r Academy of Management Journal
2019, Vol. 62, No. 5, 1609–1642.
https://doi.org/10.5465/amj.2014.0795
MICROFOUNDATIONS OF CORPORATE SOCIAL
RESPONSIBILITY AND IRRESPONSIBILITY
CATHERINE T. SHEA
Carnegie Mellon University
OLGA V. HAWN
University of North Carolina, Chapel Hill
This study examines the importance of social perception of corporate social responsibility
(CSR) and irresponsibility (CSI). Drawing from social psychology literature on stereotypes,
we argue that two fundamental dimensions of social perception—warmth and
competence—help explain the underlying processes and conditions under which CSR
leads to specific outcomes. We propose that firms engaging in CSR are perceived as higher
in warmth and, by default, competence; moreover, different perceptions of the organization’s warmth and competence can moderate CSR rewards and CSI penalties. To demonstrate this, we conduct three experiments. Experiment 1 links CSR with perceptions of
warmth and competence, showing that warmth perceptions mediate the relationship between CSR and important outcomes, such as purchase intentions and reputation. Experiment 2 adds information on firms’ countries of origin, revealing that CSR rewards and CSI
penalties differ depending on the (mis)alignment of CSR strategy with country stereotypes.
Experiment 3 replicates these findings using behavioral paradigms. We find that firms
from high-warmth countries (the United States, Sweden, Portugal) receive lower CSR
rewards and pay higher CSI penalties than firms from low-warmth countries (Germany,
Pakistan) but this effect is moderated by competence. Our micro–macro study advances
social evaluation, strategic CSR, and international management literatures.
McGuire, Sundgren, & Schneeweis, 1988; Orlitzky,
Schmidt, & Rynes, 2003; Russo & Fouts, 1997)
and other outcomes (Berrone & Gomez-Mejia, 2009;
Flammer, 2013; Turban & Greening, 1997; Yoon,
Gürhan-Canli, & Schwarz, 2006). Yet CSR is primarily
studied at the macro level (i.e., institutional or organizational level) compared to the micro level (i.e.,
individual level) of analysis: a recent review of the
CSR literature shows that only 4% of all studies examine the individual level, while only 5% address
CSR at two or more levels of analysis (Aguinis &
Glavas, 2012). Therefore, what is lacking in CSR research is deeper appreciation, at the individual level,
of how CSR makes an impact (Wang et al., 2016).
Accordingly, this paper addresses the need for micro
studies of CSR with a multilevel approach (Morgeson,
Aguinis, Waldman, & Siegel, 2013) in order to understand the underlying processes (i.e., mediating
effects) and conditions under which (i.e., moderating
effects) CSR leads to specific outcomes.
In particular, we still know very little about the
microfoundations of CSR (for the most recent review,
see Gond, El Akremi, Swaen, & Babu, 2017): how CSR
and, importantly, corporate social irresponsibility
Corporate social responsibility (CSR) has grown
markedly in the past decade both as an important
phenomenon in practice and as a critical field in
academia (Wang, Tong, Takeuchi, & George, 2016).
We have theorized about its institutional (Campbell,
2007; Ioannou & Serafeim, 2012) and organizational
drivers (Aguilera, Rupp, Williams, & Ganapathi,
2007; McWilliams & Siegel, 2001) as well as examined its effect on firm performance (Berman, Wicks,
Kotha, & Jones, 1999; Cochran & Wood, 1984;
The authors thank anonymous reviewers and conference participants at 2014 Academy of Management,
Academy of International Business, and Strategic Management Society annual meetings, as well as Sinziana
Dorobantu, Sanjay Patnaik, Elena Kulchina, Nel Dutt,
Elena Vidal, Amandine Ody-Brasier, Marlo Raveendran,
Zach Burns, and the brown-bag participants at Boston
University for comments on the early version of this paper.
An earlier version of this manuscript was a Finalist for
2014 AIB Haynes Prize for the Most Promising Scholar.
Please address all correspondence to Catherine Shea:
Tepper School of Business, Carnegie Mellon University,
5000 Forbes Avenue, Pittsburgh, PA, 15213, email:
ctshea@andrew.cmu.edu
1609
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express
written permission. Users may print, download, or email articles for individual use only.
1610
Academy of Management Journal
(CSI) (Lange & Washburn, 2012; Surroca, Tribó, &
Zahra, 2013) are perceived by individuals, and what
effect this perception might have on the relationship
between CSR and specific outcomes at individual
and organizational levels of analysis. For example, we
know that CSR positively affects company reputation
(Turban & Greening, 1997)—one of our outcomes—
but why and what mediates this relationship at the
individual level is still unclear. A small but growing
literature on the micro approaches to CSR shows that
involvement in CSR positively influences employee
performance, behaviors, and attitudes (Burbano,
2016), such as employee engagement (Caligiuri,
Mencin, & Jiang, 2013), organizational citizenship
behavior (Rupp, Shao, Thornton, & Skarlicki, 2013),
identification with the firm (Farooq, Rupp, & Farooq,
2017), retention (Jones, 2010), in-role performance,
and commitment, as well as attractiveness to prospective employees (Jones, Willness, & Madey, 2014;
Turban & Greening, 1997). Yet, few studies go beyond
employees (i.e., providing information on other
stakeholders) or seek to unpack the processes of CSR
evaluations by individuals more generally (Gond
et al., 2017). Our study applies to a larger number
of stakeholders—which may include employees,
customers, environmentalists, suppliers, the community as a whole, and owners/shareholders—because it
examines general individual judgments and perceptions of warmth and competence.
Warmth and competence serve as universal dimensions of social judgment (Cuddy, Fiske, & Glick,
2004; Fiske, Cuddy, & Glick, 2007; Fiske, Cuddy,
Glick, & Xu, 2002). “Warmth” is defined as perceptions related to intent, including friendliness, trustworthiness, helpfulness, sincerity, and morality,
whereas “competence” is conceptualized as perceptions related to ability, including skill, intelligence,
creativity, and efficacy (Cuddy, Fiske, & Glick, 2008).
With the help of these two dimensions, one can portray social perceptions of activities, individuals, organizations, and even countries (Cuddy, Fiske, &
Glick, 2007). Moreover, social judgment of various
degrees (i.e., high or low) of warmth and/or competence can predict a distinct emotion and behavior
toward a target; for example, being perceived as high
on warmth and/or competence is beneficial, while
being low on one of the dimensions is costly (Aaker,
Vohs, & Mogilner, 2010). Importantly, compared to
other potential mechanisms, warmth and competence help predict specific behavioral outcomes from
these social perceptions (e.g., helping those higher in
warmth; Cuddy et al., 2007), adding a more nuanced
mechanism between CSR, social perception, and
October
behavior toward the firm, which can ultimately affect
more macro-level outcomes (e.g., reputation, sales,
stakeholder and shareholder value).
While previous research mainly examines CSR, we
apply these two fundamental dimensions of social
perception to understand the microfoundations of CSR
and CSI. Naturally, our basic argument is that firms
with CSR generate a greater warmth and (by default,
without any other information) competence perception than control firms or firms engaging in CSI.
We extend this logic to argue that this perception of
warmth in turn mediates CSR and CSI effects on various outcomes (i.e., CSR rewards and CSI penalties—
the effects for CSR and CSI above and below the control
condition). Importantly, we then argue that different
perceptions of warmth and competence of the organization from other sources (in this paper, based on its
country of origin) moderate CSR rewards and CSI
penalties depending on the (mis)alignment of CSR
strategy with the (home country) stereotype. Using
three experiments with 774 participants and the
warmth and competence variation across five countries (i.e., the United States, Sweden, Germany, Portugal, Pakistan), we show that warmth and competence
perceptions explain causal effects of CSR and CSI on
purchase intentions, price, reputation, and quality assessments. Moreover, consistent with the stereotype
content model (SCM), firms from high-warmth countries pay a higher price for CSI than firms from lowwarmth countries (Germany, Pakistan), yet this effect
changes when combined with perceptions of high
competence (the United States, Sweden).
Our study has several important theoretical implications for the management literature on CSR,
social cognition, and international business, as well
as practical applications. First, we develop a deeper
understanding of the microfoundations of CSR and
CSI—by examining their social perception by individuals. We extend existing individual-level CSR
research that relies on a different set of theoretical
frameworks, such as system justification theory
(Hafenbrädl & Waeger, 2016), organizational justice,
social influence, needs, and self-determination theories (Aguinis & Glavas, 2012), by drawing from the
social psychology research on social perception and
stereotypes and arguing that CSR influences and is
influenced by two fundamental dimensions of social
perception—warmth and competence—which mediate and moderate the effects of CSR (and CSI) on
different outcomes. Our work is important because,
in comparison to objective CSR ratings, the “subjective” evaluations of CSR likely matter more for
individual reactions to CSR (Rupp et al., 2013),
2019
Shea and Hawn
which in turn lead to different outcomes (Gond et al.,
2017). For example, employees’ exposure to CSR
initiatives does not directly translate into favorable
CSR attitudes (Glavas & Godwin, 2013). Thus, it is
important to understand how the social perception
of CSR is formed more generally.
Second, we advance prior literature examining
different mediators and moderators of CSR at the
individual level, such as trust (or whether, how, and
when consumers’ perceptions of motives directly
influence consumer responses to CSR) (Vlachos,
Tsamakos, Vrechopoulos, & Avramidis, 2009), customer satisfaction (Luo & Bhattacharya, 2006), and
organizational pride and identity (Jones, 2010; Jones
et al., 2014), by conducting a multidisciplinary
multi-experimental analysis of the role of warmth
and competence in the CSR context. If we understand how the average individual perceives CSR and
CSI and how these social perceptions may coincide
(or diverge) with (from) their expectations (stereotypes) of organizations, we can understand more
about how the value from CSR is generated more
broadly for a variety of stakeholders. In addition,
since most firms try to avoid CSI, we can understand
which firms will not suffer from CSI as much as
others and why.
Third, while most prior studies focus on only one
organizational practice, we distinguish between CSR
and CSI—crucial for moving the CSR literature forward. Another important contribution to the CSR literature, prone to endogeneity and causality issues,
is empirical: by utilizing controlled experimental
methods, we are able to isolate discrete mechanisms
linking CSR/CSI to organizational outcomes. This
gives managers unequivocal insights into which “levers to pull” when activating a CSR/CSI strategy based
on their current organizational context, and, importantly, this helps specify conditions under which CSR/
CSI does not reap the expected benefits/costs. Finally,
we extend the international management literature by
showing that initial social judgments about the origin
of the firm may improve or worsen outcomes based on
the choice of CSR strategy. This has important practical implications, particularly for managers of firms
that expand abroad for the first time.
THEORY AND HYPOTHESES
Social Perception
The ability to quickly judge another individual is a
fundamental evolutionary skill. It helps us to ascertain whether “they” are a part of our social group and
1611
to assess the goodness of their motives and competence to enact these good motives (see Fiske et al.,
2007, for a review). Despite common beliefs that we
need (and use) a wealth of knowledge to form judgments about other individuals, research shows that
we make trait inferences spontaneously (Newman &
Uleman, 1993; Winter & Uleman, 1984; Winter,
Uleman, & Cunniff, 1985). For instance, when observing a fictional character, Donald, help an old
lady cross the street, we quickly (and somewhat
permanently) conclude that Donald is a kind and
helpful individual—despite our having only limited
information about him. Related work on “thin slicing” shows that a mere 30 seconds watching a college
professor teach is sufficient to predict their end-ofterm teaching evaluations (see Ambady & Rosenthal,
1992, for a review). Thus, despite the wealth of social
information at our disposal, individuals primarily
rely on quick social judgments of others, and filter
all subsequent social information based—barring a
major behavioral change—on the first impression of
someone (Newman & Uleman, 1993; Ross, Lepper, &
Hubbard, 1975).
Many of these quick social judgments about other
individuals result from stereotypes. Stereotypes are
cognitive beliefs about the characteristics of another
group (Fiske, 1998). Fiske and colleagues (Cuddy
et al., 2008; Fiske et al., 2007; Fiske et al., 2002)
identified two fundamental dimensions of all stereotypes used to evaluate people and social groups:
warmth and competence. The SCM posits that all
social groups fit into one of four quadrants based on
whether they are high or low on warmth and competence. Further research (Cuddy et al., 2008; Fiske
et al., 2007; Fiske et al., 2002; Fiske, Xu, Cuddy, &
Glick, 1999) has demonstrated the robustness of the
warmth and competence dimensions, and the fact
that they can be used to classify not only individuals
but also social groups (e.g., elderly people, Jews,
housewives, immigrants, and the homeless) and even
national cultures.
Importantly, this work on social perception has a
behavioral analog, the “behaviors from intergroup
affect and stereotypes” (BIAS) map framework
(Cuddy et al., 2007). Based on the dimensions of
warmth and competence, the BIAS map predicts not
only cognitions derived from group stereotypes but
also specific behaviors toward the group. Warmth and
competence evoke active and passive behaviors, respectively: in particular, while high-warmth targets
are helped, low-warmth targets are harmed, and
whereas high-competence targets are passively facilitated, low-competence targets are neglected (Cuddy
1612
Academy of Management Journal
et al., 2007). Furthermore, each combination of the
two trait dimensions predicts a distinct emotion or
behavior toward the target: people admire those who
are high in both competence and warmth, they feel
contempt toward those who are low competence and
low warmth, they envy those who are competent but
not warm, and they pity those who are incompetent
but warm (Fiske et al., 2002; Lee & Fiske, 2006).
Firms as Subject to Stereotypes
Although originally developed to explain personal
and social group perception, the SCM has been extrapolated to non-human objects. Kervyn, Fiske,
and Malone (2012), in their “brands as intentional
agents” framework, showed that consumers perceive
brands in the same way they perceive people. Cuddy
et al. (2007) used the SCM to map social perceptions
of European Union countries. Aaker et al. (2010)
applied the SCM to organizations, explaining forprofit and not-for-profit firms’ success and failure,
and showing that not-for-profit organizations are
associated with higher warmth and lower competence than for-profit firms.
It is not surprising that organizations are often
anthropomorphized into human beings—famously
so in the movie The Corporation and in recent decisions by the U.S. Supreme Court, such as in the
Citizens United case. In management studies, organizations are often perceived to possess actions,
thoughts, opinions (Knobe & Prinz, 2008), goals,
tastes, styles, personalities (Pfeffer, 1981), and even
attention (Ocasio, 1997). In fact, they are viewed as
social actors similar to individuals precisely because
the features that distinguish humans as actors are
functionally equivalent to the features common to
organizational actors (King, Felin, & Whetten, 2010).
Yet, when compared to human beings, organizations
are thought to have equal capability for agentic
(i.e., competence) behavior (Gray & Wegner, 2010;
Haran, 2013; Knobe & Prinz, 2008) but are less likely
to be seen as experiencing emotions and feelings
(Gray & Wegner, 2010).
Nonetheless, a large literature in marketing demonstrates that people perceive brands to have personalities (Aaker, 1997) and that people form
“relationships” with various brands and products
(Aaker & Fournier, 1995; Fournier, 1998). We build
off this previous literature by going beyond brands
and products and examining the perceptions of organizational practices. Specifically, in this paper, we
will compare perceptions of socially responsible
and irresponsible organizations, predicated on
October
social perception dimensions developed to classify
individuals and social groups. Before we discuss
how socially (ir)responsible activities may generate
social perceptions of warmth and/or competence,
we need to define CSR and CSI.
Main Effect: CSR, CSI, and Social Perception
CSR, sustainability, corporate citizenship, and
other terms are generally used to describe a portfolio
of socioeconomic activities, including environmental, social, and corporate governance actions of the
firm (Gardberg & Fombrun, 2006). Because these
voluntary actions are aimed at improving social or
ecological conditions (McWilliams & Siegel, 2001),
many observers regard CSR as an activity that benefits firms, markets, and societies (Orlitzky, 2013).
We propose that CSR represents organizational
behavior that connotes warmth, and that, as such, it
should lead to perceptions of greater warmth. Why
should CSR be associated with warmth? Individuals
and social groups associated with high warmth are
associated with behaviors that are trustworthy and
moral (Fiske et al., 2007). Additionally, warmth has
both a moral dimension (e.g., good intentions) as well
as a relational component (e.g., can successfully work
with allies) (Goodwin, 2015). By definition, CSR includes a trustworthy behavior (i.e., “responsible” in
its name); furthermore, CSR involves working with
others (e.g., stakeholders; Harrison & Freeman, 1999).
Prior literature distinguishes CSR—social actions
motivated by moral obligation—from corporate social
performance—social actions of firms—to highlight
that CSR is a moral behavior (Baron, 2009).
In turn, to distinguish between CSR and CSI, we will
follow Campbell’s (2007) threshold: if corporations
(a) knowingly do something that could harm their
stakeholders—their investors, employees, customers,
suppliers, or the local community within which they
operate—and (b) do not rectify the harm caused by
them (whenever it is discovered and brought to their
attention), the minimum behavioral standard with respect to the corporation’s relationship to its stakeholders is broken, and such corporate behavior
becomes socially irresponsible. For example, CSI behavior includes using child labor, sweatshops, and
polluting facilities in manufacturing operations. CSR
behavior, on the other hand, includes charity, volunteering, community engagement, fair labor practices,
and environmentally friendly manufacturing facilities.
In comparison to CSI and other organizational behavior, CSR behavior resembles the same set of attributes
that social psychologists traditionally associate with
2019
Shea and Hawn
being high in warmth. Therefore, as our baseline hypothesis, we propose:
Hypothesis 1. Ceteris paribus, firms engaging in CSR
(CSI) will be perceived as having higher (lower) levels
of warmth.
Mediating Effect: Spillover and Primacy Effects of
Warmth Judgments
Warmth perceptions have two effects that are
worth discussing: halo effects (Nisbett & Wilson,
1977) and the primacy of warmth over competence.
First, let us discuss the “halo effect”—that is, when
the presence (or lack) of warmth spills over into our
judgments of competence (Cuddy et al., 2008; Fiske
et al., 2007; Singh & Teoh, 2000; Tausch, Kenworthy,
& Hewstone, 2007). This effect is particularly salient
when we have no information beyond that on
warmth. Thus, we automatically assume—in the
absence of further information—that an individual
possessing warmth also possesses some degree of
competence.
In our setting, this will mean that, in comparison to
CSI firms, in the absence of further information, because CSR firms are associated with higher warmth
they will also be associated with higher competence.
We argue that this will be the case because the limited information on CSR practices suggests to the
evaluator that the organization mastered at least
one skill and therefore is intelligent and competent.
Competence is conceptualized as perceptions related to ability, including skill, intelligence, creativity, and efficacy; therefore, in comparison to firms
engaging in CSI demonstrating their inability to be a
corporate citizen and behave in socially responsible
ways, the perception of competence will be greater
for firms engaging in CSR.
This prediction is consistent with recent work in
consumer research: products of companies engaged
in prosocial activities are perceived as performing
better—due to the moral undertone of the company’s motivation for engaging in socially responsible behavior; more importantly, this effect
holds even when consumers can directly observe
and experience the product and when the acts of
social goodwill are unrelated to the company’s core
business (Chernev & Blair, 2015). Another alternative explanation for why CSR engagement may be
perceived to be high in competence is that, in comparison with CSI, it can improve firm reputation,
performance, and other outcomes (Choi & Wang,
2009; Yoon et al., 2006) traditionally associated with
competence.
1613
A counterargument will require a discussion of the
purpose of the firm (Friedman, 1970), market actors as
social evaluators, and more information on other organizational practices, activities, or performance. For
example, if evaluators were market actors (particularly
in earlier years), they might have assumed that firms
engaging in CSR were less competent because CSR
investment requires diverting scarce resources from
other more strategic or core business activities of the
firm. However, in comparison to CSI, recent strategy
work shows that CSR engagement is now making organizations more competent because they perform
better even in financial markets (Cheng, Ioannou, &
Serafeim, 2013; Eccles, Ioannou, & Serafeim, 2014;
Ioannou & Serafeim, 2015). Therefore, due to the “halo
effect” of warmth, as well as other alternative explanations listed above, we propose:
Hypothesis 2. Ceteris paribus, firms engaging in CSR
(CSI) will be perceived as having higher (lower) levels
of competence.
Second, let us discuss the primacy effect of
warmth judgment. Although the SCM is predicted on
two dimensions—warmth and competence—when
it comes to forming a social perception, a large body
of research highlights the primacy of warmth
(Cacioppo, Gardner, & Berntson, 1997; Kenworthy
& Tausch, 2008; Wojciszke, Bazinska, & Jaworski,
1998). That is, we anchor on perceptions of warmth
and adjust, albeit insufficiently, based on perceptions of competence. At the individual level, this
happens because information needed to determine
warmth (e.g., facial expressions, interpersonal skills)
is simply more readily available during the beginning of social interactions than information needed
to determine competence (e.g., skills, knowledge,
and abilities). Hence, when it comes to initial social
judgments, warmth information carries more weight
than competence information (Cuddy et al., 2008:
89–92; Singh & Teoh, 2000; Tausch et al., 2007).
At the level of organizations, we argue that this
primacy effect will be important in the relationship
between CSR and organizational outcomes, particularly those that involve a single immediate evaluation or a first encounter with the firm, because, just
like at the individual level, information needed to
determine warmth is simply more readily available
during the beginning of social interactions with this
social actor (King et al., 2010) than information
needed to determine competence. CSR studies have
long examined the effect of CSR on firm performance
(Cochran & Wood, 1984; Waddock & Graves, 1997)
among other outcomes (Sen & Bhattacharya, 2001;
1614
Academy of Management Journal
Turban & Greening, 1997), generally finding a significant
positive effect (Margolis, Elfenbein, & Walsh, 2009).
We distinguish between outcomes that involve a
single immediate evaluation (i.e., a purchase intention or reputation assessment based on the first
encounter with the firm) and those involving an
ongoing evaluation (e.g., by market analysts and
investors who collect information over time, accumulate more information contributing to a competence perception of the firm, and value competence
over warmth in their judgment).
We argue that, for first-time encounters, the
warmth perception that CSR generates could in fact
act as the mechanism by which CSR affects these
outcomes. In particular, we suggest that, in the absence of further information, the primacy of warmth
over competence will play a mediating role in the
relationship between CSR and the outcome of the
first encounter. We previously argued that, when
individuals evaluate CSR, they perceive greater
levels of warmth and, by default, competence than
when they evaluate CSI (Hypotheses 1 and 2). Now,
we argue that, when observers make single immediate evaluations of CSR/CSI behavior, they will pay
more attention to the warmth than to the competence
perception. There are three reasons for this. First,
cognitively, people prove more sensitive to warmth
information than to competence information. Second, they judge warmth faster than they do competence (Cuddy et al., 2008: 90). Third, higher warmth
leads to higher levels of helping behavior (Cuddy
et al., 2007); in an organizational context, help can be
conceptualized as positive evaluations and willingness to purchase from the firm. Therefore:
Hypothesis 3. Warmth perceptions will mediate the
relationship between CSR/CSI and outcomes that
involve single immediate evaluation.
Moderating Effect: Stereotype Fulfillment
and Violation
While Hypotheses 1–3 posit CSR as a source of
warmth and, by default, competence perception,
and warmth as a potential mediator between CSR
and outcomes that involve single immediate evaluation, we next discuss how other sources of
warmth perception in an organizational context can
interact with CSR and potentially moderate the relationship between CSR/CSI and organizational
outcomes. In doing so, we assume, based on prior
literature, that engaging in CSR leads to rewards
(Margolis et al., 2009) and engaging in CSI leads to
penalties for the focal organization (Salaiz, 2016).
October
Thus, we theorize about organizational contexts in
which the rewards/penalties to a CSR/CSI strategy
are amplified/mitigated, answering the following
question: “Is it always useful to engage in CSR and/
or avoid CSI?”
We propose that the answer to this question is
grounded in the literature on stereotype fulfillment
and violation. When we evaluate people from other
social groups or cultures, stereotypes act as cognitive
shortcuts. For instance, we meet a rule-abiding, formal German and our stereotype of German people
is confirmed (i.e., high competence, low warmth;
Cuddy et al., 2007). But, what happens when we
meet a bubbly, scatterbrained German? Social judgments broadly fall into two categories: assimilation
and contrast (Biernat, 2005; Newman & Uleman,
1993; Sherif & Hovland, 1961). “Assimilation” occurs when we judge the target according to the held
stereotype (e.g., rule-abiding, formal German), and
“contrast” occurs when we differentiate the target
from our traditionally held stereotype (e.g., bubbly,
scatterbrained German). In the absence of contradictory evidence, assimilation is relatively automatic (Dijksterhuis, Spears, & Lépinasse, 2001);
otherwise, contrast takes place.
The effect of contrasting social judgments can
be both positive and negative. For instance,
professionals—perceived as highly competent—
receive differential treatment based on their gender
after having a baby (Cuddy et al., 2004). Professional
women are seen as less competent and do not experience a boost in perceived warmth from motherhood to make up for their newly perceived lack of
competence. Men, on the other hand, receive increased perceptions of warmth and maintain their
competence, gaining a net benefit from parenthood
(Cuddy et al., 2004). Likewise, agentic women—
violating the female stereotype—experience a backlash when they apply for feminized jobs, whereas
prototypical women do not (Rudman & Glick, 1999,
2001). These studies suggest that additional information on other sources of warmth and/or competence perception for the subject of interest (e.g.,
gender) can help explain the benefits and detriments
of certain behavior. The mechanism is that additional
information may strengthen the original stereotype
through assimilation, or change it through contrast.
In order to examine under what conditions CSR
and CSI behaviors generate greater (lower) outcomes, we add another piece of information about
the firm that exogenously changes perceptions of its
warmth and competence. As discussed in Studies 2
and 3, we achieve this by adding cues on the firm’s
2019
Shea and Hawn
origin; Cuddy and colleagues (2007) portrayed stereotypes of countries on the two dimensions of
warmth and competence, finding significant variation. In this section, to further our understanding of
CSR as a warmth strategy, we mainly focus on the
warmth dimension of the firm (or its origin); this
helps to disentangle the mediating or supplemental
warmth effects, if any, of CSR. This also helps address a potential reverse-causality issue: if firms
perceived to be high in warmth are the ones who
engage in CSR in the first place, this section of our
theory (and analysis) helps to identify the leftover
effect, if any, of CSR as a warmth strategy.
We are interested in how social perception of the
firm based on other sources of information (i.e., its
origin) moderates CSR and CSI outcomes. In particular, if firms that engage in CSR are perceived to
be higher in warmth and this warmth perception
mediates the effect of CSR on our outcomes (Hypotheses 1 and 3), does an exogenous increase
(decrease) in warmth of the firm help improve (attenuate) these outcomes for firms engaging in CSI
(CSR)? We suggest that, because of assimilation
with the existing stereotype (i.e., a match between
low-warmth country of origin and low-warmth
practice of CSI), firms perceived to be low in
warmth will be forgiven for CSI behavior, and,
therefore, will not be punished as harshly. Highwarmth firms, on the other hand, assimilate with
CSR, not CSI, so, when they engage in socially irresponsible practices, contrast occurs and the penalty is high (i.e., a mismatch between high-warmth
country of origin and low-warmth practice of CSI
results in greater punishment).
A similar logic applies to CSR: if a high-warmth
firm engages in CSR, this demonstrates stereotypical
behavior for this kind of a firm (i.e., assimilation
occurs); hence, the benefits are mediocre. If a lowwarmth firm engages in CSR, on the other hand, that
contradicts the stereotype and, as a result, generates
greater rewards from the surprised but positive reaction of the firm’s observers. Importantly, stereotype match and violation based on the fit between the
social perception of the firm and its behavior (CSR/
CSI) help advance our understanding of the cognitive mechanisms behind the efficacy of CSR investments, suggesting that not all firms engaging in
CSR/CSI reap the same benefits/costs.
Hypothesis 4. The higher the perception of warmth of
the firm, the higher the penalties for CSI.
Hypothesis 5. The higher the perception of warmth of
the firm, the lower the rewards for CSR.
1615
Our key hypothesis is that CSR is predominantly
perceived as influencing warmth. However, individuals do not only evaluate organizations on warmth
but also competence, and thus we need to consider
competence in our predictions. Prior research found
that, although not-for-profit organizations were perceived to be high in warmth, consumers were less
willing to buy from them unless they perceived them
to be highly competent (Aaker et al., 2010). Therefore,
we speculate that the presence of high (low) levels
of perceived competence in addition to high (low)
warmth may affect the above relationships.
As we consider CSR and CSI, it is important to examine how the valence of these actions interacts with
both warmth and competence. While warmth has a
moral dimension (e.g., good intentions) and a relational component (e.g., can successfully work with
allies) (Goodwin, 2015), competence relates directly to
ability to execute a plan, positive or negative. In the
case of CSI, organizations are behaving negatively, and
we know that negative news generates stronger observer reactions than positive events in the CSR context
(Lange & Washburn, 2012). As per Hypothesis 1, these
negative events should bear negatively on warmth,
particularly the moral aspects of warmth (Goodwin,
Piazza, & Rozin, 2014), thus canceling out the positive
effects of warmth on organizational outcomes. Moreover, negative behaviors are diagnostic of competence
(Fiske et al., 2002), or lack thereof, and, per Hypothesis
2, CSI should decrease competence evaluations. Thus,
a firm caught engaging in CSI will be perceived to be
low on both competence and warmth.
However, when additional information on warmth
and competence stemming from other sources than
organizational behavior (e.g., origin) is revealed, it
may change the observers’ reaction to CSI, and, in
fact, if the levels of both warmth and competence are
high from this additional source, it could potentially
buffer the firm against negative effects of CSI (Koh,
Qian, & Wang, 2014; Shiu & Yang, 2017). This is once
again due to the stereotype fulfillment and violation
mechanism: observers expect high-competence–
high-warmth firms to solve CSI issues much faster
than low-competence–high-/low-warmth firms, and
therefore do not punish them as harshly for CSI.
Therefore, high competence and warmth of the home
country of the firm that engages in CSI will buffer it
from the negative effects of its poor behavior when
we assess both dimensions of social perception
(compared to firms high only on warmth).
In turn, in the case of CSR, if a firm from a lowcompetence–low-warmth country is found to engage
in CSR, observers may perceive it as an exemplar in
1616
Academy of Management Journal
that country (due to the stereotype violation on both
warmth and competence dimensions). As a result, this
perception will increase their helping behaviors and
positive evaluations toward such firms because they
are doing the “most good” (namely, helping those who
need it the most). Moreover, because CSR engagement
is costly—as it may divert scarce resources from other
more strategic or core business activities of the firm
(Friedman, 1970)—engaging in CSR when low in both
warmth and competence provides a stronger positive
signal compared to firms low only on warmth. To
summarize, we predict that the simultaneous presence of high warmth and high competence will lower
the negative impact of CSI (attenuating Hypothesis 4),
while the simultaneous presence of low warmth and
low competence will amplify the positive effect of
CSR (strengthening Hypothesis 5).
Hypothesis 6. The perception of competence of the
firm moderates the relationships in Hypotheses 4 and
5, such that (a) the higher the perception of warmth
and competence of the firm (as opposed to lower), the
lower the penalties for CSI, and (b) the lower the
perception of warmth and competence (as opposed to
higher), the higher the rewards to CSR.
METHODS
Given the plethora of empirical challenges in CSR
research, it was important to conduct experiments
to examine and establish the causal links between
CSR/CSI, social perceptions, and different outcomes. Experiments allowed us to make causal
statements based on control of the environment
(Shadish, Cook, & Campbell, 2002), making them
ideal for testing our theory. They helped isolate our
theoretical mechanisms and, therefore, advance
theory (Haslam & McGarty, 2004; Mook, 1983).
Moreover, as we test our hypotheses using multiple
paradigms and outcomes, multiple experiments
allow for conceptual replication and extension,
highlighting the robustness of the effects. We conducted three experiments.
In Study 1, we manipulated CSR and CSI to examine perceptions of warmth and competence as
well as such outcomes as reputation and purchase
intentions. In addition, thanks to the temporal order
in our experimental design, we were able to infer
warmth mediation. In Study 2, owing to the differences in the domain of the country of origin—
organizational context—we were able to manipulate
warmth and competence of the organization, and
examine them as moderators in the relationship between CSR, CSI, and firm outcomes. Specifically, we
October
examined four countries representing each quadrant
of the warmth–competence BIAS map (i.e., the United
States, Germany, Portugal, and Pakistan) and three
conditions (i.e., CSR, CSI, and control). Finally, in
Study 3, we replicated and extended these effects in a
behavioral experiment measuring helping and purchasing behaviors using a within-subjects design.
STUDY 1
Study 1 examines the relationship between CSR
activities and perceptions of warmth and competence (Hypotheses 1 and 2). It also examines whether
warmth and competence serve as mediating mechanisms between CSR and important organizational
outcomes (Hypothesis 3), such as purchase intentions (Sen & Bhattacharya, 2001) and reputation
(Wagner, Lutz, & Weitz, 2009).
Sample and Procedures
Participants and design. One hundred and two
participants (66 males, meanage 5 31, SD 5 10.11)
were recruited from Amazon’s Mechanical Turk
(Buhrmester, Kwang, & Gosling, 2011; Horton, Rand,
& Zeckhauser, 2011; see also O’Reilly, Robinson,
Berdahl, & Banki, 2015, for recent management research using this data source; Paolacci & Chandler,
2014) in exchange for $0.30 in Amazon credits. All
participants were based in the United States and
were employed full-time. All participants passed
attention filters embedded in the survey (i.e., “I will
choose ‘disagree’ to demonstrate that I am paying
attention”), and all participants had a unique IP address located within the United States. We used a
one-factor design that assigned participants to either
a CSR, CSI, or control condition randomly.
Procedure. Participants were told that they would
read a brief business scenario and provide their
opinions. Participants completed the study online.
CSR manipulation. Participants read about
“Company X,” a fictitious company that engaged in
either CSR or CSI activities, involving fair/unfair
manufacturing processes overseas—the vignette had
been established in prior literature (Sen & Bhattacharya,
2001). We also included a control condition with
no information on CSR (see Appendix A for full
vignettes).
Measures
Warmth and competence. Immediately following
the CSR manipulation, participants rated whether
2019
Shea and Hawn
1617
significantly different levels of warmth, F(2, 99) 5
133.43, p , .000, hp2 5 .73 (see Table 1). Results of
planned comparisons indicated that participants
assigned to the CSR condition (M 5 4.33, SD 5 0.58)
were more likely to view the firm as warm than were
participants assigned to the CSI condition (M 5 2.02,
SD 5 0.61), F(1, 67) 5 265.25, p , .000, hp2 5 .80, d 5
3.95, 95% CI [3.74, 4.17], or to our control condition
(M 5 3.20, SD 5 0.59), F(1, 67) 5 66.71, p , .000, hp2 5
.50, d 5 1.96, 95% CI [1.76, 2.18]. Moreover, participants
in the control condition perceived the firm as having
significantly higher levels of warmth than did those in
the CSI condition, F(1, 64) 5 64.32, p , .000, hp2 5 .50,
d 5 2.00, 95% CI [1.79, 2.22]. The means are displayed in
Table 1 and provide support for Hypothesis 1.
Hypothesis 2: Competence. A one-way ANOVA
showed that participants who were assigned to CSR,
CSI, or control manipulations were more likely to
judge the firm as possessing significantly different
levels of competence, F(2, 99) 5 15.65, p , .000, hp2 5
.24 (see Table 1). Results of planned comparisons indicated that participants assigned to the CSR condition (M 5 4.14, SD 5 0.48) were more likely to view
the firm as competent than were participants assigned
to the CSI condition (M 5 3.30, SD 5 0.88), F(1, 67) 5
24.68, p , .000, hp2 5 .27, d 5 1.21, 95% CI [1.03,
1.52], and the control condition (M 5 3.93, SD 5 0.50),
F(1, 67) 5 3.18, p 5 .08, hp2 5 .05, d 5 0.44, 95% CI
[0.26, 0.61], while participants in the control condition perceived the firm as having significantly higher
levels of competence than did participants in the CSI
condition, F(1, 64) 5 12.70, p 5 .001, hp2 5 .17, d 5
1.28, 95% CI [1.10, 1.46]. The means are displayed in
Table 1 and provide support for Hypothesis 2.
Reputation. To check the validity of our assumption that CSR generates rewards and CSI generates
penalties (Margolis et al., 2009; Salaiz, 2016), we ran
the same analysis on Reputation. Participants who
were assigned to CSR, CSI, or control manipulations
they believed Company X to possess various attributes. Specifically, they rated whether Company X
was tolerant, warm, good natured, and sincere (Fiske
et al., 2002), and competent, confident, independent,
competitive, and intelligent (Fiske et al., 2002). These
items were randomly embedded in other, unrelated
attribute items. Confirmatory factor analysis supported a two-factor solution that explained 72.2% of
the variance. The four warmth items loaded onto
Factor 1 with loadings ranging from .80 to .85. The five
competence items loaded onto Factor 2 with loadings
ranging from .77 to .84. The internal consistency coefficient of the warmth and competence items, measured using Cronbach’s alpha, were .93 and .86
respectively. In the control condition, warmth and
competence were not correlated (r 5 .19, p 5 .26).
Additional dependent variables. After rating
Company X on the dimensions of warmth and competence, participants answered questions about two
additional dependent variables. First, participants
indicated their purchase intentions. Specifically,
they were asked: “If the products from Company X
were available for purchase, what is your likelihood
of purchasing a product from Company X?” (1 5 not
at all likely, 7 5 very likely) (see Chandon, Morwitz,
& Reinartz, 2005, for a discussion of external validity). Additionally, participants rated their perceptions of the firm’s reputation: favorable, good,
pleasant, positive (reputation a 5 .97; Homer, 1995;
Wagner et al., 2009). Then, participants completed
another manipulation check, regarding the content
of the CSR scenario, and provided demographic data.
Results
Hypothesis 1: Warmth. A one-way analysis of
variance (ANOVA) showed that participants who
were assigned to CSR, CSI, or control manipulations
were more likely to judge the firm as possessing
TABLE 1
Summary of Results for Experiment 1
Study 1 (one-way ANOVA)
Condition
CSR
CSI
Control
Warmth (1–5 scale)
Competence (1–5 scale)
Purchase Intentions (1–7 scale)
Reputation (1–5 scale)
4.33 (0.58)***
2.02 (0.61)***
3.2 (0.59)***
4.14 (0.48)***
3.3 (0.88)***
3.93 (0.5)*
5.61 (1.15)***
2.7 (1.38)***
4.82 (1.03)**
4.57 (0.63)***
1.84 (0.73)***
3.48 (0.72)***
Notes: Table 1 shows means with standard deviations in brackets. CSR 5 corporate social responsibility; CSI 5 corporate social irresponsibility (see Appendix A for scenarios).
*p , .10
**p , .05
***p , .01
1618
Academy of Management Journal
perceived significantly different levels of firm reputation, F(2, 99) 5 134.71, p , .000, hp2 5 .73 (see Table 1).
Results of planned comparisons indicated that participants assigned to the CSR condition (M 5 4.57, SD 5
0.63) perceived the firm as having a better reputation
than did participants assigned to the CSI condition
(M 5 1.84, SD 5 0.73), F(1, 67) 5 279.94, p , .000, hp2 5
.81, d 5 4.07, 95% CI [3.85, 4.33], or to our control
condition (M 5 3.48, SD 5 0.72), F(1, 67) 5 45.12, p ,
.000, hp2 5 .40, d 5 1.64, 95% CI [1.41, 1.90]. Participants in the control condition perceived the firm as
having a significantly better reputation than did participants in the CSI condition, F(1, 64) 5 83.73, p , .000,
hp2 5 .57, d 5 2.30, 95% CI [2.04, 2.56]. Results from this
test support our assumption that rewards are above and
penalties are below the control condition—see Table 1.
Purchase intentions. Similarly, a one-way ANOVA
showed that participants who were assigned to CSR,
CSI, or control manipulations were more likely
to report different levels of purchase intentions,
F(2, 99) 5 52.15, p , .000, hp2 5 .513 (see Table 1).
Results of planned comparisons indicated that participants assigned to the CSR condition (M 5 5.61, SD 5
1.15) were more likely to have higher purchase intentions than participants assigned to the CSI condition (M 5 2.70, SD 5 1.38), F(1, 67) 5 91.11, p , .000,
hp2 5 .58, d 5 2.33, 95% CI [1.92, 2.82] or to our control
condition (M 5 4.82, SD 5 1.03), F(1, 67) 5 8.49, p 5
.005, hp2 5 .11, d 5 0.74, 95% CI [0.32, 1.10]. Moreover, participants in the control condition had higher
purchase intentions than participants in the CSI
condition, F(1, 64) 5 47.57, p , .000, hp2 5 .43, d 5
1.77, 95% CI [1.40, 2.26]. These results support our
assumption that CSR generates rewards, and CSI
penalties (i.e., rewards are above and penalties are
below the control condition—see Table 1).
Hypothesis 3: Warmth and competence as
mediators. Using Hayes’s (2013) process macros and
the bootstrap method, we ran mediational analyses,
predicting two dependent variables: firm reputation
and purchase intentions. We used the bootstrap model
of Preacher and Hayes (2004) to estimate the indirect effect of warmth and competence based on 10,000
bootstrap samples. This method is preferred to the traditional method proposed by Baron and Kenny (1986),
as it does not rely on the assumption that the sampling
distribution of the mediation effect is normal. We undertook multiple sets of analyses. First, we looked at
mediation of CSR versus Control, and CSI versus Control for warmth while controlling for competence1 to
1
We footnote results for mediations with both warmth
and competence as simultaneous mediators.
October
predict reputation and purchase intentions.2 Next,
we looked at a sequential mediation model between
warmth, competence, reputation, and purchase intentions. This allowed us to examine the distinct
effects of warmth and competence in both socially
responsible and irresponsible firms.
Results for reputation. We first constructed two
mediation models in which CSR versus control, CSI
versus control predicted reputation, with warmth as
a mediator (controlling for competence).3 We then
examined the indirect pathway comparing CSR and
control (i.e., CSR influences reputation through increased warmth). The 95% CI for the bootstrap estimation of the indirect pathway did not overlap with
0 when examining warmth in the CSR condition, b 5
0.91, 95% CI [0.59, 1.19], indicating mediation.4
Next, we examined the indirect pathway comparing
CSI and control (i.e., CSI influences reputation
through decreased warmth controlling for competence). The 95% CI for the bootstrap estimation of the
indirect pathway did not overlap with 0 when examining warmth in the CSI condition, b 5 20.95,
95% CI [21.28, 20.67].5 This indicates that warmth
mediates the relationship between CSR/CSI and firm
reputation, further substantiating Hypothesis 3.
Results for purchase intentions. We constructed
two mediation models in which CSR versus control,
CSI versus control predicted purchase intentions,
with warmth as mediator controlling for competence.6 First, we examined the indirect pathway
comparing CSR and control (i.e., CSR influences
2
We code such that the control condition is the reference group in each separate analysis.
3
When examined as joint predictors with the three experimental conditions, warmth, b 5 0.75, t 5 2.63, p 5
.009, predicted reputation, while competence, b 5 0.04, t 5
0.44, p 5 .44, did not.
4
According to the same analysis where Competence
serves as a joint mediator with Warmth, a significant indirect effect emerges for warmth, b 5 0.83, 95% CI [0.51,
1.10], but the 95% CI for competence contains 0, b 5 0.05,
95% CI [20.02, 0.16], suggesting warmth, not competence,
mediates this relationship.
5
According to the same analysis where Competence
serves as a joint mediator with Warmth, a significant indirect effect emerges for warmth, b 5 20.93, 95% CI
[21.28, 20.66], but the 95% CI for competence contains 0,
b 5 20.02, 95% CI [20.15, 0.12], suggesting warmth, not
competence, mediates this relationship.
6
When examined as joint predictors with the three experimental conditions, both warmth, b 5 0.50, t 5 2.48, p 5
.01, and competence, b 5 0.73, t 5 3.99, p 5 .001, predicted
purchase intentions.
2019
Shea and Hawn
Purchase Intentions through increased warmth
controlling for competence). The estimation of the
indirect effect did not overlap with 0, b 5 0.83, 95%
CI [0.15, 1.46], indicating mediation of purchase intentions by warmth.7 Next, we completed the same
analyses for CSI versus control (i.e., CSI influences
Purchase Intentions through decreased warmth
controlling for competence). The estimation of the
indirect effect did not overlap with 0, b 5 20.98,
95% CI [21.84, 20.33], indicating mediation of
purchase intentions by warmth.8
Sequential mediation. We constructed sequential
mediation models including reputation (a cognitive
evaluation) to predict purchase intentions (a behavioral intention). We examined purchase intentions
as the final dependent variable as it is the most distant variable and represents a behavioral intention.
We also tested primacy of warmth assumptions by
adding in competence as a sequential mediator following warmth. We examined the indirect pathway
(i.e., CSR/CSI influences purchase intentions through
warmth, competence, and reputation; PROCESS,
Model 6). Using three mediators, seven potential indirect effects were calculated. The 95% CI for the
bootstrap estimation of the indirect pathway between
CSR → Warmth → Competence → Purchase Intentions excluded 0, b 5 0.34, 95% CI [0.09, 0.64], indicating sequential mediation. Likewise, the 95% CI
for the bootstrap estimation of the indirect pathway
between CSR → Warmth → Reputation → Purchase
Intentions excluded 0, b 5 0.64, 95% CI [0.19, 1.11],
indicating sequential mediation as well. No other indirect effects excluded 0.
Next, we examined the indirect pathway with CSI.
The 95% CI for the bootstrap estimation of the indirect pathway between CSI → Warmth → Competence → Purchase Intentions excluded 0, b 5 20.49,
95% CI [20.82, 20.19], indicating sequential mediation.
Likewise, the 95% CI for the bootstrap estimation of
the indirect pathway between CSI → Warmth →
7
According to the same analysis where Competence
serves as a joint mediator with Warmth, the 95% CI for the
bootstrap estimation of the indirect pathway did overlap
with 0 when examining warmth and competence together
in the CSR condition, bwarmth 5 0.43, 95% CI [20.18,
1.008]; bcompetence 5 0.20, 95% CI [20.02, 0.53].
8
According to the same analysis where Competence
serves as a joint mediator with Warmth, the 95% CI for the
bootstrap estimation of the indirect pathway did overlap
with 0 when examining warmth and competence together
in the CSI condition, bwarmth 5 20.47, 95% CI [21.29,
0.07], but not for competence, bcompetence 5 20.50, 95% CI
[20.86, 20.19].
1619
Reputation → Purchase Intentions excluded 0,
b 5 20.81, 95% CI [21.42, 20.43], indicating sequential mediation as well. No other indirect effects
excluded 0. We ran equivalent models changing the
order of the potential mediators. Importantly, analyses in which competence preceded warmth in the
model did not produce indirect effects excluding 0,
suggesting the primacy of warmth as a mediating
mechanism in our data set. Overall, our results indicate that warmth consistently mediates the relationship between CSR/CSI and purchase intentions (on
its own and when predicting Purchase intentions in
conjunction with Reputation), whereas competence
does not consistently mediate it, supporting Hypothesis 3.
Discussion
Study 1 provides support for Hypotheses 1 through
3. When “Company X” engaged in CSR activities, it
was judged to be significantly higher in warmth than
both control and CSI firms (supporting Hypothesis 1).
In addition, per traditional halo effects in psychological research (Ross & Nisbett, 1991) as well as other
alternative explanations, we observed a positive relationship between CSR and competence ratings
(supporting Hypothesis 2). However, although ratings
of competence differed between CSR and CSI conditions, the control condition was not significantly different from the CSR condition, p . .08, indicating that
competence may not be directly related to the presence of CSR activities but is rather related to their
absence (i.e., CSI). This finding is consistent with
Fiske et al. (2007), which found that negative behaviors are more indicative of competence while positive
behaviors are more indicative of warmth. Next, we
established further evidence of CSR as a warmth
strategy in our mediational analyses. Specifically,
warmth mediates the relationship between CSR/CSI
and the control condition to predict reputation and
purchase intentions (supporting Hypothesis 3). When
examined as a sequential mediator, warmth is necessary to mediate the relationship between CSR/CSI
and purchase intention while including measures of
competence and reputation. This evidence confirms
previous research on the primacy of warmth when
forming social judgments (Cuddy et al., 2008; Fiske
et al., 2007; Singh & Teoh, 2000; Tausch et al., 2007).
Although this study represents, to our knowledge,
one of the first empirical links between corporate
strategy and social perception, many questions remain. These results are about “Company X”—a fictitious organization used to maintain experimental
1620
Academy of Management Journal
control. Typically, we know more about an organization when making a purchase and evaluating firm
reputation (e.g., country of origin, size, status/brand
of the firm). Much of this auxiliary information can
be related to the dimensions of warmth and competence. Our results so far suggest that, just like stereotypes of organizations (Aaker et al., 2010),
stereotypes of organizational practices indeed exist,
and warmth and competence are in fact organizing
dimensions that help individuals categorize companies and their strategies (i.e., CSR or CSI). Moreover, warmth perceptions mediate the relationship
between CSR/CSI and outcomes. However, if this is
the case, do warmth and competence perceptions of
organizations from other sources of evaluation color
the way in which CSR and CSI are evaluated? Could
they shift individual willingness to buy or evaluation
of firm reputation or other behavioral outcomes of
CSR and CSI? Experiments 2 and 3 were designed to
address these questions.
October
Study 1 examined the relationship between a firm’s
CSR activities and perceptions of warmth and competence, establishing CSR as a warmth strategy via
mediation. Study 2 shifts our focus to the warmth–
competence moderating effect: in particular, do
warmth and competence perceptions of the organization from other sources (i.e., country of origin)
moderate the effects of CSR (CSI) on different outcomes? In other words, under what conditions do
firms achieve greater rewards (penalties) for CSR
(CSI) in terms of favorable (unfavorable) evaluations? To test this, we manipulated the perception of
the organization by changing a firm’s country of origin to reflect different levels of warmth and competence (i.e., high or low). Just as the national origin
of immigrants guides majority members’ perceptions
of them (Lee & Fiske, 2006), we expect the country of
origin to affect the social perception of the firm and of
its strategic choices.
attention filters embedded in the survey. We used a
three-factor design that randomly assigned participants to a CSR state (i.e., CSR, CSI, or control),
country-of-origin warmth (i.e., high, low), and competence (i.e., high, low). We also included a pure
control condition that did not manipulate country of
origin; its inclusion did not change the significance
or pattern of our results. However, because this experiment tests Hypotheses 4 through 6 using warmth
and competence perceptions of the countries of origin and the control condition does not specify any
country (and, hence, warmth or competence), we did
not include it in the final analyses.
Procedure. Participants completed the study
online. They were asked to read a brief business
scenario and to provide their opinions by rating the
warmth and competence of the firm’s country of origin as well as their perceptions of the firm based on
the CSR vignette.
CSR manipulation. Participants read the same
vignettes as in Study 1 (Sen & Bhattacharya, 2001),
modified to manipulate warmth and competence via
country of origin.
Warmth and competence manipulations. Based
on previous research, we chose the in-group
to represent the high-warmth, high-competence
country (in our sample, the United States, as all
participants were based in the United States) (e.g.
Cuddy et al., 2008). We chose Germany (low
warmth, high competence) and Portugal (high
warmth, low competence) as the two countries farthest from each other on the BIAS map (see Cuddy
et al., 2007). Finally, because there was no European low–low country and it has the lowest ranking
on the BIAS map, we chose Pakistan as our lowwarmth, low-competence country (see Cuddy et al.,
2007). To manipulate warmth and competence, we
changed the name of the organization in the vignette to include the country of origin: thus,
U.S. Tech Corp., Pakistan Tech Corp., German Tech
Corp., Portugal Tech Corp., and Company X for the
control condition.
Sample and Procedures
Measures
Participants and design. Five hundred and
seventy-two participants (357 males, meanage 5
32.13, SD 5 11.12) were recruited from Amazon’s
Mechanical Turk (Buhrmester et al., 2011) in exchange for $0.50 in Amazon credits. To avoid the
out-of-group bias (e.g., rating a U.S. firm differently if
you were based outside the United States), all participants were based in the United States and passed
Measures of warmth and competence. Participants completed the same warmth and competence
items as in Study 1 (warmth, a 5 .94; competence,
a 5 .84) (Fiske et al., 2002). In our pure control
condition (no CSR/CSI, or country of origin), we
performed a confirmatory factor analysis, which
supported a two-factor solution that explained
75.69% of the variance. The four warmth items
STUDY 2
2019
Shea and Hawn
loaded onto Factor 1 with loadings ranging from .73
to .97. The five competence items loaded onto Factor
2 with loadings ranging from .40 to .92. Warmth and
competence were not significantly correlated in our
control condition, r 5 .35, p 5 .12.
Additional dependent variables. After rating the
firm on the dimensions of warmth and competence,
participants answered the same questions constituting our two dependent variables as in Study 1.
Participants then completed a manipulation check
regarding the content of the CSR scenario and provided demographic information.9
Results
Data analysis strategy. To test Hypotheses 4 and
5, we analyzed the two-way interaction effects between Warmth (high, low) and CSR/CSI conditions
(vs. control), collapsing across firm competence. To
test Hypothesis 6, we conducted pairwise comparisons regarding whether changes in competence
moderate the effects of warmth on CSR/CSI outcomes. Table 2 presents our results.
Reputation. Table 2a presents the means with
standard errors across all conditions predicting
reputation. Table 2b collapses across our warmth
conditions. We ran a three-way ANOVA to analyze
the effect of Condition (CSR, CSI, Control) by
Warmth (high, low) and Competence (high, low) on
Reputation. A significant main effect emerged for
CSR condition, F(2, 494) 5 613.94, p 5 .000, hp2 5
.71. Interaction effects emerged between Warmth 3
Condition, Competence 3 Condition, and Warmth 3
Competence, Fs . 2.4, ps , .10, hp2 . .009. These
effects were qualified by a significant three-way interaction, F(2, 494) 5 4.37, p 5 .01, hp2 5 .017. The
main effect of CSR condition once again confirms our
initial assumption about rewards to CSR and penalties to CSI: firms engaging in CSR (M 5 4.27, SD 5
0.69) attained higher levels of reputation than firms
engaging in CSI (M 5 1.67, SD 5 0.79), F(1, 351) 5
1068.95, p 5 .00, hp2 5 .75, d 5 3.52, 95% CI [3.41,
3.63], and control firms (M 5 3.58, SD 5 0.69), F(1,
331) 5 83.64, p 5 .00, hp2 5 .20, d 5 1.00, 95% CI
[0.90, 1.11). Control firms had higher reputation than
CSI firms, F(1, 326) 5 530.55, p 5 .00, hp2 5 .62, d 5
2.58, 95% CI [2.48, 2.70] (Figure 1a and Figure 1b).
To test Hypothesis 4 (comparing CSI outcomes
across high- and low-warmth firms), a two-way
9
In our pure control condition, purchase intentions and
reputation were not significantly correlated, r 5 .29, p 5
21.
1621
TABLE 2
Summary of Results for Experiment 2
TABLE 2a
Mean Levels of Reputation by Country (Hypothesis 6)
United States
(high, high)
Germany (low, high)
Portugal (high, low)
Pakistan (low, low)
CSI
Control
CSR
1.60 (0.11)
3.54 (0.10)
4.16 (0.12)
1.82 (0.15)
1.51 (0.11)
1.74 (0.12)
3.85 (0.08)
3.75 (0.10)
3.20 (0.14)
4.25 (0.10)
4.34 (0.10)
4.35 (0.09)
Notes: Reputation was measured on a 5-point scale. Standard
errors in brackets.
TABLE 2b
Mean Levels of Reputation by Warmth
(Hypotheses 4 and 5)
High Warmth
Low Warmth
CSI
Control
CSR
1.56 (0.07)
1.78 (0.09)
3.63 (0.07)
3.52 (0.09)
4.25 (0.08)
4.30 (0.07)
Notes: Reputation was measured on a 5-point scale. Standard
errors in brackets.
TABLE 2c
Mean Levels of Purchase Intentions by Country
(Hypothesis 6)
United States
(high, high)
Germany (low, high)
Portugal (high, low)
Pakistan (low, low)
CSI
Control
CSR
2.32 (0.19)
4.70 (0.13)
5.43 (0.18)
2.41 (0.25)
2.32 (0.35)
2.74 (0.21)
4.81 (0.16)
4.75 (0.17)
4.11 (0.18)
5.40 (0.18)
5.36 (0.21)
5.52 (0.20)
Notes: Purchase Intentions was measured on a 7-point scale.
Standard errors in brackets.
TABLE 2d
Mean Levels of Purchase Intentions by Warmth
(Hypotheses 4 and 5)
High Warmth
Low Warmth
CSI
Control
CSR
2.28 (0.14)
2.57 (0.16)
4.73 (0.11)
4.45 (0.12)
5.39 (0.13)
5.46 (0.14)
Notes: Purchase Intentions was measured on a 7-point scale.
Standard errors in brackets.
ANOVA was run to analyze the relative effect of
Condition (CSI, control) by Warmth (high, low)
on Reputation. A main effect for CSI emerged,
1622
Academy of Management Journal
FIGURE 1a
Results of Study 2: Average Effect of Warmth and
Condition on Reputation
5
4.5
Reputation
4
3.5
3
2.5
2
1.5
1
CSR
Control
High Warmth
CSI
Low Warmth
FIGURE 1b
Results of Study 2: Average Effect of Warmth and
Condition on Purchase Intentions
Purchase Intentions
7
6
5
4
3
2
1
CSR
Control
High Warmth
CSI
Low Warmth
F(1, 324) 5 533.72, p 5 .000, hp2 5 .62, and this effect
was qualified by a significant interaction between
CSI x Warmth, F(1, 324) 5 4.37, p 5 .04, hp2 5 .01,
suggesting moderation. No significant differences
emerged in the control condition between high (M 5
3.63, SD 5 0.62) and low (M 5 3.52, SD 5 0.75)
warmth firms, F(1, 152) 5 1.12, p 5 .29, hp2 5 .007,
d 5 0.17, 95% CI [0.03, 0.44]. However, when highwarmth firms engaged in CSI (M 5 1.56, SD 5 0.70),
they received lower reputation evaluations than lowwarmth firms (M 5 1.78, SD 5 0.88), F(1, 172) 5 3.58,
p 5 .06, hp2 5 .02, d 5 0.31, 95% CI [0.15, 0.45]. This
provides support for Hypothesis 4 for reputation:
high-warmth firms face a larger reputation penalty
(Δ2.07) for engaging in CSI than low-warmth
firms (Δ1.74).
October
To test Hypothesis 5 (comparing CSR outcomes
across high- and low-warmth firms), a two-way
ANOVA was run to analyze the relative effect of
Condition (CSR, control) by Warmth (high, low) on
Reputation. The interaction was not significant,
F(1, 328) 5 1.18, p 5 .28, hp2 5 .004, and, thus, we
do not find support for Hypothesis 5 for reputation:
high- and low-warmth firms had equivalent outcomes for engaging in CSR compared to control
firms.
Next, to test Hypothesis 6 (comparing CSI outcomes for Germany–Pakistan, United States–
Portugal), a Condition (CSI, control) by Competence
(high, low) ANOVA was run within both high- and
low-warmth conditions. At low levels of country
warmth (Germany–Pakistan), a Condition (CSI,
control) by Competence (high, low) ANOVA
revealed an interaction between CSI and competence, F(1, 157) 5 5.00, p 5 .02, hp2 5 .03. Specifically, while firms from Germany (M 5 1.83, SD 5
0.99) and Pakistan (M 5 1.74, SD 5 0.75) do not
differ when they engage in CSI, F , .18, p . .67, a
significant difference emerges in the control condition, F(1, 72) 5 16.31, p 5 .000, hp2 5 .19, d 5 0.97,
95% CI [0.82, 1.24], such that firms from Germany
(M 5 3.85, SD 5 0.47) have higher reputation than
firms from Pakistan (M 5 3.20, SD 5 0.84). This
suggests that competence may not buffer reputation
when low-warmth firms engage in CSI in terms of
the relative losses to reputation (ΔGermany 5 2.02;
ΔPakistan 5 1.46). At high levels of country warmth
(United States–Portugal), the interaction between
CSI and competence is not significant, F , .02, p .
.88; therefore, we find no support for Hypothesis 6a
regarding reputation.
Next, to test Hypothesis 6b (comparing CSR
outcomes for Germany–Pakistan, United States–
Portugal), a Condition (CSR, control) by Competence (high, low) ANOVA was run within both high
and low warmth conditions. At low levels of country warmth (Germany–Pakistan), a Condition (CSR,
control) by Competence (high, low) revealed an
interaction between CSR and competence, F(1,
159) 5 12.59, p 5 .001, hp2 5 .07, d 5 0.98, 95% CI
[0.83, 1.13]. Specifically, while Germany (M 5 4.25,
SD 5 0.66) and Pakistan (M 5 4.35, SD 5 0.63) do
not differ when they engage in CSR, F , .18, p . .67,
the interaction is driven by the relative differences
(ΔGermany 5 0.41; ΔPakistan 5 1.14) gained from engaging in CSR versus control (Germany: M 5 3.84,
SD 5 0.47; Pakistan: M 5 3.20, SD 5 0.84). When
engaging in CSR, a firm from Pakistan, a low–low
country, made higher relative gains compared to
2019
Shea and Hawn
that from Germany (low-warmth–high-competence
country), suggesting that initial competence did not
have an additive effect, but, rather, low–low firms
are seen as exemplars and get higher returns to CSR,
supporting Hypothesis 6b. At high levels of country
warmth (United States–Portugal), however, the interaction between CSR and competence is not
significant, F , .02, p . .88, suggesting that the
additive benefits of competence are limited to lowwarmth firms engaging in CSR (which violates the
stereotype of low–low organizations).
Purchase intentions. Table 2c presents the means
with standard error bars across all conditions predicting purchase intentions. Table 2d includes
a separate plot collapsing across our warmthcompetence conditions. First, a three-way ANOVA
was run to analyze the effect of Condition (CSR, CSI,
Control) by Warmth (high, low) and Competence
(high, low) on Purchase Intentions. While the threeway interaction did not reach statistical significance, F(2, 494) 5 2.00, p 5 .14, hp2 5 .008, we
tested our hypotheses using a series of two-way
ANOVAs reported below. A significant main effect
emerged for CSR condition, F(2, 494) 5 266.50, p 5
.000, hp2 5 .52. Once again, the main effect of CSR
condition supported our assumption: firms engaging in CSR (M 5 5.43, SD 5 1.28) attained higher
levels of purchase intentions than firms engaging in
CSI (M 5 2.43, SD 5 1.42), F(1, 351) 5 434.86, p 5
.00, hp2 5 .55, d 5 2.23, 95% CI [2.04, 2.44], and
control firms (M 5 4.59, SD 5 1.01), F(1, 331) 5
42.82, p 5 .00, hp2 5 .43, d 5 0.73, 95% CI [0.54,
0.88]. Control firms had greater purchase intentions
than CSI firms, F(1, 327) 5 245.99, p 5 .00, hp2 5
.12, d 5 1.76, 95% CI [1.61, 1.97].
To test Hypothesis 4 (comparing CSI outcomes
across high and low warmth), a two-way ANOVA
was run to analyze the effect of Condition (CSI,
control) by Warmth (high, low) on Purchase Intentions. A main effect for CSI emerged, F(1, 324) 5
246.41, p 5 .000, hp2 5 .432, and this effect was
qualified by a significant interaction between CSI 3
Warmth, F(1, 324) 5 4.41, p 5 .04, hp2 5 .01. When
firms high (M 5 2.28, SD 5 1.30) or low (M 5 2.57,
SD 5 1.52) in warmth engage in CSI, purchase intentions are not significantly different, F(1, 172) 5
1.93, p 5 .16, hp2 5 .01, d 5 0.21, 95% CI [0.12,
0.48]. However, marginally significant differences emerged between high (M 5 4.73, SD 5 0.94)
and low (M 5 4.45, SD 5 1.07) warmth firms in the
control condition, F(1, 152) 5 2.95, p 5 .09, hp2 5
.02, d 5 0.28, 95% CI [0.07, 0.52]. Thus, when highwarmth firms shift from no strategy to CSI (Δ2.45),
1623
compared to low-warmth firms (Δ1.88), they face a
bigger deficit in purchase intentions. This provides
support for Hypothesis 4.
To test Hypothesis 5 (comparing CSR outcomes
across high and low warmth), a two-way ANOVA
was run to analyze the effect of Condition (CSR,
control) by Warmth (high, low) on purchase intentions. The interaction was not significant, F(1,
328) 5 1.84, p 5 .18, hp2 5 .006, and does not provide
support for Hypothesis 5 for purchase intentions.
Next, to test Hypothesis 6a (comparing CSI
outcomes for Germany–Pakistan, United States–
Portugal), a condition (CSI, control) by Competence
(high, low) ANOVA was run within both high and
low warmth conditions. At low levels of country
warmth (Germany–Pakistan), an interaction between CSI and competence emerged, F(1, 157) 5
6.18, p 5 .01, hp2 5 .04. Specifically, while Germany (M 5 2.41, SD 5 1.64) and Pakistan (M 5 2.74,
SD 5 1.38) do not differ in purchase intentions
when they engage in CSI, F , 1.05, p . .30, a significant difference emerges in the control condition, F(1, 72) 5 8.68, p 5 .004, hp2 5 .11, d 5 0.69,
95% CI [.38, 1.05], such that Germany (M 5 4.81,
SD 5 0.95) has higher purchase intentions than
Pakistan (M 5 4.11, SD 5 1.09). When low-warmth
firms engage in CSI, competence does not buffer the
negative impact leading to greater relative decreases
in purchases intentions (ΔGermany 5 2.40; ΔPakistan 5
1.37). At high levels of warmth, the interaction between CSI and competence is not significant, F ,
.02, p . .91. Therefore, just like for reputation, we
do not find support for Hypothesis 6a for purchase
intentions.
Next, to test Hypothesis 6b (comparing CSR
outcomes for Germany–Pakistan, United States–
Portugal), a Condition (CSR, control) by Competence (high, low) ANOVA was run within both
high and low warmth conditions. At low levels of
warmth (Germany–Pakistan), an interaction between CSR and competence emerged, F(1, 159) 5
5.11, p 5 .03, hp2 5 .03, d 5 0.56, 95% CI [0.28,
0.87]. Specifically, while Germany (M 5 5.40, SD 5
1.18) and Pakistan (M 5 5.52, SD 5 1.34) do not
differ when they engage in CSR, F , .21, p . .64, the
interaction is driven by the relative differences
gained (ΔGermany 5 .59; ΔPakistan 5 1.41) versus
control (Germany: M 5 4.81, SD 5 0.95; Pakistan:
M 5 4.11, SD 5 1.09). At high levels of warmth, the
interaction between CSR and competence is not
significant, F , .12, p . .77. Thus, our findings
suggest that low-warmth–low-competence firms
receive a larger relative benefit when engaging in
1624
Academy of Management Journal
CSR—supporting Hypothesis 6b—even though
high-warmth firms do not see differences with the
addition of competence.
Discussion
There are several implications of Study 2, with
the main one being that we find no support for Hypotheses 5 and 6a for both outcomes (reputation
and purchase intentions). Hypotheses 4 and 6b
demonstrated consistent results for both outcomes
for firms from a low–low country (Pakistan), suggesting that, in comparison to Hypothesis 5, which
only considers warmth perception, both high competence and warmth help a low–low company
achieve higher rewards for CSR. Thus, the effectiveness of a CSR strategy is contingent on the
warmth and competence perception of the organization’s country of origin (Hypothesis 6b), while
firms from high-warmth countries received harsher
evaluations for engaging in CSI (Hypothesis 4). We
theorize that these effects are driven by assimilation
(CSI) and contrast (CSR) effects from the stereotypes associated with low-warmth countries. Importantly, this study provides insight into why
some organizations fare better or worse in observer
reactions to their corporate strategies. The attributes of the firm—conceptualized as countryof-origin warmth and competence—influenced
important organizational perceptions, such as reputation and purchase intentions.
In addition, Study 2 helps explore the question of
endogeneity regarding whether high-warmth organizations are more likely to engage in CSR behavior
in the first place—to earn greater benefits from it. We
find consistent marginal support for Hypothesis 4
that states that firms with higher warmth will pay a
higher cost for CSI, and no support for Hypothesis 5
about lower rewards for CSR. Likewise, we do not
find support for the argument that the costs for engaging in irresponsible behavior are lower for organizations from high–high countries. In fact, firms
from Pakistan (a low-warmth and low-competence
country) experienced the greatest rewards for CSR
engagement, due to stereotype violation. To conclude, Study 2 demonstrates that a firm’s country of
origin (manipulating warmth and competence perceptions) influences the success or failure of its CSR
strategies relative to its baseline state. Specifically,
we find that CSR can supplement for low-warmth
(and low-competence) country of origin, and that
low-warmth country of origin shields organizations
from harsher judgments of CSI.
October
STUDY 3
Studies 1 and 2 showed how CSR and CSI affect the
perceived warmth and competence of an organization, and that these effects are amplified depending on
the firms’ country of origin. Study 3 aims to replicate
and extend these findings while addressing various
shortcomings of Studies 1 and 2. First, while Studies 1
and 2 measure perceptions, Study 3 uses behavioral
measures (Colquitt, 2008), adding external validity to
our previous measures. Specifically, participants
used their own money to purchase products from
firms with differing social strategies and also provided help (e.g., feedback) to the organizations.
Second, to ensure that our high-warmth–highcompetence results are not an artifact of an in-group
bias, Study 3 uses Sweden as the high-warmth–
high-competence country.10 Third, we switched our
CSR manipulations to a fact-sheet format that included information on both labor and environmental
10
Prior to running Study 3, we sought to validate previous
research on the BIAS map (Cuddy et al., 2007) with regard to
where countries fall on the warmth and competence dimensions. In addition, we sought to replace the United States
in the final study. Using a unique sample of 95 Mechanical
Turk workers, we assessed the warmth and competence of
seven countries (Germany, Iran, Pakistan, Portugal, Spain,
Sweden, and the United States) previous research had classified into one of the four quadrants on the BIAS map. We
first asked about the warmth and competence of each
country (Fiske et al., 2002) and then had participants rank
order the countries in terms of their warmth and competence. Pairwise comparisons, rank order, and correlational
evidence support the classification in our experimental
paradigm. First, ranked from highest to lowest warmth, were
Sweden, the United States, Spain, Portugal, Germany,
Pakistan, and Iran; and, second, ranked from highest to
lowest competence were Germany, the United States, Sweden, Spain, Portugal, Iran, and Pakistan. Looking at correlations between warmth and competence, the United States,
r 5 .55, p , .000, and Sweden, r 5 .70, p , .000, are seen as
high warmth and competence. Given that our study comprises U.S. participants, we used Sweden to mitigate ingroup bias concerns and “Made in America” preferences in
our Study 3. Pakistan is viewed as low warmth and competence, r 5 .50, p , .000, as compared to Iran, r 5 .21, p , .000,
given the high correlation between warmth and competence
as well as low means. Germany is viewed as high competence and low warmth, r 5 .21, p 5 .04, given the smaller
correlation. Portugal is viewed as high warmth and low
competence. r 5 .48, p , .000, as compared to Spain, r 5 .63,
p , .000; although not as cleanly manipulated, we believe
that Portugal represents the better manipulation of low
competence and high warmth, given the smaller correlation
coefficient and previous research on the BIAS map.
2019
Shea and Hawn
performance to ensure that the effects were not idiosyncratic to our Study 1 and 2 materials. And, finally,
to provide a conservative test of our hypotheses, we
modified our design to assess these behaviors within
subjects; that is, all participants engaged with firms
from four countries within CSR and CSI conditions.
Sample and Procedures
Participants. We recruited 100 participants from
a paid student research lab at a private Midwestern
university. We excluded one participant due to expletive language in all of our written prompts, and 11
participants who failed the manipulation check. Including these participants in analyses does not alter
the direction of effects.
Design. Study 3 employed a mixed design methodology; in particular, 3 (CSR, CSI, control condition) by 2
(Order: Pricing first, Feedback first) between-subjects
factors across a four-factor country-of-origin warmth by
competence (HH 5 Sweden; HL 5 Germany; LH 5
Portugal; LL 5 Pakistan) within-subjects factor. We
randomized the order of our country of origin manipulation as a repeated measure across all conditions.
Procedure. The study was advertised as a 30-minute product development study. Participants received
$8 in exchange for their participation and an additional
$1 for a purchase decision. The study asked participants to test, evaluate, and make purchase decisions
about a series of pens. We chose pens as an experimental stimulus because pens have been used successfully in previous research (e.g., Shah & Wolford,
2007). We purchased four types of pens priced
$0.78–$0.83 from a wholesaler. The pens did not differ
on measures of quality or looks. Upon arrival to the lab,
each participant was seated at a computer station
where they received a plastic bag containing four
“prototype” pens to evaluate as well as $1 in coins for
the pricing task (in addition to the $8 payment). They
also received a notepad to test out each pen. Participants completed the entire study at a computer station.
Warmth–Competence manipulation. To manipulate country of origin we had the pens labeled as
“Made in Sweden,” “Made in Germany,” “Made in
Portugal,” and “Made in Pakistan” in a 1-millimeter
font. We then used neutral labels on each pen to serve
as logos and translated the word “fine” into each of the
four languages to put on the logo. Each participant
received and rated each of these four pens as a withinsubjects variable.
Social responsibility manipulation. Participants
were randomly assigned one of three experimental
conditions: CSR, CSI or control. We provided them
1625
with background information about each of the four pen
manufacturers whose products they would be testing.
Embedded within neutral performance information
was information on environmental impact and labor
practices of each of the manufacturers. In the CSR condition, participants read that all four pen manufacturers
were going out of their way to promote strong environmental protection and repair, as well as employee
standards. In the CSI condition, pen manufacturers
were stated to be poor in environmental standards and
not desirable in terms of their labor practices. In the
control condition, participants were provided with
neutral information. All companies were established in
their industries and in good financial standing. Each
condition had the same number of categories to describe
the organization.11 Please see Appendix B.
11
As a manipulation check, participants rated both
the warmth and competence of the pen manufacturing
organizations using a single-item measure (“The pen
manufacturing organizations seem to be warm/competent”). A three-cell (CSR, CSI, control conditions) betweensubjects analysis of variance on perceptions of Warmth
revealed a significant effect of social responsibility condition, F(2, 81) 5 5.73, p 5 .005, hp2 5 .12. Specifically, individuals in the CSR condition (M 5 4.56, SD 5 1.41) had
significantly higher perceptions of warmth than individuals in the CSI condition (M 5 3.33, SD 5 1.42), F(1, 53) 5
10.18, p 5 .002, hp2 5 .16, d 5 0.88, 95% CI [0.38, 1.39].
Individuals in the Control condition (M 5 3.97, SD 5 1.18)
had marginally higher perceptions of warmth than individuals in the CSI condition, F(1, 57) 5 3.44, p 5 .07, hp2 5 .06,
d 5 0.50, 95% CI [0.08, 1.01]. Individuals in the CSR condition had marginally higher perceptions of warmth than
individuals in the control condition, F(1, 52) 5 2.82, p 5 .09,
hp2 5 .05, d 5 0.46, 95% CI [20.04, 0.88]. This replicates our
previous studies that found CSR and CSI are linked with
perceptions of warmth, supporting Hypothesis 1. A threecell (CSR, CSI, control conditions) between-subjects analysis
of variance on perceptions of Competence revealed a significant effect of social responsibility condition, F(2, 81) 5
5.14, p 5 .008, hp2 5 .11. Specifically, individuals in the CSR
condition (M 5 5.56, SD 5 0.82) had significantly higher
perceptions of competence than individuals in the CSI
condition (M 5 4.70, SD 5 1.26), F(1, 53) 5 8.56, p 5 .005,
hp2 5 .14, d 5 0.82, 95% CI [0.53, 1.27]. Individuals in the
Control condition (M 5 5.31, SD 5 0.93) had significantly
higher perceptions of competence than individuals in the
CSI condition, F(1, 57) 5 4.44, p 5 .04, hp2 5 .07, d 5 0.56,
95% CI [0.23, 1.01]. Individuals in the CSR and control
conditions did not differ significantly in perceptions of
competence, F(1, 52) 5 1.08, p 5 .30, hp2 5 .02, d 5 0.29, CI
[0.00, 0.62]. This replicates our previous studies that found
CSR and CSI are linked with perceptions of competence
while control conditions oftentimes do not differ from CSR,
supporting Hypothesis 2.
1626
Academy of Management Journal
Measures
Participants completed a feedback task set and a
purchasing task. The two tasks were counterbalanced.12 When relevant, we controlled for order.
Each participant completed each dependent measure across the four country of origin manipulations.
Feedback task. The first set of tasks examined the
time and effort that participants devoted to providing
feedback to each pen manufacturing organization.
Participants were told that the pens were prototypes
and that the organizations would like feedback on the
pen designs. We encouraged participants to try out
each pen in order to form an opinion of each product.
Participants were given a blank essay box to provide
feedback to the organization on their pen design. We
measured the number of words written (M 5 21.87,
SD 5 10.42) as a measure of helping (adapted from
Porath & Erez, 2007). We also measured the subjective
quality of each pen (“The pen is of high quality,” “The
pen is desirable”; as 5 .72–.76).
Purchasing task. The second task was a purchasing task using the Becker–DeGroot–Marschak method
(Becker, DeGroot, & Marschak, 1964), which has been
used to reveal true reservation prices in a variety
of contexts (Burbano, 2016; Kahneman, Knetsch, &
Thaler, 1991; Lerner, Small, & Loewenstein, 2004).
In this task, participants were given an additional
dollar and were told that they could use this money to
purchase the pens from the experiment and to keep
whatever money they did not use. Participants were
told that the computer program would randomly select one pen and a random price at which they had the
opportunity to purchase it. If their stated price was
below the randomly generated price, they would not
have the opportunity to purchase the pen. If their
stated price was above the randomly generated price,
they would have the opportunity to purchase the pen.
Participants had to successfully recall these instructions prior to continuing to ensure that they
understood the experimental paradigm. This goes
above traditional willingness-to-pay measures by
forcing participants to forgo some of their bonus pay
to make a purchase. Likewise, the random lottery
nature of the task ensures that participants reveal
their true purchase prices (Becker et al., 1964). Participants stated a purchase price for each of the four
pens on a sliding scale between $0.01 and $1.00 (M 5
12
Pre-tests indicated that licensing effects may occur
when tasks were sequential (i.e., “I did not help the organization, so I supplement by overpaying for their product,”
or vice versa).
October
$0.33, SD 5 $0.17). No pens significantly differed
from this price and no pen was consistently priced as
the “top” pen across the entire sample. After participants stated their price for each pen, a random
payment price was generated, they took the pen (if
purchased), and left the purchase price at the computer station at the end of the study. Participants then
completed a manipulation check about the social
practices of the pen manufacturers, as well as demographic questions.
Results
Data analysis strategy. We ran a series of repeated measures ANOVA across dependent measures. We first ran a CSR condition (three: CSR, CSI,
Control) by Order (two: Pricing first, Feedback first)
across the four country of origin manipulations
(separately as well as grouped together by high and
low warmth) as a repeated measures ANOVA.13 To
directly test hypotheses, we re-ran models with a
series of planned comparisons to probe statistically
significant mean differences both within and between conditions (see Table 3). Assumption of
sphericity was not violated in any of the analyses;
therefore, we report the sphericity-assumed results.
Number of words. A three-cell (CSI, CSI, control
condition) by two (Order: Pricing first, Feedback first)
repeated measures ANOVA was run on the number of
words written for the high- and low-warmth firms, as
well as the Sweden, Germany, Portugal, and Pakistan
country of origin conditions. No significant differences
emerged in the models, Fs , 2.02, ps . .11. When
examined as a between-subjects analysis, a significant
effect emerged for the social responsibility condition,
F(2,78) 5 3.46, p 5 .04, hp2 5 .08. Specifically, individuals in the CSR condition (M 5 26.28, SD 5
14.10) wrote marginally more words than individuals
in the CSI condition (M 5 20.53, SD 5 7.22), F(1,53) 5
3.81, p 5 .05, hp2 5 .07, d 5 0.52, 95% CI [24.52,
3.11], and individuals in the control condition (M 5
19.45, SD 5 8.48), F(1,52) 5 4.80, p 5 .03, hp2 5 .09,
d 5 0.60, 95% CI [24.45, 3.63]. The CSI and control
conditions did not significantly differ, F , .3, p . .60.
Although we did not see variation within our country
of origin manipulation, individuals offered more help
to organizations that engaged in CSR, providing behavioral replication and extension of Hypothesis 1 as
13
We re-ran analyses excluding the control condition as
it did not produce any significant effect, nor were there any
significant interactions between the control condition and
other conditions. We report both sets of ANOVAs below.
2019
Shea and Hawn
TABLE 3
Summary of Results for Experiment 3
1627
FIGURE 2a
Results of Study 3: Average Effect of Warmth and
Condition on Perceived Quality
TABLE 3a
Mean Levels of Purchase Price by Country (Hypothesis 6)
Control
CSR
0.43 (0.07)
0.25 (0.03)
0.37 (0.08)
0.31 (0.05)
0.35 (0.06)
0.26 (0.05)
0.32 (0.06)
0.29 (0.05)
0.31 (0.08)
0.35 (0.06)
0.27 (0.06)
0.37 (0.05)
Notes: Purchase Price was measured on a $1 scale. Standard
errors in brackets. CSI, Control, and CSR are between-subjects
factors, while country of origin is a within-subjects factor.
4.000
Subjective Quality
Sweden (high, high)
Germany (low, high)
Portugal (high, low)
Pakistan (low, low)
CSI
Evaluation Task First
3.500
3.000
2.500
2.000
1.500
1.000
CSI
High Warmth
TABLE 3b
Mean Levels of Purchase Price by Warmth (Hypotheses 4
and 5)
CSI
Control
CSR
4.000
0.40 (0.06)
0.28 (0.05)
0.33 (0.06)
0.28 (0.05)
0.29 (0.06)
0.36 (0.06)
3.500
Notes: Purchase Price was measured on a $1 scale. Standard
errors in brackets. CSI, Control, and CSR are between-subjects
factors, while warmth is a within-subjects factor.
TABLE 3c
Mean Levels of Subjective Quality by Country (Hypothesis
6)
Sweden (high, high)
Germany (low, high)
Portugal (high, low)
Pakistan (low, low)
CSI
Control
CSR
3.19 (0.22)
2.88 (0.16)
2.88 (0.21)
2.54 (0.18)
3.09 (0.21)
2.94 (0.25)
3.03 (0.19)
3.12 (0.27)
3.00 (0.26)
3.00 (0.20)
2.83 (0.19)
3.58 (0.25)
Notes: Subjective quality was measured on a 5-point scale.
Standard errors in brackets. CSI, Control, and CSR are betweensubjects factors, while country of origin is a within-subjects factor.
TABLE 3d
Mean Levels of Subjective Quality by Warmth (Hypotheses 4 and 5)
High Warmth
Low Warmth
Control
CSR
3.04 (0.20)
2.71 (0.21)
3.06 (0.18)
3.03 (0.19)
2.92 (0.21)
3.29 (0.22)
Notes: Subjective quality was measured on a 5-point scale.
Standard errors in brackets. CSI, Control, and CSR are betweensubjects factors, while warmth is a within-subjects factor.
Low Warmth
3.000
2.500
2.000
1.500
1.000
CSI
Control
High Warmth
CSR
Low Warmth
well as support for our initial assumption about CSR
generating rewards.
Subjective quality. To test Hypotheses 4 and 5, a
two-cell (CSI or CSI) by two (Order: Pricing first,
Feedback first) repeated measures ANOVA was run on
subjective quality in high- and low-warmth firms.14
The three-way interaction was significant, F(1, 51) 5
4.71, p 5 .04, hp2 5 .08. When participants completed
the feedback tasks first,15 an interaction emerged, F(1,
23) 5 6.51, p 5 .02, hp2 5 .22, suggesting moderation.
While no differences between high- and low-warmth
14
CSI
CSR
Pricing Task First
Subjective Quality
High Warmth
Low Warmth
Control
Including all CSR/CSI/control conditions, the threeway interaction is not significant, F(2, 78) 5 1.68, p 5 .19,
hp2 5 .041.
15
When participants completed the pricing task first, a
main effect for factor emerges, F(1, 28) 5 4.45, p 5 .04,
hp2 5 .137, such that participants had a preference for highwarmth firms (M 5 3.20, SD 5 0.64) over low-warmth firms
(M 5 2.88, SD 5 0.70) across conditions.
1628
Academy of Management Journal
FIGURE 2b
Results of Study 3: Average Effect of Warmth and
Condition on Purchase Price
Purchase Price
Evaluation Task First
0.500
0.450
0.400
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0.000
CSI
Control
High Warmth
CSR
Low Warmth
Pricing Task First
Purchase Price
0.500
0.400
0.300
0.200
0.100
0.000
CSI
Control
High Warmth
CSR
Low Warmth
firms emerged in the CSR condition, t(11) 5 1.62, p 5
.13, within the CSI condition, paired samples t tests
revealed that high-warmth firms (M 5 3.09, SD 5 0.45)
had marginally higher quality ratings than low-warmth
firms (M 5 2.71, SD 5 0.50), t(12) 5 2.10, p 5 .06,
failing to provide support for Hypothesis 4. However,
exploring the moderation effect between conditions,
no significant differences emerged for high-warmth
firms, F , .3, p . .59, while low-warmth firms saw
significant gains between CSI (M 5 2.71, SD 5 0.50)
and CSR (M 5 3.29, SD 5 0.54), F(1, 23) 5 7.78, p 5 .01,
hp2 5 .25, d 5 1.16, 95% CI [0.86, 1.44]. Taken together,
the significant moderation effect suggests that, while
low-warmth firms receive equivalent outcomes for
CSR, but not CSI, they significantly increase their
outcomes as they changed strategies from CSI to CSR
(Δ 5 .58), while high-warmth firms did not (Δ 5 .12).
This provides indirect support for Hypothesis 5 that
low-warmth firms reap higher gains from CSR.
To test Hypothesis 6, a two-cell (CSI or CSI) by
two (Order: Pricing first, Feedback first) repeated
October
measures ANOVA was run on subjective quality in
the Sweden, Germany, Portugal, and Pakistan
country of origin conditions.16 The three-way interaction was significant, F(3, 153) 5 3.46, p 5 .02,
hp2 5 .06. When the feedback task came first, a significant interaction emerged between CSR/CSI and
country of origin, F(3, 69) 5 3.90, p 5 .01, hp2 5 .25.
We report paired samples t tests across countries of
origin within the feedback task first condition.
In the CSI condition, a paired samples t test indicated that Pakistan (M 5 2.54, SD 5 0.66) had a significantly lower subjective quality than Sweden (M 5
3.19, SD 5 0.78), t(12) 5 2.85, p 5 .02, d 5 0.94, 95% CI
[0.50, 1.31]. No significant effects were observed between other countries, all ts , 1.67, ps . .12. This does
not provide clear support for Hypothesis 6a.
In the CSR condition, a paired samples t test indicated
that Pakistan (M 5 3.58, SD 5 0.87)17 had a significantly
higher subjective quality than Portugal (M 5 2.83, SD 5
0.65), t(11) 5 2.37, p 5 .04, and a marginally significant
higher subjective quality than Sweden (M 5 3.00, SD 5
0.90), t(11) 5 1.83, p 5 .09. No significant effects were
observed between other countries, all ts , 1.74, ps .
.11.18 This provides support for Hypothesis 6b, which
predicted higher rewards to CSR for low–low firms.
These results—obtained within subject—suggest that a
low-warmth–low-competence firm gains significantly
higher benefits for doing CSR than high-warmth firms.
Purchase price. To test Hypotheses 4 and 5, a twocell (CSI, CSI) by two (Order: Pricing first, Feedback
first) repeated measures ANOVA was run on Purchase Price in the high- and low-warmth firms.19 A
marginal effect emerged for Condition, F(1, 51) 5 3.08,
16
Including all CSR/CSI/control conditions, the threeway interaction is marginally significant, F(6, 234) 5 1.83,
p 5 .09, hp2 5 .045.
17
Probing this interaction between conditions, a onefactor ANOVA on Pakistan’s subjective quality revealed a
significant effect, F(1,23) 5 11.48, p 5 .003, hp2 5 .333, d 5
1.37, 95% CI [1.06, 1.61]. Specifically, subjective quality
for the Pakistani pen was significantly higher in the CSR
condition (M 5 3.58, SD 5 0.87) than in the CSI condition
(M 5 2.54, SD 5 0.66).
18
While not directly testing a hypothesis, a one-factor
ANOVA between CSR and CSI conditions on Pakistan’s
subjective quality revealed a significant effect, F(1,23) 5
11.48, p 5 .003, hp2 5 .333. Specifically, subjective quality
for the Pakistani pen was significantly higher in the CSR
condition (M 5 3.58, SD 5 0.87) than in the CSI condition
(M 5 2.54, SD 5 0.66).
19
Including all CSR/CSI/control conditions, the threeway interaction is marginally significant, F(2, 78) 5 2.68,
p 5 .07, hp2 5 .064.
2019
Shea and Hawn
p 5 .09, hp2 5 .06. This effect was qualified by a
significant three-way interaction, F(1, 51) 5 5.69, p 5
.02, hp2 5 .10. When participants completed the
feedback tasks first,20 an interaction emerged, F(1,
23) 5 7.04, p 5 .01, hp2 5 .23, suggesting moderation.
Exploring these interactions, in the CSI condition,
paired samples t tests revealed that high-warmth
firms (M 5 0.40, SD 5 0.07) and low-warmth firms
(M 5 0.28, SD 5 0.03) had significantly different
purchase prices, t(12) 5 2.44, p 5 .02, with highwarmth firms having significantly higher prices,
which does not support Hypothesis 4. However, in
the CSR condition, we observed no differences between high- and low-warmth firms, t(11) 5 1.35, p 5
.20. Looking between conditions, no significant differences emerged for high- or low-warmth firms, F ,
1.8, p . .19. Thus, the interaction term reaches significance, driven by the difference in prices in the
CSI condition, with the means crossover: when highwarmth firms switch from CSI to CSR, their purchase
price decreased (Δ 5 .11), while low-warmth firm’s
price increased (Δ 5 .08). Taken together, these results suggest that low-warmth firms closed the gap as
they changed strategies from CSI to CSR, while highwarmth firms surprisingly saw a decrease in terms of
pricing (between conditions), providing indirect
support for Hypothesis 5.
To test Hypothesis 6, a two (Condition: CSI or CSI)
by two (Order: Pricing first, Feedback first) repeated
measures ANOVA was run on Purchase Price in the
Sweden, Germany, Portugal, and Pakistan country of
origin conditions.21 The three-way interaction was
significant, F(3, 153) 5 2.61, p 5 .05, hp2 5 .05.
Probing this interaction further within Order conditions, when purchase price was determined after the
feedback task, a significant interaction emerged between CSR/CSI and country of origin, F(3, 69) 5 3.45,
p 5 .03, hp2 5 .13.22 In the CSI condition, a paired
samples t test indicated that Sweden (M 5 0.43, SD 5
0.25) had a significantly higher price than Germany
(M 5 0.25, SD 5 0.12), t(12) 5 2.69, p 5 .02, d 5 0.94,
95% CI [0.83, 0.99]; likewise, Sweden had a higher
price than Pakistan (M 5 0.31, SD 5 0.17), t(12) 5 2.70,
20
When participants completed the pricing task first,
participants had higher purchase prices, F(1, 28) 5 3.74,
p 5 .06, hp2 5 .117, for high-warmth firms (M 5 0.37, SD 5
0.20) than low-warmth firms (M 5 0.31, SD 5 0.20).
21
Including all social responsibility conditions, a marginally significant interaction effect between Order and
country of origin emerged, F(3, 234) 5 2.56, p 5 .08, hp2 5
.028.
22
Including control, F(6, 114) 5 1.99, p 5 .07, hp2 5 .095.
1629
p 5 .02, d 5 0.58, 95% CI [0.47, 0.65]. No significant
effects were observed between any other conditions,
ts , 1.87, ps . .15. In the CSR condition, no significant differences in purchase price emerged, all ts ,
1.55, ps . .14, providing no evidence to support
Hypothesis 6b. The finding that our high-warmth,
high-competence country of origin (Sweden) was able
to maintain a higher purchase price than the two lowwarmth countries while acting irresponsibly provides
some support to Hypothesis 6a.
Discussion
Study 3 replicates and adds nuance to effects observed in Studies 1 and 2. When it came to helping
behaviors, participants in the CSR condi...
Purchase answer to see full
attachment