, SE =
.50, β = .08, p = .498.
Among low identiﬁed Americans, the two-way (afﬁrmation
status × article type) interaction was signiﬁcant, B = 1.52, SE = .66,
β = .30, p = .023, and the pattern of means conformed to predictions.
Control participants based their evaluations on the evidentiary data, as
their evaluations of Obama were signiﬁcantly higher when the economy
(evidentiary information) was rising (Estimated M = 5.46, SE = .39)
than when it was sinking (Estimated M = 4.50, SE = .29), B = −.96,
SE = .47, β = − .21, p = .045. This pattern was eliminated among
7
Speaking to the interplay between American identiﬁcation and political orientation,
Step 3 of the analysis also revealed a two-way interaction between political orientation
and American identiﬁcation (B = −13, SE = .06, β = −.13, p = .022). Simple effects tests
showed that among more politically conservative participants (+1 SD on political orientation), there was no relationship between American identiﬁcation and approval of
Obama (B = −.02, SE = .17, β = −.01, p = .913), but among more liberal participants
(–1 SD on political orientation), higher American identiﬁcation predicted higher approval
of Obama (B = .40, SE = .16, β = .28, p = .014). This ﬁnding hints that the three-way effect in Model 1 may have been muddied by the uncontrolled inﬂuence of political orientation. When political orientation was examined on its own as a moderator of the
experimental manipulations, it was not a signiﬁcant moderator. However, as seen in Table 2, when political orientation was included in the four-way model that also included
(and thereby controlled for) American identiﬁcation, its effect on the manipulations
strengthened to reveal a marginal three-way interaction (political orientation × afﬁrmation status × article type), B = .57, SE = .32, β = .20, p = .078. Although no simple slopes
approached signiﬁcance, the overall pattern suggested that left-leaning participants – participants who shared a political identity with the president – responded like highly identiﬁed Americans, while right-leaning participants responded like low identiﬁed
Americans. That is, when American identity was controlled, political orientation behaved
similarly to it. It is notable that this effect only emerged in the model in which American
identiﬁcation was included, as it suggests that controlling the effects of a salient
(American) identity might reveal the workings of less salient identities in the social
context.
84
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
Table 1
Bivariate correlations among measured variables included in Study 3 regression models.
Non-White
Non-White
Age
Female
Education
American ID
Political orientation
Obama eval.
Age
1.00
−.26⁎⁎⁎
−.09
−.10
−.03
−.09
.21⁎⁎
1.00
.28⁎⁎⁎
.20⁎⁎
.25⁎⁎⁎
.02
−.03
Female
Education
1.00
.00
.03
−.17⁎
.18⁎⁎
1.00
.01
.18⁎⁎
−.15⁎
American ID
1.00
.30***
−.04
Political orientation
Obama eval.
1.00
−.62⁎⁎⁎
1.00
⁎ p b .05.
⁎⁎ p b .01.
⁎⁎⁎ p b .001.
afﬁrmed participants with low American identiﬁcation (Estimated Ms =
4.79 and 5.35, SEs = .37 and .35, for rising and sinking economy, respectively), B = .56, SE = .50, β = .12, p = .263, suggesting that among lowidentiﬁed participants, afﬁrmation actually reduced the processing of evidentiary information.
Consistent with hypotheses, the afﬁrmed participants with high
American identiﬁcation and the control participants with low
American identiﬁcation processed the article information in similar
ways—as they were both more swayed by evidence than normative information. We argue that this is because for both groups, people's selfconcept was less linked to salient American group norms, and thus
both groups of participants were free to evaluate Obama based on the
merits of the country's economic performance rather than on salient
ingroup norms.
General discussion
Three studies support a self-integrity explanation for why people
conform to political group norms, even when evidence suggests that
Table 2
Unstandardized coefﬁcients (B), standard errors (SE), standardized coefﬁcients (β) and
signiﬁcance levels for regression models in Study 3. Model 1 does not control for political
orientation; Model 2 does.
Model 1
B
(Constant)
White vs. Non-White
Age
Male vs. Female
Education
American identiﬁcation
Afﬁrmation status
Article type
Political orientation
American ID × Afﬁrmation
status
American ID × Article type
Afﬁrmation status × Article
Type
Article type × Political
orientation
Afﬁrmation status × Political
orientation
American ID × Political
orientation
American ID × Afﬁrmation
status × Article type
Article type × Afﬁrmation
status × Political orientation
American ID × Article type ×
Political orientation
American ID × Afﬁrmation
status × Political orientation
⁎ p b .05.
⁎⁎ p b .01.
⁎⁎⁎ p b .001.
Model 2
SE
β
5.12
1.18
.00
.88
−.22
−.31
.43
−.31
–
.31
.29
.34
.01
.30
.13
.20
.41
.42
–
.27
.23
.01
.19
−.11
−.22
.10
−.07
–
.15
.47
−.10
.26
.58
.23
−.02
−.76
.37
−.25
p
⁎⁎⁎
⁎⁎
⁎
B
SE
5.35
.74
−.01
.35
−.01
−.07
.08
−.31
−.56
.47
.24
.28
.01
.25
.11
.17
.33
.34
.17
.23
.15
−.06
.08
.00
−.05
.02
−.07
−.38
.23
.41
.18
.22
.46
.20
.03
−.24
.23
−.12
−.68
.24
−.33
⁎⁎
−.24
.10
−.25
⁎
−.84
.31
−.27
⁎⁎
.57
.32
.20
.05
.11
.04
.21
.11
.15
β
p
⁎⁎
⁎⁎
⁎
the group norms are wrong. This research proceeded from the assumption that people respond differently to normative and evidentiary information because each type of information serves different psychological
functions. Whereas normative information is often used for belonging
and acceptance in social groups, evidentiary information is often used
for accuracy and independence in judgment. In service of the desire to
maintain a global sense of self-integrity, people should respond to
which ever type of information aligns with the most salient aspect of
their identity. When collective identity is salient, people conform to
group norms and tend to ignore evidentiary information because only
the normative information is relevant to maintaining self-integrity.
But when a valued collective identity is not salient, concerns about
individual-level identity may be more likely to guide judgment. In
such circumstances, self-integrity may be maintained not by following
group norms, but by attending to evidentiary information that could
lead to an accurate, evidence-based judgment. Support for this reasoning was obtained in two primary ways.
First, following the social identity approach, when a relevant collective identity was made salient, participants in the control condition responded to normative information describing the ingroup's
beliefs about President Obama (Studies 1 and 2) but not to evidentiary information describing how the economy was performing under
his policies (Study 2). By contrast, participants in the afﬁrmation
condition were consistently untethered from the norms of the
ingroup, as neither their self-feelings nor their approval of Obama
were swayed in the direction of the polls. Afﬁrming global selfintegrity prior to judgment produced a qualitative shift in the types
of information people found self-relevant: afﬁrmed participants
judged the president based on evidentiary information (Study 2)
and were inattentive to ingroup-normative information (Studies 1
and 2). This effect persisted over four months and continued to affect
attention to novel political information (Study 2 follow-up), suggesting the effects of the afﬁrmation persisted and propagated to
produce a general change in the identity relevance of normative
information.
Second, participants' level of identiﬁcation with America moderated
the self-relevance of normative and evidentiary information. Following
the idea that people attend to different types of information to maintain
their self-integrity, participants who do not strongly identify with the
group should be relatively inattentive to its norms. Instead, low identiﬁers may be attentive to information serving individual-level goals,
such as goals of accuracy and independence in judgment. In Study 3, individual differences in social identiﬁcation with America did indeed
moderate participants' responses to normative and evidentiary information in the predicted fashion. Control participants with low
American identiﬁcation were inattentive to national polls about President Obama but were responsive to evidentiary information regarding
Obama's economic performance. Together these results suggest that social identity processes help determine which types of information are
relevant to the self-concept, ﬁndings consistent with prior research
showing that social identiﬁcation moderates responsiveness to salient
ingroup norms (Terry & Hogg, 1996).
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
85
Fig. 4. Estimated means and standard errors for evaluations of President Obama's economic policies as functions of afﬁrmation status and article type, plotted at low (−1 SD) and high (+1
SD) American identiﬁcation from Study 3 without controls for political orientation (top), and with controls for political orientation (bottom).
These ﬁndings provide the ﬁrst direct evidence that afﬁrmations of
self-integrity are a means to attenuate conformity and encourage independent, evidence-based decision-making. However, the ﬁndings also
suggest afﬁrmations of global self-integrity should not always lead to reductions in conformity and increases in evidence-based processing.
That is, afﬁrmations do not attenuate conformity per se. They do so
only when people's immediate sense of self-integrity is tied to their collective identity, since it is these situations where people may be more
concerned with “going along” than with “getting it right.” This reasoning is suggested by the results of Study 3. There, afﬁrmation appeared
to shift attention to either normative or evidentiary information, depending on the type of information people were attending to in order
to maintain global self-integrity. For example, Americans in the control
condition who were weakly identiﬁed with America appeared to resist
the group norms and follow the evidence (Study 3). For low-identiﬁed
Americans, a lack of investment in the group should correspond with
a lack of reliance on ingroup norms and, perhaps, an increased concern
with individual level goals of accuracy. Indeed, afﬁrmation appeared to
attenuate evidence-based processing among the low-identiﬁed participants. Although preliminary, this ﬁnding is informative. For example,
the ﬁnding that afﬁrmation can actually reduce the use of evidentiary
information means that afﬁrmation does not necessarily reduce heuristic processing and increase deliberative processing of evidence (see
Jacobson, Mortensen, & Cialdini, 2011): it does so primarily when the
norms of a valued group are salient. Likewise, research suggests that
feelings of love and social connectedness may underpin the effects of afﬁrmation on the processing of threatening information (e.g., Burson,
Crocker, & Mischkowski, 2012; Crocker, Niiya, & Mischkowski, 2008).
The Study 3 ﬁndings suggest that if this is the case, such feelings do
not necessarily make people less concerned with normative information, since afﬁrmation appeared to make low identiﬁed Americans
more concerned with normative information.
More generally, we suggest that among people whose sense of selfintegrity depends on being independent, accurate, or neutral (e.g., a
judge or referee), an afﬁrmation of self-integrity prior to judgment
could actually attenuate accuracy concerns. Consistent with this idea,
Cohen et al. (2007) found that when participants were instructed to
be even-handed in their judgment, control participants heeded the
instructions whereas afﬁrmation participants were actually more biased. Self-afﬁrmation prior to judgment is not simply a de-biasing strategy that should always make people more likely to make evidencebased decisions. Rather, when self-integrity depends on making
evidence-based decisions, afﬁrmation could actually make people less
likely to rely on evidence. In short, afﬁrmation is a tool to free people
from self-evaluative concerns, for good or for ill (Cohen & Sherman,
2014).
Effects over time
The long-term effects observed in Study 2 showed that brief afﬁrmations propagated their impact over the surprisingly long period of four
months (cf. Cohen, Garcia, Purdie-Vaughns, Apfel, & Brzustoski, 2009;
Cohen & Sherman, 2014; Sherman et al., 2013). Evidence from the
follow-up study suggests that this perseverance occurred through two
processes. First, the newly acquired attitude persisted, such that the attitude change seen at Time 1 continued with no additional manipulations at Time 2. Afﬁrmed participants resisted normative information
and attended to evidentiary information at Time 1, and they continued
to show these effects at Time 2. This ﬁnding is consistent with research
showing that attitude change persists when it emerges from systematic
processing of the central merits of the attitude object (Correll et al.,
2004; Petty & Cacioppo, 1986). Second, the afﬁrmation continued to
exert effects on the processing of new political information. When participants were randomly assigned to a read a new article about Obama
86
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
and the economy, participants who had been afﬁrmed four months earlier were still less likely to respond to polling information and more likely to respond to evidentiary economic data when evaluating Obama.
This ﬁnding suggests that afﬁrmation may have produced a general
shift in the self-concept's relationship to political information. When
American norms were salient, afﬁrmation seemed to turn down the relevance of normative information and turn up the relevance of evidentiary information.
Although speculative, one explanation for these ﬁndings is that the
afﬁrmation — timed to immediately precede the presentation of selfrelevant political information (Critcher, Dunning, & Armor, 2010)produced a subtle shift in participants' self-narratives in the political
realm (Wilson, 2011). Perhaps when afﬁrmed subjects became
untethered from group norms, they came to see themselves as more
open-minded or rational. Subsequent information may have been
viewed through this shifted lens, with people giving more attention to
evidentiary information when making political judgments. In one
study, afﬁrmed participants in a politically charged context rated themselves as more open-minded immediately after responding to persuasive political appeals (Binning et al., 2010). Research also suggests that
the cumulative impact of a relatively small change in the nature of information processing can compound signiﬁcantly over time (Cohen et al.,
2009; Garcia & Cohen, 2012; McGuire, 1960; Sherman et al., 2013).
The effect of afﬁrmation may not end after the encoding of the original
message. When timed appropriately, afﬁrmation may trigger new processes (a change in self-narratives, perhaps) that carry the effect forward over time (Cohen & Sherman, 2014).
Untethering threats from the self
In a review, Sherman and Hartson (2011) and Sherman (2013) proposed a general model for understanding the effects of values afﬁrmations. According to one facet of the model, afﬁrmation enables people
to untether threats from the self by promoting self-evaluations that
are independent of the potential threat (Cook, Purdie-Vaughns, Garcia,
& Cohen, 2012; Sherman & Kim, 2005; Sherman et al., 2013). The
present studies found support for this aspect of the model in the analysis
of self-feelings (Study 1) and feelings of self-worth (Study 2). In Study 1,
Democrats felt better about themselves when Obama was rising in the
polls than when he was falling, but the afﬁrmation made people psychologically immune to this information, as afﬁrmed participants resisted
the inﬂuence from the polls. In Study 2, the overall pattern of results
on self-feelings indicated that Republicans in the control condition felt
better when Obama was doing poorly, and this pattern was marginally
reversed in the afﬁrmation condition.
Taken together, along with other research showing how afﬁrmation
buffers the rise and fall of self-feelings in response to collective events
(e.g., Adams et al., 2006; Čehajić-Clancy et al., 2011; Sherman & Kim,
2005), the studies suggest that afﬁrmation can reduce the negative impact of collective events on the self. Whereas previous research has
tended to infer the presence of group norms based on characteristics
of the situation, in the present research group norms were directly manipulated. In doing so, this research provides the ﬁrst experimental evidence that self-afﬁrmation affects the way the people respond to
salient group norms.
Implications for the psychology of political approval
The present ﬁndings shed light on a variety of research illustrating
conformity in political contexts. As seen among control participants, research suggests that presidential approval is inﬂuenced by salient social
norms (Ceci & Kain, 1982; Hayes, Matthes, Hively, & Eveland, 2008;
Marsh, 1985). In an experimental study on how people evaluate political debate performances, exposure to the opinion “worm” on the bottom of the television screen – which ostensibly reﬂected the real time
opinion of undecided voters – strongly predicted who was deemed to
win the debate (Davis, Bowers, & Memon, 2011). More broadly, the
post-election glow in approval that US presidents experience after assuming ofﬁce (Stimson, 1976 id="cf95) and the rally-around-thepresident effect that presidents experience in times of conﬂict (see
Kam & Ramos, 2008), might partly feed off conformity to group norms
in times of uncertainty and fear (also see Landau et al., 2004;
Schmeichel & Martens, 2005).
Political science research has also found that approval can be inﬂuenced by evidentiary information, albeit with some notable qualiﬁcations. In an analysis of US presidential approval data from 1955 to
2005, a president's economic performance was related to their approval
levels, but only among participants for whom the president was in the
political outgroup (Democrat or Republican), and thus, when social
identity was not shared with the president (Lebo & Cassino, 2007).
When social identity was shared with the president, information
about the economy's performance was unrelated to approval of the
president. Such ﬁndings are supportive of the present theorizing that
people tend to be more receptive to evidentiary information and less responsive to group norms when social identity concerns are attenuated.
Limitations
An important limitation of the present work is that much of the theoretical rationale was tested indirectly. For example, across all studies
the role of self-integrity on political conformity was assessed by
afﬁrming participants (or not), presenting them with normative information, and then observing the effects of this manipulation on opinions
of President Obama. From the results we inferred that the afﬁrmation
manipulation untethered participants from normative pressure, but
participants' perceptions of the national norms was not consistently
measured across studies or, when it was measured (in Study 2), it did
not mediate the observed effects. This recommends caution in drawing
conclusions about the impact of afﬁrmation on group norms because
the mediating role of perceived norms was not observed directly.
Similarly, in Study 3 social identiﬁcation with America was measured and then used as a moderator of the experimental manipulations.
Although social identiﬁcation moderated the effects as expected, we did
not directly test if social identiﬁcation changed the perceived signiﬁcance of social norms or evidentiary information. The present ﬁndings
therefore make a novel theoretical contribution but also point to directions for new research. One important direction is to directly test for the
processes theorized to mediate the effects of the afﬁrmation manipulation on participants' attitudes.
On and off the political bandwagon
When a candidate or position is perceived to be popular in the polls,
the salient group norms can increase the target's popularity among the
masses, which can increase others' perceptions of the target's popularity, and so on in a continuing cascade (Bikhchandani, Hirshleifer, &
Welch, 1992; Salganik & Watts, 2008). The present studies suggest
that such recursive processes may depend not, as some scholars have
suggested, on intellectual or moral shortcomings of the public (e.g. Le
Bon, 1897; see Ewen, 1996, for a review). Rather, the effects of political
messages on the public appear to depend critically on perceivers' identiﬁcation with the group and their immediate sense of self-integrity in
the social context.
After completing a brief afﬁrmation exercise and being exposed to a
political message, perceivers' judgments of political information
changed in predictable and signiﬁcant ways. The results suggest that
theory-driven interventions in the political domain can attenuate the
power of public pressure on political beliefs. Being detached from
group norms, either via self-afﬁrmation or low identiﬁcation, may
make people more concerned with independent and evidence-based
decision-making. During situations where unchecked conformity can
lead to groupthink, unjust social policies, or ill-conceived wars, a well-
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
timed reminder of self-integrity may be one means to break the spell of
the group and promote independent decision-making.
Acknowledgments
The authors thank Ishani Banerji, Danieli Evans, Kimberly Hartson,
David Hoffman, Dan Kahan, Heejung Kim, Jonathan Kunstman, Brenda
Major, Anthony Scroggins, and Dimitri Voisin for their comments on
previous versions of this manuscript. This research received ﬁnancial
support from the Stanford Graduate School of Business and from the
University of California, All-Campus Consortium on Research for
Diversity.
References
Abrams, D., Wetherell, M., Cochrane, S., Hogg, M.A., & Turner, J. C. (1994). Knowing what
to think by knowing who you are: Self‐categorization and the nature of norm formation, conformity and group polarization. British Journal of Social Psychology, 29(2),
97–119.
Adams, J. (1794). A defence of the Constitutions of Government of the United States of
America, Vol. 3, London: John Stockdale, Piccadilly.
Adams, G., Tormala, T. T., & O'Brien, L. T. (2006). The effect of self-afﬁrmation on perception of racism. Journal of Experimental Social Psychology, 42, 616–626.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions.
Thousand Oaks, CA: Sage Publications.
Anderson, J. W. (2009, April 27). Polls: Obama scores high marks. Politico. Retrieved from
http://www.politico.com/news/stories/0409/21740.html.
Asch, S. E. (1951). Effects of group pressure upon the modiﬁcation and distortion of judgments. Groups, leadership, and men, 222–236.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (pp. 222–236). New York: Academic Press.
Berinsky, A. J., Huber, G. A., & Lenz, G. S. (2012). Evaluating online labor markets for experimental research: Amazon. com's Mechanical Turk. Political Analysis, 20(3),
351–368.
Bernays, E. L. (1945). Attitude polls — Servants or masters? Public Opinion Quarterly, 9,
264–268.
Bikhchandani, S., Hirshleifer, D., & Welch, I. (1992). A theory of fads, fashion, custom, and
cultural change as informational cascades. Journal of Political Economy, 100,
992–1026.
Binning, K. R. (2007). “It's us against the world”: How distrust in Americans versus
people-in-general shapes competitive foreign policy preferences. Political
Psychology, 28, 777–799.
Binning, K. R., & Sherman, D. K. (2011). Categorization and communication in the face of
prejudice: When describing perceptions changes what is perceived. Journal of
Personality and Social Psychology, 101, 321–336.
Binning, K. R., Sherman, D. K., Cohen, G. L., & Heitland, K. (2010). Seeing the other side:
Reducing political partisanship via self-afﬁrmation in the 2008 presidential election.
Analyses of Social Issues and Public Policy, 10, 276–292.
Bolsen, T., Druckman, J. N., & Cook, F. L. (2014). The inﬂuence of partisan motivated reasoning on public opinion. Political Behavior, 36, 1–28.
Boudreau, C., & McCubbins, M.D. (2010). The blind leading the blind: Who gets polling information and does it improve decisions? Journal of Politics, 72(2), 513–527.
Brown, J.D., & Dutton, K. A. (1995). The thrill of victory, the complexity of defeat: Selfesteem and people's emotional reactions to success and failure. Journal of
Personality and Social Psychology, 68(4), 712.
Burson, A., Crocker, J., & Mischkowski, D. (2012). Two types of value-afﬁrmation: Implications for self-control following social exclusion. Social Psychological and Personality
Science, 3, 510–516.
Calmes, J. (2014, January 27). Obama's puzzle: Economy rarely better, approval rarely
worse. Retrieved from: www.nytimes.com/2014/01/28/us/politics/obamas-puzzleeconomy-rarely-better-approval-rarely-worse.html
Campbell, J.D., & Fairey, P. J. (1989). Informational and normative routes to conformity:
The effect of faction size as a function of norm extremity and attention to the stimulus. Journal of Personality and Social Psychology, 57(3), 457.
Ceci, S. J., & Kain, E. L. (1982). Jumping on the bandwagon with the underdog: The impact
of attitude polls on polling behavior. Public Opinion Quarterly, 46, 228–242.
Čehajić-Clancy, S., Effron, D. A., Halperin, E., Liberman, V., & Ross, L. D. (2011). Afﬁrmation,
acknowledgment of in-group responsibility, group-based guilt, and support for reparative measures. Journal of Personality and Social Psychology, 101, 256–270.
Chaiken, S., Giner-Sorolla, R., & Chen, S. (1996). Beyond accuracy: Defense and impression
motives in heuristic and systematic information processing. In P.M. Gollwitzer, & J. A.
Bargh (Eds.), The psychology of action: Linking cognition and motivation to behavior
(pp. 553–578). New York: Guilford Press.
Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic processing within
and beyond the persuasion context. In J. S. Uleman, & J. A. Bargh (Eds.), Unintended
thought (pp. 212–252). New York: Guilford Press.
Chen, S., Shechter, D., & Chaiken, S. (1996). Getting at the truth or getting along:
Accuracy- versus impression-motivated heuristic and systematic processing. Journal
of Personality and Social Psychology, 71, 262–275.
87
Chung, R. (2012). The freedom to publish opinion poll results: A worldwide update of
2012. Downloaded from: http://wapor.org/wp-content/uploads/2012/12/WAPOR_
FTP_2012.pdf
Cialdini, R. B., & Goldstein, N. J. (2004). Social inﬂuence: Compliance and conformity.
Annual Review of Psychology, 55, 591–621.
Cohen, G. L. (2003). Party over policy: The dominating impact of group inﬂuence on political beliefs. Journal of Personality and Social Psychology, 85, 808–822.
Cohen, G. L., Aronson, J., & Steele, C. M. (2000). When beliefs yield to evidence: Reducing
biased evaluation by afﬁrming the self. Personality and Social Psychology Bulletin, 26,
1151–1164.
Cohen, G. L., Garcia, J., Purdie-Vaughns, V., Apfel, N., & Brzustoski, P. (2009). Recursive processes in self-afﬁrmation: Intervening to close the minority achievement gap. Science,
324, 400–403.
Cohen, G. L., & Sherman, D. K. (2014). The psychology of change: Self-afﬁrmation and social psychological intervention. Annual Review of Psychology, 65, 333–371.
Cohen, G. L., Sherman, D. K., Bastardi, A., McGoey, M., Hsu, A., & Ross, L. (2007). Bridging
the partisan divide: Self-afﬁrmation reduces ideological closed-mindedness and inﬂexibility in negotiation. Journal of Personality and Social Psychology, 93, 415–430.
Coleman, S. (2007). Popular delusions: How social conformity molds society and politics.
Youngstown, NY: Cambria.
Condon, S. (2010, February 11). Poll: Economy brings down Obama's job approval rating.
CBS News. Downloaded from: http://www.cbsnews.com/news/poll-economy-bringsdown-obamas-job-approval-rating/.
Cook, J. E., Purdie-Vaughns, V., Garcia, J., & Cohen, G. L. (2012). Chronic threat and contingent belonging: Protective beneﬁts of values afﬁrmation on identity development.
Journal of Personality and Social Psychology, 102, 479–496.
Correll, J., Spencer, S. J., & Zanna, M. P. (2004). An afﬁrmed self and an open mind: Selfafﬁrmation and sensitivity to argument strength. Journal of Experimental Social
Psychology, 40, 350–356.
Coultas, J. C. (2004). When in Rome… An evolutionary perspective on conformity. Group
Processes & Intergroup Relations, 7, 317–331.
Critcher, C. R., Dunning, D., & Armor, D. A. (2010). When self-afﬁrmations reduce defensiveness: Timing is key. Personality and Social Psychology Bulletin, 36, 947–959.
Crocker, J., Niiya, Y., & Mischkowski, D. (2008). Why does writing about important values
reduce defensiveness? Self-afﬁrmation and the role of positive other-directed feelings. Psychological Science, 19, 740–747.
Davis, C. J., Bowers, J. S., & Memon, A. (2011). Social inﬂuence in televised election debates: A potential distortion of democracy. PLoS One, 6(3), e18154.
Deustch, M., & Gerard, H. (1955). A study of normative and informational social inﬂuences upon individual judgment. Journal of Abnormal and Social Psychology, 51,
629–636.
Dunning, D. (2003). The zealous self-afﬁrmer: How and why the self lurks so pervasively
behind social judgment. In S. J. Spencer, S. Fein, M. P. Zanna, & J. M. Olson (Eds.),
Motivated social perception: Ontario symposium on personality and social psychology,
Vol. 9. (pp. 45–72). Mahwah, NJ: Lawrence Erlbaum.
Ewen, S. (1996). PR! A social history of spin. New York: Basic Books.
Fast, N. J., Heath, C., & Wu, G. (2009). Common Ground and Cultural Prominence How
Conversation Reinforces Culture. Psychological Science, 20, 904–911.
Federal Reserve Bank of Philadelphia (2010, February 12). Survey of professional forecasters: First quarter 2010. Downloaded from: http://www.phil.frb.org/researchand-data/real-time-center/survey-of-professional-forecasters/2010/spfq110.pdf
Fein, S., & Spencer, S. J. (1997). Prejudice as self-image maintenance: Afﬁrming the self
through derogating others. Journal of Personality and Social Psychology, 73, 31–44.
Gaertner, L., Sedikides, C., Vevea, J. L., & Iuzzini, J. (2002). The “I,” the “we,” and the
“when”: A meta-analysis of motivational primacy in self-deﬁnition. Journal of
Personality and Social Psychology, 83, 574–591.
Gallup (2014). Gallup daily: Obama job approval. Retrieved from: http://www.gallup.
com/poll/113980/gallup-daily-obama-job-approval.aspx
Garcia, J., & Cohen, G. L. (2012). A social psychological perspective on educational intervention. In E. Shaﬁr (Ed.), The behavioral foundations of policy. (pp. 329–350)
Princeton, NJ: Princeton University Press.
Grifﬁn, D. W., & Harris, P. R. (2011). Calibrating the response to health warnings: Limiting
both overreaction and underreaction with self-afﬁrmation. Psychological Science, 22,
572–578.
Grube, J. W., Morgan, M., & McGree, S. T. (1986). Attitudes and normative beliefs as predictors of smoking intentions and behaviours: A test of three models. British Journal of
Social Psychology, 25(2), 81–93.
Hardin, C. D., & Higgins, E. T. (1996). Shared reality: How social veriﬁcation makes the
subjective objective. In R. M. Sorrentino, & E. T. Higgins (Eds.), Handbook of motivation
and cognition (pp. 28–84). New York: Guilford Press.
Harris, P. R., & Epton, T. (2009). The impact of self-afﬁrmation on health cognition, health
behaviour, and other health-related responses: A narrative review. Social and
Personality Psychology Compass, 3, 932–978.
Harris, P. R., & Epton, T. (2010). The impact of self-afﬁrmation on health-related cognition
and health behaviour: Issues and prospects. Social and Personality Psychology
Compass, 4, 439–454.
Harris, P. R., & Napper, L. (2005). Self-afﬁrmation and the biased processing of threatening
health-risk information. Personality and Social Psychology Bulletin, 32, 1250–1263.
Hayes, A. F., Matthes, J., Hively, M. H., & Eveland, W. P., Jr. (2008). In search of the opinion
climate: A new (and novel) test of the spiral of silence theory. Unpublished
manuscript.
Hogg, M.A., & Reid, S. A. (2006). Social identity, self‐categorization, and the communication of group norms. Communication Theory, 16(1), 7–30.
Huo, Y. J., Binning, K. R., & Molina, L. E. (2010). Testing an integrative model of respect:
Implications for social engagement and well-being. Personality and Social Psychology
Bulletin, 36, 200–212.
88
K.R. Binning et al. / Journal of Experimental Social Psychology 56 (2015) 73–88
Insko, C. A., Drenan, S., Solomon, M. R., Smith, R., & Wade, T. J. (1983). Conformity as a
function of the consistency of positive self-evaluation with being like and being
right. Journal of Experimental Social Psychology, 19, 341–358.
Jackson, J. M., & Saltzstein, H. D. (1958). The effect of person-group relationships on conformity processes. Journal of Abnormal and Social Psychology, 57(1), 17.
Jacobson, R. P., Mortensen, C. R., & Cialdini, R. B. (2011). Bodies obliged and unbound: Differentiated response tendencies for injunctive and descriptive social norms. Journal of
Personality and Social Psychology, 100, 433–448.
Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and ﬁascos. Boston:
Houghton -Mifﬂin.
Jones, J. M. (2010, February 2). In U.S., unemployment jumps to top problem status. Gallup
Politics. Retrieved from http://www.gallup.com/poll/125846/Unemployment-JumpsTop-Problem-Status.aspx.
Kahan, D.M., Jenkins‐Smith, H., & Braman, D. (2011). Cultural cognition of scientiﬁc consensus. Journal of Risk Research, 14, 147–174.
Kam, C. D., & Ramos, J. M. (2008). Joining and leaving the rally: Understanding the surge
and decline in presidential approval following 9/11. Public Opinion Quarterly, 72,
619–650.
Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion
Quarterly, 24, 163–204.
Klein, W. M. P., Harris, P. R., Ferrer, R. A., & Zajac, L. E. (2011). Feelings of vulnerability in
response to threatening messages: Effects of self-afﬁrmation. Journal of Experimental
Social Psychology, 47, 1237–1241, http://dx.doi.org/10.1016/j.jesp.2011.05.005.
Koole, S. L., Smeets, K., Van Knippenberg, A., & Dijksterhuis, A. (1999). The cessation of rumination through self-afﬁrmation. Journal of Personality and Social Psychology, 77,
111–125.
Kunda, Z. (1987). Motivated inference: Self-serving generation and evaluation of causal
theories. Journal of Personality and Social Psychology, 53(4), 636.
Landau, M., Soloman, S., Greenberg, J., Cohen, F., Pyszczynski, T., & Arndt, J. (2004). Deliver
us from evil: The effects of mortality salience and reminders of 9/11 on support for
George W. Bush. Personality and Social Psychology Bulletin, 30, 1136–1150.
Latanè, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help? New
York: Appleton-Century Crofts.
Le Bon, G. (1897). The crowd: A study of the popular mind (2nd ed.). New York: Macmillan.
Lebo, M. J., & Cassino, D. (2007). The aggregated consequences of motivated reasoning
and the dynamics of partisan presidential approval. Political Psychology, 28(6),
719–746.
Leeper, T. J., & Slothuus, R. (2014). Political parties, motivated reasoning, and public opinion formation. Political Psychology, 35, 129–156.
Liberman, A., & Chaiken, S. (1992). Defensive processing of personally relevant health
messages. Personality and Social Psychology Bulletin, 18, 669–679.
Marques, J. M., Yzerbyt, V. Y., & Leyens, J. P. (1988). The black sheep effect: Judgmental extremity towards ingroup members as a function of group identiﬁcation. European
Journal of Social Psychology, 18, 1–16.
Marsh, C. (1985). Back on the bandwagon: The effect of opinion polls on public opinion.
British Journal of Political Science, 15, 51–74.
McGuire, W. J. (1960). Cognitive consistency and attitude change. Journal of Abnormal and
Social Psychology, 60, 345.
McQueen, A., & Klein, W. M. P. (2006). Experimental manipulations of self-afﬁrmation: A
systematic review. Self and Identity, 5, 289–354.
Nolan, J. M., Schultz, P. W., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2008). Normative social inﬂuence is underdetected. Personality and Social Psychology Bulletin, 34,
913–923.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral
routes to attitude change. New York: Springer-Verlag.
Rokeach, M. (1973). The nature of human values. New York: Free press.
Rokeach, M. (1973). The nature of human values. New York: Free press. Ross, L., & Ward,
A. (1996). Naïve realism in everyday life: Implications for social conﬂict and misunderstanding. In T. Brown, E. Reed, & E. Turiel (Eds.), Values and knowledge
(pp. 103–135). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Saad, L. (2009, May 11). Obama Approval Picks Up in May. Gallup Politics. Retrieved from
http://www.gallup.com/poll/118306/Obama-Approval-Picks-Up-May.aspx.
Salganik, M. J., & Watts, D. J. (2008). Leading the herd astray: An experimental study of
self-fulﬁlling prophecies in an artiﬁcial cultural market. Social Psychology Quarterly,
71, 338–355.
Schmeichel, B. J., & Martens, A. (2005). Self-afﬁrmation and mortality salience: Afﬁrming
values reduces worldview defense and death-thought accessibility. Personality and
Social Psychology Bulletin, 31, 658–667.
Schoen, D. E., & Rasmussen, S. (2009, March 13). Obama's poll numbers are falling to
earth. The Wall Street Journal. Retrieved from http://online.wsj.com/article/
SB123690358175013837.html.
Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The
constructive, destructive, and reconstructive power of social norms. Psychological
Science, 18(5), 429–434.
Sherif, M. (1936). The psychology of social norms.
Sherman, D. K. (2013). Self-afﬁrmation: Understanding the effects. Social Psychology and
Personality Compass, 7, 834–845.
Sherman, D. K., & Cohen, G. L. (2006). The psychology of self-defense: Self-afﬁrmation
theory. In M. P. Zanna (Ed.), Advances in experimental social psychology, Vol. 38. (pp.
183–242). San Diego, CA: Academic Press.
Sherman, D. K., & Hartson, K. (2011). Reconciling self-defense with self-criticism: Selfafﬁrmation theory. In M. Alicke, & C. Sedikides (Eds.), The handbook of selfenhancement and self protection (pp. 128–151). New York: Guilford Press.
Sherman, D. K., Hartson, K. A., Binning, K. R., Purdie-Vaughns, V., Garcia, J., TaborskyBarba, S., et al. (2013). Deﬂecting the trajectory and changing the narrative: How
self-afﬁrmation affects academic performance and motivation under identity threat.
Journal of Personality and Social Psychology, 104, 591–618, http://dx.doi.org/10.1037/
a0031495.
Sherman, D. K., & Kim, H. S. (2005). Is there an “I” in “team”? The role of the self in groupserving judgments. Journal of Personality and Social Psychology, 88, 108–120.
Sherman, D. A. K., Nelson, L. D., & Steele, C. M. (2000). Do messages about health risks
threaten the self? Increasing the acceptance of threatening health messages via
self-afﬁrmation. Personality and Social Psychology Bulletin, 26, 1046–1058.
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology undisclosed ﬂexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359–1366.
Simon, H. A. (1954). Bandwagon and underdog effects and the possibility of election predictions. Public Opinion Quarterly, 18, 245–253.
Snyder, M., & DeBono, K. G. (1985). Appeals to image and claims about quality: Understanding the psychology of advertising. Journal of Personality and Social Psychology,
49, 586–597.
Steele, C. M. (1988). The psychology of self-afﬁrmation: Sustaining the integrity of the
self. In L. Berkowitz (Ed.), Advances in experimental social psychology, Vol. 21. (pp.
261–302). New York: Academic Press.
Stimson, J. A. (1976). Public Support for American Presidents A Cyclical Model. Public
Opinion Quarterly, 40(1), 1–21.
Surowiecki, J. (2004). The wisdom of crowds. New York: Doubleday.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conﬂict. The Social
Psychology of Intergroup Relations, 33, 47.
Tajfel, H., & Turner, J. (1986). The social identity theory of intergroup behavior. In S.
Worchel, & W. Austin (Eds.), The psychology of intergroup relations (pp. 7–24). Chicago, IL: Nelson Hall.
Terry, D. J., & Hogg, M.A. (1996). Group norms and the attitude-behavior relationship: A
role for group identiﬁcation. Personality and Social Psychology Bulletin, 22(8),
776–793.
The World Bank (2014). GDP growth (annual %). Downloaded from: http://data.
worldbank.org/indicator/NY.GDP.MKTP.KD.ZG
Turner, J. C., Hogg, M.A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering
the social group: A self-categorization theory. Basil Blackwell.
Turner, J. C., & Reynolds, K. J. (2011). Self-categorization theory. In A. W. Kruglanski, E. T.
Higgins, & P. A.M. van Lange (Eds.), Handbook of theories of social psychology: Volume
two (pp. 399–417). London: Sage.
Tye, L. (1998). The father of spin: Edward Bernays and the birth of public relations. New
York: Holt.
Wilson, T. D. (2011). Redirect: The surprising new science of psychological change. New
York: Little, Brown, and Company.
Health Psychology
2014, Vol. 33, No. 5, 000
© 2014 American Psychological Association
0278-6133/14/$12.00 http://dx.doi.org/10.1037/hea0000101
Message Framing for Health: Moderation by Perceived Susceptibility and
Motivational Orientation in a Diverse Sample of Americans
John A. Updegraff
Cameron Brick
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Kent State University
University of California, Santa Barbara
Amber S. Emanuel
Roy E. Mintzer
University of Florida
University of Southern California
David K. Sherman
University of California, Santa Barbara
Objective: The present study examined how gain- and loss-framed informational videos about oral health
influence self-reported flossing behavior over a 6-month period, as well as the roles of perceived
susceptibility to oral health problems and approach/avoidance motivational orientation in moderating
these effects. Method: An age and ethnically diverse sample of 855 American adults were randomized
to receive no health message, or either a gain-framed or loss-framed video presented on the Internet.
Self-reported flossing was assessed longitudinally at 2 and 6 months. Results: Among the entire sample,
susceptibility interacted with frame to predict flossing. Participants who watched a video where the frame
(gain/loss) matched perceived susceptibility (low/high) had significantly greater likelihood of flossing at
recommended levels at the 6-month follow-up, compared with those who viewed a mismatched video or
no video at all. However, young adults (18 –24) showed stronger moderation by motivational orientation
than by perceived susceptibility, in line with previous work largely conducted with young adult samples.
Conclusion: Brief informational interventions can influence long-term health behavior, particularly when
the gain- or loss-frame of the information matches the recipient’s beliefs about their health outcome risks.
Keywords: perceived risk, message framing, oral care, oral health, persuasive communication
Supplemental materials: http://dx.doi.org/10.1037/hea0000101.supp
relative effectiveness of gain- and loss-framed health messages
depends on characteristics of the message recipient. In this study,
conducted over 6 months among a large, age and ethnically diverse
sample of Americans, we examine two classes of psychological
characteristics—the motivational orientation of the message recipient and beliefs about one’s health risk—that have each received
considerable support as moderators of successful message framing
(Rothman & Updegraff, 2010).
Health behavior interventions often seek to educate people
about the consequences of a health behavior. Information about
consequences can be framed as either gains or losses. Gain-framed
messages emphasize the benefits of adherence, such as “Flossing
daily can lower your risk for periodontitis.” Loss-framed messages, on the other hand, emphasize the costs of nonadherence
such as “Not flossing daily can increase your risk for periodontitis.” Although no overall advantage of gain- versus loss-framed
messages exists in many domains of health behavior including oral
health (for meta-analysis, see Gallagher & Updegraff, 2012), the
Motivational Orientation
People differ in the degree to which they monitor for and
respond to favorable versus unfavorable outcomes (Carver &
White, 1994; Elliot & Thrash, 2002). Approach/avoidance theorists (Carver & White, 1994; Elliot & Thrash, 2002) view this
disposition as the extent to which people are motivated to approach
favorable outcomes (approach orientation) or avoid unfavorable
outcomes (avoidance orientation). Regulatory focus theorists (Higgins, 1997) view this disposition as the extent to which people are
sensitive to the presence or absence of positive events (promotion
regulatory focus) or to the presence or absence of negative events
(prevention regulatory focus). Both views converge on the
prediction that approach-oriented and promotion-focused people should respond more favorably to gain-framed appeals,
whereas avoidance-oriented and prevention-focused people
John A. Updegraff, Department of Psychology, Kent State University;
Cameron Brick and David K. Sherman, Department of Psychological and
Brain Sciences, University of California, Santa Barbara; Roy E. Mintzer,
Herman Ostrow School of Dentistry of the University of Southern California; Amber S. Emanuel, Department of Community Dentistry and
Behavioral Science, University of Florida.
Funding for this project was provided by a grant from the National
Institute of Dental and Craniofacial Research (R21 DE019704).
Correspondence concerning this article should be addressed to John A.
Updegraff, Department of Psychology, Kent State University, Kent, OH
44242-0001. E-mail: [email protected]
1
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2
UPDEGRAFF, BRICK, EMANUEL, MINTZER, AND SHERMAN
should respond more favorably to loss-framed appeals, because
congruently framed messages yield more regulatory fit (Lee &
Aaker, 2004) and lead to greater elaboration (Updegraff, Sherman,
Luyster & Mann, 2007).1
Support for these predictions was first shown in the domain of
oral health (Mann, Sherman, & Updegraff, 2004; Sherman, Mann,
& Updegraff, 2006; for a review, see Sherman, Updegraff, &
Mann, 2008). Similar effects of matching message frame to motivational orientation have now been observed for HPV vaccination (Gerend & Shepherd, 2007; Nan, 2012), physical activity
(Latimer, Rivers, et al., 2008), calcium consumption (Gerend &
Shepherd, 2013), fruit and vegetable consumption (Latimer,
Williams-Piehota, et al., 2008), and smoking prevention (Zhao &
Pechmann, 2007). With few exceptions (Latimer, Rivers, et al.,
2008; Latimer, Williams-Piehota, et al., 2008), moderation by
motivational orientation has only been shown in adolescent or
young adult populations. Meta-analyses show the health behavior
of younger people to be driven by different factors than that of
middle-aged and older adults (e.g., Albarracin, Johnson, Fishbein,
& Muellerleile, 2001; Hagger, Chatzisarantis, & Biddle, 2002).
Adolescents’ physical activity, for example, is more strongly related to personality traits such as achievement orientation and
sensation-seeking than it is to specific beliefs about the behavior
such as attitudes, outcome expectations, or perceived benefits of
physical activity (Sallis, Prochaska, & Taylor, 2000; see Ulleberg
& Rundmo, 2003, for similar patterns in the domain of adolescent
risky driving behavior). Young adults typically have fewer health
problems than their older counterparts, and unrealistic optimism
about health risks (cf. Renner, Knoll, & Schwarzer, 2000; Weinstein, 1984). As such, young adults’ beliefs about their health
status may be less dominant in shaping their responses to health
communications than their dispositional motivations.
Perceived Susceptibility
Perceived susceptibility is another key moderator of people’s
responses to framed health messages. When people perceive themselves as susceptible to a health condition, they may be especially
vigilant against possibility of negative health outcomes; in regulatory focus terms, they are prevention-focused concerning the
health issue. In contrast, when people perceive low susceptibility
to a health condition, they may be especially eager to achieve and
maintain positive outcomes; they are promotion-focused concerning the health issue. This regulatory focus-based account (cf. Lee
& Aaker, 2004) predicts that gain-framed messages should be
more effective for those who perceive low susceptibility to a health
problem, but loss-framed messages should be more effective for
those who perceive high susceptibility.
In a study that manipulated young adults’ perceptions of risk for
mononucleosis, Lee and Aaker (2004) found that a gain-framed
message promoted more interest in a preventive supplement
against mononucleosis than a loss-framed message among people
told that they were at low risk; in contrast, a loss-framed message
promoted more interest for those told that they were at high risk.
Naturally occurring beliefs about susceptibility likewise moderate
message framing effects, as shown in studies of middle-to-olderaged women considering mammography screening (Gallagher,
Updegraff, Rothman, & Sims, 2011), middle-aged adults considering colorectal cancer screening (Ferrer, Klein, Zajac, Land, &
Ling, 2012), and young adults considering HIV-testing (Hull,
2012). Hull (2012) also showed that perceived susceptibility interacted with message framing to increase participants’ elaboration
of the message, which mediated the effects of framing on intentions to get tested.
Most studies that have reported a moderating role of people’s
beliefs about susceptibility have found it among middle- to olderadult samples (Ferrer et al., 2012; Gallagher et al., 2011), perhaps
because beliefs about susceptibility to health issues become more
salient with age. In the domain of oral health, most message
framing studies have utilized adolescent or young adult samples
(Mann et al., 2004; Sherman et al., 2006; Uskul, Sherman, &
Fitzgibbon, 2009), perhaps limiting the ability to identify beliefs
about susceptibility as moderators of message framing effects,
although that it should be noted that these aforementioned studies
did not assess susceptibility so it remains an open question.
Study Objectives
Dental caries, gingivitis, and periodontitis are widespread health
problems, and may increase one’s risk for serious medical conditions such as cardiovascular disease (Beck & Offenbacher, 2005;
Loos, Craandijk, Hoek, Wertheim-van Dillen, & van der Velden,
2000). Many Americans remain nonadherent to practices that can
prevent the onset of oral health problems (McCaul, Glasgow, &
Gustafson, 1985), with over 70% of American adults not flossing
daily and over 30% not flossing at all (CDC/NCHS, 2010). To
address this issue, we evaluated an online message framing intervention to improve oral health behaviors.
We recruited a large sample of Americans, diverse in both age
and ethnic/racial background, randomly assigned participants to
view an Internet-administered gain- or loss-framed educational
video about the importance of dental flossing (or no video), and
observed the impact of these videos on flossing over 6 months. We
predicted that gain-framed messages would be more persuasive for
approach-oriented and low-susceptibility people but loss-framed
messages would be more persuasive for avoidance-oriented and
high-susceptibility people. We also expected that motivational
orientation would moderate effects among the young adults in our
sample, but expected that perceived susceptibility might moderate
among a more age-diverse sample. The no message control condition allowed us to examine whether “mismatched” messages
provided any benefit compared with no message at all. When
behavioral effects were observed, a secondary aim was to explore
potential psychological mediators, including attitudes and intentions.
Method
Participants
Participants were recruited by Knowledge Networks®, a survey
research firm that maintains a large panel of Americans. Details of
1
Regulatory focus theory makes further distinctions, including (a)
promotion-oriented people should be more responsive to gains representing
the presence of positive outcomes rather than the absence of negative
outcomes and (b) prevention-oriented people should be more responsive to
losses representing the presence of negative outcomes rather than the
absence of positive outcomes.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
MESSAGE FRAMING FOR HEALTH
this panel can be found here: http://www.knowledgenetworks
.com/knpanel/. Knowledge Networks utilizes probability-based
sampling with a mix of random-digit-dialing and address-basedsampling to recruit and select members into the panel. Households
that have a home computer and Internet access earn incentive
points (redeemable for cash) for completing surveys. Households
that do not have a computer and access to the Internet are provided
with free monthly Internet access and an Internet-enabled device
(either laptop or WebTV) in return for completing a short survey
weekly. The typical survey commitment for panel members is one
survey per week or four per month with a duration of 10 to 15 min
per survey, and typical payment is approximately $5 per month.
Panel members have the option to not respond to any surveys or
individual survey items. Participants in this study were drawn from
panel members who preferred to complete surveys in English and
were invited via an e-mail that did not immediately reveal the
focus on oral health until the informed consent process. The
inclusion criteria were flossing less than twice a day at baseline,
staying on the assigned video page for the full duration of the
video (! 5 min), and providing at least one follow-up (see Figure
1 for CONSORT Diagram; Moher, Schulz, & Altman, 2001).
Eight-hundred fifty-five participants met inclusion criteria. The
sample was 46.6% female, with average age of 45, ranging from
18 to 89. The sample was comparably distributed between Hispanic (28.7%), Black/African American (26.8%), White (22.3%),
and Asian (22.2%) participants.2 Due to panel limitations, a sizable
proportion of American Indians, Alaska Natives, and Native Hawaiians could not be recruited, although a few participants who
identified as Hispanic also identified as Native American or
Alaska Native (N " 5) or as Native Hawaiian (N " 1). Thirty-two
percent had a high school education or less, 29.1% had some
college, and 39.0% had a bachelor’s degree or higher. Median
household income was approximately $50,000 per year.
Procedure
All surveys were presented online. All participants completed
baseline measures of oral health behaviors, perceived susceptibility, and motivational orientation. Participants were then randomly
assigned to a gain-framed video (n " 352), loss-framed video (n "
334), or no video control condition (n " 169), with an a priori
decision to run approximately half as many in the control condition
as in each of the framed conditions. Next, participants reported
intentions to engage in oral health behaviors. After 2 and 6 months,
individuals were again e-mailed (and phoned, if needed) to complete an online follow-up survey about their recent oral health
behaviors. Participants who remained at the 6-month survey were
no different from those who dropped out on any demographic or
health variable including flossing behavior at 2 months, all ps !
.20. However, there was more dropout in the no video control
condition than in the video conditions at 2 months (#2(1) " 3.11,
p " .08) and 6 months (#2(1) " 7.79, p $ .01). Importantly, there
was no significant difference in dropout between the gain- and
loss-framed conditions at either 2 months (p " .85) or 6 months
(p " .73).
Measures and Materials
Past flossing. Frequency of flossing at baseline was assessed
with a single item used in a prior national survey of oral health
3
behavior (Davidson, Rams, & Andersen, 1997): “In general, how
often do you floss your teeth?” with response options of never,
once per month, a few (2–3) times per month, once per week, a few
(2– 6) times per week, once per day, and two or more times per
day.
Motivational orientation. Motivational orientation was assessed using the Behavioral Inhibition System (BIS) and Behavioral Activation System (BAS) scales (Carver & White, 1994),
which include 13 items that assess approach (BAS) orientation
(% " .86) and seven items that assess avoidance (BIS) orientation
(% " .73). Consistent with procedures used elsewhere (Mann et al.,
2004; Sherman et al., 2006), we computed a measure of predominant motivational orientation by subtracting the mean score on the
avoidance items from the mean score on the approach items, so
higher values represent stronger approach relative to avoidance
orientation. The BIS/BAS has been validated in adult community
samples (Jorm et al., 1998).
Perceived susceptibility. Perceived susceptibility to oral
health problems was assessed by a five-item scale (Batchelor &
Sheiham, 2002) on the perceived likelihood of having a filling,
getting gum disease, or having other oral health problems within
the next 12 months and the next 5 years (% " .86).
Framed videos. Each of the videos were created for this
study, were 5 1/2 min in length, and featured a female dentist who
described the health consequences of regular brushing, flossing,
and dental visits. Most of the video covered the importance of
flossing, including a section with the dentist and a patient that
instructed viewers how to floss according to American Dental
Association recommendations (ADA, 2013; spool method, at least
once per day). Approximately 40% of the script consisted of
framed statements. Pilot testing validated the manipulation of
framing and showed that the videos were easy to understand and
pay attention to (see supplementary materials for more details and
video scripts).
Potential mediators. At the end of the baseline survey and the
2-month survey, several possible mediating beliefs were assessed.
Intention to floss was assessed by a single item “How many times
do you plan to floss your teeth in the upcoming week?” with
options ranging from 0 –14 (see also Sherman et al., 2006). Attitude toward flossing was assessed with a single item developed for
this study that asked “What is the value of flossing your teeth?”
(1 " extremely worthless, 5 " extremely valuable). Self-efficacy
specifically related to dental flossing was assessed with seven
items used in prior research (Sherman et al., 2006), % " .95.
Flossing at 2 and 6 months. To obtain a sensitive measure of
flossing behavior at each of the follow-ups, we asked participants
to report how many times in the past week they flossed their teeth,
on a scale ranging from 0 –14. This measure shows significant
correlation with clinical indices of oral health (CDC/NHCS, 2010).
Analytic Strategy
We used generalized estimating equations (GEE; Liang &
Zeger, 1986) in Stata 12 to examine the influence of the framed
2
A secondary purpose of recruiting equal numbers of Whites, Blacks,
Asians, and Hispanics was to examine whether cultural exposure moderates differential effectiveness of gain versus loss frames (e.g., Uskul et al.,
2009). This is the focus of a forthcoming report (Sherman et al., 2014).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4
UPDEGRAFF, BRICK, EMANUEL, MINTZER, AND SHERMAN
Figure 1. CONSORT diagram of eligibility, allocation, and attrition. (Note: One participant was missing
outcome data at 2 months but remained in the study at 6 months; hence the difference in the 2-month and final
analysis N).
video messages on flossing across the 6-month study period. GEE
is an extension of the generalized linear model for regressions
involving repeated outcomes. By analyzing both follow-up waves
simultaneously, Type I error is reduced. When a significant effect
was found across follow-ups, we used multiple regression to
identify the influence of framing separately at each of the waves.
Missing data other than attrition was negligible (see Figure 1);
thus, missing values were not imputed. As our hypotheses predicted interactions between perceived susceptibility, motivational
orientation, and message framing, we use procedures (outlined by
Aiken & West, 1991) to test moderation. Continuous measures
were mean-centered prior to computing product terms with the
message framing variable. To ensure that standard errors and all
associated significance tests were robust to any violations of
normality in the residuals, we report bootstrapped standard errors.
Baseline demographic differences in flossing were also examined.
Females (M " 4.25, SD " 1.73) flossed more than males (M "
3.69, SD " 1.86), p $ .001. Asians (M " 4.32, SD " 1.77) flossed
significantly more than Caucasians (M " 4.00, SD " 1.72),
Hispanics (M " 3.91, SD " 1.85), and Blacks (M " 3.63, SD "
1.86) combined, p " .001; Blacks flossed significantly less than all
other groups combined, p " .002. Higher age (r " .13) and income
(r " .16) were associated with greater flossing. These variables
were retained as covariates in all further analyses.
Results
Comparison of Video and Control Groups
Table 1 presents descriptive data by experimental condition. We
first examined the effects of our videos on subsequent flossing by
comparing those who viewed either of the framed videos with
those in the control condition, while controlling for past flossing.
Participants who saw a video reported flossing significantly more
(M " 3.82, SE " .057) than those who did not see a video (M "
3.48, SE " .11), B " .29, SE " .12, p " .017, Cohen’s d " .15.
Moderators of Message Framing Effects
Next we examined the relative effectiveness of the gain- and
loss-framed videos and their interaction with perceived susceptibility and motivational orientation (see Table 2 for results of GEE
MESSAGE FRAMING FOR HEALTH
Table 1
Demographic and Analytical Variable by Condition
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Age
Female
Race: White
Race: Black
Race: Hispanic
Race: Asian
Median income
Susceptibility
Motivational orientation
Flossing at baseline!
Flossing at 2 months
Flossing at 6 months
Dental visit by 6 months!!
Gain video
Loss video
No video
(N " 334)
(N " 352)
(N " 169)
43.69 (14.81) 45.46 (15.80) 44.25 (15.84)
46%
23%
25%
29%
23%
$47,100
2.35 (.97)
.08 (.57)
2.97 (2.79)
3.68 (3.33)
4.02 (3.69)
47%
46%
22%
27%
29%
22%
$47,500
2.39 (1.08)
.07 (.58)
3.03 (2.84)
3.79 (3.64)
3.86 (3.85)
54%
49%
21%
30%
27%
22%
$47,900
2.50 (1.11)
.09 (.58)
3.06 (2.91)
3.47 (3.46)
3.68 (3.77)
48%
!
Values reported in this table derive from converting the baseline measure
into a days/week metric; as such, direct comparisons between baseline and
subsequent flossing measures should be interpreted cautiously. !! Dental
visitation was not significantly predicted by framing condition. Furthermore, it did not predict overall levels of flossing, nor interact with the video
intervention to significantly predict flossing at either the 2-month or
6-month follow-up.
analysis), while also controlling for past flossing and demographics. There was no overall effect of gain versus loss framing (p "
.53). However, there was a significant Frame & Perceived Susceptibility interaction (p " .03), indicating that the influence of
message framing on flossing depended on the recipient’s level of
perceived susceptibility.3 For participants with low ('1 SD) perceived susceptibility, those who viewed the gain-framed video
flossed significantly more across the 6-month period (M " 3.97,
SE " .12) than those who viewed the loss-framed video (M "
3.60, SE " .11), z " 2.32, p " .02. For participants with high ((1
SD) perceived susceptibility, this pattern was reversed; those who
viewed the loss-framed video flossed somewhat but not significantly more (M " 3.96, SE " .13) than those who viewed the
gain-framed video (M " 3.75, SE " .13), z " 1.22, p " .22.
The 2- and 6-month follow-ups were separately examined for
the Frame & Susceptibility effect. At 2 months, the Frame &
Susceptibility was not significant. At the 6-month follow-up, the
Frame & Susceptibility interaction was significant (p " .004, see
Figure 2).4 Consistent with the pattern observed in the GEE
analyses, among the participants with low perceived susceptibility
('1 SD), those who viewed the gain-framed video flossed more
(M " 4.35, SE " .25) than those who viewed the loss-framed
video (M " 3.44, SE " .25), z " 2.56, p " .01. Among participants at high perceived susceptibility ((1 SD), this pattern was
different, with those who viewed the loss-framed video flossing
somewhat but not significantly more (M " 4.19, SE " .26) than
those who viewed the gain-framed video (M " 3.83, SE " .24),
z " 1.01, p " .31.
To examine the potential practical effect of this finding, we
examined the proportion of participants flossing once a day or
more at the 6-month follow-up as a function of match between
frame and perceived susceptibility. People considered as matches
(N " 276) were either (a) those with susceptibility at or above 3
and who received a loss-framed message, or (b) those with susceptibility below 3 and who received a gain-framed message;
5
mismatches (N " 290) were the remaining participants. Indeed, a
significant matching effect was observed, as shown in Figure 3. A
greater proportion of participants who received a matched message
flossed at recommended rates (34.5%, SE " 2%) compared with
those with a mismatched message (27.4%, SE " 2%), odds ratio
(OR) " 1.58, SE " .35, p " .04. Incorporating the no video group
into analyses, planned contrasts showed that matched participants
were more likely to floss at recommended levels than those in the
mismatched and no video groups combined (OR " 1.50, SE " .29,
p " .03), and those in the mismatched group were no more likely
to floss at recommended levels than those in the no video group
(30.3%, SE " 3%; OR " .80, SE " .21, p " .40).
Moderation Among Young Adults
As Table 2 shows, there was no significant moderation of
framing by motivational orientation. Because these results are at
odds with past findings of A Frame & Motivational Orientation
interaction found in mostly young adult samples, we ran an additional GEE model (as described earlier) restricted to participants
younger than 25 (N " 62; 35 females), an age cutoff frequently
used in meta-analytic reviews (e.g., Hagger et al., 2002; Sheeran &
Orbell, 1998). Within this subsample, the Frame & Motivational
Orientation interaction was significant (B " 2.60, SE " 1.20, z "
2.16, p " .031) but the Frame & Susceptibility interaction was not
significant (B " '.14, SE " .36, z " '.40, p " .69).5 Follow-up
regression analyses showed that the Frame & Motivational Orientation predicted flossing at 6 months (B " 5.17, SE " 2.03, z "
2.55, p " .011) but not at 2 months (B " .12, SE " 1.83, z " .07.
p " .95).6 Among approach-oriented young adults, the gainframed video led to significantly greater flossing at 6 months (M "
3.04, SE " .56) than the loss-framed video (M " 0.26, SE " .12),
z " 4.99, p $ .001. However, among avoidance-oriented young
adults, the loss-framed video led to greater flossing (M " 4.34,
SE " .92) than the gain-framed video (M " 2.29, SE " .43), z "
2.13, p " .03. Thus, young adults in this sample exhibited a pattern
of framing effects consistent with prior work using young adults.
Mediation of Behavioral Effects
Because behavioral effects of framing and perceived susceptibility were observed across the 6-month study period, we also
examined whether attitudes, intentions, or self-efficacy might have
3
This Perceived Risk & Framing interaction was not moderated by
ethnicity.
4
To ensure that this finding was not attributable to attrition, we also
conducted a last-observation-carried-forward analysis in which 2-month
flossing values were carried forward to 6-months for participants who
dropped out at 6-months. In this analysis, The Frame & Susceptibility
interaction was likewise significant (B " '.54, p " .014).
5
We also examined the age-dependence of framing effects by including
age as a moderator variable using the full sample. A marginally significant
Frame & Motivational Orientation & Age interaction was found (p "
.086), showing that the congruency effect became stronger with decreasing
age. In contrast, the Frame & Susceptibility & Age interaction was not
significant (p " .31), suggesting that it was largely robust across the age
spectrum of our sample.
6
Both BIS and BAS contributed to this interaction. As BIS increased,
loss frames became more effective (BIS & Frame B " '5.45, p " .04). As
BAS increased, gain frames became somewhat more effective (BAS &
Frame B " 4.63, p " .14).
6
UPDEGRAFF, BRICK, EMANUEL, MINTZER, AND SHERMAN
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Table 2
Generalized Estimation Equation Model Predicting Reported Flossing Across Both the
2- and 6-Month Follow-Ups, Among Participants in the Gain and Loss Video Conditions
Prior flossing
Female
Age
Race: White
Race: Black
Race: Asian
Household income
Gain frame
Susceptibility
Motivational orientation (MO)
Gain Frame & Susceptibility
Gain Frame & MO
Intercept
B
SE
Z
p
1.25
0.41
0.02
'.72
'.27
'.70
'0.03
0.07
0.18
'0.01
'0.29
'0.05
'2.12
0.03
0.12
0.00
0.16
0.17
0.17
0.02
0.11
0.09
0.15
0.13
0.22
0.32
37.89
3.34
6.19
'4.40
'1.55
'4.22
'1.70
0.63
1.98
'0.08
'2.22
'0.23
'6.63
$.01
$.01
$.01
$.01
.12
$.01
.09
.53
.05
.93
.03
.82
$.01
95% CI
1.18
0.17
0.02
'1.03
'.60
'1.03
'0.06
'0.16
0.01
'0.30
'0.54
'0.48
'2.75
1.31
0.65
0.03
'.40
0.07
'0.38
0.00
0.30
0.36
0.28
'0.03
0.38
'1.49
Note. N " 686 with 1,251 observations. Wald #2(12) " 1812.66. The reference group for the race indicator
variables was the numerically largest group (Hispanic). As this model predicts flossing across 6 months while
controlling for baseline flossing, all other coefficients including those for demographic variables represent
change in flossing behavior from baseline through follow-up.
mediated this effect, assessed immediately after the video and at
the 2-month follow-up. None of these tests were significant, suggesting that the interaction between framing and perceived susceptibility was not explained by changes in participants’ stated
attitudes, intentions, or efficacy regarding flossing.
However, another pathway by which perceived susceptibility
and framing might interact to predict flossing is by changing how
much a mediator predicts subsequent flossing: in other words, by
influencing the “intention-behavior gap” (Sheeran, 2002). To test
this possibility, we conducted a regression analysis that predicted
flossing at the 6-month follow-up from 2-month intentions to floss,
framing condition, perceived susceptibility, and their 2- and 3-way
interactions, controlling for flossing at the 2-month follow-up. The
3-way interaction was key, as it represented the interaction between intentions and the 2-way Framing & Susceptibility interaction on subsequent flossing, and it was significant, B " '.19,
SE " .08, p " .02, indicating that the interaction between susceptibility and framing moderated the link between intentions and
behavior.
To simplify interpretation, we collapsed the Frame & Susceptibility interaction into a single variable representing match versus
mismatch, as described earlier. Among those who viewed a
matched video, intentions significantly predicted subsequent flossing, B " .26, SE " .05, p $ .001. Among those who viewed a
mismatched video, intentions predicted subsequent flossing (B "
.13, SE " .05) but to a significantly lesser extent (difference test
Figure 2. Interaction between perceived susceptibility and message frame on self-reported flossing (times/
week) at the 6-month follow-up.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
MESSAGE FRAMING FOR HEALTH
7
Figure 3. Predicted likelihood of daily flossing at the 6-month follow-up, depending on whether participant
received video matched or mismatched to perceived susceptibility, or no video. Bars represent standard errors
of estimates. Planned contrasts show matched video to increase likelihood of daily flossing relative to
mismatched and no video combined, and mismatched video to be no different than no video.
t " 2.58, p " .01). Thus, the interaction between framing and
perceived susceptibility may have influenced long-term flossing
not by raising intentions, but by leading to intentions that more
reliably predicted behavior.
Discussion
Consistent with recent approaches to health message framing,
we found that the effectiveness of gain- and loss-framed messages
depended on individual differences of the message recipient. Most
generally, individuals’ perceived susceptibility to health problems
shaped their responses to framed appeals. This moderation by
susceptibility has been found in other studies of screening behaviors (Ferrer et al., 2012; Gallagher et al., 2011; Hull, 2012), but has
never extended to preventive behaviors or been demonstrated over
as extended of a timeframe as in this study. We showed significantly greater odds of adhering to the American Dental Association’s recommended rate of flossing at the 6-month follow-up
among participants who received a brief intervention in which the
frame was matched to their beliefs about susceptibility, as compared with those who received an intervention with a mismatched
frame or no intervention at all (these groups did not differ significantly). Given that framing is an inevitable part of most health
educational interventions, these findings attest to the benefits of
strategically tailoring the frame of a health behavior intervention to
the relevant health beliefs of the recipient.
Theoretical Implications
This is the first study to show that naturally occurring beliefs
about susceptibility shape people’s responses to framed messages
in promoting an illness prevention behavior such as dental flossing. Framing effects were most pronounced among participants
with low perceived susceptibility, underscoring the importance of
using framing as a way to motivate healthy behavior among people
who may not yet perceive a risk. Although the pattern of framing
was somewhat reversed among participants with high perceived
susceptibility, it was not significant. Prior studies that have examined the interaction between perceived susceptibility and framing
have found the effects either driven by those with low perceived
susceptibility (Hull, 2012, HIV intentions; Lee & Aaker, 2004,
impressions of mono-fighting supplement ads), or by those with
high perceived susceptibility (Ferrer et al., 2012, colorectal cancer
screening intentions; Gallagher et al., 2011, mammography screening). No study has found significant effects of framing at both low
and high levels of susceptibility. Thus, although mounting evidence shows the role of perceived susceptibility in moderating
effects of framed messages, more work is needed to predict exactly
where along the susceptibility spectrum framing effects emerge.
Perhaps differences in how people arrive at perceptions of risk
may explain some of this variability (Gallagher et al., 2011). With
HIV and oral health, people may come to a clear awareness of their
low risk status because they have not engaged in known behavioral
risk factors for HIV or because they show no symptoms of oral
health problems (e.g., pain, bleeding gums). In contrast, risk factors for breast and colorectal cancer are less controllable and less
visible than for HIV or oral health, so even people who report low
perceived susceptibility may not be fully convinced of their risk
status (Gallagher et al., 2011). There are many other factors that
could solidify a person’s perceived susceptibility to a given health
threat— objective risk feedback, genetic testing, regular engage-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8
UPDEGRAFF, BRICK, EMANUEL, MINTZER, AND SHERMAN
ment in preventive behavior—so future research may benefit from
examining the interaction of message framing and perceived susceptibility in contexts that involve any of these factors.
Differential moderation by age. Consistent with prior research (see Sherman et al., 2008 for review), motivational orientation moderated framing effects among young adults but not
among the full sample. Furthermore, perceived susceptibility did
not moderate framing effects among young adults but did among
the full sample. It may be that beliefs about perceived susceptibility to health problems are not particularly grounded beliefs for
young adults, as they typically experience few health problems and
have unrealistic optimism about health risks (cf. Renner, Knoll, &
Schwarzer, 2000; Weinstein, 1984). In situations where people’s
beliefs about a health issue are not particularly strong—as may be
the case with young adults and oral health—their responses to
framed messages may be more influenced by dispositional factors
such as approach/avoidance motivation orientation (Rothman, Wlaschin, Bartels, Latimer, & Salovey, 2008) that lead to a greater
depth of processing of strongly constructed health arguments (Updegraff et al., 2007). Thus, tailoring framed messages to approach/
avoidance motivations may be a useful strategy for interventions
targeting young adults, but may not meet with as much success
when targeting older individuals.
Framing and the intention-behavior gap. Like many prior
framing studies (see Gallagher & Updegraff, 2012, for review), we
did not find evidence that our framing effects were directly mediated by changes in intentions or attitudes toward flossing. Instead,
the Frame & Susceptibility interaction may have been due to a
bridging of the intention– behavior gap. When people viewed a
video that matched their perceived susceptibility to oral health
problems, their intentions more reliably predicted behavior than
when the video was mismatched. This finding is consistent with
research by Hull (2012), who found that the interaction between
perceived susceptibility and framing on intentions to HIV test was
mediated by elaborative processing of the message. When elaboration on a persuasive message is high, attitudes should be more
stable over time and more predictive of behavior than when
elaboration is low (Briñol & Petty, 2006). Thus, our findings
reinforce the argument that matching the frame of a health message to personal factors such as susceptibility or motivational
orientation is a simple technique that increases a recipient’s attention and elaboration on a health message.
Health Implications
These findings have implications for oral health research as
well. Relatively few behavioral interventions have shown longterm improvements in oral health care among adults (Watt &
Marinho, 2005; but see Halvari & Halvari, 2006; Stewart et al.,
1991; Tedesco, Keffer, Davis, & Christersson, 1992 for exceptions). Our Internet-delivered framed videos produced more selfreported flossing behavior over a 6-month period than a no intervention control. Thus, these findings attest to the feasibility of
using the Internet in future oral health interventions. Internetdelivered interventions can be less expensive and wider-reaching
than intensive, in-person treatments.
The long-term behavior change we observed should be interpreted within the context of the manipulation’s brevity: It was just
a 5-min educational video. Other oral health interventions have
been more intensive, utilizing in-person consultation with dental
hygienists over multiple sessions (Halvari & Halvari, 2006; Stewart et al., 1991; Tedesco et al., 1992). These interventions have
shown greater self-reported flossing behavior at 6-month followups compared with no intervention controls, corresponding to
effect sizes ranging from small (d " .24 in Tedesco et al., 1992)
to large (d " 1.2 in Halvari & Halvari, 2006 and d " 2.4 in Stewart
et al., 1991). In comparison, we found a small but significant effect
of our videos on self-reported flossing behavior (d " .15, compared to no video), with the most notable effects emerging when
the frame of the video matched people’s beliefs about perceived
susceptibility (d " .19, compared with no video). Thus, our brief
intervention produced effects comparable with one study of an
in-person intervention (Tedesco et al., 1992), but not as strong of
an effect as interventions that included more than 60 min of
interaction with an oral health professional. However, a tailored,
Internet-administered intervention can have a far greater reach
than in-person oral health interventions. A central challenge for
Internet-administered intervention will be getting and holding an
audience’s attention; we show that matching the frame to beliefs
about susceptibility is one strategy that may help. Thus, matching
an intervention’s frame to beliefs about susceptibility could be
profitably incorporated into other types of tailored interventions,
whether those interventions be brief or extensive, in-person or
online.
Limitations
There are several limitations of this research that should be
noted and addressed in future work. There was considerable attrition in this study. Although there was no difference in attrition
between the gain and loss video conditions, there was more attrition in the no video control condition than in the video conditions.
Thus, comparisons involving the control condition should be interpreted with some caution. As our study relied on self-reports of
flossing, future research may benefit by examining the effects of a
message framing intervention on more objective measures of behavior or on clinical indices of oral health. Furthermore, attitudes
and intentions were assessed with single-item measures. Although
we excluded a few participants who skipped through the video
early, we could not verify that all participants watched the video;
this may have attenuated the influence of the framed videos.
Lastly, our study focused on a general population of Americans,
but due to panel limitations the study did not include a sizable
proportion of American Indians, Alaska Natives, and Native Hawaiians. Future research should evaluate the use of framed interventions in such populations as well as in populations with higher
rates of oral health problems both in the U.S. and internationally
(see Iranians; Pakpour, Yekaninejad, Sniehotta, Updegraff, &
Dombrowski, in press).
Conclusion
Framing is an inevitable part of interventions that seek to
educate people about the consequences of their health behaviors,
and interventions can benefit by strategically using particular
message frames. This study integrates and reconciles two bodies of
work in the health message framing literature, and yields the
following recommendations for the strategic use of framed mes-
MESSAGE FRAMING FOR HEALTH
sages. Among young adults, motivational orientation shapes responses to
framed messages. However, among middle- and older-adult samples, $!----!interventions should focus on perceived susceptibility.
Future research should also examine the issue of how best to
integrate multiple moderators in creating models to examine heterogeneous treatment effects for health psychology interventions.
Doing so will build on the present demonstration that long-term
changes in American adults’ health behavior can be achieved by
matching the frame of a single Internet-administered intervention
message to people’s relevant health beliefs.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Thousand Oaks, CA: Sage.
Albarracin, D., Johnson, B. T., Fishbein, M., & Muellerleile, P. A. (2001).
Theories of reasoned action and planned behavior as models of condom
use: A meta-analysis. Psychological Bulletin, 127, 142–161. doi:
10.1037/0033-2909.127.1.142
American Dental Association (ADA). (2013). Flossing. Retrieved from
www.mouthhealthy.org/en/az-topics/f/flossing
Batchelor, P., & Sheiham, A. (2002). Does perceived risk of oral problems
influence the use of dental services in university entrants? Community
Dental Health, 19, 116 –119.
Beck, J. D., & Offenbacher, S. (2005). Systemic effects of periodontitis:
Epidemiology of periodontal disease and cardiovascular disease. Journal
of Periodontology, 76, 2089 –2100. doi:10.1902/jop.2005.76.11-S.2089
Briñol, P., & Petty, R. E. (2006). Fundamental processes leading to attitude
change: Implications for cancer prevention communications. Journal of
Communication, 56, S81–S104. doi:10.1111/j.1460-2466.2006.00284.x
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral
activation, and affective responses to impending reward and punishment.
Journal of Personality and Social Psychology, 67, 319 –333. doi:
10.1037/0022-3514.67.2.319
Centers for Disease Control (CDC) & National Center for Health Statistics
(NCHS). (2010). National health and nutrition examination survey data.
Hyattsville, MD: CDC.
Davidson, P. L., Rams, T. E., & Andersen, R. M. (1997). Socio-behavioral
determinants of oral hygiene practices among USA ethnic and age
groups. Advances in Dental Research, 11, 245–253. doi:10.1177/
08959374970110020701
Elliot, A. J., & Thrash, T. M. (2002). Approach-avoidance motivation in
personality. Journal of Personality and Social Psychology, 82, 804 –
818. doi:10.1037/0022-3514.82.5.804
Ferrer, R. A., Klein, W. M. P., Zajac, L. E., Land, S. R., & Ling, B. S.
(2012). An affective booster moderates the effect of gain- and lossframed messages on behavioral intentions for colorectal cancer screening. Journal of Behavioral Medicine, 35, 452– 461. doi:10.1007/s10865011-9371-3
Gallagher, K. M., & Updegraff, J. A. (2012). Health message framing
effects on attitudes, intentions, and behavior: A meta-analytic review.
Annals of Behavioral Medicine, 43, 101–116. doi:10.1007/s12160-0119308-7
Gallagher, K. M., Updegraff, J. A., Rothman, A. J., & Sims, L. (2011).
Perceived susceptibility to breast cancer moderates the effect of gainand loss-framed messages on use of screening mammography. Health
Psychology, 30, 145–152. doi:10.1037/a0022264
Gerend, M. A., & Shepherd, M. (2013). Message framing, it does a body
good: Effects of message framing and motivational orientation on young
women’s calcium consumption. Journal of Health Psychology, 18,
1296 –1306. doi:10.1177/1359105312463587
Gerend, M. A., & Shepherd, J. E. (2007). Using message framing to
promote acceptance of the human papillomavirus vaccine. Health Psychology, 26, 745–752. doi:10.1037/0278-6133.26.6.745
9
Hagger, M. S., Chatzisarantis, N. L., & Biddle, S. J. (2002). A metaanalytic review of the theories of reasoned action and planned behavior
in physical activity: Predictive validity and the contribution of additional
variables. Journal of Sport & Exercise Psychology, 24, 3–32.
Halvari, A. E. M., & Halvari, H. (2006). Motivational predictors of change
in oral health: An experimental test of self-determination theory. Motivation and Emotion, 30, 294 –305. doi:10.1007/s11031-006-9035-8
Higgins, E. T. (1997). Beyond pleasure and pain. American Psychologist,
52, 1280 –1300. doi:10.1037/0003-066X.52.12.1280
Hull, S. J. (2012). Perceived risk as a moderator of the effectiveness of
framed HIV-test promotion messages among women: A randomized
controlled trial. Health Psychology, 31, 114 –121. doi:10.1037/a0024702
Jorm, A. F., Christensen, H., Henderson, A. H., Jacomb, P. A., Korten,
A. E., & Rodgers, B. (1998). Using the BIS/BAS scales to measure
behavioural inhibition and behavioural activation: Factor structure, validity and norms in a large community sample. Personality and Individual Differences, 26, 49 –58. doi:10.1016/S0191-8869(98)00143-3
Latimer, A. E., Rivers, S. E., Rench, T. A., Katulak, N. A., Hicks, A.,
Hodorowski, J. K., Higgins, E. T., & Salovey, P. (2008). A field
experiment testing the utility of regulatory fit messages for promoting
physical activity. Journal of Experimental Social Psychology, 44, 826 –
832. doi:10.1016/j.jesp.2007.07.013
Latimer, A. E., Williams-Piehota, P., Katulak, N. A., Cox, A., Mowad, L.,
Higgins, E. T., & Salovey, P. (2008). Promoting fruit and vegetable
intake through messages tailored to individual differences in regulatory
focus. Annals of Behavioral Medicine, 35, 363–369. doi:10.1007/
s12160-008-9039-6
Lee, A. Y., & Aaker, J. L. (2004). Bringing the frame into focus: The
influence of regulatory fit on processing fluency and persuasion. Journal
of Personality and Social Psychology, 86, 205–218. doi:10.1037/00223514.86.2.205
Liang, K-Y., & Zeger, S. L. (1986). Longitudinal data analysis using
generalized linear models. Biometrika, 73, 13–22. doi:10.1093/biomet/
73.1.13
Loos, B. G., Craandijk, J., Hoek, F. J., Wertheim-van Dillen, P. M., & van
der Velden, U. (2000). Elevation of systemic markers related to cardiovascular diseases in the peripheral blood of periodontitis patients. Journal of Periodontology, 71, 1528 –1534. doi:10.1902/jop.2000.71.10
.1528
Mann, T., Sherman, D., & Updegraff, J. A. (2004). Dispositional motivations and message framing: A test of the congruency hypothesis in
college students. Health Psychology, 23, 330 –334. doi:10.1037/02786133.23.3.330
McCaul, K., Glasgow, R., & Gustafson, C. (1985). Predicting levels of
preventive dental behaviors. Journal of the American Dental Association, 111, 601– 605.
Moher, D., Schulz, K. F., & Altman, D. G. (2001). The CONSORT
statement: Revised recommendations for improving the quality of reports of parallel-group randomised trials. The Lancet, 357, 1191–1194.
doi:10.1016/S0140-6736(00)04337-3
Nan, X. (2012). Communicating to young adults about HPV vaccination:
Consideration of message framing, motivation, and gender. Health Communication, 27, 10 –18. doi:10.1080/10410236.2011.567447
Pakpour, A. H., Yekaninejad, M. S., Sniehotta, F. F., Updegraff, J. A., &
Dombrowski, S. U. (in press). The effectiveness of gain- versus lossframed health messages in improving oral health in Iranian secondary
schools: A cluster-randomized controlled trial. Annals of Behavioral
Medicine.
Renner, B., Knoll, N., & Schwarzer, R. (2000). Age and body make a
difference in optimistic health beliefs and nutrition behaviors. International Journal of Behavioral Medicine, 7, 143–159. doi:10.1207/
S15327558IJBM0702_4
Rothman, A. J., & Updegraff, J. A. (2010). Specifying when and how gainand loss-framed messages motivate healthy behavior: An integrated
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10
UPDEGRAFF, BRICK, EMANUEL, MINTZER, AND SHERMAN
approach. In G. Keren (Ed.), Perspectives on framing (pp. 257–278).
London, UK: Psychology Press/Taylor & Francis.
Rothman, A. J., Wlaschin, J., Bartels, R. D., Latimer, A. E., & Salovey, P.
(2008). How persons and situations regulate message framing effects:
The study of health behavior. In A. J. Elliot (Ed.), Handbook of approach and avoidance motivation (pp. 475– 486). Mahwah, NJ: Erlbaum.
Sallis, J. F., Prochaska, J. J., & Taylor, W. C. (2000). A review of
correlates of physical activity of children and adolescents. Medicine and
Science in Sports and Exercise, 32, 963–975. doi:10.1097/00005768200005000-00014
Sheeran, P. (2002). Intention—Behavior relations: A conceptual and empirical review. European Review of Social Psychology, 12, 1–36. doi:
10.1080/14792772143000003
Sheeran, P., & Orbell, S. (1998). Do intentions predict condom use?
Meta-analysis and examination of six moderator variables. British Journal of Social Psychology, 37, 231–250. doi:10.1111/j.2044-8309.1998
.tb01167.x
Sherman, D. K., Brick, C., Binning, K. R., Mintzer, R. E., McCully, S. E.,
& Updegraff, J. A. (2014). Culture and health communication: The role
of cultural exposure in health message framing. Manuscript in preparation.
Sherman, D. K., Mann, T., & Updegraff, J. A. (2006). Approach/avoidance
motivation, message framing, and health behavior: Understanding the
congruency effect. Motivation and Emotion, 30, 164 –168. doi:10.1007/
s11031-006-9001-5
Sherman, D. K., Updegraff, J. A., & Mann, T. (2008). Improving oral
health behavior: A social psychological approach. Journal of the American Dental Association, 139, 1382–1387.
Stewart, J., Jacobs-Schoen, M., Padilla, M., Maeder, L., Wolfe, G., &
Hartz, G. (1991). The effect of a cognitive behavioral intervention on
oral hygiene. Journal of Clinical Periodontology, 18, 219 –222. doi:
10.1111/j.1600-051X.1991.tb00418.x
Tedesco, L., Keffer, M., Davis, E. L., & Christersson, L. A. (1992). Effect
of a social cognitive intervention on oral health status, behavior reports,
and cognitions. Journal of Periodontology, 63, 567–575. doi:10.1902/
jop.1992.63.7.567
Ulleberg, P., & Rundmo, T. (2003). Personality, attitudes and risk perception as predictors of risky driving behaviour among young drivers. Safety
Science, 41, 427– 443. doi:10.1016/S0925-7535(01)00077-7
Updegraff, J. A., Sherman, D. K., Luyster, F. S., & Mann, T. L. (2007).
The effects of message quality and congruency on perceptions of tailored health communications. Journal of Experimental Social Psychology, 43, 249 –257. doi:10.1016/j.jesp.2006.01.007
Uskul, A. K., Sherman, D. K., & Fitzgibbon, J. (2009). The cultural
congruency effect: Culture, regulatory focus, and the effectiveness of
gain- vs. loss-framed health messages. Journal of Experimental Social
Psychology, 45, 535–541. doi:10.1016/j.jesp.2008.12.005
Watt, R. G., & Marinho, V. C. (2005). Does oral health promotion improve
oral hygiene and gingival health? Periodontology, 37, 35– 47. doi:
10.1111/j.1600-0757.2004.03796.x
Weinstein, N. D. (1984). Why it won’t happen to me: Perceptions of risk
factors and susceptibility. Health Psychology, 3, 431– 457. doi:10.1037/
0278-6133.3.5.431
Zhao, G., & Pechmann, C. (2007). The impact of regulatory focus on
adolescents’ response to antismoking advertising campaigns. Journal of
Marketing Research, 44, 671– 687. doi:10.1509/jmkr.44.4.671
Received October 29, 2013
Revision received March 23, 2014
Accepted March 26, 2014 !