Journal of Positive School Psychology http://journalppw.

com
2022, Vol. 6, No. 8, 5665-5672

A Review On Employees Emotional Intelligence At The

Workplace
Dr. Sharfras Navas1* , Dr. M. Vijayakumar 2 , Najmudeen Sulthan 3

1*
Researcher, King Saud University, Riyadh, Saudi Arabia.
2
Professor, K.S.R. School of Management, K.S.Rangasamy college of Technology, Tiruchengodu -637209,
Tamilnadu, India.
3
Business Research Scholar, Westford University College, United Arab Emirates/ UCAM Universidad
Católica San Antonio de Murcia, Spain.

ABSTRACT

Goal: This paper aims to analyze to which extend the emotional intelligence (EI) may play a major role
in employees' emotional experiences at work.
Design / Methodology / Approach: This study conducts a systematic literature review and 342 results
were retrieved, of which 52 were selected for the analysis. The results containing emotional intelligence
at workplace and the productivity after bringing emotional intelligence (EI) to the workplace.
Results: The analysis provides an overview of the "super salespeople" who had an incredible capacity
to discern what consumers cared about and build a trusting relationship with them. This study finds that
business leaders realised how important these "emotionally intelligent" individuals are to a company's
success as emotional intelligence is crucial in the workplace. While cognitive talents are thought to help
people find jobs, emotional intelligence is seen to help people advance in their careers. It is seen to be
twice as valuable as analytical and technical skills in the job.
Limitations of the investigation: The systematic literature review is restricted to the Elsevier SCOPUS
database. SCOPUS has excellent coverage regarding relevant journals for this study. Despite SCOPUS
being a prominent scientific database, other databases might reveal additional studies and further
insights.
Practical implications: The findings of this study provide insights on how emotionally intelligent"
individuals are to a company's success

Originality / Value: In recent years, emotional intelligence (EI) has seen a growing interest by scholars.
To the best of the authors' knowledge, the number of published literature reviews on Emotional
Intelligence (EI) is scarce. This study provides an overview of how emotional intelligence help business
at workplace.

Keywords: Emotional intelligence; workplace; productivity; emotional competence; Systematic

Review;

INTRODUCTION well-being, motivation, job satisfaction, and

Emotions are produced by both inner and
exterior human life. Emotions pervade human
social affairs and can have a substantial impact
on people's work experiences, affecting their
performance. Emotions, on the other hand, have
been pitted against reason in a Cartesian duality
because of their irrational, physical origin.
Emotional labour is the manifestation of those
emotions in the job that are judged acceptable
by the community in order to meet
Dr. Sharfras Navas
organisational standards. Employees may be
required, if not forced, to adjust their emotional
expressions in the workplace as part of their
professional job to improve the organization's
task, performance, and efficiency.
However, studies have shown that positive
individual outcomes are also important.
Employees may engage in and acquire
automatic responses with little participation of
cognitive functioning and effort as a result of a
gradual and ongoing learning experience in the
workplace. Thus, EI may play a major role in
employees' emotional experiences at work, not
only because it can allow effective workplace
functioning and positive outcomes, but also
because it can assist people improve their self-
identity by meeting the expectations that come
with it(Carminati, 2021).
Emotional intelligence
Figure 1 : Emotional Intelligence model by Daniel Goleman(Goleman, 1995)
5666
"The ability to accurately perceive, appraise,
and express emotion; the ability to access and/or
generate feelings when they facilitate thought;
the ability to understand emotion and emotional
knowledge; and the ability to regulate emotions
to promote emotional and intellectual growth,"
according to the definition of emotional
intelligence. Later, Goleman (1998) defined it
as “the capacity for recognizing our own
feelings and those of others, for motivating
ourselves, and for managing emotions well in
ourselves and in our relationships.”(Dirican &
Erdil, 2020).
Due of its importance in the workplace,
emotional intelligence has become a common
topic among educators, counsellors, and
business leaders. Specifically, accurate self-
assessment and conflict management are just
two examples of major emotional intelligence
influences in organisations that might result in
increased organisational
effectiveness(Ashkanasy et al., 2003).
5667
Motivation, empathy, sociability, warmth, and
optimism are all mentioned in popular
definitions of EI(Mayer et al., 2001).
Approaches to EI can be divided into two
categories: (2) mixed models (Mayer &
Salovey, 1997); and (1) ability models (Mayer
et al., 2001). EI is conceptualised in ability
models in a similar way to cognitive
intelligence (i.e., IQ). EI is thought to grow
during time, be connected with IQ tests, and be
measured through a performance-based
test(Rosete & Ciarrochi, 2005).
Emotional Competencies
The Emotional Intelligence Quotient is a set of
skills that demonstrate the capacity to recognise
one's own actions, feelings, and urges and to
control them appropriately in different
situations. Emotional empathy, attention to, and
discrimination of one's emotions; accurate
recognition of one's own and others' moods;
mood management or control over emotions;
response with appropriate (adaptive) emotions
and behaviours in various life situations
(especially to stress and difficult situations);
and balancing of honest expression of emotions
against courtesy, consideration, and respect are
all considered to be part of "emotional
intelligence" (i.e., possession of good social
skills and communication skills). The five
emotional intelligence qualities such as social
competencies, personal competencies, self-
discipline, motivation and self-expectation have
been shown to contribute more to job success
than technical skills, cognitive capacity, and
traditional personality attributes
combined(Posky, 2017)
METHODS
The systematic review method proposes to map
and assess a specific body of literature to
recognize potential research gaps and produce a
reliable knowledge stock(Tranfield et al.,
2003). No geographical restrictions were used,
Journal of Positive School Psychology
and the considered period for data collection
was papers published until January 2021.
The academic databases selected were Science
Direct, Scopus and Web of Science. To achieve
the highest level of applicability, only peer-
reviewed complete articles written in English
and published in International Journals were
selected, whereas master and doctoral
dissertations, notes, conference papers were
excluded in this review. The inclusion criteria
were papers that contained on the title, abstract
or keywords one or more of the emotional
intelligences. The results were downloaded in
RIS format, and the RIS data was exported to
Zotero bibliography software, as the remaining
procedures were carried out there.
In all, 342 articles were collected, 52 were
selected for the analysis. We examined the
content of each paper as a means to ensure that
the article helps to answer the main research
question of how emotional intelligence helps in
the workplace atmosphere.
RESULTS AND DISCUSSION
Emotional differences among gender
Emotional intelligence (EI) is said to have a lot
to offer businesses in the twenty-first century,
including better performance and productivity.
EI predicts group performance and
organisational civic behaviour, for example. Its
popularity has changed workplace perspectives
on effective management methods, and EI
competencies are taking centre stage in staff
selection, according to the thriving management
consultant business.
Medicine, nursing, pharmacy, the legal
profession, leaders, managers, executives, and
students, to name a few industries and
professional groupings, all offer EI training.
Emotion is no longer seen as an irrational force,
but rather as something that can be managed and
made functional with the right training. EI can
be measured in test-like and self-report
situations to produce a score that can be used to
guide competency training, team building,
Dr. Sharfras Navas
learning and development programmes, and
career development planning. In psychological
accounts, differences in emotional intelligence
between men and women have been examined.
Women appear to be winning in the EI game
since the notion challenges more masculinized,
male-dominated industries. However, because
psychological framing has dominated EI
research, a more sophisticated sociological
perspective that recognises structural issues has
been overlooked. To yet, sociological critiques
of EI have focused on a variety of topics, but
gender has been overlooked. For example,
accounts have looked into the commodification
and instrumentalization of EI at work, its
potential for being used in more manipulative
ways, its elitist position, which undercuts
people's subjective experiences and emotional
valuations, its cultural insensitivity, and
incursion into identity formation as a form of
self-surveillance(Thory, 2013).
Workplace
The workplace is a relational setting in which
positive and negative behaviours are displayed,
each having a different impact on the business
and its people. The topic of "Can employee’s
emotional intelligence influence these
behaviours?" has picked researchers' interest.
Because emotions are one of the most
fundamental factors influencing human
behaviour, emotional intelligence may have a
significant impact in employee behaviour.
Emotional intelligence has been found in other
studies to improve employees' positive
discretionary behaviours while reducing their
negative discretionary behaviours in the
workplace(Dirican & Erdil, 2020).
Workplace Environment
The workplace environment is always
changing, reflecting its fluidity and
unpredictability. Not only in terms of age, but
also in terms of nationality, the workforce is
becoming increasingly diversified. A more
productive workplace is associated with a better
5668
working environment. Workplace unity builds a
stronger organisation that can survive the
rigours of today's fast-paced corporate climate.
Psychologists, educators, leadership theorists,
and business executives are all interested in
Emotional Intelligence. Workplaces now have a
diverse workforce as a result of globalisation.
People are formed in the workplace. There must
be relationships between persons. Employees
with strong interpersonal skills should be
valued as such. When it comes to the workplace,
emotional intelligence refers to the ability to
recognise, express, analyse, and manage
emotions in a professional and successful
manner; the workplace has evolved into a place
where individuals or employees can find their
entire sense of purpose(Chin et al., 2011).
Individuals, leaders, and teams' EI has emerged
as a significant determinant in workplace
performance. A company's ability to respond to
customers or other stakeholders with more
creative and innovative products and solutions
is enhanced by diversity. EI is essential in each
situation where people interact. Differences in
beliefs, language, behaviour, tastes, and
customs are constantly bombarding people
everywhere they encounter, but especially in the
workplace. The ideal workplace is one that is
guided by deliberate actions that are enriched by
purpose, contributions, and
meaning(Gardenswartz et al., 2010).
The workplace is a place where feelings are
abundant. Several studies have found that
emotionally intelligent people have a
favourable impact on several parts of the
workplace. Workplace EI appears to be the
deciding factor. Several social scientists have
conducted empirical study on organisational,
group, team, and individual job performance,
and found that all of the aforementioned
components of the workplace are positively
connected to EI. The study of emotion in
organisations focuses on the individual's
emotion as well as the emotions that are noticed
in the workplace and the environment in which
they are exposed. We talked about facial
5669 Journal of Positive School Psychology

expressions expressing various emotions in Emotions at workplace

various cultural situations. As a result, EI Although culture has been shown to be a
appears to be a crucial factor in determining significant force in businesses, as it has the
workplace success. The study of emotion in the ability to alter people's beliefs, behaviours, and
workplace focuses on the individual's emotions emotions at work. Emotional intelligence (EI),
as well as the emotions that are seen in the a construct that may give a realistic means of
workplace and the environment in which they studying how the emotional capacities of teams
are exposed(Thomas & Kamalanabhan, 2012). and people contribute to workplace outcomes,
has recently been the focus of research on the
Workplace learning relevance of emotions in the
workplace(Downey et al., 2011).
Workplace learning refers to learning that
occurs while performing a job or that occurs
Stress relationship with EI
predominantly within the framework of the
The Organizational Stress Screening Tool
workplace. This term refers to a variety of non-
(ASSET), a new organisational screening tool
formal learning techniques, both planned and
that is an enhanced version of the well-known
unplanned, such as mentorship, coaching, team
and widely used Occupational Stress Indicator,
learning, special tasks, expanding job scope and
was used to assess workplace stress (OSI).
challenge, self-directed learning, and the
However, OSI is designed largely for use with
abundance of incidental learning that occurs
white-collar and managerial personnel, and it
from ordinary work experience. As a result, it
takes a long time to complete. Additional
has been discovered that workplace learning is
measures to cross-validate findings of the
particularly crucial for developing the wide
association between EI (e.g., observers'
variety of abilities required in today's
evaluations, 360 degree feedback), workplace
workplace, particularly in respect to problem-
stress (e.g., electro-physiological markers of
solving, creativity, and teamwork. Most
stress) and organisational commitment will be
importantly, such methods are said to be better
beneficial in future studies (e.g., absenteeism,
than traditional training approaches in
turnover, etc.)(Nikolaou & Tsaousis, 2002).
facilitating the transfer of tacit and procedural
Today, one of the most common reasons people
knowledge, which are crucial components of
seek executive coaching is to improve their
job performance, among organisational
emotional intelligence: strong EQ abilities may
members.
provide one a competitive advantage in the
office and make work life significantly more
The artificial separation that has frequently
enjoyable(Fernandez, 2007).
defined approaches to both learning and
employment is challenged by workplace
Bringing Emotional Intelligence to the
learning, so that involvement in the workplace
Workplace
and learning are regarded as integrally
When the book Emotional Intelligence was first
interwoven within the same process.
published in 1995, many business leaders
Individuals make sense of emotional knowledge
recognised that personal traits like tenacity,
by continuing reflection on emotional
self-control, and the ability to get along with
experiences and discourse within various
others have a significant impact on
workplace social structures, according to the
performance. They could point to "super
research. Sharing and conveying work-related
salespeople" who had an incredible capacity to
tales is recognised as a key strategy for
discern what consumers cared about and build a
strengthening social bonds and enhancing
trusting relationship with them. They might
individual and collective learning in the subject
point to customer service representatives who
of workplace learning(Clarke, 2006).
excelled in calming irate consumers and helped
Dr. Sharfras Navas
them be more reasonable about their complaints
about the product or service. They could also
point to bright executives who excelled at
everything but getting along with people, or
managers who were technically smart but
couldn't handle stress, and whose careers were
blocked as a result of these flaws.
Business leaders realised how important these
"emotionally intelligent" individuals are to a
company's success. But what about the
Figure 2 : Bringing Emotional Intelligence to Workplace
Sources : (Adler et al., 1998).
CONCLUSION
Emotional intelligence is crucial in the
workplace while cognitive talents are thought to
help people find jobs. Emotional intelligence is
seen to help people advance in their careers. It
is seen to be twice as valuable as analytical and
technical skills in the job. EI has the potential
to make a huge difference in our work lives.
That is why it is critical to comprehend what it
is and how it affects the workplace. People
differ in their personalities, emotional
capacities, and strengths, and these differences
5670
numerous employees who lack these crucial
emotional skills? Is it feasible for people to
improve their social and emotional skills? This
is a question that many company leaders are
unsure about. When asked about the value of
emotional intelligence at work, the dean of a
prominent business school, for example,
warmly agreed. Hundreds of experts are now
selling workshops and seminars to assist people
improve their emotional and social skills(Adler
et al., 1998).
can have a significant impact on how they
perform. As a result, it appears that EI has an
impact on a growing number of workplace
behaviours, including stress management.
Individual differences have also been
highlighted in research on more recent and
emerging workplace challenges such as
occupational stress and organisational
transformation. Workplace EI, is about
attaining one's goals "through the ability to
control one's own feelings and emotions, to be
attentive to and influence other critical
individuals, and to balance one's motives and
5671
drives with conscientiousness and ethical
behaviour." EI may play a moderating impact in
aggressive behaviours and job stress.
REFERENCES
1. Adler, M., Cherniss, C., Goleman, D.,
Emmerling, R., & Cowan, K. (1998).
Bringing Emotional Intelligence to the
Workplace: A Technical Report Issued
by the Consortium for Research on
Emotional Intelligence in
Organizations.
2. Ashkanasy, N. M., Ashton-James, C.
E., & Jordan, P. J. (2003).
PERFORMANCE IMPACTS OF
APPRAISAL AND COPING WITH
STRESS IN WORKPLACE
SETTINGS: THE ROLE OF AFFECT
AND EMOTIONAL
INTELLIGENCE. In P. L. Perrewe &
D. C. Ganster (Eds.), Emotional and
Physiological Processes and Positive
Intervention Strategies (Vol. 3, pp. 1–
43). Emerald Group Publishing
Limited.
https://doi.org/10.1016/S1479-
3555(03)03001-4
3. Carminati, L. (2021). Emotions,
Emotion Management and Emotional
Intelligence in the Workplace:
Healthcare Professionals’ Experience
in Emotionally-Charged Situations.
Frontiers in Sociology, 6, 640384.
https://doi.org/10.3389/fsoc.2021.6403
84
4. Chin, S., Anantharaman, R. N., Faculty
of Business and Law, Multimedia
University, Melaka, Malaysia, Tong,
D., & Faculty of Business and Law,
Multimedia University, Melaka,
Malaysia. (2011). The Roles of
Emotional Intelligence and Spiritual
Intelligence at the Workplace. Journal
of Human Resources Management
Research, 1–9.
https://doi.org/10.5171/2011.582992
Journal of Positive School Psychology
5. Clarke, N. (2006). Developing
emotional intelligence through
workplace learning: Findings from a
case study in healthcare. Human
Resource Development International,
9(4), 447–465.
https://doi.org/10.1080/136788606010
32585
6. Dirican, A. H., & Erdil, O. (2020). The
influence of ability-based emotional
intelligence on discretionary workplace
behaviors. Journal of Human Behavior
in the Social Environment, 30(3), 369–
382.
https://doi.org/10.1080/10911359.2019
.1687388
7. Downey, L. A., Roberts, J., & Stough,
C. (2011). Workplace culture
emotional intelligence and trust in the
prediction of workplace outcomes.
International Journal of Business
Science & Applied Management
(IJBSAM), 6(1), 30–40.
8. Fernandez, C. S. P. (2007). Emotional
Intelligence in the Workplace. Journal
of Public Health Management and
Practice, 13(1), 80–82.
9. Gardenswartz, L., Cherbosque, J., &
Rowe, A. (2010). Emotional
intelligence and diversity: A model for
differences in the workplace. Journal of
Psychological Issues in Organizational
Culture, 1(1), 74–84.
https://doi.org/10.1002/jpoc.20002
10. Goleman, D. (1995). Emotional
Intelligence. Bantam Books.
11. Mayer, J. D., & Salovey, P. (1997).
What is emotional intelligence? In
Emotional development and emotional
intelligence: Educational implications
(pp. 3–34). Basic Books.
12. Mayer, J. D., Salovey, P., Caruso, D.
R., & Sitarenios, G. (2001). Emotional
intelligence as a standard intelligence.
Emotion, 1(3), 232–242.
https://doi.org/10.1037/1528-
3542.1.3.232
Dr. Sharfras Navas 5672

13. Nikolaou, I., & Tsaousis, I. (2002). https://doi.org/10.1111/1467-

EMOTIONAL INTELLIGENCE IN 8551.00375
THE WORKPLACE: EXPLORING
ITS EFFECTS ON OCCUPATIONAL
STRESS AND ORGANIZATIONAL
COMMITMENT. The International
Journal of Organizational Analysis,
10(4), 327–342.
https://doi.org/10.1108/eb028956
14. Posky, M. (2017, October 15). The
Importance of Emotional Intelligence
in the Workplace: Why It Matters More
Than Personality. ZERORISK HR.
https://www.zeroriskhr.com/post/the-
importance-of-emotional-intelligence-
in-the-workplace-why-it-matters-
more-than-personality
15. Rosete, D., & Ciarrochi, J. (2005).
Emotional intelligence and its
relationship to workplace performance
outcomes of leadership effectiveness.
Leadership & Organization
Development Journal, 26(5), 388–399.
https://doi.org/10.1108/014377305106
07871
16. Thomas, S., & Kamalanabhan, T. j.
(2012). Measuring emotional
intelligence of the workplace.
International Journal of Business
Innovation and Research, 6(4), 418–
432.
https://doi.org/10.1504/IJBIR.2012.04
7275
17. Thory, K. (2013). A Gendered Analysis
of Emotional Intelligence in the
Workplace Issues and Concerns for
Human Resource Development.
Human Resource Development
Review, 12, 221–244.
https://doi.org/10.1177/153448431246
9100
18. Tranfield, D., Denyer, D., & Smart, P.
(2003). Towards a Methodology for
Developing Evidence-Informed
Management Knowledge by Means of
Systematic Review. British Journal of
Management, 14(3), 207–222.
 sustainability
Article
Emotional Labor, Rapport, and Word of Mouth in
Fitness Organizations
Ye Hoon Lee 1 and Seunghyun Hwang 2, *
Citation: Lee, Y.H.; Hwang, S.
Emotional Labor, Rapport, and Word
of Mouth in Fitness Organizations.
Sustainability 2022, 14, 9968. https://
doi.org/10.3390/su14169968
Received: 15 July 2022
Accepted: 11 August 2022
Published: 12 August 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Sustainability 2022, 14, 9968. https://doi.org/10.3390/su14169968
1 Division of Global Sport Industry, Hankuk University of Foreign Studies, Yongin 17035, Korea
2 Department of Leisure Sport, Kyungpook National University, Daegu 37225, Korea
* Correspondence: hwangsh@knu.ac.kr; Tel.: +82-54-530-1437
Abstract: The fitness service industry requires employees to interact with members, placing great
importance on their emotional expression to manage service quality and the subsequent organi-
zational outcomes. Emotional labor may be a significant psychologically related factor that plays
a critical role in building pleasant interactions between a fitness organization’s employees and its
members. Thus, grounded in the emotional contagion theory, this study examines the predictability
of perceived emotional labor strategies on the development of a rapport between a fitness organi-
zation’s employees and its members, and the subsequent effect of this rapport on word of mouth
regarding the fitness organization. A total of 353 fitness members in the United States participated
in the modified versions of paper-and-pencil surveys pertaining to the proposed variables. The
findings showed that perceived surface acting, or superficial interaction, is negatively related to
rapport, whereas perceived deep acting, a more profound form of interaction, is positively related to
rapport. Furthermore, rapport was positively associated with word of mouth. This result indicates
the significance of emotional labor in the service delivery of fitness employees and developing a
positive relationship with fitness members.
Keywords: behavioral intention; fitness center; perceived authenticity; personal bond; emotional
contagion theory
1. Introduction
Fitness employees who provide frontline services to members play a vital role in
building competitive advantage in fitness organizations, as they have the potential to form
pleasant relationships with members [1]. Indeed, the previous literature has reported
that strong personal bonds between service employees and customers have significant
implications for favorable organizational outcomes, such as customer satisfaction and
behavioral intentions [2]. Such personal bonds seem more relevant in fitness contexts than in
any other service context, as well as in tangible product contexts, because fitness employees
generally work with repeat customers who have multi-month/year memberships, rather
than with one-time visitors [3]. However, despite the potential importance of personal
bonds in fitness services, scholars have neglected to identify the strategies that might
improve such bonds between fitness employees and members.
According to a prior study, five factors strengthen personal bonds in the customer con-
text: trust, familiarity, interest, intimacy, and rapport [4]. Among them, rapport, described
as a pleasant relationship resulting from service employee-customer interaction [5], was
found to improve mutual trust and the quality of interaction between service providers
and members and to generate positive word of mouth (WOM) [2,4,6]. WOM may work
as an important source of information in the field of fitness services because customers
seldom notice differences in the quality of equipment and programs among fitness organi-
zations. Additionally, fitness organizations seldom use traditional promotional tools, such
as advertising and one-on-one personal selling, which further stresses the importance of
positive WOM as a marketing strategy [3].
https://www.mdpi.com/journal/sustainability
Sustainability 2022, 14, 9968 2 of 12

Despite the potential roles of rapport and WOM in the growth of fitness organizations,
relatively little research has examined the antecedents of these two important customer out-
comes. Thus, further study could allow us to understand how to enhance these outcomes,
which in turn could help fitness organizations to sustain their businesses. Emotional labor,
defined as an employee’s deliberate and purposeful efforts to express certain emotions,
as required by their organization, during their interactions with customers [7] may play
a critical role in this process, as the service quality is often influenced by the employee’s
emotional displays during employee–customer interaction [8]. Although there has been
much research on the role of emotional labor in individual outcomes, such as physical
and psychological wellbeing in various occupations (e.g., [9,10]), scholars have neglected
the role of emotional labor in key customer-focused outcomes. Further, no study has
investigated the effect of employees’ emotional labor on customer outcomes in fitness
service contexts.
Given this research gap, this study attempts to examine the relationships between the
two types of perceived emotional labor strategies among fitness members—perceived rap-
port and WOM. More specifically, this study examines the relationship between
(a) perceived emotional labor strategies and rapport, (b) perceived emotional labor strate-
gies and WOM, (c) rapport and WOM, and the mediating role of (d) rapport alone in the
emotional labor strategies–WOM relationship.

2. Literature Review
2.1. Emotional Labor as an Internal Regulation Process
Emotional labor is defined as an employee’s regulation of both felt emotions and
experienced emotions in their efforts to comply with a company’s emotional display rules
to achieve organizational goals [7]. This definition suggests that employees learn to regulate
their internal emotions and emotional expressions due to the implicit or explicit display
rules imposed by organizations. For example, flight attendants who display positive
emotions toward annoying customers due to their training are said to be engaging in
emotional labor.
Grandey and Gabriel [9] have identified two emotional labor strategies for managing
employees’ emotions. The first is surface acting, referring to superficial interaction wherein
one modifies one’s external expressions, such as a smile or a frown, without modifying
one’s internal feelings as a way to abide by a certain display rule. Hochschild [7] proposed
a second strategy of deep acting, in which one tries to change one’s feelings according
to the display rules in a more profound interaction, meaning that the person attempts to
practice the feeling that is appropriate for the moment. Thus, surface acting only manages
observable expressions, like facial expressions, whereas deep acting is considered to be
“acting in good faith” because of the attempt to change one’s internal emotional state to
meet organizational expectations [10].
Previous studies have reported that different types of emotional labor strategies are
significantly related to key individual outcomes in different ways. For example, ample
evidence exists to demonstrate that surface acting is significantly associated with job
burnout and job satisfaction among athletic coaches [11], physical education teachers [12],
and fitness employees [13]. In conjunction with key customer outcomes, research has
shown that deep acting is a source of favorable customer outcomes, such as customer
satisfaction [14] and perceived customer orientation [15], whereas surface acting is not
significantly associated with these outcomes. Based on the emotional contagion theory,
research has shown that inauthentic displays can affect customers’ service experience and
influence their opinions [15,16], including judgments concerning the expresser’s honesty,
pleasantness, and likeability [16,17], as well as trust and cooperation [18].

2.2. Emotional Labor and Rapport

Rapport, which in French means “to lay a bridge”, refers to the feeling of mutual
closeness in human relationships [19]. This term was first used by the Austrian physician
Sustainability 2022, 14, 9968 3 of 12

Mesmer to describe empathy between therapists and patients during hypnotherapy. Today,
it is widely used in a variety of settings, such as teacher–student relationships in educa-
tion [20], employee–customer relationships in marketing [21], and interactions between
psychotherapists and patients [22]. Looking closely at previous studies, Norfolk, Birdi,
and Patterson [23] described rapport as the expression of positive emotions that customers
generally feel while interacting with salespeople and conceptualized it as a comfortable
relationship between two parties. A good rapport leads to a permissible atmosphere or
relationship in which both parties speak openly and without any problems. In this way,
rapport can be defined as the process of understanding each other’s viewpoints and cre-
ating an atmosphere of trust, based on sympathy and mutual understanding. Although
various elements of rapport have been presented by scholars, Gremler and Gwinner [5]
looked at it from a relationship-oriented point of view and divided it into two categories:
pleasant relationships and personal ties.
Regarding the relationship between emotional labor strategies and rapport, Ford and
Etienne [24] noted that courteous and sincere service is an important factor in forming
rapport among service recipients. Surface acting is perceived as inauthentic by customers
because it creates a discrepancy between what employees internally feel and what they
display externally (i.e., emotional dissonance) [7]. Employees displaying surface acting
are expected to suppress their inner feelings and fake their outward expressions, mak-
ing customers believe they are inauthentic [16]. Conversely, as deep acting does not
generate emotional dissonance, it will yield authentic emotional displays [25]. When em-
ployees perform deep-acting exchanges, they try to match their inner feelings with their
outward expression by changing their perception of the situation, which in turn creates
congruence between felt emotions and expressed emotions. This process develops a pos-
itive perception regarding deep acting among customers and increases their satisfaction
with their interactions with an organization’s employees [9]. Thus, this study posits the
following hypotheses:

Hypothesis 1. Perceived surface acting is negatively related to rapport.

Hypothesis 2. Perceived deep acting is positively related to rapport.

2.3. Rapport and WOM

WOM is a communication effect derived from the information regarding a product or
service passed on from one person to another [26]. According to Westbrook [27], WOM is
an experienced process wherein people deliver concerned messages to other individuals.
Thus, it is highly likely that a messenger can change, simplify, exclude, and even beautify
some content during the delivery process, while information derived by WOM can leave a
strong impression on people’s minds and people can easily believe the content. Moreover,
due to immediate feedback, WOM greatly influences service recovery, corporate reputation,
customer satisfaction, and product sales. Indeed, companies make great efforts to attract
new customers to generate profits, and, as the previous literature has noted, one of the
most effective ways to acquire new customers is the positive WOM effect on existing
customers who have experienced the service [28,29]. Above all, customers believe that
information from family members or friends around them is more reliable than corporate
promotional materials [30,31]. Numerous scholars have empirically studied the effect of
WOM as an outcome variable of customer satisfaction and found that it plays an important
role in generating favorable organizational outcomes [32,33]. Furthermore, research has
shown that respondents who were given positive oral communications about new products
were twice as likely to purchase the product than respondents who received negative
communications [34].
Regarding the relationship between rapport and WOM, the previous literature has
noted that employee-customer rapport has significant implications for customers’ future
behavioral intentions [6,35]. The rationale is that customers with high rapport may have
 behavioral intentions [6,35]. The rationale is that customers with high rapport may have
positive expectations regarding a future service experience with an employee [16] and
positive emotions about that organization [2,6]. Indeed, empirical evidence supports this
notion, in that rapport between tour guides and tourists is positively associated with
WOM
Sustainability [6],
2022, 14, 9968while Macintosh [2] found that customers with high rapport are more likely to
4 of 12

spread positive information about the service. Accordingly, it is posited that:

Hypothesis 3. Rapport is positively
positive related
expectations to WOM.
regarding a future service experience with an employee [16] and
positive emotions about that organization [2,6]. Indeed, empirical evidence supports this
notion,
2.4. Emotional Labor in thatRapport,
Strategies, rapport between
and WOM tour guides and tourists is positively associated with
WOM [6], while Macintosh [2] found that customers with high rapport are more likely to
Taken together, thispositive
spread study information
posits thatabout
the different types
the service. of emotional
Accordingly, labor
it is posited strategies
that:
employed by fitness employees have different implications for their rapport with fitness
members, which in Hypothesis
turn have3. Rapport is positively
positive related to with
associations WOM.WOM. As fitness employees’
adoption of surface acting is more likely to increase the feeling of inauthenticity, fitness
2.4. Emotional Labor Strategies, Rapport, and WOM
members will experience a low level of rapport with them. Conversely, because deep
Taken together, this study posits that the different types of emotional labor strategies
acting entails a employed
high degree of employees
by fitness sincerity,have fitness members
different who
implications encounter
for their fitness
rapport with fitness
employees’ deep-acting
members,behavior
which in will
turndevelop a good
have positive rapport with
associations with them.
WOM. In Asturn,
fitnessrapport
employees’
with a high level adoption of surface
of positive acting will
emotions is more
leadlikely to increase
fitness the feeling
members of inauthenticity,
to engage fitness
in positive
members will experience a low level of rapport with them.
WOM about the particular fitness organization. Thus, we hypothesize the following: Conversely, because deep
acting entails a high degree of sincerity, fitness members who encounter fitness employees’
Hypothesis 4. deep-acting
Rapport mediatesbehavior will develop
the perceived a good
surface rapport with them.
acting–WOM In turn, rapport with a high
relationship.
level of positive emotions will lead fitness members to engage in positive WOM about the
Hypothesis 5. Rapport mediates
particular fitnessthe perceived deep
organization. Thus,acting–WOM
we hypothesizerelationship.
the following:

2.5. Development ofHypothesis 4. Rapport mediates the perceived surface acting–WOM relationship.
Research Model
Building upon the above
Hypothesis discussions,
5. Rapport mediates this study deep
the perceived established
acting–WOM an relationship.
integrated research
model based on both the existing literature and empirical research, to analyze the
2.5. Development
relationship between perceivedof Research Modellabor, rapport, and WOM among fitness
emotional
members (see Figure 1). Building upon the above
More specifically, discussions,
rapport this study established
is hypothesized an integrated
to be negatively research
related
model based on both the existing literature and empirical research, to analyze the relation-
to perceived surface acting and positively related to deep acting. As a result of generating
ship between perceived emotional labor, rapport, and WOM among fitness members (see
inauthentic emotional
Figuredisplays, surface acting
1). More specifically, rapport may lead members
is hypothesized to experience
to be negatively related toaperceived
low
level of rapport, whereas deepand
surface acting acting mayrelated
positively resulttoindeep
a high level
acting. As aofresult
rapport becauseinauthentic
of generating of its
emotional displays, surface acting may lead members to experience
potential to generate authentic emotional expression. Additionally, as fitness members’ a low level of rapport,
whereas deep acting may result in a high level of rapport because of its potential to generate
rapport can generate positive expectations and emotions, it is posited to be positively
authentic emotional expression. Additionally, as fitness members’ rapport can generate
related to WOM. positive expectations and emotions, it is posited to be positively related to WOM.

Figure 1. IntegratedFigure
research model. research model.
1. Integrated
Sustainability 2022, 14, 9968 5 of 12

3. Method
3.1. Procedures and Sample
Participants were recruited from local fitness centers through the convenience sam-
pling method. Upon their acceptance, a paper-and-pencil questionnaire with a cover letter
was distributed to the participants, explaining the purpose of the study and instructions
about data collection. Informed consent was obtained from the participants, which assured
confidentiality, and the participants were encouraged to answer the questions as honestly
as possible. All procedures followed in this study involving human participants were in
accordance with the ethical standards of the institutional and/or national research com-
mittee and the 1964 Helsinki Declaration and its later amendments or comparable ethical
standards (protocol number: 064).
A total of 372 respondents initially participated in the survey but we removed 19 ques-
tionnaires from consideration because of missing values. Thus, 353 validated participants
(response rate = 94.9%) were registered from five local fitness centers in the southeastern
United States. Among them, 206 respondents were male respondents (58.4%), 144 were
female respondents (40.8%), and 3 (0.8%) did not identify their gender. The participants
were between 13 and 76 years of age, with the average age being 34.29 years, with a stan-
dard deviation of 12.61. The major ethnic groups in this study were Caucasian (65.2%) and
African American (20.7%). Most of the participants were high school (37.7%) or college
(53.1%) graduates, and some had a master’s degree or above (9.2%). A few participants
were daily users of their fitness club (9.9%), while others visited four to six times per week
(25.8%), or two to three times per week (32.5%), and some were weekly users (11.3%), with
an average of 36.18 months as registered members of the fitness organizations (Table 1).

Table 1. General characteristics of the respondents.

Demographic Variables Frequency (n) Percentage (%)

Gender
Male 206 58.4
Female 144 40.8
Missing values 3 0.8
Ethnicity
American Indian 4 1.1
African American 73 20.7
Asian and Pacific Islander 6 1.7
Caucasian 230 65.2
Hispanic 27 7.6
Multiple ethnicities 13 3.7
Education
Graduated from high
133 37.7
school
Graduated from college 187 53.1
Master’s degree 33 9.2
Visit frequency in a week
Daily 35 9.9
Four to six times 91 25.8
Two to three times 115 32.5
Weekly 40 11.3
Missing values 72 20.5

3.2. Instrument
3.2.1. Perceived Surface Acting and Deep Acting
We used two three-item measures from Groth et al. [15], originally developed by
Brotheridge and Lee [36], to measure the perceptions of surface and deep acting of fit-
ness employees. The items were modified to fit this study’s context as they involved
customers’ interactions with fitness employees. The six items measured surface acting
(e.g., “Employees showed feelings to me that are different from what s/he actually felt”)
Sustainability 2022, 14, 9968 6 of 12

and deep acting (e.g., “Employees worked hard to feel the emotions that s/he needed to
show to me”) on a response scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Groth et al. [15] showed composite reliability scores of 0.90 and 0.92 for surface acting and
deep acting, respectively.

3.2.2. Rapport
Gremler and Gwinner [5] developed a six-item measure with two dimensions—personal
connection and enjoyable interaction—for measuring rapport between customers and
employees and demonstrated acceptable psychometric properties. For the present study,
we adapted only three items from the category of enjoyable interaction from the customers’
perspective (e.g., “Employees create a feeling of warmth in our relationship”) and used
a response scale ranging from 1 (strongly disagree) to 7 (strongly agree). Arnold and
colleagues [37] showed an average variance-extracted (AVE) score of 0.67, while Hwang
and Lee [6] showed a composite reliability score of 0.93 and an AVE score of 0.70.

3.2.3. WOM
Three items that were cited from Hennig-Thurau et al. [38] and Hwang and Lee [6]
were modified to fit the context of this study and were used to measure the WOM of the
fitness-club customers. The respondents were asked to indicate the degree to which they
tried to provide their opinions about their fitness club, on a response scale ranging from
1 (strongly disagree) to 7 (strongly agree). A sample item is “I would like to recommend
this facility to my friends.” Hwang and Lee [6] reported adequate statistical fitness for a
three-item structural model with higher factor loadings (above 0.90), a composite reliability
score of 0.95, and an AVE score of 0.88.

4. Data Analysis
A two-step data analysis was performed. In the first step, descriptive statistics, includ-
ing checking normality and bivariate correlations between the latent variables (i.e., deep
acting, surface acting, rapport, and WOM), were calculated to test key assumptions about
the data. The proposed model was examined via structural equation modeling (SEM) in the
second step. As Anderson and Gerbing [39] suggested, confirmatory factor analysis (CFA)
was first conducted to explore the validity and reliability of the measurement model by
examining factor loadings, composite reliability, and AVE scores. Teo et al. [40] suggested
factor loadings of 0.40, composite reliability of 0.70, and an AVE score of 0.50 as the cutoff
point for adequate psychometric properties. Next, the hypothetical paths for three alterna-
tive models, namely, the direct-effects model (Model A), the fully mediated model (Model
B, as shown in Figure 1), and the partially mediated model (Model C), were specified based
on the measurement model, using Mplus (Version 4.21, Los Angeles, CA, USA). To estimate
the parameters and test the model fit, the covariance matrix and the maximum likelihood
estimation method were used, and the x2 difference test was performed to compare Models
A, B, and C. Regarding the mediation effects, a bias-corrected bootstrapping method is
recommended to establish confidence intervals for the mediation effect [41,42]. Despite
chi-square values of less than 3.0, representing an acceptable fit [43], other fitness indices,
such as the root mean square error of approximation (RMSEA), comparative fit index (CFI),
and Tucker–Lewis index (TLI), which is also known as the non-normed fit index (NNFI),
were used to evaluate the model fit. Hu and Bentler [44] suggested that the CFI and TLI
values should be higher than 0.90 and the RMSEA should be lower than 0.08 to create an
acceptable structural model.

5. Results
5.1. Preliminary Analyses
Skewness and kurtosis were checked and confirmed for the normality assumption
of the 12 items (Table 2). The skewness ranged from −1.02 to 0.66, and −0.42 and 0.82
were the minimum and maximum values of kurtosis, respectively, which satisfied the
Sustainability 2022, 14, 9968 7 of 12

criteria set (i.e., less than an absolute value of 2 as skewness and of 7 as kurtosis) for a
normal distribution [45]. Table 1 displays the descriptive statistics and the correlation
coefficients of the latent variables. The internal consistency of the items on the latent
variables (i.e., Cronbach’s alpha) was higher than the acceptable level (0.70). Regarding
the relationships among the variables, surface acting showed negative relationships with
deep acting (r(352) = −0.24, p < 0.001) and WOM (r(352) = −0.24, p < 0.001), while deep
acting showed positive associations with rapport (r(352) = 0.56, p < 0.001) and WOM (r(352)
= 0.43, p < 0.001). The strength of the relationship between rapport and WOM was highest,
r(352) = 0.80, p < 0.001.

Table 2. Descriptive statistics and correlation between the focal factors.

1 2 3 4 Mean SD Skewness Kurtosis Cronbach Alpha

SA 3.22 1.57 0.66 −0.42 0.89
DA −0.24 ** 4.72 1.35 −0.56 −0.23 0.85
RP −0.13 ** 0.56 ** 5.23 1.32 −0.96 0.68 0.88
WOM −0.24 ** 0.43 ** 0.80 ** 5.50 1.27 −1.02 0.82 0.93
Note: SA = surface acting. DA = deep acting. RP = rapport. WOM = word of mouth. Statistical significance at 0.01
is marked **.

5.2. Measurement Model

CFA was performed to confirm the measurement model of surface and deep acting,
rapport, and WOM as latent variables. With a total of 12 observations for the first measure-
ment, each of the three observed measures was specified for each latent variable, and the
results indicated a good model fit: χ2 = 104.309 (df = 47, p < 0.001), CFI = 0.98, TLI = 0.97,
RMSEA = 0.06, SRMR = 0.04. As shown in Table 3, all standardized loadings were higher
than 0.70, and composite reliability and AVE satisfied the cutoff point for a psychometrically
sound measurement model.

Table 3. Results from the measurement model.

Standardized
Factors SE Estimate/SE CR AVE
Estimate
Perceived Surface Acting 0.89 0.73
The employee just pretended to have the emotions s/he displayed to me. 0.88 0.07 100.13 ***
The employee put on a “mask” in order to display the emotions his/her boss wants
0.89 0.07 100.85 ***
him/her to display
The employee showed feelings to me that are different from what s/he actually felt. 0.80 0.07 110.82 ***
Perceived Deep Acting 0.85 0.66
The employee tried to actually experience the emotions s/he had to show to me. 0.75 0.06 110.85 ***
The employee worked hard to feel the emotions that s/he needed to show to me. 0.83 0.05 140.17 ***
The employee made a strong effort to actually feel the emotions that s/he needed to
0.84 0.04 190.19 ***
display toward me.
Rapport 0.88 0.71
The employees create a feeling of “warmth” in our relationship. 0.83 0.03 320.04 ***
The employees relate well to me. 0.86 0.02 390.35 ***
I have a harmonious relationship with the employees. 0.85 0.03 270.85 ***
Word of Mouth 0.91 0.78
I am likely to spread positive word of mouth about this fitness center. 0.89 0.02 470.35 ***
I would recommend this fitness center to my friends. 0.90 0.01 940.08 ***
If my friends were looking for a fitness center, I would tell them to try this facility. 0.86 0.01 720.22 ***
Note: SE = standard error. CR = composite reliability. AVE = average variance extracted. Statistical significance at
0.001 is marked ***.

5.3. Structural Equation Modeling and Mediation Test

In the Model A specification, WOM was only directly regressed on surface acting,
deep acting, and rapport. For the full mediation of rapport, as shown in the hypothesis,
WOM was regressed only in terms of rapport, which in turn was regressed on surface
acting and deep acting (Model B). In Model C, the direct paths from surface acting and
deep acting to WOM were set to examine the partial mediation effect of rapport on the
Sustainability 2022, 14, 9968 8 of 12

relationship between surface acting and deep acting and WOM. As Table 4 shows, the most
parsimonious model was the partial mediation model. However, the RMSEA value was
0.086, which was not a cutoff point, as suggested by Hu and Bentler [44]. One covariance
between latent variables (i.e., surface acting–deep acting) and one covariance between
error variances (i.e., items 4–7) was set based on the model modification indices for model
improvement. As a result, the RMSEA value dropped to 0.074.

Table 4. Fit statistics of the estimated structural models.

Model χ2 df χ2 /df CFI TLI RMSEA ∆χ2 /df ∆CFI ∆TLI ∆RMSEA
A 183.78 47 3.91 0.907 0.869 0.139
B 108.67 47 2.31 0.958 0.941 0.094 10.6 0.051 0.072 0.045
C-1 94.61 45 2.10 0.966 0.950 0.086 00.21 0.008 0.009 0.008
C-2 78.49 43 1.82 0.976 0.963 0.074 00.28 0.010 0.013 0.012
Note. A = the direct model, B = the fully mediated model, and C = the partially mediated model.

In the final mediation model (Figure 2), all direct paths except the prediction from deep
acting to WOM were significant. These results warrant the acceptance of hypotheses 1–3.
Deep acting showed a higher predictive strength. Surface acting negatively predicted rap-
port (−0.22) and WOM (−0.22), while 0.82 and 0.81 were the coefficients for the prediction
from deep acting to rapport and WOM, respectively. In addition, we found a significant
correlation between deep acting and surface acting (0.23). Apart from the direct paths,
the mediating effects of rapport were significant for hypotheses 4 and 5. The mediating
Sustainability 2022, 14, x FOR PEER REVIEW 9 of 13
effect of rapport in the relationship between surface acting and WOM was −0.17, and that
between deep acting and WOM was 0.66.

Figure2.2. The final

Figure final partial
partialmediation
mediationmodel
modelwas
wasobtained. Values
obtained. are standardized
Values coefficients.
are standardized ** p
coefficients.
< 0.01, * p < 0.05.
** p < 0.01, * p < 0.05.

6.6.Discussion
Discussion
The
Thepurpose
purposeofofthis study
this studywaswasto examine
to examinethe relationships between
the relationships perceived
between emo-
perceived
tional labor labor
emotional strategies, rapport,
strategies, and WOM
rapport, and WOM among fitnessfitness
among members in the in
members United States.
the United
All the hypotheses
States. were supported;
All the hypotheses perceived
were supported; rapport
perceived was negatively
rapport associated
was negatively with
associated
perceived surfacesurface
with perceived acting and
actingpositively associated
and positively with deep
associated withacting.
deep Furthermore, rapport
acting. Furthermore,
was positively
rapport associatedassociated
was positively with WOM. Mediation
with analyses also
WOM. Mediation supported
analyses the hypothesis
also supported the
that
hypothesis that surface acting and deep acting are significantly associated through
surface acting and deep acting are significantly associated with WOM with WOM the
mediating
through the effect of rapport.
mediating The results
effect of thisThe
of rapport. study have of
results significant implications
this study for both
have significant
scholars and practitioners.
implications for both scholars and practitioners.
First,
First,thethefindings
findingssuggest
suggest that
that different
different emotional labor strategies
emotional labor strategies as
as perceived
perceivedby by
fitness
fitnessmembers
membersgenerate
generatedifferent
different customer
customer outcomes in variousvariousways.
ways.More
Morespecifically,
specifically,
fitness
fitnessmembers’
members’rapportrapportisisnegatively
negatively related
related to
to perceived surface acting
acting and
and isispositively
positively
related to perceived deep acting. That is, when fitness members perceive
related to perceived deep acting. That is, when fitness members perceive that fitness that fitness em-
ployees suppress their internal feelings and fake their outward emotional
employees suppress their internal feelings and fake their outward emotional expressions, expressions, they
are lessare
they likely
lesstolikely
have positive
to have feelings
positive toward
feelingstheir relationships
toward with those with
their relationships employees.
those
Conversely,
employees. fitness employees’
Conversely, fitnessauthentic
employees’displays, derived
authentic from their
displays, adoption
derived fromof their
deep
adoption of deep acting, may lead clients to develop a stronger rapport. This is in line
with the previous literature, which reported a positive association between employees’
adoption of deep acting and positive service outcomes, such as customer satisfaction,
perceived customer orientation, and service quality [14,15]. Because emotional labor can
Sustainability 2022, 14, 9968 9 of 12

acting, may lead clients to develop a stronger rapport. This is in line with the previous
literature, which reported a positive association between employees’ adoption of deep
acting and positive service outcomes, such as customer satisfaction, perceived customer
orientation, and service quality [14,15]. Because emotional labor can impact the authenticity
of the resulting emotional display [15], employees’ emotional labor may result in interper-
sonal processes that affect service judgments. This is the first study in the fitness service
context, demonstrating that the way emotions are displayed (i.e., faking vs. authentic) can
generate more favorable customer outcomes than the simple emotions themselves. Thus,
the findings of this study provide a deeper understanding of the association between emo-
tional displays and customer outcomes in sports marketing, as it is not only the different
types of emotions (i.e., positive or negative) that affect customer outcomes but also how
these emotions are displayed. This finding also contributes to the emotional labor literature
by generalizing the relationship between deep acting and important customer outcomes.
Second, this study found the role of rapport with fitness employees to be an im-
portant factor regarding fitness members’ tendency to recommend fitness organizations
to others. This finding is consistent with the studies published by Hwang and Lee [6]
and Macintosh [2]. This implies that when fitness members have pleasant interactions
with fitness employees, they are more likely to disseminate positive information about
the fitness centers to others. The positive relationship between rapport and WOM may
be due to the positive emotions experienced by members and the customer satisfaction
derived from rapport [4]. Gremler and Gwinner [4] noted that when people have a close
relationship with others, they have more positive feelings for each other, which leads to
dedicational behaviors. Additionally, Collier, Barnes, Abney, and Pelletier [46] indicated a
special experience at the service contact point to be a leading variable of positive WOM and
noted that the positive emotions formed through this personal experience led customers to
convey word of mouth as a form of information provision for others. Indeed, empirical
evidence has shown that rapport is directly linked to WOM and is indirectly associated with
WOM through positive emotions and customer satisfaction [2,46]. This finding significantly
contributes to the service marketing literature, as it finds an important antecedent of WOM.
WOM is considered rather difficult to control from a managerial perspective and is simply
dependent on customer satisfaction [43]. This study contributes to marketing research
in that customer service and efforts to build rapport can have significant and important
impacts on WOM.
The current study also found that rapport fully mediated the relationship between
perceived surface acting and WOM, as well as between perceived deep acting and WOM.
Thus, when fitness members notice sincerity and authenticity in employees’ behavior
toward them, they feel a closeness in their interactions, which in turn leads them to talk
positively about the organization to others. Conversely, when they feel that the fitness
employees’ smiles are fake and are only meant to satisfy them, they will be less likely to
feel pleasantness in their relationships with employees, reducing the probability of positive
WOM. This finding adds new knowledge to the emotional labor literature, as this is the
first study to identify a significant path from perceived emotional labor to positive WOM.
This study also identifies the underlying psychological mechanism of rapport, which aids
in a deeper understanding of this relationship.

6.1. Practical Implications

This study has various implications. It shows that not only the formation of rapport
but also the emotional work of fitness employees should be considered important variables
in forming a positive WOM effect. This means that employees who perform emotional
labor should be managed and encouraged to use their emotional performance strategically.
According to the study’s results, deep acting is positively associated with key customer
outcomes, thereby increasing organizational effectiveness, whereas surface acting is highly
relevant to negative customer outcomes. In other words, the performance of emotional
labor for improving employee–customer relationships is ineffective in the case of superficial
Sustainability 2022, 14, 9968 10 of 12

behavior (i.e., surface acting), which may impede service delivery and eventually result in
deteriorating service quality. Therefore, in the process of conducting emotional labor, it
is necessary to use the deep acting method rather than surface acting. To accomplish this,
fitness organizations should implement systematic education and training sessions that
can help employees to accept organizational emotional display rules and apply them to
themselves. In addition, efforts are needed to create and sustain this organizational culture
so that the norms of emotional labor can be established as the core values and traditions of
the organization.
Furthermore, it proposes a new perspective: that beyond the typical existing brand
promotions or advertising strategy, fitness organizations should focus on the management
of fitness employees who have direct interactions with members to spread positive in-
formation about the fitness center. To this end, it can be said that fitness organizations
may implement events designed to shape and sustain positive relationships between their
employees and members. For example, they can plan active promotional events that foster
a pleasant meeting between fitness employees and members to form positive relationships.
This event formation strategy may induce pleasant emotions in the relationship between
visiting customers, and pleasant memories in the future of employees who have been
encountered can be transferred to pleasant emotions in the region and stores. Additionally,
fitness organizations can conduct promotional events on their blogs or homepages, wherein
members can describe their memories of positive interactions with the employees. By
recalling positive experiences with the employees, they may induce or strengthen positive
feelings about the organization.

6.2. Limitations and Future Research Suggestions

Despite the significant theoretical and practical implications, this study also has limita-
tions that need to be considered when interpreting the results. First, this cross-sectional
study could not infer cause-and-effect relationships between variables. Future studies need
to adopt a (quasi-) experimental design to verify the role of perceived emotional labor in
key member outcomes.
Another limitation is based on the process through which the data were collected
solely from fitness members, although it is the fitness employees who perform emotional
labor. Therefore, the study can be more objective when collecting data from emotional labor
performers (i.e., fitness employees). However, this study has a limitation in that it asks
all questions about independent (i.e., emotional labor strategies) and dependent variables
(i.e., rapport and WOM) to the members of the fitness organizations. In future research,
data should be collected from both emotional labor performers and receivers to make the
research more objective.
Furthermore, this study only used rapport and WOM as mediators and outcome
variables in fitness employee–member dynamics. Numerous customer outcome factors
can predict the success of organizations, such as customer satisfaction, perceived service
quality, and pleasant emotions. Other factors can also be proposed that can mediate the
perceived emotional labor–customer outcome variables, such as the detection ability of
customers, customer orientation, and the quality of the relationship. Thus, future work
should consider adding these variables to further understand the role of emotional labor in
influencing customer outcomes.
Finally, as mentioned above, the previously published literature has identified the
five factors of trust, familiarity, interest, intimacy, and rapport as the antecedents of
personal bonds in the employee–customer interface. This study only used rapport as
the main variable and applied it to the fitness context. Future research may use the re-
maining variables and investigate their implications for important customer outcomes in
fitness organizations.
Sustainability 2022, 14, 9968 11 of 12

7. Conclusions
Based on the previously published literature that demonstrated the positive role of
customer experience on the recipients’ memories, satisfaction, and future behavioral in-
tentions, this study chose employees’ adoption of emotional labor strategies as important
variables of customer experience and examined their roles in the promotion of rapport
between the fitness employees and fitness members and the generation of positive WOM.
This study successfully demonstrated that fitness employees’ authentic emotional displays,
produced by deep acting, were positively associated with rapport, which, in turn, was
positively associated with WOM. The findings of this study demonstrate that it is important
to check the fitness of employees’ competence in communicating with members and devel-
oping positive interactions with them, along with the importance of taking the required
measures to develop the same. In this regard, it is necessary to re-examine the relevant
skills and educate fitness employees when hiring and training them. Simple technical skills
and understanding how to use the fitness equipment are only basic knowledge, whereas
communication skills and the ability to display their emotions authentically are additional
skills that are required and that prove their competence.

Author Contributions: Conceptualization, Y.H.L.; data collection, Y.H.L.; methodology, S.H.; writing
(original draft preparation), Y.H.L., S.H.; review and editing, S.H. All authors have read and agreed
to the published version of the manuscript.
Funding: This work was supported by the Hankuk University of Foreign Studies Research Fund
(of 2022).
Institutional Review Board Statement: This is mentioned in the manuscript.
Informed Consent Statement: Informed consent was obtained from all subjects in the study.
Data Availability Statement: The data presented in the study are available upon request from the
corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Chelladurai, P. Managing Organizations for Sport & Physical Activity: A Systems Perspective, 4th ed.; Holcomb Hathaway Publishers:
Scottsdale, AZ, USA, 2014.
2. Macintosh, G. The role of rapport in professional services: Antecedents and outcomes. J. Serv. Mark. 2009, 23, 70–78. [CrossRef]
3. Maguire, J.S. Fit and flexible: The fitness industry, personal trainers and emotional service labor. Sociol. Sport J. 2001, 18, 379–402.
[CrossRef]
4. Gremler, G.; Brown, S. Service loyalty: Antecedents, components, and outcomes. In Winter Educators’ Conference: Marketing Theory
and Application; Grewal, D., Pechmann, C., Eds.; American Marketing Association: Chicago, IL, USA, 1998; pp. 165–166.
5. Gremler, D.D.; Gwinner, K.P. Customer-Employee Rapport in Service relationships. J. Serv. Res. 2000, 3, 82–104. [CrossRef]
6. Hwang, J.; Lee, J.H. Relationships among senior tourists’ perceptions of tour guides’ professional competencies, Rapport,
Satisfaction with the Guide Service, Tour Satisfaction, and Word of Mouth, rapport. J. Travel Res. 2019, 58, 1331–1346. [CrossRef]
7. Hochschild, A. The Managed Heart: Commercialization of Human Feeling; University of California Press: Berkeley, CA, USA, 1983.
8. Parasuraman, A.; Zeithaml, V.A.; Berry, L.L. A conceptual model of service quality and its implications for future research.
J. Mark. 1985, 49, 41–50. [CrossRef]
9. Grandey, A.A.; Gabriel, A.S. Emotional labor at a crossroads: Where do we go from here? Annu. Rev. Organ. Psychol. Organ.
Behav. 2015, 2, 323–349. [CrossRef]
10. Grandey, A.A. Emotion regulation in the workplace: A new way to conceptualize emotional labor. J. Occup. Health Psychol. 2000,
5, 95–110. [CrossRef] [PubMed]
11. Lee, Y.H.; Chelladurai, P. Emotional intelligence, emotional labor, coach burnout, job satisfaction, and turnover intention in sport
leadership. Eur. Sport Manag. Q. 2018, 18, 393–412. [CrossRef]
12. Lee, Y.H. Emotional labor, teacher burnout, and turnover intention in high-school physical education teaching. Eur. Phys. Educ.
Rev. 2019, 25, 236–253. [CrossRef]
13. Lee, Y.H.; Woo, B. Emotional intelligence, emotional labor, and emotional exhaustion among Korean fitness employees. J. Glob.
Sport Manag. 2017, 2, 65–78. [CrossRef]
14. Hur, W.; Moon, T.; Jung, Y.S. Customer response to employee emotional labor: The structural relationship between emotional
labor, job satisfaction, and customer satisfaction. J. Serv. Mark. 2015, 29, 71–80. [CrossRef]
Sustainability 2022, 14, 9968 12 of 12

15. Groth, M.; Hennig-Thurau, T.; Walsh, G. Customer reactions to emotional labor: The roles of employee acting strategies and
customer detection accuracy. Acad. Manag. J. 2009, 52, 958–974. [CrossRef]
16. Frank, M.G.; Ekman, P.; Friesen, W.V. Behavioral markers and recognizability of the smile of enjoyment. J. Pers. Soc. Psychol. 1993,
64, 83–93. [CrossRef] [PubMed]
17. Surakka, V.; Hietanen, J.K. Facial and emotional reactions to Duchenne and non-Duchenne smiles. Int. J. Psychophysiol. 1998, 29,
23–33. [CrossRef]
18. Krumhuber, E.; Manstead, A.S.; Cosker, D.; Marshall, D.; Rosin, P.L.; Kappas, A. Facial dynamics as indicators of trustworthiness
and cooperative behavior. Emotion 2007, 7, 730–735. [CrossRef] [PubMed]
19. Tickle-Degnen, L.; Rosenthal, R. The nature of rapport and its nonverbal correlates. Psychol. Inq. 1990, 1, 285–293. [CrossRef]
20. Lammers, W.J.; Byrd, A.A. Student gender and instructor gender as predictors of student-instructor rapport. Teach. Psychol. 2019,
46, 127–134. [CrossRef]
21. Kim, K.; Baker, M.A. How the employee looks and looks at you: Building customer-employee rapport. J. Hosp. Tour. Res. 2019, 43,
20–40. [CrossRef]
22. Leach, M.J. Rapport: A key to treatment success. Complement. Ther. Clin. Pract. 2005, 11, 262–265. [CrossRef] [PubMed]
23. Norfolk, T.; Birdi, K.; Patterson, F. Developing therapeutic rapport: A training validation study. Qual. Prim. Care 2009, 17, 99–106.
24. Ford, W.S.Z.; Etienne, C.N. Can I Help You?: A framework for the interdisciplinary research on customer service encounters.
Manag. Commun. Q. 1994, 7, 413–441. [CrossRef]
25. Cropanzano, R.; Weiss, H.M.; Elias, S.M. The impact of display rules and emotional labor on psychological well-being at
work. In Emotional and Physiological Processes and Positive Intervention Strategies (Research in Occupational Stress and Well Being);
Perrewé, P.L., Ganster, D.C., Eds.; Elsevier Science/JAI Press: Amsterdam, The Netherlands, 2004; pp. 45–89.
26. Arndt, J. Role of product-related conversations in the diffusion of a new product. J. Mark. Res. 1967, 4, 291–295. [CrossRef]
27. Westbrook, R.A. Product/consumption-based affective responses and postpurchase processes. J. Mark. Res. 1987, 24, 258–270.
[CrossRef]
28. Anderson, E.W. Customer satisfaction and word of mouth. J. Serv. Res. 1998, 1, 5–17. [CrossRef]
29. Roy, S.K. Consequences of customer advocacy. J. Strateg. Mark. 2013, 21, 260–276. [CrossRef]
30. Allsop, D.T.; Bassett, B.R.; Hoskins, J.A. Word-of-mouth research: Principles and applications. J. Advert. Res. 2007, 47, 398–411.
[CrossRef]
31. Verlegh, P.; Moldovan, S. What drives word of mouth: A multidisciplinary perspective. Adv. Con. Res. 2008, 7, 49–51.
32. Khare, A.; Labrecque, L.I.; Asare, A.K. The assimilative and contrastive effects of word-of-mouth volume: An experimental
examination of online consumer ratings. J. Retail. 2011, 87, 111–126. [CrossRef]
33. Moe, W.W.; Trusov, M. The value of social dynamics in online product ratings forums. J. Mark. Res. 2011, 48, 444–456. [CrossRef]
34. Villanueva, J.; Yoo, S.; Hanssens, D.M. The impact of marketing-induced versus word-of-mouth customer acquisition on customer
equity growth. J. Mark. Res. 2008, 45, 48–59. [CrossRef]
35. Gremler, D.D.; Gwinner, K.P.; Brown, S.W. Generating positive word-of-mouth communication through customer-employee
relationships. Int. J. Serv. Ind. Manag. 2001, 12, 44–59. [CrossRef]
36. Brotheridge, C.M.; Lee, R.T. Testing a conservation of resources model of the dynamics of emotional labor. J. Occup. Health Psychol.
2002, 7, 57–67. [CrossRef]
37. Arnold, J.; Hall, K.; Baker, T. Authenticity & rapport in frontline service: Interactional justice for unique customers. In Reimagining:
The Power of Marketing to Create Enduring Value; Fowler, J.G., Weiser, J., Eds.; Society for Marketing Advances: San Antonio, TX,
USA, 2016; pp. 326–327.
38. Hennig-Thurau, T.; Gwinner, K.P.; Gremler, D.D. Understanding relationship marketing outcomes: An integration of relational
benefits and relationship quality. J. Serv. Res. 2002, 4, 230–247. [CrossRef]
39. Anderson, J.C.; Gerbing, D.W. Structural equation modeling in practice: A review and recommended two-step approach. Psychol.
Bull. 1988, 103, 411–423. [CrossRef]
40. Teo, T.; Lee, C.B.; Chai, C.S.; Wong, S.L. Assessing the intention to use technology among pre-service teachers in Singapore and
Malaysia: A multigroup invariance analysis of the Technology Acceptance Model (TAM). Comput. Educ. 2009, 53, 1000–1009.
[CrossRef]
41. Cheung, G.W.; Lau, R.S. Testing mediation and suppression effects of latent variables: Bootstrapping with structural equation
models. Organ. Res. Methods 2008, 11, 296–325. [CrossRef]
42. Preacher, K.J.; Hayes, A.F. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator
models. Behav. Res. Methods 2008, 40, 879–891. [CrossRef] [PubMed]
43. Tabachnick, B.; Fidell, L. Using Multivariate Statistics, 6th ed.; Allyn & Bacon Inc.: Boston, MA, USA, 2013.
44. Hu, L.; Bentler, P.M. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives.
Struct. Equ Model. 1999, 6, 1–55. [CrossRef]
45. West, S.; Finch, J.; Curran, P. Structural equation models with nonnormal variables: Problems and remedies. In Structural Equation
Modeling: Concepts, Issues and Applications; Hoyle, R.H., Ed.; Sage: Newbery Park, CA, USA, 1995; pp. 56–75.
46. Collier, J.E.; Barnes, D.C.; Abney, A.K.; Pelletier, M.J. Idiosyncratic service experiences: When customers desire the extraordinary
in a service encounter. J. Bus. Res. 2018, 84, 150–161. [CrossRef]
 Journal of Personality and Social Psychology Copyright 1989 by the American Psychological Association. Inc.
1989. Vol. 56. No. 3,471-477 0022-3514/89/$00.75

Explanatory Style Across the Life Span]

Evidence for Stability Over 52 Years
Melanie O. Burns and Martin E. P. Seligman
University of Pennsylvania

Analyzed explanatory style across the life span. 30 Ss whose average age was 72 responded to ques-
tions about their current life and provided diaries or letters written in their youth, an average of 52
years earlier. A blind content analysis of explanatory style derived from these 2 sources revealed that
explanatory style for negative events was stable throughout adult life (r = .54, p < .002). In contrast,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

there appeared to be no stability of explanatory style for positive events between the same 2 time
This document is copyrighted by the American Psychological Association or one of its allied publishers.

periods. These results suggest that explanatory style for negative events may persist across the life
span and may constitute an enduring risk factor for depression, low achievement, and physical ill-
ness.

The reformulated learned helplessness model (Abramson,
Seligman, & Teasdale, 1978) suggests that individuals con-
fronted with negative events try to explain those events. Accord-
ing to the model, causal explanations can be analyzed along
three dimensions. Those individuals who characteristically
make internal, stable, and global explanations for negative
events will be at greater risk for depressive deficits in the face of
those events. In other words, people who blame themselves and
who believe that bad events will endure in time and will affect
many areas of their lives are more likely to become depressed.
A recent meta-analysis of 104 cross-sectional studies involv-
ing 15,000 subjects suggests support for this model. Sweeney,
Anderson, and Bailey (1986) reported a highly reliable relation
of moderate size between the postulated pessimistic explana-
tory style and depression. In addition, there are an accumulat-
ing number of longitudinal studies that examine the possibility
that early pessimistic explanatory style may be a risk factor for
later depression, health, and achievement problems (see 'Peter-
son & Seligman, 1984, and Seligman, Kamen, and Nolen-
Hoeksema, 1988, for reviews; see Peterson, Seligman, & Vail-
lant, 1988, and Seligman & Schulman, 1986, for recent data).
These studies have used two techniques for assessing explana-
tory style. The first technique is the Attributional Style Ques-
tionnaire (ASQ). This questionnaire asks individuals to imag-
This research was partially supported by a National Science Founda-
tion Graduate Fellowship to Melanie O. Burns and by NIMH Grant
MH19604, NIMH Grant MH40142, NIA Grant AG05590, and a Mac-
Arthur Foundation Research Network on Determinants and Conse-
quences of Health Promoting and Health Damaging Behavior Grant to
Martin E. P. Seligman.
We are grateful to Camilo Castellon for help with the development of
this project, to Clark McCauley for statistical advice, to Peter Schulman
for assistance with data analysis, and to the many undergraduate re-
search assistants whose efforts have contributed to this study.
Correspondence concerning this article should be addressed to Mar-
tin E. P. Seligman, Department of Psychology, University of Pennsylva-
nia, 3815 Walnut Street, Philadelphia, Pennsylvania 19104.
ine that a hypothetical event has happened to them, to provide
a cause for each event, and to rate each cause on a scale ranging
from 1 to 7 for intemality, stability, and globality (Seligman,
Abramson, Semmel, & von Baeyer, 1979). The second tech-
nique is the Content Analysis of Verbatim Explanations
(CAVE). This approach uses independent, trained judges to rate
verbatim causal statements extracted from spoken or written
material on a scale ranging from 1 to 7 for the same three di-
mensions. The CAVE technique has demonstrated high inter-
rater reliability and good intrasubject consistency (Peterson,
Luborsky, & Seligman, 1983). Ratings derived from the CAVE
also correlate significantly with ratings on the ASQ (Peterson,
Bettes, & Seligman, 1985), although validity coefficients tend to
be modest, in the range of .30 (Peterson & Seligman, 1984). In
addition, a comparison of explanatory styles of depressed and
anxious outpatients (Riskind, Castellon, & Beck, in press) pro-
vides some evidence for the discriminant validity of explana-
tory style as measured by the CAVE technique. A discriminant
function analysis based on explanatory style for negative events
correctly classified 92% of the depressed patients and 58% of
the anxious patients. On the basis of these results, it appears
that depression, but not anxiety, is associated with a character-
istic explanatory style.
In a recent study, explanatory style assessed in young adults
by the CAVE technique predicted health later in life. Using 99
members of the Harvard classes of 1939-1945, Peterson et al.
(1988) found that explanatory style at age 25 predicted health
at ages 45 and 65, partialing out health at age 25: The more
pessimistic the individual's style at age 25, the poorer his health
at age 45. The results of this study and of Seligman and Schul-
man's (1986) study of explanatory style and achievement sug-
gest that having a pessimistic explanatory style may have patho-
logical consequences even in normal (nondepressed) popula-
tions. As yet, little research has been done on possible
mechanisms by which explanatory style at one time might lead
to poor health, poor achievement, or depression years later.
How might explanatory style early in life affect health,
achievement, or depression later in life? One plausible hypothe-

471
 472 MELANIE O. BURNS AND MARTIN E. P. SELIGMAN
sis is that explanatory style is a stable trait, a relatively constant
aspect of a given individual's personality. If explanatory style is
stable, then any behavioral or affective deficits related to poor
explanatory style might persist throughout life and constitute
an enduring risk factor for depression, low achievement, and
poor health. However, there have been few attempts to analyze
the stability of explanatory style over time. Our goal in this
study, then, was to test the stability of explanatory style by using
the CAVE technique to rate material written by the same indi-
viduals at two different times separated by approximately 50
years.
Method
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Subjects and Materials
To locate material written by subjects at two times 50 years apart, we
decided to recruit older subjects who had saved diaries or letters written
when they were in their teens or twenties. Announcements requesting
volunteers for the study were posted on bulletin boards in recreational
centers and life-care residences for older citizens. Subjects were also re-
cruited through contact with the Pennsylvania State Bureau of Aging,
which resulted in a number of small newspaper stories, and through
contact with other organizations in which a large proportion of the
members were likely to be more than 55 years of age. Subjects, who
were required to be older than 55 at the time of the study, were asked to
loan to the researchers for photocopying and extracting a diary or letters
that they had written when they were 17-30 years of age.
All subjects completed a Life Style and Attitude Survey in which they
were asked to record, using at least 750 words, the circumstances of
their present life, current interests, hobbies, friends, problems, and so
forth. The questionnaire consisted of three questions:
Question 1: Think of four ways your life style and attitudes might
have changed since you were in your twenties. Vbu might include,
for example, difference in hobbies, social relationships, political or
religious beliefs. Please describe each change in as much detail as
possible, focusing on the way your life is now. How do you feel
about these changes? Use at least 250 words.
Question 2: Try to remember something upsetting that happened
to you within the last year. (Examples could be an argument with
a friend, an illness or accident, a financial problem, etc.) Describe
in detail the negative event and what you did to resolve the prob-
lem. Please use at least 250 words.
Question 3: Describe a recent holiday, party, or other occasion you
shared with friends or relatives. (Examples might include birth-
days, vacation trips, holiday dinners, etc.) What made the occasion
especially pleasant? What difficulties were there in preparing for
the celebration? Would you ever again want to have a get-together
exactly like the one you choose to describe? Please use at least 250
words.
Subjects also filled out a Beck Depression Inventory (BDI; Beck,
Ward, Mendelson, Mock, & Erbaugh, 1961) to assess depressed mood
and signed a consent form.' Inquiries about participation in the study
as well as completion of the surveys were usually handled by mail. All
subjects were paid $20 for their participation in the project.
A total of 82 individuals responded to requests for volunteers that
were posted in approximately 40 locations. Of these, 56 individuals
seemed to fulfill the initial requirements of the study and were invited
to participate. The invitations resulted in 30 subjects who returned all
required materials, 10 who returned only some of the forms, and 16
nonrespondents. Therefore, the sample on which analyses were made
was composed of 30 individuals: 5 men and 25 women. Ages at which
material at Time 1 was written ranged from 16 to 35; the average age
was 20.5 years (SD = 4.8). Age at Time 2 ranged from 57 to 90 and
averaged 72.3 (SD = 9.0). The amount of time between material written
at Time 1 and responses to the questionnaire at Time 2 ranged from 32
to 69 years. The average time between the two measures was 51.8 years
(SD = 9.8) in the final sample.
Procedure
Causal explanations were extracted by several trained extractors as
well as by Melanie O. Burns. The procedure used to extract event-expla-
nation units was the same as described in Peterson and Seligman (1986)
and as detailed in the training manual for extracting and rating sponta-
neous explanations (Schulman, Castellon, & Seligman, 1987). Using
this procedure, a trained extractor reads the material until an event is
located. The extractor then looks for an explanation for that event. To
be extracted, the event-explanation unit must satisfy three conditions.
First, the event must be good or bad from the subject's point of view.
This restriction is necessary because the reformulated learned helpless-
ness model specifies quite different relations between explanatory style
for good events or for bad events and depression. We therefore at-
tempted to use only events that raters could unambiguously evaluate as
positive or negative. Second, the subject must express his or her own
explanation for that event and not simply agree or quote another per-
son's explanation. Third, the unit must contain a clear causal relation
between the explanation and the event, and not simply be a description
of an event, sequence of events, or a justification for an event. This con-
dition was usually met by relying on the inclusion of words such as
therefore or because in the text and was sometimes met by analyzing the
context in which a statement was embedded. It should be noted that
this third condition does not imply that we were testing a theory about
the necessary and sufficient conditions for the generation or identifica-
tion of all causal explanations. Rather, we were using explanatory style
as a hypothetical construct (see Peterson & Seligman, 1984) that has
served in past research as an organizing principle for the evaluation
of the relation between pessimistic statements and depression. Finally,
because previous research has suggested that extractors usually agree in
the identification of event-explanation units, with interextractor reli-
abilities in the range of .80-.90 (Peterson et al., 1985; Peterson & Selig-
man, 1986), in our study we used only one extractor to identify event-
explanation units from a given written source.
Material was randomized by time period over extractors so that event-
explanation units included in material written by a subject at Time 1
were usually not extracted by the same researcher as material written
by the same subject at Time 2. The individuals who extracted event-
explanation units included all units within a sample of writing except
those that were ambiguous or unusually difficult to understand if read
out of context. When a diary contained a great deal more material than
was necessary to ensure reliable ratings, extractions were made from
several periods throughout the year the diary spanned (e.g., from the
first week in every third month). Typically only 1,000 words are consid-
1
We had initially planned to ask all subjects to complete the Attribu-
tional Style Questionnaire (ASQ) along with the other measures, and in
fact the first 12 subjects did. However, to ensure the continued coopera-
tion of our subjects, we rejected this plan when several of the subjects
complained about completing the ASQ and reported feelings of confu-
sion and anxiety about the instructions despite our best efforts to ex-
plain what was required. The ASQ has not been validated for use with
an older population, and our initial attempt to use it with individuals
over age 65 suggests that some revision of the instrument may be neces-
sary in order to ensure validity in this age group.
 EXPLANATORY STYLE IS STABLE 473

Table 1
Ratings by Three Judges of Event-Explanation Units from Two Times
Dimension

Subject and time Judge I S G C

Subject 1
Time 1
Event: I'm a little disappointed. A 1 2 2
Explanation: No mention was made of an B 1 1 2
invitation to the party. C 1 1 1
M 1 1.3 1.6 1.3
Time 2
Event: I experienced the news that my son A 1 4 3
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

was leaving as a little death. B 1 2 2

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Explanation: It was quite unexpected. C 2 2 2

M 1.3 2.7 2.3 2.1
Subject 2
Time 1
Event: I'm feeling so discouraged. A 7 6 6
Explanation: I'm really no good for B 7 6 7
anything. C 7 7 7
M 7 6.3 6.7 6.7
Time 2
Event: I don't try as hard and I probably care A 7 5 5
less. B 7 6 6
Explanation: I feel powerless. C 7 6 7
M 7 5.7 6 6.2
Subject 3
Time 1
Event: I'm almost ashamed to go. A 6 2 2
Explanation: I'm simply a sight in all my old B 7 2 2
clothes. C 4 5 . 2
M 5.7 3 2 3.6
Time 2
Event: I was quite embarrassed. A 7 2 2
Explanation: To have been so absent-minded B 7 3 3
not to think it through the first time. C 7 3 2
M 7 2.7 2.3 4.0
Note. Explanations were chosen to illustrate similarities between ratings at Time 1 and Time 2 and not
because they were necessarily prototypical. I = Internal; S = Stable; G = Global; C = the composite formed
by averaging the I, S, and G measures across all three raters.

ered necessary to generate 5-10 negative and 5-10 positive units, an
amount usually sufficient to produce good intrasubject consistency.
An average of 9.7 explanations for negative events (range, 1-22; SD =
4.5) and an average of 7.6 explanations for positive events (range, 3-16;
SD = 3.2) were extracted from each subject's diary or letters. Slightly
smaller numbers of explanations were obtained from responses to the
Time 2 questionnaires: For negative events at Time 2, the average num-
ber of explanations was 5.9 (SD = 2.8); for positive events at Time 2,
the average was 6.5 (SD = 3.2).
Ratings of explanatory style were carried out by trained judges as
described in Peterson et al. (1983). Each judge rated each explanation
on a scale ranging from 1 to 7 along internal, stable, and global dimen-
sions. Judges were trained to use the entire 7-point scale and to discrimi-
nate between points on the scale using criteria developed by Schulman,
Castellon, and Seligman (1987). Whether judges did in fact use the en-
tire scale was assessed in this study by an analysis of the number of times
an explanation was assigned a particular numerical rating for each of
the three dimensions. These analyses demonstrated an approximately
rectangular distribution for the frequency of ratings at each point on the
scale for each dimension, although there was a "bulge" on the internal
dimension only, with approximately three times as many events being
given a rating of 1. With this exception, ratings were approximately
evenly distributed over all points on the scale for all three dimensions.
In addition, each judge decided whether each event-explanation unit
represented a positive or negative experience for the subject. Ratings of
event-explanation units on which there was agreement as to valence
were then averaged across the three raters, and a composite score was
derived for each dimension (internal, stable, and global) for each sub-
ject. When there was disagreement about valence, the unit was dis-
carded. In addition, the internal, stable, and global categories were
summed for positive and negative events at each time, and composite
positive (Coposl and Copos2) and composite negative (Conegl and
Coneg2) scores were recorded. As the Sweeney et al. (1986) meta-analy-
sis of 104 studies has shown, these composite measures are important
because previous research has demonstrated that those individuals who
make causal explanations for negative events that are scored as high on
 474 MELANIE Q BURNS AND MARTIN E. P. SELIGMAN

all three dimensions are more likely to experience depression, achieve- Table 2
ment problems, or illness than are individuals with high scores on only Intraclass Correlations
one or two dimensions (Sweeney et al., 1986). We also used a summary
statistic derived by subtracting the composite negative score from the Event Time 1 Time 2
composite positive score from each time period (CPCN). Finally, a mea-
sure of hopelessness was derived by adding the stable and global com- Negative
posites for negative events (Hopeless Scale). High scores on this scale Composite .64 .21
reflect the degree to which subjects expect that their present misery will Hopeless .68 .36
Internal .49 -.06
continue over time and that the causes of bad events affect all areas of Stable .52 .41
their lives, capturing the ordinary meaning of hopelessness. Conversely, Global. .78 .54
a measure of hopefulness was derived by adding the stable and global Positive
composites for positive events, reflecting the expectation that good Composite .64 .61
events will endure and enhance all areas of life (Hopeful Scale). Hopeful .69 .61
All ratings of explanatory style were made by judges who were blind Internal .48 .58
Stable .58 .68
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

to the identity of each subject. Blindness of the judges was ensured by

Global .76 .56
This document is copyrighted by the American Psychological Association or one of its allied publishers.

randomizing each subject's event-explanation units with material writ-

ten by other subjects and by identifying each unit by only a code num-
ber. The random presentation of material written by several subjects
prevents raters from developing a response set toward a given individual
on the basis of their assessment of the kinds of responses typically made
by that subject. In other words, raters are unable to identify any given
subject and must therefore rate each explanation as a discrete unit. Pre-
vious research has demonstrated that these ratings can be made reliably,
with interjudge agreement in the range of .70-.95 (Peterson & Seligman,
1986).
Results
Explanatory style for negative events demonstrated reason-
able stability over an average time period of 52 years. The corre-
lation between explanations for negative events at Time 1 and
explanations for negative events at Time 2 was significant (r =
.54, n = 30, p < .002). However, there appeared to be little if
any temporal stability in explanatory style for positive events.
Interrater Reliabilities
All event-explanation units were rated by three raters. To as-
sess agreement between raters, we calculated the alphas (Cron-
bach, 1951) separately for ratings of negative and positive events
at each time period. In general, interrater agreement was good.
Alphas for three raters were as follows: for negative events at
Time 1, a = .81; for positive events at Time 1, a = .76; for
negative events at Time 2, a = .77; and for positive events at
Time 2, a = .81. For subsequent analyses, the ratings from three
raters were averaged for each explanation rated. A sample of
judges' ratings of an event-explanation unit from Time 1 and
an event-explanation unit from Time 2 for each of three sub-
jects can be found in Table 1.
Intrasubject Consistency
Did each subject have a style at each time? Because the num-
ber of explanations available for each subject tended to vary
considerably, the usual techniques for estimating an intraclass
correlation, which assume that responses are recorded for equal
items on equal length tests, are not fully applicable. Therefore,
we analyzed ratings from different situations for each subject
in an analysis of variance (ANOVA), following the methods of
Cronbach, Rajaratnam, and Gleser (1963) and Shrout and
Fleiss (1979),2 and used the resulting sums of squares in the
formula ICC (1, k) = 1 - (MS^/MSt,) in order to estimate the
consistency of each dimension and each composite score at
each time period. The results suggested that the stable and
global dimensions and three of the four composite scores were
consistent, with intraclass correlations equaling .50 or greater
for most categories. The internal dimension was less consistent,
with correlations ranging from —.06 to .58 within each of the
two times. In addition, the consistency of Coneg2 was low, per-
haps owing to the low reliability of the internal dimension that
is one of its components. With this exception, there was at least
modest consistency of explanatory style within-subjects across
different event-explanation units at each time. These results are
shown in Table 2.
Stability of Style Across Time
The composite explanatory style for negative events at Time
1 was significantly correlated with that at Time 2 (r = .54, p <
.002). The relation between these measures was significant and
of similar magnitude even when controlling for variation in the
age of the respondents at Time 2 or the number of years between
the written sources of Conegl and Coneg2. The composite posi-
tive score at Time 1 was not significantly correlated with the
same measure at Time 2(r= .13, ns) The results of a correla-
tional analysis between the subcomponents of the Time 1 and
Time 2 measures can be found in Table 3. The stable dimension
seemed to account for most of the observed correlation across
time of explanatory style for negative events, but by itself it was
only marginally significant (r = .32, p < .10). The internal and
global dimensions of explanatory style appeared to be less sta-
ble across time. Correlations between the two time periods for
2
Shrout and Fleiss (1979) suggested a formula for determining the
interrater reliability of a composite score derived from independent
judges' ratings of various targets. Their method treats each target as a
cell and uses a one-way analysis of variance to determine interjudge
agreement within each cell. We have adapted this technique in order to
determine intrasubject consistency by treating each event-explanation
unit as an independent rating and performing a similar analysis with
the 30 subjects as "targets."
 EXPLANATORY STYLE IS STABLE 475

Table 3 negative events remains reasonably stable over time, there are
Stability of Explanatory Style Across 52 Years either important differences in the kinds of events explained or
the kinds of explanations offered in the two different sources of
Time 1 Time 2 written material (diaries vs. questionnaires), or possibly there
Event M SD M SD is a strong historical or maturational trend of unknown source
in the cohort.
Negative
Composite 8.6 1.5 10.4 1.6 .54**
Hopeless 2.8 0.6 3.6 0.5 .49* Discussion
Internal 3.0 0.9 3.2 1.0 .17
Stable 3.0 0.7 4.0 0.7 .32 Explanatory style for negative events seems to be stable
Global 2.6 0.7 3.3 0.7 .16 across many years of adult life. In contrast, explanatory style
Positive for positive events demonstrates no stability across the adult life
Composite 8.8 1.8 11.8 2.2 .13 span. These findings are of interest in light of recent reports
Hopeful 0.7 .09
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2.8 0.6 3.9

Internal 3.0 1.0 3.9 1.2 .19 of a relation between explanatory style for negative events and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Stable 3.1 0.8 4.4 1.1 .09
Global 2.6 0.7 3.4 0.7 .15
CP-CN 0.2 1.6 1.3 2.8 .11
Note. Hopeless and hopeful are the sums of the stable and global catego-
ries divided by 2. CP - CN is the composite for negative events sub-
tracted from the composite for positive events.
*p<.01. **p<.002.
the Hopelessness scale were nearly equal in magnitude (r = .49,
p < .01) to correlations between Conegl and Coneg2.
Scores on the BDI (at Time 2) for the 28 subjects who com-
pleted the measure were in the predicted direction but were not
significantly correlated with the composite negative explana-
tory style at Time 2 (r = .26, p < .17). The BDI scores were
significantly correlated with the explanatory style for positive
events (r = -.40, p < .04), as predicted by the reformulated
learned helplessness model. The CPCN variable was also sig-
nificantly correlated with BDI scores (r = -.46, p < .02). These
correlations must be interpreted with caution, however. This
study was not designed to test the relation between explanatory
style and depression: No subjects were clinically depressed, and
the ranges of explanatory styles and of BDI scores were quite
restricted; the BDI, for example, ranged from 0 to 16, with an
average score of 4.8 (SD = 3.7). In other words, this is essentially
a normal sample of nondepressed older adults.
Tests for Differences at Time 1 and Time 2
Inspection of the data suggested that scores on all the dimen-
sions at Time 1 were substantially lower than scores on all di-
mensions at Time 2. Although there was no a priori rationale
for expecting all scores to be systematically higher at one time
than the other, we performed paired t tests for negative and posi-
tive composites in order to explore for possible group trends in
explanatory style across historical time. Differences in magni-
tude were significant in both the Coneg and the Copos mea-
sures, with composites at Time 2 substantially higher than those
at Time 1. This means that individuals were making more inter-
nal, stable, and global explanations in their descriptions of both
good and bad events in the material used from Time 2: for nega-
tive events, /(29) = 6.5, p < .0001; for positive events, ?(29) =
6.15, p < .0001, in paired comparisons of means at the two
times. So although the rank ordering of explanatory style for
depression (Sweeney et al., 1986), between explanatory style
and poor health (Peterson et al., 1988), and between explana-
tory style and low achievement (Seligman & Schulman, 1986).
If explanatory style for negative events is a stable aspect of adult
functioning, then individuals with dysfunctional explanatory
styles may be at increased risk for depression, poor health, and
low achievement throughout their lives.
The stability of explanatory style (r = .54) appears to be lower
than the stability coefficients reported by Costa, McCrae, and
Arenberg (1980) for variables derived from the Guilford-Zim-
merman Temperament Survey. These researchers used a large
sample of adult men to analyze the stability over a 12-year pe-
riod of variables such as general activity and masculinity. Re-
sulting correlations between measures at the two times usually
exceeded .70 for all age groups. However, the stability of explan-
atory style compares favorably with the stability of other per-
sonality variables across longer and more directly comparable
time periods. For example, Conley (1984) reported correlations
in the range of .30-.40 for variables such as neurotic tendency
and introversion over a period of 45 years. Similarly, Leon, Gil-
lum, Gillum, and Gouze (1979) analyzed the stability over a 30-
year period of scores on 13 scales of the MMPI and reported
correlations ranging from .28 for Hypochondriasis to .74 for
Social Introversion.
We wonder about the relation between explanatory style and
these more traditional personality constructs. Can explanatory
style be considered a personality "trait"? There are three cri-
teria against which to judge the "traitness" of explanatory style:
stability across time, consistency within a subject, and consis-
tency across domains. The current study most directly ad-
dresses the first criterion. On this criterion (stability across
time), explanatory style—at least for negative events—is stable
across many years of adult life and therefore appears to be a
trait. On the second—intrasubject consistency—explanatory
style displays some consistency for both negative and positive
events and thus appears to be a trait. Whereas some researchers
have suggested that individuals do not actually have a character-
istic "style" for explaining diverse events (Cutrona, Russell, &
Jones, 1984), our results suggest reasonable consistency of ex-
planatory style for different events for three of the four compos-
ites. The third criterion—consistency across situations—is only
peripherally explored in this study. Quite different sorts of
events were discussed in the teenage years and in old age: Typi-
cally, children, grandchildren, and health are discussed in later
 476 MELANIE Q BURNS AND MARTIN E. P. SELIGMAN
years, as opposed to dating, school, and work in early life. De-
spite these differences, style for negative events remained stable
across 52 years. Overall then, we judge explanatory style for
negative events, to be a trait: it is stable across time, consistent
within individuals, and shows some cross-situational consis-
tency.
Despite this stability, there is strong indication of differences
in mean scores at the two times in the absence of any rationale
that readily predicts this difference. This finding contrasts with
results of longitudinal studies of other personality variables,
which generally find that individuals retain similar mean rat-
ings over time (see Costa & McCrae, 1986, for a review). Al-
though it is possible that this result is indicative of some impor-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
tant differences in cognitive functioning as individuals age, our
This document is copyrighted by the American Psychological Association or one of its allied publishers.
data do not allow us to analyze this question. Alternatively, the
most parsimonious explanation for this result may be that the
differences in means are an artifact of using quite dissimilar
written sources for extraction of event-explanation units. At
Time 1, subjects were recording in diaries or letters an ongoing
record of day-to-day happenings that might tend to produce
causal explanations that are unstable and relatively situation-
specific. In contrast, to tell us about the circumstances of their
present life at Time 2, subjects tended to make sweeping gener-
alizations about their lives over a period of years. It may be that
our instructions at Time 2 generated explanations that were
more stable and global than were diary explanations.
Our small (n = 30) and relatively homogeneous sample sug-
gests a need for caution in interpreting these findings. One fac-
tor that might inflate the stability of explanatory style for nega-
tive events is that the individuals in this sample maintained reg-
ular and frequent records of their own ongoing thoughts and
experiences through letters and diaries. It is possible that writ-
ing diaries or letters reinforces stylistic tendencies, and there-
fore generates a more stable style than would infrequent writ-
ing. In addition, we cannot rule out the remote possibility that
the stability of explanatory style in this sample is at least par-
tially an artifact of the use of subjects who had coped well
enough with the vicissitudes of their lives to survive to a mean
age of 72, and who were therefore not motivated to change their
style (Moss & Susman, 1980). Finally, the stability of the com-
posite explanatory style for negative events is high relative to
the theoretical upper limit imposed by the consistency of the
composite at Time 2 and relative to the stabilities of the underly-
ing internal, stable, and global dimensions. Although this does
not invalidate the results, it suggests the need for larger and
more diverse samples in future studies.
Despite these cautions, we think the results are not likely to
be artifacts for two reasons. First, the finding of stability over
time for explanatory style for negative events converges with the
results of another study (Seligman & Elder, 1986) that used a
sample of 28 women from the Berkeley-Oakland growth study.
Content analysis of oral interviews conducted 27 years apart
revealed significant stability for explanatory style for negative
events over the time period analyzed (r = .38) and no stability of
explanatory style for positive events. It is of interest that, despite
differences in methodology and sample characteristics, these re-
sults are very similar to the results of the present study. There-
fore, it seems unlikely that the stability of explanatory style for
negative events is an artifact of the use of diaries and letters.
Second, although the material comes largely from healthy sub-
jects, several of the subjects reported that they had not always
coped well in the past and had in fact experienced many psycho-
logical problems. For example, one woman loaned us material
she had written shortly after hospitalization for a psychotic de-
pression, and another subject started that he had recently
started taking lithium as therapy for recurrent depression.
Taken together, the stability and consistency we found suggest
that individuals may maintain a characteristic explanatory style
for bad events across situations and throughout their lives. This
style may be an enduring risk factor for depression, low achieve-
ment, and poor health.
References
Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. (1978). Learned
helplessness in humans: Critique and reformulation. Journal of Ab-
normal Psychology, 87, 32-48.
Beck, A. X, Ward, C. H., Mendelson, M., Mock, J. E., & Erbaugh, J.
(1961). An inventory for measuring depression. Archives of General
Psychiatry, 4, 561-571.
Conley, J. J. (1984). Longitudinal consistency of adult personality: Self-
reported psychological characteristics across 45 years. Journal of Per-
sonality and Social Psychology, 47, 1325-1333.
Costa, P. T., & McCrae, R. R. (1986). Personality stability and its im-
plications for clinical psychology. Clinical Psychology Review, 6,407-
423.
Costa, P. T., McCrae, R. R., & Arenberg, D. (1980). Enduring disposi-
tions in adult males. Journal of Personality and Social Psychology,
38, 793-800.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of
tests. Psychometrika, 16, 297-334.
Cronbach, L. J., Rajaratnam, N., & Gleser, G. C. (1963). Theory of
generalizability: A liberalization of reliability theory. British Journal
of Statistical Psychology, 16, 137-162.
Cutrona, C. E., Russell, D., & Jones, R. D. (1984). Cross-situational
consistency in causal attributions: Does attributional style exist?
Journal of Personality and Social Psychology, 47, 1043-1058.
Leon, G. R., Gillum, B., Gillum, R., & Gouze, M. (1979). Personality
stability and change over a 30-year period—middle age to old age.
Journal of Consulting and Clinical Psychology, 47, 517-524.
Moss, H. A., & Susman, E. J. (1980). Longitudinal study of personality
development. In O. G. Brim, Jr. & J. Kagan (Eds.), Constancy and
change in human development (pp. 530-595). Cambridge, MA: Har-
vard University Press.
Peterson, C., Bettes, B. A., & Seligman, M. E. P. (1985). Depressive
symptoms and unprompted causal attributions: Content analysis.
Behavior Research and Therapy, 23, 379-382.
Peterson, C., Luborsky, L., & Seligman, M. E. P. (1983). Attributions
and depressive mood shifts: A case study using the symptom-context
method. Journal of Abnormal Psychology, 92, 96-103.
Peterson, C., & Seligman, M. E. P. (1984). Causal explanations as a risk
factor for depression: Theory and evidence. Psychological Review, 91,
347-374.
Peterson, C., & Seligman, M. E. P. (1986). Content analysis of verbatim
explanations: The CAVE technique for assessing explanatory style.
Unpublished manuscript, University of Pennsylvania, Philadelphia.
Peterson, C., Seligman, M. E. P., & Vaillant, G. (1988). Pessimistic ex-
planatory style is a risk factor for physical illness: A thirty-five-year
longitudinal study. Journal of Personality and Social Psychology, 55,
23-27.
 EXPLANATORY STYLE IS STABLE
Riskind, J. H., Castellon, C. S., & Beck, A. T. (in press). Spontaneous
causal explanations in unipolar depression and generalized anxiety:
Content analysis of dysfunctional-thought diaries. Cognitive Therapy
and Research.
Schulman, P., Castellon, C., & Seligman, M. E. P. (1987). Guidelines
for extracting and rating spontaneous explanations. Unpublished
manuscript, University of Pennsylvania, Philadelphia.
Seligman, M. E. P., Abramson, L. Y., Semmel, A., & von Baeyer, C.
(1979). Depressive attributional style. Journal of Abnormal Psychol-
ogy, 88, 242-247.
Seligman, M. E. P., & Elder, G. (1986). Learned helplessness and life
span development. In A. Sorenson, F. Weinert, & L. Sherrod (Eds.),
Human development and the life course: Multidisciplinary perspec-
tives (pp. 377-427). Hillsdale, NJ: Erlbaum.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Seligman, M. E. P., Kamen, L. P., & Nolen-Hoeksema, S. (1988). Ex-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
477
planatory style across the lifespan: achievement and health. In R. M.
Lemer, E. M. Hetherington, & M. Perlmutter (Eds.), Child develop-
ment in lifespan perspective (pp. 91-114). Hillsdale, NJ: Erlbaum.
Seligman, M. E. P., & Schulman, P. (1986). Explanatory style as a pre-
dictor of productivity and quitting among life insurance sales agents.
Journal of Personality and Social Psychology, 50, 832-838.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in as-
sessing rater reliability. Psychological Bulletin, 86, 420-428.
Sweeney, P. D., Anderson, A., & Bailey, S. (1986). Attributional style in
depression: A meta-analytic view. Journal of Personality and Social
Psychology, 50. 974-991.
Received July 21,1987
Revision received August 16,1988
Accepted August 17, 1988 •
 Journal of Management Inquiry
Volume XX Number X
Month XXXX xx-xx
© 2008 Sage Publications
10.1177/1056492608320580
Learning From the Toxic Trenches http://jmi.sagepub.com
hosted at
http://online.sagepub.com

The Winding Road to Healthier Organizations–and to

Healthy Everyday Leaders
Joan V. Gallos
University of Missouri–Kansas City

This article offers an insider’s look at the power and consequences of high levels of workplace emotion in order to deepen
understanding of the workplace relationships and practices needed for individual and organizational success. It begins with
a definition of terms, illustrating the linkages among toxic emotions, organizational effectiveness, leading, and change. It
explores insights and strategies for managing the dynamics and proposes five steps for sustaining healthy leaders in an
increasingly pressurized work world. The article advocates new models for education and training, as well as renewed atten-
tion to the development of theories and structures that promote individual and organizational health.

Keywords: leadership; resilience; toxic emotions; change management; workplace emotions; organizational health

T his article begins with a caution. Handling strong

emotions in the workplace—dealing over time with
others’ frustration, anger, and disappointment resulting
from organizational life in a competitive world of scarce
resources and nonstop change—can be hazardous to
body and soul. It exaggerates feelings of managerial
overload, diminishes creativity, and makes it harder to
resolve everyday dilemmas. It can lull those exposed to
the workplace affect into a complacency that keeps
people and organizations locked in patterns that are pro-
ductive for neither—and that block the development of
structures and strategies for a healthy workplace. It can
test and tax leaders—even those who feel forewarned
and who readily accept the challenge of the work. I speak
from experience—and research confirms that I am not
alone (Frost, 2003, 2005; Frost & Robinson, 1999;
Kellerman, 2004; Lipman-Blumen, 2005; Offermann,
2004; Stark, 2003). I share my learning from life in the
toxic trenches so that others can better prepare to forge
the workplace relationships and practices needed for
individual and organizational success.
This reflection is shaped by my work as a university
administrator and by the experiences of others who have
labored in the toxic trenches in universities and else-
where. It offers an insider’s look at the power and conse-
quences of high levels of workplace emotion and at the
speed with which they develop in response to everyday
business decisions. The purpose is to deepen understanding
of the phenomenon and its impact and to encourage the
development of practices that support organizational
health. Equally important, this reflection explores strate-
gies that support and sustain the health of everyday
leaders—individuals at all levels of the organization who
formally or informally choose to respond to the leader-
ship challenge. I am supported in this reflection by lan-
guage and theory that enable me to name and probe my
experiences. Abraham Maslow (1968) reminded us that
naming—bringing our tacit and subjective experiences
into the world for public exploration and testing—is a
critical step in understanding human behavior and in
developing healthy social structures. Shared personal
reflections promote public exploration. Both are essen-
tial for the creation of positive organizations and of the
processes that enable individuals to lead and to learn
in them.
This article begins with a definition of terms, illustrat-
ing the linkages among toxic emotions, organizational
effectiveness, leading, and change. It explores insights and
strategies for better managing the dynamics and proposes
five steps for sustaining healthy leaders in an increasingly
Author’s Note: My thanks to colleague and friend Dr. Brian Paul for
our regular conversations on the meaning of healthy leadership and
followership and for sharing perspectives and experiences from his
work in clinical psychology.

1
2 Journal of Management Inquiry
pressurized work world. The article ends by advocating
new models for education and training as well as renewed
attention to the development of theories and structures that
promote individual and organizational health.
Handling Organizational Toxins:
Setting the Stage, Defining the Issues
This article is an outgrowth of two projects: a study of
healthy organizations and reflection on work as dean
during a particularly challenging chapter in a campus
history. My dean experiences differed markedly from my
experience in other administrative roles, and I wanted to
understand that—and my own leadership—better.
Experiences that “grab our attention” are clues to some-
thing fundamental we need to know (Dotlich, Noel. &
Walker, 2008).
In the course of the two projects, I happened upon the
concept of toxic emotions in the workplace (Frost, 2003,
2005; Frost & Robinson, 1999; Stark, 2003). I identified
experientially and spiritually with what I read and knew
I had found “something fundamental.” Under the best of
situations, handling the emotional undercurrents that
accompany organizational change and growth takes
time, skill, and care. Conditions of fast-paced change or
unrealistic expectations—self or other imposed—test
and consume. As an informed student of organizations
with a few successful administrative runs under my belt,
I had approached the deanship with confidence and
appreciation for the complexity of the situation. I knew
the school and its history. I respected the human side of
enterprise, knew the human resource literature at the core
of the organizational theory base, appreciated the links
between loss and change, and understood the emotional
nature of organizational decision making (e.g., Ashford,
Rothbard, Piderit, & Dutton, 1998; Ashforth &
Humphrey, 1995, Ashkanasy, Härtel, & Daus, 2002,
Fineman, 2000; Huy, 2002; Maitlis & Ozcelik, 2004).
All that helped, but did not fully prepare me for things
that I found: What it is like, for example, to live—not just
understand—the systemic struggles in leading from the
organizational middle (Gallos, 2002; Oshry, 1995; Sales,
2006), the power and the durability of strong emotion
once rooted into an organization’s culture, and the non-
rational temptation to ignore the buildup of systemic
affect and the personal toll in working with and in it for
the sake of moving an organizational agenda. In fact,
theory and past successes may have made me slow to
recognize the full meaning of what was happening
around me and too quick to assume that strategies and
skills which had worked in the past would work well
again (Watkins, 2008). Bennis (2003) reminded us that
context matters—and in today’s fast-paced, competitive
world of nonstop change, context matters more than ever.
The daily pressures of work life in a sea of disappoint-
ments, frustration, and complaints, for example, encour-
age an implicit model of leadership as firefighting. They
make it harder to keep the whole herd moving roughly
west and focused on the larger mission. They also make it
tough to see or to embrace opportunities for shaping a sup-
portive work culture—using, for example, the leverage of
position to facilitate open communications, conflict resolu-
tion, and understanding among diverse individuals and
groups. Those who see the possibilities can assist their
units in developing compassionate cultures and collabora-
tive efforts that benefit multiple stakeholders. Leaders who
miss that end of being—or needing— an organizational
toxin handler.
The concept of organizational toxin handler may
seem specialized and exotic; however, those who do the
work are quick to understand the everyday nature of the
role and its function. Organizations as social systems are
populated by individuals who respond with a range of
emotions to the challenges, disruptions, and demands
experienced every day at work. Implicit in the organiza-
tional work contract is the expectation that individuals
will absorb and manage their own reactions. In most sit-
uations, they do. They may process pressures and disap-
pointments by talking with friends, going to the gym,
grumping through a bad day at the office, or employing
spiritual practices like meditation or prayer (Delbecq,
2008). Organizational productivity depends on this
informal microprocessing—and is helped when periods
of frustration and employee negativity are brief, low
in intensity, or staggered in occurrence among the
workforce.
There are, however, times when the emotional cards
are dealt too quickly, widely, or often. Situations like
nonstop change, massive turnover, reorganization, bud-
get crunches, downsizing, rigid and unfair policies, or
abusive bosses can raise the emotional ante beyond the
ability of individuals to self-manage and absorb. In those
circumstances, emotions accumulate and can impair the
informal system’s capacities for managing the pain. The
result is unhelpful for all: emotional overload, disrupted
productivity, and a toxic work environment.
A look into the toxicity, however, often reveals certain
individuals taking it upon themselves to ease the over-
load and to keep people focused on their work. Frost
(2003) labeled these people as organizational toxin han-
dlers. We recognize them by their work:

• The coworker who patiently listens to complaints
and who offers solace, good advice, and hope;
• The administrative assistant who informally coun-
sels people on how and when to pitch new ideas to
a hard-driving or idiosyncratic boss;
• The boss who shields subordinates from the shift-
ing demands, unfair critiques, and changing prior-
ities of a volatile senior executive;
• The supervisor who quietly assumes work duties
for employees in need of extra time and care
because of family or health concerns;
• The inner-city school teacher who recognizes that
hungry children do not learn well and brings food
to class.
Naming the phenomenon, as one toxin handler told me,
casts these deeds in a larger light. They are more than
personal acts of kindness or compassion. They are every-
day leadership-in-action: tacit organizational productiv-
ity mechanisms. This kind of informal organizational
leadership is important—and often more pervasive than
recognized.
It is important, however, to distinguish toxin handling
from toxic leadership. Toxic leaders (Kellerman, 2004;
Lipman-Blumen, 2005; Whicker, 1996) are destructive
individuals who abuse their power, role, and followers
for immoral or narcissistic purposes. Toxin handlers are
exemplary workers. They are distinguished by their
empathy and emotional intelligence (Goleman, 1995),
and they foster productivity by ministering to others in
need. As our university, for example, was reeling from
“transformational change” at breakneck speed—and
from the loss of campus leaders discouraged by the
direction and heavy-handedness of the process—a
number of individuals across the campus and throughout
the hierarchy stepped in to support, coach, and advise
others about how to stay focused and keep the university
going. Toxin handling can be episodic, like defending a
coworker from public criticism or representing group
concerns to an unreasonable boss. It may be recurring. In
organizations with chronic dysfunction—those charac-
terized by imposed and unrealistic performance goals or
cultures of blame, unhealthy competition, dishonesty, or
irresponsibility (Frost & Robinson, 1999)—handling
high levels of emotions can be daily work. Toxin han-
dlers, I can confirm, quickly learn to accept the high
levels as a given, a kind of background noise to the work
that must be done. More important, they become hard-
ened to the dangers and often toil away oblivious to the
increasing personal and professional toll.
Gallos / Toxic Trenches 3
The word toxic is strong and evocative, and it is
tempting to limit applicability to dramatic events like
massive layoffs or death of a coworker. Although these
kinds of situations raise intense affect, the power of the
concept is in its pervasiveness. Everyday decisions, poli-
cies, mistakes, and pressures inevitably trigger some
employee pain—and institutional structures, reward sys-
tems, and norms of rationality encourage people and
organizations to push on despite it (Mumby & Putnam,
1992). Ignoring the pain, however, does not eliminate it.
As dean, I found myself and my peers acting as if it
would, focusing on the quick wins and lists of accom-
plishments that influential change models (e.g., Kotter,
2006) tell us compensate for the evoked loss and fear.
They don’t: Anguish just goes underground and can
erupt at unexpected times and in unanticipated places—
a toxic shock to all when it does. Leadership training
rarely emphasizes the central role of counseling-related
skills and sensibilities. In reality, they are invaluable.
Toxic Shock in the Workplace: Multiple
Paths to Overload
It is important for leaders to appreciate both the speed
and the variety of social processes that contribute to
toxic overload. The leadership literature largely ignores
this important set of issues. Toxic diffusion, for example,
can be deliberate efforts by pained individuals to seek
what they see as justifiable recourse or revenge—
repeated public venting of strong emotions or stories
about the circumstances that evoked them, false accusa-
tions, attacks on colleagues or processes, filing of
unfounded grievances or lawsuits, or more insidious
strategies like scapegoating, anonymous letters or post-
ings, gossip, isolating targeted individuals, or vandalism.
The affective impact is fueled by conflicts between those
who sympathize with and those who oppose the toxic
acts and by widespread feelings of powerlessness in
knowing how to resolve the mounting tensions.
Less deliberate but equally powerful is the day-to-day
sharing of gripes and complaints from individuals seek-
ing support or a shoulder to cry on. Toxicity spreads with-
out clear systemic recognition of its source. Research
documents the unconscious contagion of negative emo-
tions (Hatfield, Cacioppo, & Rapson, 1994; Joiner, 1994;
Restak, 2003) and a rapid rise of systemic affect from
empathetic identification with peers who suffer. Maitlis
and Ozcelik (2004) found this true even when others see
the cause as fair and justified. It is common, for example,
4 Journal of Management Inquiry
for many to feel job insecurity when a coworker is fired,
even when everyone has known for a long time that the
discharged employee was performing badly.
Outside interventions can also fuel organizational tox-
icity, especially when they contain implied critiques
from powerful others, distorted information, or data
beyond the system’s coping capacity (Argyris, 1985).
Nonroutine external evaluations, externally driven fact-
findings, and actions that focus undue public attention on
emotion-laden issues raise the stakes and the emotions
that surround all complex decision making. Although
data and an “objective” perspective might seem like a
plausible route to organizational rationality, they can be
anything but. System stress and anguish escalate—as
does the workload—with external impression manage-
ment now added to the plate. External interventions
divert system attention from ongoing challenges—and
may surface data or ask for responses beyond the sys-
tem’s current capacities and resources. They can also
decontextualize and exaggerate isolated events and
actions, and thereby magnify the time, attention, and
emotional investment given to blips in the flow of events
and common bumps on the organizational road.
Organizations also foster toxic diffusion through dis-
torted acculturation practices that socialize new
members—and remind old-timers—to accept or perpet-
uate toxin spreading. Cultures teach whatever cultures
are. An unhealthy, emotion-saturated culture reinforces
its own pathologies and promotes its version of “accept-
able ways” of viewing the workplace or behaving in it.
Organizational socialization can aim at controlling or
promoting role-specific workplace emotions (Van
Maanen & Kunda, 1989)—and in the process teach dis-
torted affective responses, coping strategies, and com-
munication patterns as the right stuff for professional
success. Tacit expectations of differential behaviors and
acceptable levels of emotion for faculty and administra-
tors in universities are a classic example (e.g., Bedeian,
2002; Gallos, 2002).
Finally, it is important to recognize the snowball
effect. Rarely do these processes exist alone, and the
combinations and interactions magnify the impact.
Diffusion processes feed on each other, setting off emo-
tional chain reactions. Individuals and the collective sys-
tem weaken under the cumulative buildup. Both can lose
critical responsive and defensive capacities: Under
stress, everyone regresses to a lower form of functioning.
Optimism is harder to muster—and so is the creativity
needed to generate options for stemming the toxic
spread. Survival and hope rest in periodic moves from
the dance floor to the balcony (Heifetz, 1994) for a new
perspective on the situation and for respite from the din.
Travel Through the Toxic Trenches: In
Search of Grounded Understanding
A snippet of organizational life illustrates the short
path from everyday business decisions to toxic mael-
strom. For a grounded feel of this, imagine yourself as
dean of a professional school in a large, public univer-
sity. For those outside the academy, deans are classic
middle managers. Like their business counterparts, they
have limited authority and resources yet face demands
and pressures from multiple constituents and from up
and down in the hierarchy. The workload is diverse and
heavy. Vaill (1996) noted the large number of problems
and projects delegated to deans—in many cases because
no one knows what else to do.
As dean, you are expected to increase enrollments,
strengthen academic programs, and enhance community
outreach in ways consistent with the campus strategic plan.
You undertake the work in a context of leadership instabil-
ity. You are the fifth dean in 7 years. In your 1st year on the
job, you have already worked with two chancellors and two
chief academic officers. Your school faces lingering ques-
tions about its quality; you share some of the same con-
cerns. Faculty and staff are cordial, yet reticent—maybe
depressed. Maccoby (2004) might diagnose fear of attach-
ment. These individuals bristle from what they see as a
history of unfair attacks, and they hope you will bring
recognition for their efforts. Some fear taking on new
responsibilities and are uncertain what that might entail.
Others wonder if their skills are sufficient to meet new
opportunities. Unit and campus efforts to link rewards with
quality have made even strong performers nervous.
Griping in the hallways has already begun.
As if the system were not stretched enough, add major
state funding cuts. The university proposes an early retire-
ment program and will reach its revenue goals only if a
high number of senior employees leave. Deans will be
allowed to refill some lost positions, but there are no
promises about how this will be done, nor will there be
options to hire senior people for any new openings.
Talented individuals with options quickly accept the pack-
age and interview for other jobs, fueling peer jealousy.
Others agree to simply retire. Programs are unequally hit:
All scramble to regroup in light of anticipated changes.
Feelings of loss and confusion are palpable. Campus goals
for enhancement of services, offerings, and quality are still
there, as are lingering critiques of the unit. Do more with
less is the campus mandate: Demonstrated unit progress is
a criterion for release of all future funds.
Faculty and staff stream through your office—vent-
ing, mourning, celebrating, raging, complaining, advo-
cating, negotiating, begging. Many want answers and

resources you don’t have. Although no one directly tells
you, many blame you for the current mess—you are,
after all, in charge and responsible for the school. Some
who chose retirement experience buyer’s remorse. At
least one contacts a lawyer, works unsuccessfully to void
the signed contract, and feels betrayed by an unyielding
university. Others negotiate part-time roles as transition.
Uncertainty abounds: All the holding on and letting go
complicates planning. Faculty and staff pressure you to
begin hiring immediately. You’d like to, but can’t. Saying
that fuels more angst. Everyone (including you) sees
pressures to advance strategic goals as unfair; however,
none of the deans have been unable to secure any slack,
and pushing back with your boss risks losing credibility
and future resources for the school. The chancellor
advises everyone to hunker down and focus on making
quality improvements.
There is nothing particularly extraordinary in this sce-
nario. Details may change, but the expectations and
problems are not unusual for middle managers—nor is
the widespread affect they trigger. You have been asked
to take a unit forward during times of financial retrench-
ment and are accountable for strategic progress.
Addressing financial realities while advancing the orga-
nization necessitates major change—and change brings
loss and a deluge of human emotion. Those retiring, for
example, face disruption in everything from where and
how to spend their days to how they will express impor-
tant values and needs (Gallos, 2007; Marris, 1986).
People staying on the job have lost colleagues and
friends, as well as established work patterns and norms.
Tacit knowledge for a host of jobs is gone, as are per-
sonal files, informal information networks, and institu-
tional memory that has facilitated work. Stress, bigger
workloads, and insecurity characterize life for those left
behind—as do potential health concerns. Research from
the Finnish Institute of Occupational Health (“Job Cuts,”
2004), for example, found that workers age 19 to 62 who
survived massive staff downsizing were five times more
likely to die from heart disease or stroke in the 3 to 5
years after the cuts than employees from stable work-
places. Maitlis and Ozcelik (2004) attributed this to the
toxic nature of survivor’s guilt. Brockner, Grover, Reed,
& Lee Dewitt (1992) noted the potential for fear and pro-
jection—there but for the grace of God go I. With system
stress, worry, and uncertainty so high, the scene is ripe
for toxic overload. The dean stands in the middle of it all.
Remember too the school’s unique history that exac-
erbates its current state: Accumulated pain from revolv-
ing-door leadership and external critiques magnify the
affect. Although every new leader hopes for a clean slate
and unlimited opportunities, reality is less kind. Every
Gallos / Toxic Trenches 5
leader inherits an organizational history, often with long-
buried, toxic land mines. Pent-up frustration may fade
from organizational consciousness and records, but it
still resides in the social system—as do memories of
ineffective containment strategies employed by past
leaders. New pain rekindles unresolved past emotions.
The toxic beat goes on.
The result is a toxic version of a Cohen and March
(1974) institutional “garbage can.” The social system
begins to act as if hardwired for collective angst:
Emotions are stronger, quicker to trigger, more random,
more disruptive. Anger, frustration, and a host of other
feelings randomly circulate around and through the orga-
nizations. They attach unselectively to available issues,
persons, or projects, undermining decision making and
authority structures while raising the emotional ante for
all. Toxic garbage cans may seem functional to those
involved. This may be the way many have related to the
organization for a long time. Well-meaning leaders may
encourage the venting: Dumping pain is one way to
release frustration and shared powerlessness. On the
other hand, toxic dumping contaminates a broad range of
organizational processes and outcomes, impacts inno-
cent bystanders, and weakens collective capacities to
break the escalating spiral. People feel caught in a whirl-
wind rather than positioned to heal systemic rawness.
Under such conditions, it is easier to embrace a task,
rather than a process, focus. In the case at hand, attacks on
the school and the catch-22 requirement for progress
before resources kept the focus on rational planning.
Setting measurable goals is tangible—and psychologically
safe. Acknowledging the massive amount of individual and
systemic healing needed feels daunting; devising strategies
for handling the Pandora’s box of long-term, pent-up pain
and denial equally so. Kubler-Ross (1969, 1975) reminded
that loss involves grieving—and grief resembles clinical
descriptions of mental illness. It takes months, even years,
to work through deep cycles of grief. It takes longer to do
that and to simultaneously enhance programs and perfor-
mance. Avoidance begins to feel like a rational response
and, I attest, is easy to sustain.
Handling Organizational Toxins:
Leader as Toxin Magnet
Certain roles are strategically positioned to attract
workplace affect. Although anyone can assume a toxin
handler role, leaders are natural toxin magnets. As dean
in the scenario, you remain in the center of the fray.
There are better and worse ways of handling it; however,
escape is not an option. Organizational hierarchies make
6 Journal of Management Inquiry
leaders designated links to power structures that solve
(and cause) problems, organizational symbols of hope
and progress, and convenient outlets for frustration when
things go wrong. Maccoby (2004) and Kets de Vries
(2003) reminded us of the ease with which we all trans-
fer early-life disappointments, ambivalence about power,
and disdain for dependency onto leaders. Part of the
unstated contract between leader and followers is the
leader’s willingness to accept followers’ current and past
emotion-laden projections—and more.
Followers, after all, give leaders their authority in
exchange for the leader’s willingness to hold their fears
and needs. This is especially true under conditions of
confusion, complexity, or overload.
In times of distress, we turn to authority. To the
breaking point, we place our hopes and frustrations
upon those whose presumed knowledge, wisdom,
and skill show the promise of fulfillment. Authorities
serve as repositories for our worries and aspirations,
holding them, if they can, in exchange for the powers
we give them. (Heifetz, 1994, p. 69)
By creating holding environments, leaders serve as
buffers whose job includes institutional stress manage-
ment and deep understandings of followers’ needs.
In his study of caregivers, William Kahn (2005) iden-
tified a paradox in compassionate service that is relevant
to leading. Caring professionals who serve individuals in
need require simultaneous openness to and distance from
those they seek to aid. They need clear boundaries to sus-
tain objectivity, protect themselves from the stress of the
work, and nurture essential autonomy in others. At the
same time, good caregivers, like good leaders, need to
understand others at a deep level to respond in appropri-
ate ways to the unique realities of their situation over
time. This only happens when caregivers “take in” those
in need—fully grasp others’ fears, capabilities, limita-
tions, and needs. Learned skills in “detached concern”
enable clinicians to bound this process; however, they
still risk “the strain of absorption”—accumulated stress
from close relationships with those in need, recognition
of others’ pain, and the “constant waves of emotion” that
wash up against them in the course of their everyday
work (Kahn, 2005). Over time, “compassion fatigue”
(Figley, 1993, 1995, 2003) takes a toll. It is easy for care-
givers—and even easier for leaders—to ignore this.
Leaders face internal and external pressures to produce
and dynamics that keep them focused largely on follower
needs. Gardner (1990) acknowledged a universal
ambivalence toward leaders that predispose them to be
other-centered. People want leaders who are powerful
and capable of results. At the same time, they hate
dependence and giving power to others—or in others
taking it even at their insistence. The ambivalence can
push followers to blindly up-the-dependence-ante and
then punish leaders who don’t—or can’t—deliver
quickly enough. The stage is set for an escalating cycle
of pressure and angst for all.
Shared conceptions of heroic leadership—the solitary
superhero whose brilliance and strength save the day—
support a leader’s stoic acceptance of the added pres-
sures. So does the reality that all leaders serve at the will
of their followers. Rising expectations bring the potential
for rising disappointment. Leaders, after all, are only
human. And mounting frustrations can lead followers
across a dangerous line where disappointment morphs
into ill-will. Research (Dossey 1993, 1997) points to the
physiological consequences of negative wishes directed
toward another. Plain and simple, toxic projections can
harm, and leaders face a multiplier effect. People in emo-
tional pain can unconsciously blame toxin handlers for
their inability to provide them relief and, therefore, wish
them ill. They may also hold leaders responsible for their
suffering and project harm toward them. When toxin
handling and leading overlap, the leader becomes the
focus of collective double negativity.
Handling Toxins: Organizational
Supports and Strategies
Organizations benefit from those who willingly tend
to the needs of others at work. They help advance insti-
tutional goals, retain employees, solve problems, and
facilitate a better quality of work life. Over time, how-
ever, organizations risk losing these helpful employees:
They burn out. When toxin handling intersects with race
and gender, organizations also jeopardize their capacities
to attract and retain a diverse workforce. Lingering gen-
der stereotypes support tacit societal expectations for
women to assume emotional caretaker roles at work.
Many do, and research suggests they may face higher
levels of burnout from it (Taylor et al., 2000).
Organizations can alleviate wear and tear on toxin
handlers—and leaders—through policies, programs,
structures, and cultures that promote attention to the
human side of organizations. The payoff is twofold: (a)
support and compassion for loyal employees (Kanov et
al., 2004) and (b) increased productivity for all (Pfeffer
& Veiga, 1999). Organizational options might include
things like

• education to raise awareness of toxic emotions
and teach stress management;
• structural audits to identify jobs and positions
most susceptible to toxic overload;
• recovery breaks and “safe zones”—nonpressur-
ized work areas for short-term stints in a different
work environment (Frost, 2003);
• assessment and reward systems that promote a
balance between long- and short-term goals
(Pfeffer & Veiga, 1999), financial results and
organizational development, and personal and
professional needs;
• organizational rituals, ceremonies, and “arenas”
for legitimizing, addressing, and safely processing
workplace emotions (Bolman & Deal, 2003);
• “surrogate handlers”—short-term employees or
consultants during periods of institutional stress
or change (Frost & Robinson, 1999);
• on-site exercise facilities and midday programs
like yoga, tai chi, meditation, book groups, walk-
ing groups—or even a company choir—to break
the stress cycle and impart healthy work-life man-
agement skills.
These and other strategies imply the importance of a
people-centered organizational culture and of institu-
tional commitments to strong human resource manage-
ment practices and policies.
The Handler’s Journey: Five Steps
to Healthy Leadership
As a recovering toxin handler, I am struck by three
things: (a) the openness and willingness with which I took
on the work, (b) the power and durability of the experience,
and (c) the fact that leadership always looks simpler from
the outside. Knowledge of the dynamics explored in this
reflection while in the midst of them would have offered
welcomed alternatives for managing self and situation—
and a measure of sustenance during those dark nights of the
soul. May this article provide that for others. Individuals can
also take active steps to support themselves. Healthy toxin
handling—healthy leading—rests in attention to five key
areas: boundaries, biology, balance, beauty, and bounce.
Boundaries
It is awfully important to know what is and what is not
your business.
Gertrude Stein
Gallos / Toxic Trenches 7
An important set of skills and understandings relates to
boundary management. Paraphrasing Gertrude Stein,
toxin handlers need to distinguish between their own
business and the baggage and work of others. Managing
boundaries occurs at a number of critical interfaces:
self–other, professional–personal, self–work role,
leader–follower. Monitoring and managing each interface
is ongoing. Repeated exposure to workplace pain makes
that easy to forget. It also predisposes leaders to remain
blind to boundary breaches and their consequences.
Managing the self–other interface requires the capa-
city to feel and express empathy for others without
absorbing their pain. Although this may seem like an
academic distinction, it is not. Leaders burn out when
they act like “psychic sponges” (Borysenko, 1988),
soaking up the emotions of those around them whether
they realize it or not. Counselors and caregivers are
trained in clinical detachment to limit their psychic
engagement. They also learn to recognize when they
have crossed the line, accumulated too much exposure to
others’ pain, or inadequately cleared themselves of the
unavoidable “affective residue” of their work (Kahn,
2005). Toxin handlers may have empathy, organizational
commitment, and a willingness to charge into the affec-
tive fray. Few, however, have the clinical skills to protect
themselves from the toxicity or to recognize when it
begins to extract a personal toll.
Goleman (1995) asserted the human predisposition to
absorb the feelings of others. People vary in their empathic
capacity; however, cognitive scientists have identified
hard-wiring in the brain that makes certain people more
attuned to others (Restak, 2003). Caring leaders may fall
into this category. In her research, Maureen O’Sullivan
(“Researcher Finds Fibbers,” 2004) found individuals so
interpersonally sensitive they can detect liars by noting
extremely subtle emotions that flick across a fibber’s face.
These “wizards,” as O’Sullivan called them, pick up clues
that a majority of others never notice, employing a natural
rapid cognition (Gladwell, 2005) that surpasses the skills
of trained professionals. For such emotionally open and
sensitive individuals, toxin handling without clinical train-
ing can be dangerous.
Before his death, Peter Frost (2003) shared his story
to illustrate the creeping dangers of toxic contamination.
As associate dean, he repeatedly interacted with frus-
trated faculty, the majority of those who come to a dean’s
office. Frost remained unaware of the ways that those
repeated exposures weakened his defenses, despite his
experience as a private sector manager, professor, and
organizational scholar. Only in hindsight did he notice a
change in himself over his years in administration. Frost
8 Journal of Management Inquiry
reported taking in more, rather than less, of others’ pain
and feeling it more deeply and longer. He increasingly
replayed emotion-laden exchanges in his mind, search-
ing for better ways to handle and respond to problems
and complaints. The clinical psychology literature pro-
vides explanations for these phenomena. I can appreciate
the consequences: a growing sensitivity that increases
the probability of burnout. Frost described sleeplessness
yet feeling devoid of energy at the end of each day. Data
from more than 70 toxin handlers add depression, heart
palpitations, ulcers, pneumonia, and heart disease, as
well as other manifestations of stress and weakened
immune systems (Frost & Robinson, 1999). Toxin han-
dlers whom I have interviewed also spoke of anxiety, dis-
torted judgments, back and joint pain, loss of appetite,
increased alcohol use, a “hardened” feeling, and quick-
ness to anger. As one young handler explained, “I had a
complete meltdown and couldn’t understand what was
happening. I ended up taking some time off to put myself
back together. I guess I had what they used to call a ner-
vous breakdown.”
None of this is surprising to those who have labored in
the toxic trenches. Repeated exposure to strong emotions
and negativity results in brain changes that make individ-
uals more vulnerable to emotional pain—and their bod-
ies quicker to trigger unhealthy responses with each
repeated experience (Goleman, 1995; Restak, 2003). A
Journal of Advancement in Medicine study (Rein,
McCraty, & Atkinson, 1995) confirmed that simply
remembering strong emotions creates physiological
changes that can compromise the immune system.
Discussing stressful events with friends may even be
counterproductive (Rose, Carlson, & Waller, 2007).
An equally important boundary to manage is the distinc-
tion between self and work role. Toxin magnets need to
remember that others’ reactions to them are largely
responses to their role and actions taken in it, or they may be
transferences, projections, and other psychic assaults from
distressed others (Kahn, 2005). This does not exempt people
from reflecting on better and worse ways to lead. Rather, it
cautions them not to be blinded by what can seem like per-
sonal attacks, act precipitously because of them, internalize
system-level conflicts, or assume too much responsibility for
solving organizational concerns. Managing this boundary
goes to the heart of professional effectiveness.
Heifetz (1994) underscored the developmental nature
of leading: growing a group’s adaptive capacities for
tackling complex challenges. His advice is simple and
prudent: Create adequate supports and then give the
work back to the people. In the words of Gertrude Stein,
know what is and what is not your business! Without a
developmental focus, leaders are left holding the system’s
toxic baggage longer than necessary or healthy. By blur-
ring boundaries between leader–follower work, I under-
challenged others in solving their problems and too
willingly carried the weight of projects on my shoulders.
Too many toxin handlers do the same.
Biology
To keep the body in good health is a duty . . . otherwise we
shall not be able to keep our mind strong and clear.
Gautama Siddharta
Remaining vigilant to boundary management takes con-
centration and stamina. Both come from conscious atten-
tion to self-care and good health. Common-sense strategies
like exercising, eating sensibly, staying hydrated, limiting
caffeine and alcohol, maintaining blood sugar levels with
well-paced meals and healthy snacks, and developing reg-
ular sleep patterns are essential for managing the demands
of the work. So is early attention to mild, stress-related
symptoms, like sleeplessness or back pain—possible warn-
ing signs of a toxic boundary breach.
Borysenko (1988) reminded how easy it is under
stress—physical or emotional, personal or professional, real
or imagined—to settle into a primitive fight–flight stance
while ignoring healthier routes to strength and stamina.
Humans are well adapted for imminent danger through
complex physiological responses involving hormones and
inflammatory chemicals that ready the body for rapid
defense. They are largely unprepared for life’s steady
stream of low-level annoyances that evoke the same bio-
chemistry. Fight–flight reactions continuously bathe mind
and body in the stress hormone cortisol, raising blood sugar
to levels perfect for emergency action (Stoppler, 2007).
Adrenal glands ramp up heart, breathing, muscle tension,
and blood pressure—just right for limbs that need extra
energy for battle or escape. Chronic activation of this phys-
iological stress response, however, wears a body down. One
toxin handler joked to me that his gray hair—dark before he
assumed the role of shoring up an abusive boss and protect-
ing others from her quick temper—was a battle scar. He
may be right. Toxin handlers, take heed. Siegel (1993)
warned that wounded healers cannot minister well or
muster the care and empathy necessary to attend to others.
Balance
There is a proper dignity and proportion to be observed
in the performance of every act of life.
Marcus Aurelius

Strong boundaries and health require balance—retaining
one’s equilibrium and perspective in the face of chal-
lenge. Balance flows from grounded appreciation of
life’s richness and from willingness to attend to the
diverse needs of mind, body, and soul. Theorists have
differed on the path to balance. Freud (1975) proposed
love and work. Terr (1999) added play. Rohrlich (1980)
saw a resolution of tensions among work, family, and
leisure. Lawrence and Nohria (2002) identified four
innate human drives—to acquire, bond, learn, and
defend. The path, however defined, shapes choices and
becomes a measure of satisfaction, meaning, and joy.
Toxin handlers fair best when they identify their life
priorities and take specific steps to keep in balance. Two
options include deliberately counterbalance stress with
relaxation and neutralize toxic affect with positive emo-
tions. Toxin handlers may not be able to stop the waves,
but they can learn to surf (Kabat-Zinn, 1994). A relax-
ation response is a learned state of mental calm where
heart, breathing, muscle tension, and blood pressure
rates drop (Benson & Klipper, 1976). Soothing, simple
repetitive activities, like deep breathing, yoga, tai chi,
repeated prayers, practicing a musical instrument, or
even knitting can elicit the physiological benefits. The
Mayo Clinic (2007) offers opportunity for a midwork
respite with a short, online meditation video. New Age
guru Ram Dass suggests crocheting (Lipstein, 1992).
Leaders can also balance a toxic maelstrom with con-
scious focus on positive sentiments, such as appreciation,
love, care, forgiveness, and compassion. Researchers at
Institute of HeartMath (http://www.heartmath.org/) have
reported immediate benefits: positive changes in heart
rhythms as well as neural, hormonal, and biochemical
reactions that drop blood pressure, muscle tension, and
stress hormones. Scientists at UCLA found that opti-
mism and hope strengthened immune functioning
(Benson, Corliss, & Cowley, 2004). Forgiveness—let-
ting go of resentment for a perceived offense—decreases
blood pressure, cortisol, and other negative hormones
associated with heart disease, immunity disorders, and
more (A Campaign for Forgiveness Research, n.d.;
Lewis & Adler, 2004).
Beauty
Without art, the crudeness of reality would make the
world unbearable.
George Bernard Shaw
Maintaining balance also suggests identifying activities
and events that feed the soul and counter exposure to life’s
Gallos / Toxic Trenches 9
dark sides. The beauty and recuperative power of the arts
make them an obvious choice. Heifetz (1994) discussed the
importance of sanctuary—finding respite for perspective
and rejuvenation. Looking back at my dean days, the arts
were my sanctuary. I relished music and devoured fiction,
and I led a creative writing and theater-based project on
teen health and launched a Center for the Healing Arts at
our local public hospital during a postadministrative sab-
batical. The arts are powerful medicine.
There is considerable evidence, for example, that
music enhances health and healing. Music’s ability to
alter emotions has long been known experientially (e.g.,
Jourdain, 2002; Storr, 1992) and documented scientifi-
cally in the music therapy literature and elsewhere (e.g.,
McCraty, Barrios-Choplin, Atkinson, & Tomasino,
1998). Music’s benefits are diverse: masking unpleasant
sounds and feelings; slowing brain waves and equalizing
neuronal firing patterns; reducing heartbeat, pulse rates,
blood pressure, muscle tension, and stress-related hor-
mones; increasing endorphin levels; stimulating diges-
tion; boosting immunity; increasing memory and
sensitivity to imagery; and more (Campbell, 1999).
Oliver Sacks (1973) found that music eased symptoms in
Parkinson’s patients and playing the harp improved their
motor and functioning skills. Jourdain (2002) asserted
that music—any kind enjoyable to the listener—guides
the brain to higher than normal levels of integration of
the right and left lobes. Although music cannot repair the
brain—medical symptoms kept at bay by playing the
harp, for example, return after several days of not play-
ing—it reorganizes it in ways that other life experiences
do not. Music’s rhythms and patterns speak in primitive
ways to the brain. They sooth and “lift us from our
frozen mental habits” (Jourdain, 2002, p. 303). An ele-
vating sanctuary may be exactly what the stress doctor
orders.
The health sciences have historically used literature
and creative writing for caregiver renewal and education.
William Carlos Williams, Anton Checkhov, W. Somerset
Maugham, and John Keats were all trained as physicians.
Prestigious journals like the Journal of the American
Medical Association and the Annals of Internal Medicine
regularly publish literary works by physicians. Many
health care professionals are poets (e.g., Breedlove,
1998; Campo, 1994, 1996)—perhaps a suggested avoca-
tion for organizational leaders, too. Robert Coles (1989)
asserts that fiction and storytelling deepen inner life for
those who work on life’s emotional boundaries. They
nurture skills in observation, analysis, empathy, and self-
reflection—essential capacities for professional effec-
tiveness in any environment. And good fiction is healthy
escapism.
10 Journal of Management Inquiry
Bounce
Never take life seriously. Nobody gets out alive anyway.
Anonymous
Healthy toxin handling requires resilience: the abil-
ity to adapt and strengthen in the face of challenge,
trauma, or stress. Clinical detachment is never fool-
proof. And even the best trained clinicians suffer from
repeated exposure to those in pain (Kahn, 2005).
Managing boundaries, balance, biology, and beauty
helps, but eventually all feel the impact from affect-
laden work. Staying on top of the signs and taking
quick action sustain health and effectiveness—so do
strategies for building personal resilience. How can
leaders increase their odds of quickly bouncing back in
the face of toxic setbacks?
The American Psychological Association launched a
public education campaign on resilience building in the
wake of the September 11 attacks on the World Trade
Center (American Psychological Association, n.d.;
Murray, 2003). Resilience and bounce are learned skills:
recognizing choices for how to interpret and respond to
events, keeping things in perspective, trusting one’s
instincts, practicing new behaviors, and reflecting on the
consequences. Learning to reframe helps—organiza-
tional struggles, for example, are gifts to support learn-
ing. So are mistakes, plenty of laughter, and a good nap!
Mistakes provide needed interruptions for a pulse check
on choices—and on the status of one’s bounce. The ben-
efits of a good laugh speak for themselves, and a nap is
opportunity for physical rejuvenation and for psycholog-
ical diagnosis (Seigel, 1998). Rest restores physical
exhaustion from hard work but does nothing to relieve
burnout. Asked the secret of compassionate care, the
Dalai Lama answered, “Water. And sleep.” (Miller,
2005). Resilience, after all, comes from learning to
“wear life loosely” (Siegel, 1998).
In Over Our Heads: Toward Healthy
Leadership and Organizations
Developmental psychologist Robert Kegan (1994)
made a poignant observation. Over a growing portion of
our adult lives, there is a mismatch between the com-
plexity of modern culture’s “curriculum”—all that we
need to know and understand to function effectively and
productively—and the human capacity to grasp it. The
result is increased stress and need for more sophisticated
levels of human consciousness and learning to satisfy
contemporary expectations for love and work. Looking
through a developmental lens, modern life is just too
darn hard. Whether we realize it or not, Kegan con-
cluded, we are literally in over our heads. Toxic emotions
in the workplace is a case in point. How well have we
structured organizations and trained leaders and institu-
tional citizens to cope with this reality? Recovering toxin
handlers would conclude, not well at all. May this recog-
nition—and appreciation for the pain that accompanies
the work of too many individuals in modern organiza-
tions—give us pause.
Do we, in fact, foster approaches to leadership and orga-
nizations that promote toxic workplace dynamics and then
ignore those who rise to the challenge of managing them?
Place people recklessly in the line of affective fire without
know-how to protect themselves? The answer in many
cases is yes. What then are the implications for those who
prepare to assume higher levels of leadership and responsi-
bility? For those who teach current and future leaders and
citizens? For scholars who study the world of work and
propose the structures and models that shape modern orga-
nizations? For professional schools whose curricula
overemphasize theory, rationality, and quantitative skills
and fail in their mission to develop humane and ethical
leaders (Bennis & O’Toole, 2005)? This reflection sug-
gests the need for added attention in two key areas: (a) real-
istic understandings of the human side of enterprise and of
ways to manage it, and (b) the development of capacities
needed to handle the increasing complexities of modern,
global life. What does each imply for quality leadership
and management development?
Understanding Humanity: Rich,
Realistic, and Grounded Curricula
Preparing organizational citizens to productively deal
with the emotional realities of the workplace requires
learning on four levels: (a) conceptual understanding of
the phenomena, its origins, and functions; (b) diagnostic
capacities to recognize the situation-in-action; (c) strate-
gies for response; and (d) skills for effective implemen-
tation. All this implies a learning program—in the
classroom and in the boardroom—rich in attention to a
broad range of topics on human and interpersonal rela-
tions, grounded in cases and life scenarios, and balanced
between skill building and personal growth. It also points
to the need for diverse teaching tools—use of fieldwork,
conceptual material, discussions, personal cases, role
plays, experiential activities, journaling, and other forms
of active learning. On a macrolevel, it calls for a renewed
commitment to reflective practice and for a reframing of
effective leadership development.

Management education and leadership development
are at their best when they offer opportunities to reflect
on the “grand dilemmas of human existence” as they pre-
sent themselves in context (March & Weil, 2005). Major
issues in organizational life—understanding others and
the environment, motivating and influencing, acknowl-
edging enduring differences and managing the diversity,
handling conflict, dealing with scarce resources, generat-
ing productive alternatives to complex problems, and the
list goes on—are often indistinguishable from the larger
issues of life. We do leaders a disservice then when we
train them to rely too heavily—too exclusively—on
management formulas and models. Leaders have no
choice but to grapple with the paradoxes and complexi-
ties of human nature. They need opportunities to explore
and develop a range of options—and strengths—for
responding to them.
The topic of this reflection also suggests that learning
activities need to be realistic and balanced in addressing
humanity in its fullness. This includes attention to the
less attractive sides of human nature and organizational
life. Heifetz and Linsky (2002), for example, identified
the dangers in leading, from careers derailed by
unfounded public attacks to stress-related illness and
assassinations—character or otherwise. Allcorn (1994)
found increased workplace aggression, whereas
Susskind and Field (1996) provided strategies for deal-
ing with increasingly angry publics. Pritchard, Griffin,
and O’Leary-Kelly (2004) confirmed that violence, dis-
crimination, abuse, retaliation, and other forms of inci-
vility are powerful influences in modern organizations.
Ignoring or denying such dynamics will not make them
go away, but we know that positive dynamics like care,
compassion, and resilience—underlying themes in this
article—can counter the negativity and are powerful
processes for enhancing productivity and health (e.g.,
Baker & Dutton, 2007; Dutton, Lilius, & Kanov, 2007;
Dutton, Worline, Frost, & Lilius, 2006; Heaphy &
Dutton, 2008; Morgam Roberts, Dutton, Spreitzer,
Heaphy, & Quinn, 2005). (Additional research and rele-
vant working papers on the topic are also available
through the University of Michigan’s Center for the
Positive Organization Sciences at http://www.bus.umich
.edu/Positive/Center-for-POS/Publications-Working-
Papers.htm)
Finally, the messy sides of human and organizational
behavior raise the importance of multidisciplinary and
problem- or case-centered approaches to learning. Issues
like toxic emotions at work do not fit neatly into a disci-
plinary box—neither do solutions and strategies for han-
dling them. Complex human phenomena require insights
Gallos / Toxic Trenches 11
from across the pluralistic base of organizational
theory—understandings about structures and processes,
individuals and groups, power and politics, symbols and
meaning making (Bolman & Deal, 2003)—and beyond.
For centuries, great literature has been the primordial
source of learning about human nature and social change
(Gallos, in press). Management education is rich, realis-
tic, and grounded when it brings a full range of disci-
plines and understandings to bear.
An Essential Metacurriculum:
Encouraging Developmental Growth
The aforementioned recommendations point to an
essential metacurriculum for management education:
encouraging individuals to develop the increasingly
sophisticated cognitive and socioemotional capacities
needed to handle an increasingly complex world.
Individuals at different developmental stages perceive
and structure their world differently (Gallos, 1989,
1993). These different patterns of psychological organi-
zation lead to different capabilities for self-reflection, rel-
ative thinking, acceptance of personal causality, and
tolerance for ambiguity. A metacurriculum to encourage
developmental growth requires attention to individual dif-
ference across situations and roles. It offers leaders-in-
training repeated opportunities to experience and describe
the world as they see it and compare their worldview with
others; to increase capacities for understanding self,
other, and broad social issues; to take personal responsi-
bility for choices and actions; to formulate personal
frameworks for managing complexity; and to develop
“more complicated” intellectual and ethical reasoning
(Weick, 1979)—important capacities for navigating the
winding path to healthy leadership and organizations.
Project-based learning and structured reflections are
essential for this kind of learning.
In Closing: A Call for Action
This article points to the power and consequences of
high levels of workplace emotion and toward strategies
for healthier ways to cope. On another level, it is a call to
action. It encourages scholars, practitioners, educators,
and trainers to explore the complex path to healthy lead-
ership and organizations for themselves and for those who
can benefit from their discoveries. It advocates structures
and strategies that promote caring cultures, individual
resiliency, and extraordinary performance: models that
support high productivity and high attention to human
needs at work. It encourages humane organizational
12 Journal of Management Inquiry
design and development that result in compassionate
workplaces fostering excellence and caring. It suggests
better—and different—training for those who will
implement this vision. Attention to boundaries, balance,
biology, beauty, and bounce offer a starting point for cre-
ating positive and resilient workplaces. Let the experi-
ences of organizational toxin handlers stand as reminders
of the work still needed on the winding road to healthy
organizations–and to the health of our everyday leaders.
References
A Campaign for Forgiveness Research. (n.d.). Research projects.
Retrieved June 29, 2007, from A the Campaign for Forgiveness
Research Web site: http://www.forgiving.org/campaign/research.asp
Allcorn, S. (1994). Anger in the workplace: Understanding the
causes of aggression and violence. Westport, CT: Quorum Books.
American Psychological Association. (n.d.). The road to resilience.
Washington, DC. Author. Retrieved July 3, 2007, from http://www
.apahelpcenter.org/featuredtopics/feature.php?id=6
Argyris, C. (1985). Strategy, change, and defensive routines. Boston:
Pitman.
Ashford, S. J., Rothbard, N. P., Piderit, S. K., & Dutton, J. E. (1998).
Out on a limb: The role of context and impression management in
selling gender-equity issues. Administrative Science Quarterly,
43, 23-57.
Ashforth, B. E., & Humphrey, R. H. (1995). Emotion in the work-
place: A reappraisal. Human Relations, 48(2), 97-125.
Ashkanasy, N. M., Härtel, C., & Daus, C. S. (2002). Diversity and
emotion: The new frontiers in organizational behavior research.
Journal of Management, 28(3), 307-338.
Baker, W. E., & Dutton, J. E. (2007). Enabling positive social capital
in organizations. In J. E. Dutton & B. R. Ragins (Eds.), Exploring
positive relationships at work (pp. 325-346). Mahwah, NJ:
Erlbaum.
Bedeian, A. G. (2002). The dean’s disease: How the darker side of
power manifests itself in the office of dean. Academy of
Management Learning and Education, 1(2), 164-173.
Bennis, W. (2003). On becoming a leader. New York: Perseus.
Bennis, W. G., & O’Toole, J. (2005, May). How business schools lost
their way. Harvard Business Review (Reprint R0505F).
Benson, H., Corliss, J., & Cowley, G. (2004, September 27). Brain
check: Can we teach ourselves to be healthier? Newsweek, pp.
45-47.
Benson, H., & Klipper, M. Z. (1976). The relaxation response. New
York: HarperTorch.
Bolman, L., & Deal, T. E. (2003). Reframing organizations: Artistry,
leadership, and choice. San Francisco: Jossey-Bass.
Borysenko, J. (1988). Minding the body, mending the mind. New
York: Bantam.
Breedlove, C. (1998). Uncharted lines: Poems from the journal of the
American Medical Association. New York: Ten Speed Press.
Brockner, J., Grover, S., Reed, T. F., & Lee Dewitt, R. (1992).
Layoffs, job insecurity, and survivors’ work effort: Evidence of an
inverted U relationship. Academy of Management Journal, 35(2),
413-426.
Cameron, K. S., Dutton, J. E., & Quinn, R. E. (Eds.). (2003). Positive
organizational scholarship: Foundations of a new discipline.
Berkeley, CA: Berrett-Koehler.
Campbell, D. (1999). The Mozart effect: Tapping the power of music
to heal the body, strengthen the mind and unlock the creative
spirit. New York: Avon.
Campo, R. (1994). The other man was me: A voyage to the new
world. New York: Arte Publico.
Campo, R. (1996). What the body told. Durham, NC: Duke
University Press.
Center for Positive Organizational Scholarship. (n.d.). Ann Arbor,
MI: Author. http://www.bus.umich.edu/Positive/ Accessed July
10, 2007.
Cohen, M. D., & March, J. G. (1974). Leadership and ambiguity: The
American college president. New York: McGraw-Hill.
Coles, R. (1989). The call of stories: Teaching and the moral imagi-
nation. Boston: Houghton Mifflin.
Delbecq, A. L. (2008). Nourishing the soul of the leader: Inner
growth matters. In J. V. Gallos (Ed.), Business leadership (2nd ed.,
pp. 486-503). San Francisco: Jossey-Bass.
Dossey, L. (1993). Healing words: The power of prayer and the prac-
tice of medicine. San Francisco: HarperCollins.
Dossey, L. (1997). Be careful what you pray for . . . You just might get
it: What we can do about the unintentional effects of our thoughts,
prayers, and wishes. San Francisco: HarperCollins.
Dotlich, D. L., Noel, J. L., & Walker, N. (2008). Learning for leader-
ship: Failure as a second chance. In J. V. Gallos (Ed.), Business
leadership (2nd ed., pp. 478-485). San Francisco: Jossey-Bass.
Dutton, J., Lilius, J., & Kanov, J. (2007). The transformative potential
of compassion at work. In S. Piderit, R. Fry, & D. Cooperrider
(Eds.), New designs for transformative cooperation (pp. 107-126).
Palo Alto, CA: Stanford University Press.
Dutton, J. E., Worline, M., Frost, P., & Lilius, J. (2006). Explaining com-
passion organizing. Administrative Science Quarterly, 51(1), 59-96.
Figley, C. R. (1993, January). Compassion stress: Toward its mea-
surement and management. Family Therapy News.
Figley, C. R. (Ed.). (1995). Compassion fatigue: Coping with sec-
ondary traumatic stress disorder in those who treat the trauma-
tized. New York: Brunner/Mazel.
Figley, C. R. (2003). Compassion fatigue: An introduction. Retrieved
June 29, 2007, from http://www.giftfromwithin.org/html/cmp-
fatig.html
Fineman, S. (2000). Emotion arenas revisited. In S. Fineman (Ed.),
Emotion in organizations (pp. 1-24). Thousand Oaks, CA: Sage.
Frost, P. J. (2003). Toxic emotions at work. Boston: Harvard Business
School Press.
Frost, P. J. (2005). Toxin handlers and their stories. Web site accessed
July 10, 2007.
Frost, P. J., & Robinson, S. L. (1999, July/August). The toxic handler:
Organizational hero—and casualty. Harvard Business Review
(Reprint 8571).
Freud, S. (1975). Group psychology and the analysis of the ego. New
York: Norton.
Gallos, J. V. (1989). Developmental diversity and the OB classroom:
Implications for teaching and learning. Organizational Behavior
Teaching Review, 13(4), 33-47.
Gallos, J. V. (1993). Understanding the organizational behavior class-
room: An application of developmental theory. Journal of
Management Education, 17(4), 423-439.
Gallos, J. V. (2002). The dean’s squeeze: The myths and realities of
academic leadership in the middle. Academy of Management
Learning and Education, 1(2), 174-184.
Gallos, J. V. (2007). Loss and change: A Developmental opportunity
for teaching wisdom, compassion, and respect for the human

condition using Joan Didion’s The Year of Magical Thinking.
Academy of Management Learning and Education, 6, 2.
Gallos, J. V. (in press). Artful teaching: Using the visual, creative and
performing arts in contemporary management education. In S.
Armstrong & C. Fukami (Eds.), Handbook of management learn-
ing, education and development. Thousand Oaks, CA: Sage.
Gardner, J. (1990). On leadership. New York: Free Press.
Gladwell, M. (2005). Blink: The power of thinking without thinking.
New York: Little, Brown.
Goleman, D. (1995). Emotional intelligence. New York: Bantam.
Hatfield, E., Cacioppo, J., & Rapson, R.L. (1994). Emotional conta-
gion. New York: Cambridge University Press.
Heaphy, E., & Dutton, J. E. (2008). Positive social interactions and
the human body at work: Linking organizations and physiology.
Academy of Management Review, 33(1), 137-163.
Heifetz, R.A. (1994). Leadership without easy answers. Cambridge,
MA: Harvard University Press.
Heifetz, R. A., & Linsky, M. (2002). Leadership on the line: Staying
alive through the dangers of leading. Boston: Harvard Business
School Press.
Huy, Q. N. (2002). Emotional balancing of organizational continuity
and radical change: The contribution of middle managers.
Administrative Science Quarterly, 47, 31-69.
Job cuts hurt another way. (2004, October 19). Kansas City Star,
p. E-8.
Joiner, T. E. (1994). Contagious depression: Existence specificity to
depressed symptoms and the role of reassurance seeking. Journal
of Personality and Social Psychology, 67, 287-296.
Jourdain, R. (2002). Music, the brain and ecstasy: How music cap-
tures our imagination. New York: Quill/Harper Collins.
Kabat-Zinn, J. (1994). Wherever you go there you are. New York:
Hyperion.
Kahn, W. (2005). Holding fast: The struggle to create resilient care-
giving organizations. New York: Brunner-Routledge.
Kanov, J. M., Maitlis, S., Worline, M. C., Dutton, J. E., Frost, P. J., &
Lilius, J. M. (2004). Compassion in organizational life. American
Behavioral Scientist, 47(6), 808-827.
Kegan, R. (1994). In over our heads: The mental demands of modern
life. Cambridge, MA: Harvard University Press.
Kellerman, B. (2004). Bad leadership: What it is, how it happens,
why it matters. Boston: Harvard Business School Press.
Kets de Vries, M. F. R. (2003). Leaders, fools and imposters: Essays
on the psychology of leadership. New York: Universe.
Kotter, J. (2006). Leading change: Why transformation efforts fail. In
J. V. Gallos (Ed.), Organization development (pp. 239-251). San
Francisco: Jossey-Bass.
Kubler-Ross, E. (1969). On death and dying. New York: Macmillan.
Kubler-Ross, E. (1975). Death: The final stage of growth. Englewood
Cliff, NJ: Prentice-Hall.
Lawrence, P., & Nohria, N. (2002). Driven: How human nature
shapes our choices. San Francisco: Jossey-Bass.
Lewis, J., & Adler, J. (2004, September 27). Forgive and let live.
Newsweek, p. 52.
Lipman-Blumen, J. (2005). The allure of toxic leaders: Why we
follow destructive bosses and corrupt politicians—and how we
can survive. Oxford, UK: Oxford University Press.
Lipstein, O. (1992, March). Sex and crocheting in Burma: An inter-
view with Ram Dass. Psychology Today. Retrieved from
http://psychologytoday.com/articles/pto-19920301-000028.html
Maccoby, M. (2004, September). The power of transference. Harvard
Business Review (Reprint R0409E).
Gallos / Toxic Trenches 13
Maitlis, S., & Ozcelik (2004). Toxic decision processes: A study of
emotion and organizational decision making. Organization
Science, 15(4), 375-393.
March, J. G., & Weil, T. (2005). On leadership: A short course.
London: Blackwell.
Marris, P. (1986). Loss and change. London: Routledge & Kegan.
Maslow, A. (1968). Toward a psychology of being. Princeton, NJ: Van
Nostrand.
Mayo Clinic. (2007). Need to relax? Take a break to meditate.
Retrieved March 25, 2008, from http://www.mayoclinic.com/
health/meditation/MM00623
McCraty, R., Barrios-Choplin, B., Atkinson, M., & Tomasino, D.
(1998). The effects of different types of music on mood, tension,
and mental clarity. Alternative Therapies in Health and Medicine,
4(1), 75-84.
Miller, J. (2005, September 12). A call to turn bad to good: Dalai
Lama urges seeking the positive. Kansas City Star, p. A-13.
Morgan Roberts, L., Dutton, J. E., Spreitzer, G. M., Heaphy, E. D., &
Quinn, R. E. (2005). Composing the reflected best-self portrait:
Building pathways for becoming extraordinary in work organiza-
tions. Academy of Management Review, 30(4), 712-736.
Mumby, D. K., & Putnam, L. L. (1992). The politics of emotion: A
feminist reading of bounded rationality. Academy of Management
Review, 17(3), 465-486.
Murray, B. (2003). Rebounding from losses. APA Monitor, 34(9), 43.
Offermann, L. (2004, January). When followers become toxic.
Harvard Business Review (Reprint R0401E).
Oshry, B. (1995). Seeing systems: Unlocking the mysteries of organi-
zational life. San Francisco: Berrett-Koehler.
Pfeffer, J., & Veiga, J. F. (1999). Putting people first for organiza-
tional success. Academy of Management Executive, 13(2), 37-48.
Pritchard, R., Griffin, R., & O’Leary-Kelly, A. (2004). The dark side
of organizational behavior. San Francisco: Pfeiffer.
Rein, G., McCraty, R., Atkinson, M. (1995). The physiological and
psychological effects of compassion and anger. Journal of
Advancement of Medicine, 8(2), 87-105.
Researcher finds fibbers can be found out: A few “wizards” can detect
clues to lies, professor says at 2004 AMA’s Science Reporters
Conference. (2004, October 15). Kansas City Star, p. A-4.
Restak, R. (2003). The new brain: How the modern age is rewiring
your mind. New York: Rodale.
Rohrlich, R. (1980). Work and love: The crucial balance. New York:
Harmony.
Rose, A., Carlson, W., & Waller, E. M. (2007). Prospective associa-
tions of co-rumination with friendship and emotional adjustment:
Considering the socio-emotional trade-offs of co-rumination.”
Developmental Psychology, 43(4), 1019-1031.
Sacks, O. (1973). Awakening. New York: HarperCollins.
Sales, M. J. (2006). Understanding the power of position: A diagnos-
tic model. In J. V. Gallos (Ed.), Organization development (pp.
322-343). San Francisco: Jossey-Bass.
Siegel, B. (1993). How to live between office visits. New York:
HarperCollins.
Siegel, B. (1998). Prescriptions for living. New York: HarperCollins.
Stark, M. (2003, March 10). Time to treat toxic emotions at work.
Harvard Business School Working Knowledge. Retrieved July 5,
2007, from http://hbswk.hbs.edu/pubitem.jhtml?id=3373&t=orga-
nizations
Stoppler, M. (2007). Cortisol: The stress hormone. Retrieved July 10,
2007, from http://stress.about.com/cs/cortisol/a/aa012901.html
Storr, A. (1992). Music and the mind. New York: Ballantine.
14 Journal of Management Inquiry
Susskind, L., & Field, P. (1996). Dealing with an angry public: The
mutual gains approach to resolving disputes. New York: Free Press.
Taylor, S. E., Klein, L. C., Lewis, B. P., Gruenwald, T. L., Gurung. R.
A. R., & Updegraff, J. A. (2000). Bio-behavioral responses
to stress in females: Tend-and-befriend, not fight-or-flight.
Psychological Review, 107(3), 411-429.
Terr, L. (1999). Beyond love and work: Why adults need to play. New
York: Scribner.
Vaill, P. (1996). Learning as a way of being. San Francisco: Jossey-Bass.
Van Maanen, J., & Kunda, G. (1989). “Real feelings”: Emotional
expression and organizational culture. Research in Organizational
Behavior, 11, 43-103.
Watkins, M. (2008). The first ninety days of leadership. In J. V. Gallos
(Ed.), Business leadership (2nd ed., pp. 293-300). San Francisco:
Jossey-Bass.
Weick, K. (1979) The social psychology of organizing. Reading, MA:
Addison-Wesley.
Whicker, M. L. (1996). Toxic leaders: When organizations go bad.
Westport, CT: Quoram.
Joan V. Gallos is professor of leadership and director of the
Executive MBA program at the Henry W. Bloch School of Business
and Public Administration at the University of Missouri-Kansas City
where she has also served as professor and dean of education, coor-
dinator of University Accreditation, special assistant to the
Chancellor for Strategic Planning, and director of the higher educa-
tion graduate programs. Gallos is past-president of the Organizational
Behavior Teaching Society and recent author in and editor of the
well-reviewed Organization Development and Business Leadership
texts in the Jossey-Bass Readers series.
568352
research-article2015
PPSXXX10.1177/1745691614568352Holt-Lunstad et al.Loneliness and Isolation as Mortality Risk Factors

Perspectives on Psychological Science

Loneliness and Social Isolation as Risk 2015, Vol. 10(2) 227­–237

© The Author(s) 2015
Reprints and permissions:
Factors for Mortality: A Meta-Analytic sagepub.com/journalsPermissions.nav
DOI: 10.1177/1745691614568352
Review pps.sagepub.com

Julianne Holt-Lunstad1, Timothy B. Smith2, Mark Baker1,

Tyler Harris1, and David Stephenson1
1
Department of Psychology and 2Department of Counseling Psychology, Brigham Young University

Abstract
Actual and perceived social isolation are both associated with increased risk for early mortality. In this meta-analytic
review, our objective is to establish the overall and relative magnitude of social isolation and loneliness and to examine
possible moderators. We conducted a literature search of studies (January 1980 to February 2014) using MEDLINE,
CINAHL, PsycINFO, Social Work Abstracts, and Google Scholar. The included studies provided quantitative data on
mortality as affected by loneliness, social isolation, or living alone. Across studies in which several possible confounds
were statistically controlled for, the weighted average effect sizes were as follows: social isolation odds ratio (OR) =
1.29, loneliness OR = 1.26, and living alone OR = 1.32, corresponding to an average of 29%, 26%, and 32% increased
likelihood of mortality, respectively. We found no differences between measures of objective and subjective social
isolation. Results remain consistent across gender, length of follow-up, and world region, but initial health status has an
influence on the findings. Results also differ across participant age, with social deficits being more predictive of death
in samples with an average age younger than 65 years. Overall, the influence of both objective and subjective social
isolation on risk for mortality is comparable with well-established risk factors for mortality.

Keywords
social isolation, loneliness, mortality

Several lifestyle and environmental factors are risk factors
for early mortality, including smoking, sedentary lifestyle,
and air pollution. However, in the scientific literature,
much less attention has been given to social factors dem-
onstrated to have equivalent or greater influence on mor-
tality risk (Holt-Lunstad, Smith, & Layton, 2010). Being
socially connected is not only influential for psychologi-
cal and emotional well-being but it also has a significant
and positive influence on physical well-being (Uchino,
2006) and overall longevity (Holt-Lunstad et al., 2010;
House, Landis, & Umberson, 1988; Shor, Roelfs, & Yogev,
2013). A lack of social connections has also been linked
to detrimental health outcomes in previous research.
Although the broader protective effect of social relation-
ships is known, in this meta-analytic review, we aim to
narrow researchers’ understanding of the evidence in
support of increased risk associated with social deficits.
Specifically, researchers have assumed that the overall
effect of social connections reported previously inversely
equates with risk associated with social deficits, but it is
presently unclear whether the deleterious effects of social
deficits outweigh the salubrious effects of social connec-
tions. Currently, no meta-analyses focused on social iso-
lation and loneliness exist in which mortality is the
outcome. With efforts underway to identify groups at risk
and to intervene to reduce that risk, it is important to
understand the relative influence of social isolation and
loneliness.
Living alone, having few social network ties, and hav-
ing infrequent social contact are all markers of social iso-
lation. The common thread across these is an objective
quantitative approach to establish a dearth of social con-
tact and network size. Whereas social isolation can be an
Corresponding Author:
Julianne Holt-Lunstad, Department of Psychology, Brigham Young
University, 1024 Spencer W. Kimball Tower, Provo, UT 84602-5543
E-mail: julianne_holt-lunstad@byu.edu
228
objectively quantifiable variable, loneliness is a subjective
emotional state. Loneliness is the perception of social iso-
lation, or the subjective experience of being lonely, and
thus involves necessarily subjective measurement.
Loneliness has also been described as the dissatisfaction
with the discrepancy between desired and actual social
relationships (Peplau & Perlman, 1982).
Is there a need to distinguish between social isolation
and loneliness in assessing mortality risk? People lacking
human contact often feel lonely (Yildirim & Kocabiyik,
2010); however, social isolation and loneliness are often
not significantly correlated (Coyle & Dugan, 2012;
Perissinotto & Covinsky, 2014), suggesting that these may
be independent constructs and that one may occur with-
out the other. For instance, some may be socially isolated
but content with minimal social contact or actually prefer
to be alone; others may have frequent social contact but
still feel lonely. Because of the conceptual distinction
between social isolation and loneliness, understanding
their relative influence on mortality may provide insights
into possible independent pathways by which each influ-
ences risk and, in turn, guides intervention efforts.
There are several processes by which actual and per-
ceived social isolation may influence mortality risk (also
see other reviews in this special section). Social connec-
tions, or the lack thereof, can influence health and risk of
mortality via direct and indirect pathways (see Uchino,
2006). Both loneliness and social isolation are associated
with poorer health behaviors including smoking, physi-
cal inactivity, and poorer sleep (Cacioppo et al., 2002;
Hawkley, Thisted, & Cacioppo, 2009; Theeke, 2010).
Each is also associated with health-relevant biological
processes, including higher blood pressure, C-reactive
protein, lipid profiles, and poorer immune functioning
(Grant, Hamer, & Steptoe, 2009; Hawkley & Cacioppo,
2010; Pressman et al., 2005). Researchers that have
included both social isolation and loneliness have linked
these factors independently to poorer health behaviors
and biological risk factors (Pressman et al., 2005; Shankar,
McMunn, Banks, & Steptoe, 2011). However, few
researchers have examined these concurrently, and little
is known about their relative or synergistic influence.
In this meta-analytic review, our primary aim was to
focus on the relative effects of objective and subjective
social isolation on mortality (the likelihood of death over a
given time), to determine the magnitude and nature of the
association with risk of mortality, and to identify potential
moderating variables. We reviewed studies of mortality
that included measures of loneliness, social isolation, or
living alone. Because it is important to determine the effect
of social isolation and loneliness independent of corre-
lated lifestyle (e.g., smoking, physical activity) and psycho-
logical factors (e.g., depression, anxiety), we also examined
inclusion of covariates.
Holt-Lunstad et al.
Method
Identification of studies
We identified published and unpublished studies of the
association between social relationships and mortality
using two techniques. First, we searched for studies
appearing from January 1980 to February 2014 using sev-
eral electronic databases: MEDLINE, CINAHL, PsycINFO,
Social Work Abstracts, and Google Scholar. To capture
relevant articles, we used multiple search terms, includ-
ing mortality, death, decease(d), died, dead, and
remain(ed) alive, which were crossed with synonyms of
the terms social isolation, loneliness, and living alone. To
minimize inadvertent omissions, we searched each data-
base twice, with searches ending on February 24, 2014.
Second, we manually examined the reference sections of
past reviews and of studies meeting the inclusion criteria
to locate articles not identified in the database searches.
A team of research assistants who were trained and
supervised by the authors conducted the searches.
Inclusion criteria
We included in the meta-analysis studies written in English
that provided quantitative data regarding individuals’ mor-
tality as a function of objective and subjective social isola-
tion (operational definitions of social isolation, loneliness,
and living alone are provided in Table 1). All studies needed
to be prospective in design, meaning that the researchers
measured one’s objective or subjective social isolation at the
study initiation and then followed participants over time
(typically several years) to determine who remained alive
and who was dead at the follow-up. Thus, risk for mortality
is an estimate of the extent to which social isolation, living
alone, and loneliness significantly predict the likelihood of
being dead at follow-up.
We extracted data when authors used measures includ-
ing the terms found in Table 1. In some cases, authors oper-
ationalized social isolation by contrasting the participants
from the bottom quartile or quintile on a social network or
integration measure (e.g., Social Network Index; Cohen,
Doyle, Skoner, Rabin, & Gwaltney, 1997) but otherwise did
not code data from measures of social networks/integration.
Because we were interested in the impact of social deficits
on disease, we excluded studies in which mortality was a
result of suicide or accident. We also excluded studies in
which the outcome could not be isolated to mortality (e.g.,
combined outcomes of morbidity and mortality). Although
we excluded single-case designs and reports with exclu-
sively aggregated data (e.g., census-level statistics), we
included all other types of quantitative research designs that
yielded a statistical estimate of the association between
mortality and loneliness/isolation. Figure 1 shows the flow
Loneliness and Isolation as Mortality Risk Factors 229

Table 1. Descriptive Coding of the Measures Used to Assess Objective and Subjective Isolation

Type of measure Description Example of measure

Objective
Social isolation Pervasive lack of social contact or communication, Social Isolation Scale (Greenfield, Rehm, & Rogers,
participation in social activities, or having a 2002)
confidant Social Network Index (bottom quartile; Berkman &
Syme, 1979)
Living alone Living alone versus living with others Binary item: yes, no
Number of people in household
Subjective
Loneliness Feelings of isolation, disconnectedness, and not Loneliness Scale (De Jong-Gierveld & Kamphuis, 1985)
belonging UCLA Loneliness Scale (Russell, Peplau, & Cutrona,1980)

Note: UCLA = University of California, Los Angeles.

1,384 Potentially Relevant Reports Identified

1,230 Reports Excluded Based on Title/Abstract

339 No Mortality Indicator (Including Mixed Morbidity/Mortality)
296 No Mention of Social Isolation/Loneliness
217 No Quantitative Data (Editorial/Review/Commentary)
207 Irrelevant to Social Support/Mortality Association
81 Cause of Mortality was Suicide/Violence
42 Written in a Language other than English
31 Mortality Data was not Linked to Social Isolation/Loneliness
17 Non-human Subjects

154 Full-text Reports Retrieved for Detailed Evaluation

84 Reports Excluded Based on Detailed Review

29 Social Isolation/Loneliness was not an Independent Variable
17 Mortality was not the Outcome Variable
9 Duplicate Report of Data Contained in another Report
8 Insufficient Information to Extract an Effect Size
8 Mortality Data was not Linked to Social Isolation/Loneliness
7 Contained No Quantitative Data
4 Non-human Subjects
2 Cause of Mortality was Suicide/Violence

70 Reports Included in the Meta-Analysis

Fig. 1. Reports evaluated for inclusion in the meta-analysis.

diagram containing the details of study inclusion (included
in the Supplemental Material available online).
Data abstraction
A team of research assistants and the authors performed
the data searches and coding. To reduce the likelihood
of human error in coding, a team of two raters coded
each article twice. Two different raters performed the
second coding of each article. Thus, two distinct coding
teams (four raters) coded each article. Coders extracted
several objectively verifiable characteristics of the stud-
ies: (a) the number of participants and their composi-
tion by age, gender, health status, and preexisting health
230
conditions (if any), as well as the cause of mortality; (b)
length of follow-up; (c) research design; (d) type of
social isolation (actual/perceived) evaluated; (e) num-
ber and class of covariates included in the statistical
model; and (f) exclusion of participants who were
severely ill or who died shortly after study initiation.
The latter two variables helped to address possible con-
founds (e.g., depression, health status, physical mobil-
ity, age) and reverse causality, whereby individuals with
impaired health would be more likely to report increased
social isolation or loneliness because of an inability to
engage in social contact.
For each study, we extracted the reported effect size,
making sure that odds ratio (OR) values greater than one
represented an increase in mortality as a function of
social isolation, loneliness, or living alone—and a
decrease in mortality when individuals were not isolated,
lonely, or living alone. Effect sizes less than one indicated
the opposite. To analyze the data, we temporarily trans-
formed the reported effect sizes to the natural log of the
OR and subsequently transformed them back to ORs for
purposes of interpretation.
When researchers reported multiple effect sizes within
a study at the same point in time, we averaged the sev-
eral values (weighted by standard error) to avoid violat-
ing the assumption of independent samples. We therefore
used the shifting units of analysis approach (Cooper,
1998), which minimizes the threat of nonindependence
in the data while allowing for more detailed follow-up
analyses. In a few cases in which researchers reported
multiple effect sizes across different levels of social isola-
tion (high vs. medium, medium vs. low), we extracted
only the value with the greatest contrast (high vs. low).
When a study contained multiple effect sizes across time,
we extracted the data from the longest follow-up period.
We extracted both unadjusted data and the data from the
model involving the greatest number of statistical con-
trols (although we also extracted the data from the model
utilizing the fewest number of statistical controls for a
subsequent comparison after recording the type and
number of statistical controls used within both models).
Overall, the interrater agreement for data abstraction
was adequately high for categorical variables (with
Cohen’s kappa averaging .73) and for continuous vari-
ables (with intraclass correlations for single measures
averaging .95). We resolved discrepancies across coding
teams through further scrutiny of the article until we
obtained consensus.
Results
Description of the retrieved literature
We located 79 articles reporting pertinent data, 9 of which
were excluded because they contained the same data as
Holt-Lunstad et al.
another article, resulting in 70 independent studies that
met the full inclusion criteria. The complete list of refer-
ences and a table summarizing the characteristics of
those studies (Table S1) are found in the Supplemental
Material available online. Studies typically involved older
adults, with a mean age of 66.0 years at initial data col-
lection and with a mean length of follow-up being 7.1
years. Most studies (63%) involved normal community
samples, but 37% of studies involved patients with a
medical condition, such as heart disease. See Table 2 for
further descriptive data.
Three studies included data on both loneliness and
one of the objective independent variables: two for lone-
liness and social isolation, and one for loneliness and
living alone. Using a shifting units of analysis approach
(Cooper, 1998), we included data from those distinct
measures in the analyses specific to the type of measure-
ment, but all other studies contributed a single data point
to the analyses.
Effect sizes in the 70 studies had been calculated by
researchers using a variety of methods, with some
researchers reporting unadjusted values and with other
researchers using a variety of covariates. ORs ranged
from 0.64 to 3.85, with exceptionally high heterogeneity
across studies (I2 = 97.8%, 95% CI [97.6%, 98.1%]; Q =
3,328.9, p < .0001), suggesting excessive variability in
findings across all types of data. We therefore divided the
analyses according to the number of covariates used. In
the unadjusted data group, the researchers controlled for
no other variables in the analyses. In the partially
adjusted data group, the researchers typically controlled
for one or two variables, usually age and gender. The
fully adjusted data are the model within studies with the
largest number of covariates. Effect sizes from each cat-
egory were evaluated separately, such that a single study
could contribute effect sizes to more than one category
(see Table 3).
Overall, each of the measures (social isolation, loneli-
ness, and living alone) for each type of data (unadjusted,
partially adjusted, or fully adjusted) had an OR between
1.26 and 1.83. The three measures did not differ in their
ORs for any of the three types of data, meaning that there
was no overall difference among the two objective and
one subjective factors. (Random-effects weighted analy-
ses of variance across the measures yielded all ps > .20.)
However, the type of data did matter in the analysis.
Unadjusted data yielded effect sizes of greater magnitude
than fully adjusted data (see Table 3). The differences
between unadjusted and fully adjusted data also reached
statistical significance (p < .001) when comparing data
within 27 studies in which researchers reported more
than one statistical model (e.g., unadjusted compared
with fully adjusted values) using multivariate meta-ana-
lytic methods after accounting for the .74 correlation of
Loneliness and Isolation as Mortality Risk Factors 231

Table 2. Characteristics of 70 Studies of the Association of Mortality With Subjective and Objective Measures of Social Isolation

Characteristic M Number of studies (k) %

Year of initial data collection 1,993 46
Years of participant follow-up 7 70
% deceased by the end of data collection 24.7 66
% female 52.6 67
% smokers 31.2 28
Sample size 48,673
<200 6 9
200–499 7 10
500–999 10 14
1,000–2,999 20 29
3,000–9,999 16 23
>10,000 11 16
Age of participantsa 66.0
<50 years 8 11
50–59 years 12 17
60–69 years 11 16
70–79 years 21 30
>80 years 10 14
Location of data collection
Inpatient medical treatment setting 15 21
Outpatient medical treatment setting 11 16
Community setting (normal populations) 44 63
World region of data collection
Europe 38 54
North America 19 27
Asia 7 10
Australia 3 4
Multiple regions 3 4

Note: Not all variables sum to the total number of studies because of missing data.
a
Average age category of participants at study initiation, although not all participants within the study would necessarily be in the category listed.

Table 3. Weighted Mean Effect Sizes (Odds Ratios) by Type of Measurement

Measure k OR+ SE 95% CI

Unadjusted data
Social isolation 3 1.83 0.185 [1.27, 2.63]
Living alone 20 1.51 0.072 [1.32, 1.74]
Loneliness 8 1.49 0.105 [1.22, 1.84]
Overall 31 1.53 0.035 [1.38, 1.70]
Partially adjusted dataa
Social isolation 6 1.46 0.162 [1.06, 2.00]
Living alone 8 1.55 0.132 [1.20, 2.00]
Loneliness 7 1.52 0.213 [0.99, 2.30]
Overall 21 1.51 0.117 [1.27, 1.79]
Fully adjusted datab
Social isolation 14 1.29 0.100 [1.06, 1.56]
Living alone 25 1.32 0.075 [1.14, 1.53]
Loneliness 13 1.26 0.099 [1.04, 1.53]
Overall 52 1.30 0.116 [1.16, 1.46]

Note: k = number of studies; OR+ = random-effects weighted odds ratio; CI = confidence interval.
a
Typically one or two covariates, most often age and gender. bData from the statistical model in studies that contained the most covariates; these
adjusted data yielded effect sizes that were statistically significantly (p < .05) smaller than unadjusted data.
232
effect sizes within studies. Thus, unadjusted and fully
adjusted data not only represented conceptually distinct
classes of data but also yielded findings of different
magnitude.
Moderator analyses
Given the substantial heterogeneity of the overall results
(I 2 > 80%), we analyzed the extent to which the variabil-
ity in effect sizes could be attributable to study or partici-
pant characteristics. These analyses involved only the
fully adjusted data because multiple factors predictive of
mortality had been controlled (thus minimizing possible
confounding explanations). Study and participant charac-
teristics included both categorical and continuous data,
so we report those analyses separately.
Categorical variables. We examined categorical vari-
ables using random-effects weighted analyses of vari-
ance, beginning with the type of covariates used in the
fully adjusted models. Eight studies included multiple
covariates that were directly relevant to social support,
such as marital status, social networks, and loneliness.
These eight studies had lower averaged effect sizes (OR
= 1.17) than those of 33 studies in which no covariates
directly relevant to social support were included in the
statistical model (OR = 1.27). Otherwise, the averaged
effect sizes remained of similar magnitude irrespective of
the particular covariates that were or were not included
in the models (p > .20), including covariates relevant to
depression, socioeconomic status, health status, physical
activity, smoking, gender, and age. Different combina-
tions of covariates across studies yielded similar results.
We found no substantive differences in effect sizes (p
> .15) across the other categorical variables evaluated:
world region, data collection setting, cause of mortality,
research design, health status, and medical condition at
intake. Finding no significant differences across partici-
pant health status when using the fully adjusted data was
particularly notable because of a difference that we
observed with the unadjusted data: Studies in which par-
ticipants had a medical condition and were recruited
from a medical setting had larger unadjusted average
effect sizes (OR = 1.82) than studies with ostensibly
healthy participants recruited from the general commu-
nity (OR = 1.34, p = .003). Furthermore, with the unad-
justed data, studies in which the researchers excluded
participants with terminal conditions or participants who
died shortly after baseline data collection (whose social
isolation or social support may have been affected by
their medical condition) had higher averaged effect sizes
(OR = 1.95) than the studies in which the researchers did
not report such exclusions (OR = 1.38, p < .05). Thus,
Holt-Lunstad et al.
accounting for participants’ initial health condition in the
research design resulted in systematically different find-
ings across studies. In most (81%) of the multivariate sta-
tistical models, researchers had controlled for participant
health status variables, such that we found no differences
across those conditions in the fully adjusted data. Studies
in which the researchers controlled for health status vari-
ables yielded substantially different findings from those
studies in which this was not done.
Continuous variables. We examined study and par-
ticipant characteristics involving continuous data in rela-
tion to the observed effect sizes using random-effects
weighted regression coefficients (meta-regression). We
observed no coefficients greater than the absolute value
of .20 between effect sizes and the year of initial data col-
lection, the length of follow-up, or the percentage of
female participants in each study. However, the number
of covariates included in multivariate models was moder-
ately associated with effect size (r = −.27). Visual inspec-
tion of the corresponding scatter plot indicated that when
studies included seven or more covariates, effect sizes
tended to be more homogeneous, without extremely
high values. To clarify, the inclusion of many covariates
did not substantively reduce the magnitude of the gen-
eral findings, which tended to remain in the range of
OR = 1.20–1.40, but it did eliminate all OR values greater
than 1.66.
Analyses also indicated that the association between
the effect size and the average age of participants at
intake was of a moderately strong magnitude (r = −.34
for adjusted data, and r = −.46 for unadjusted data). This
association with participant age remained of the same
magnitude when accounting for length of study follow-
up (and participants’ age at the end of the study) and
when age was or was not used as a statistical covariate.
Examination of the scatter plot and breaking down the
data into three approximately equal categories of initial
participant age helped to interpret the correlation: Studies
involving participants of an average age less than 65
years had an average effect size of OR = 1.57 for adjusted
data, and OR = 1.92 for unadjusted data; studies involv-
ing participants of an average age between 65 and 75
years had an average effect size of OR = 1.25 for adjusted
data, and OR = 1.32 for unadjusted data; and studies
involving participants of an average age greater than 75
years had an average effect size of OR = 1.14 for adjusted
data, and OR = 1.28 for unadjusted data. Adults less than
65 years of age appeared to be at greater risk of mortality
when they lived alone or were lonely compared with
older individuals in those same conditions, even after
controlling for the effect of age and other covariates on
mortality.
Loneliness and Isolation as Mortality Risk Factors 233

Studies Discussion
p < 1%
1% < p < 5% Social isolation results in higher likelihood of mortality,
5% < p < 10% whether measured objectively or subjectively. Cumulative
p > 10% data from 70 independent prospective studies, with
3,407,134 participants followed for an average of 7 years,
0
revealed a significant effect of social isolation, loneliness,
and living alone on odds of mortality. After accounting
.1 for multiple covariates, the increased likelihood of death
was 26% for reported loneliness, 29% for social isolation,
Standard error

.2 and 32% for living alone. These data indicated essentially

no difference between objective and subjective measures
of social isolation when predicting mortality.
.3 The prospective designs of these studies and the sta-
tistical models that controlled for initial health status (and
.4 several other potential confounds) provide evidence for
the directionality of the effect. Although we cannot con-
firm causality, the data show that individuals who were
.5
–1 0 1 2 socially isolated, lonely, or living alone at study initiation
Effect estimate (lnOR) were more likely to be deceased at the follow-up, regard-
less of participants’ age or socioeconomic status, length
Fig. 2. Contour-enhanced funnel plot of effect sizes (natural log of of the follow-up, and type of covariates accounted for in
the odd ratio [lnOR]) by standard error for 70 studies with measures
of the association between mortality and social isolation, loneliness, or
the adjusted models.
living alone. We caution scholars perusing the expanding research
literature on the association of social isolation and loneli-
ness with physical health against reliance on unadjusted
Likelihood of publication bias data because those data fail to account for participant
adversely influencing the results health status, a factor contributing to reverse causality
(when individuals with impaired health report increased
Publication bias occurs when the data obtained in a
loneliness or social isolation because their health condi-
meta-analysis fail to represent the entire population of
tion limits their social contacts). Averaged results with
studies because of the increased probability of nonsig-
unadjusted data were of greater magnitude than the
nificant results remaining unpublished (and therefore
results from fully adjusted models (see Table 3), particu-
less accessible for meta-analytic reviews). As can be seen
larly when participants had a preexisting health condition
in Figure 2, the data in this meta-analysis were highly
and when physically ill participants were not excluded
variable, and the distribution of effect sizes appeared
from the unadjusted analyses. In fully adjusted models
somewhat imbalanced toward the right side of the graph.
accounting for health status and in studies with physically
The distribution of the data was relatively sparse toward
ill individuals removed from analyses (thus accounting for
the bottom of the white-shaded center of the graph, the
reverse causality), social isolation and loneliness remained
area of nonsignificance. This kind of distribution can sug-
predictive of mortality. Future researchers will need to
gest that some nonsignificant studies were missing from
confirm the hypothesis that when individuals are ill (and
the meta-analysis. However, neither Egger’s regression
ostensibly needing support) their risk for mortality
test nor an alternative to that test recommended for OR
increases substantially when lacking social support.
data (Peters, Sutton, Jones, Abrams, & Rushton, 2006)
Overall, the findings from this meta-analysis are con-
reached statistical significance (p > .05), which dimin-
sistent with prior evidence that has demonstrated higher
ished the likelihood of possible publication bias. We
survival rates for those who are more socially connected
found the fail-safe N—the number of hypothetically miss-
(Holt-Lunstad et al., 2010) and extend those findings by
ing studies needed to reduce the present results to zero—
focusing specifically on measurement approaches that
to be 1,268, a number higher than the plausible number
assess the relative absence of social connections. Notably,
of studies conducted. Furthermore, using the trim and fill
the present meta-analysis included more than double the
method (Duval & Tweedie, 2000), we did not estimate
number of studies and 10 times the number of partici-
any “missing” studies; the distribution was overall fairly
pants compared with the previous meta-analysis. Thus,
symmetric relative to the average effect size. It thus
the field now has much stronger evidence that lacking
seemed unlikely that publication bias substantively
social connections is detrimental to physical health.
affected the results of this meta-analysis.
234
The average effect sizes identified in this meta-analysis
were lower than those reported previously for measures
of social networks (OR = 1.45, 95% CI [1.32, 1.59]) and
social integration (OR = 1.52, 95% CI [1.36, 1.69]) and
were much lower than complex measures of social inte-
gration (OR = 1.91, 95% CI [1.63, 2.23]; see Table 4 of
Holt-Lunstad et al., 2010). This difference may suggest
that the salubrious effects of being socially connected
may be stronger than the adverse effects of lacking con-
nections. However, it is also likely that research methods
that account for the multidimensionality of social rela-
tionships better predict mortality than measurement
focused on any single aspect of sociality, such as social
isolation. Nonetheless, identification of the relative effects
of each component may be useful in targeting those that
may be modifiable.
There is also presently no research evidence to sug-
gest a threshold effect. The aggregate results suggest
more of a continuum than a threshold at which risk
becomes pronounced. Although it is possible that indi-
viduals who are extremely lonely or socially isolated may
account for much of the elevated risk, presently too few
researchers target extremely isolated individuals in stud-
ies. Given the complexity (including objective and sub-
jective aspects) of social relationships, identifying such a
threshold seems unlikely.
Objective versus subjective isolation
Using the meta-analytic data, had we found that either
social isolation or loneliness was more predictive of mor-
tality, interventions to reduce risk could have become
more targeted. However, we presently have no evidence
to suggest that one involves more risk than the other for
mortality. Unfortunately, in the vast majority of studies,
researchers examined only one measurement approach
(social isolation, loneliness, or living alone), precluding
direct comparisons. Among the few studies in which
researchers contrasted social isolation and loneliness, the
evidence was mixed, with researchers finding that loneli-
ness was more influential in one study (Holwerda et al.,
2012), and with other researchers finding that social isola-
tion had stronger effects than loneliness in a later study
(Steptoe, Shankar, Demakakos, & Wardle, 2013). This
inconsistency may be due to differences in methodologi-
cal approaches to handling correlated psychological
states, such as depression (Booth, 2000). Our analyses
indicated that the elevated risk of mortality persisted even
when controlling for correlated components of social net-
works and multiple other factors, including depression,
with the use of covariates negating large effect sizes. In
any case, the multiple, overlapping components of social-
ity make reliance on statistical adjustment less desirable
Holt-Lunstad et al.
than direct comparisons between components, such as
loneliness and social isolation.
The equivalent effects of social isolation and loneli-
ness reported here do not indicate interchangeability of
these risk assessments. Rather, the available data suggest
that efforts to mitigate risk should consider both social
isolation and loneliness without the exclusion of the
other. Because social isolation and loneliness are often
weakly correlated (Coyle & Dugan, 2012), simply increas-
ing social contact may not mitigate loneliness. Likewise,
exclusively altering one’s subjective perceptions among
those who remain objectively socially isolated may not
mitigate risk. The evolutionary perspective of loneliness
proposed by Cacioppo and colleagues (Cacioppo et al.,
2006; Cacioppo, Cacioppo, & Boomsma, 2014) presents
loneliness as an adaptive signal, similar to hunger and
thirst, that motivates one to alter behavior in a way that
will increase survival. Accordingly, loneliness is a power-
ful motivator to reconnect socially, which, in turn,
increases survival and opportunity to pass on genes.
Consistent with this perspective, intervention attempts to
alter the signal (e.g., hunger, loneliness) without regard
to the actual behavior (e.g., eating, social connection)
and vice versa would likely be ineffective. Extending this
possibility, some data have shown that those who are
both high in loneliness and social isolation had the poor-
est immune response (Pressman et al., 2005). Therefore,
both objective and subjective measures of social isolation
should be considered in risk assessment.
It is only through direct comparisons of social isola-
tion and loneliness in the same sample that researchers
can establish independent, relative, and synergistic
effects. Consequently, it is possible that different combi-
nations of social isolation and loneliness may represent
different levels of risk. For instance, those low in both
isolation and loneliness would presumably be at lowest
risk, those high in both at highest risk, and those who are
isolated but not lonely or lonely but not isolated to be at
intermediate risk. Nonetheless, there is currently insuffi-
cient empirical evidence to test this hypothesis, highlight-
ing an important weakness of the current literature that
needs to be addressed in future research.
Isolation and aging
The data in this meta-analysis should make researchers
call into question the assumption that social isolation
among older adults places them at greater risk compared
with social isolation among younger adults. Using the
aggregate data, we found the opposite to be the case.
Middle-age adults were at greater risk of mortality when
lonely or living alone than when older adults experi-
enced those same circumstances.
Loneliness and Isolation as Mortality Risk Factors
The moderating effect of age may seem counterintui-
tive in light of data indicating that individuals more than
65 years of age are more likely to report loneliness
(Dykstra, van Tilburg, & de Jong Gierveld, 2005), but
there are at least four plausible explanations for why
middle-age adults may differ from older adults in terms
of the relevance of social networks to physical health.
First, it is possible that individuals who do not die early
may be a particularly resilient group, with different social
or health characteristics than those who die at earlier
ages. Thus, the observed difference across age could be
confounded with preexisting health status, although this
interpretation is qualified by the fact that the researchers
using multivariate statistical models accounted for partici-
pant age and health status. A second explanation involves
changes in social networks as individuals transition from
full-time employment to retirement, with decreases in
socialization in occupational and public forums that are
seen as culturally normative. This possible explanation is
supported by one study in which researchers examined
loneliness after retirement and found an effect for mental
health (anxiety and depression) but not for physical
health (functional status and number of chronic condi-
tions; Bekhet & Zauszniewski, 2012). Third, it is plausible
that individuals who are alone or lonely before retire-
ment age may be more likely to engage in risky health
behaviors or less likely to seek medical treatment early,
whereas after retirement, people may attend more assid-
uously to their physical health. Finally, it is possible that
the different results across participant age are confounded
with marital status: Older adults are much more likely to
be widows/widowers than middle-age adults. Our meta-
analysis cannot shed light on these four possible expla-
nations because the first three explanations involve
variables inadequately evaluated in the present research
literature, and the variable associated with the fourth
explanation, marital status, was not coded in our analy-
ses. Although many studies indicate that loneliness dif-
fers across marital status (Cacioppo & Patrick, 2008;
Hughes et al., 2004; Victor & Bowling, 2012) and that
marital status is significantly associated with mortality
(Roelfs, Shor, Kalish, & Yogev, 2011), we did not include
marital status as an indicator of social isolation because
being unmarried does not necessarily mean that one is
socially isolated, living alone, or lonely. Moreover, there
would be multiple qualitative differences in the social
networks of an older individual who had never been
married compared with one who had been married and
raised children but whose spouse had recently died, even
though both are living alone. Rather than include all pos-
sibly correlated variables (e.g., marital status, depression,
substance abuse), we evaluated only direct measures of
social isolation, living alone, or loneliness. Given the lim-
itations of the present meta-analysis, future researchers
235
should confirm the apparent differences across partici-
pant age and should evaluate the relative merits of the
several plausible explanations for that finding.
To better evaluate differences across age, future
researchers should involve participants from a broad
range of age groups. Most of the data in this meta-analy-
sis came from older adults. Only 24% of studies involved
people with an average age of 59 years or younger, and
only 9% of studies involved people younger than 50
years of age at intake. If future data collection with
younger adult samples confirms the age differences we
observed in this meta-analysis, then widespread beliefs
about the health risks of social isolation being greatest
among older adults are inaccurate. In any case, the meta-
analytic data, taken together with evidence for detrimen-
tal influences across the life span (Qualter et al., 2015,
this issue), suggest that future research (and possibly
interventions) should expand beyond older adults.
Conclusion
Substantial evidence now indicates that individuals lacking
social connections (both objective and subjective social
isolation) are at risk for premature mortality. The risk asso-
ciated with social isolation and loneliness is comparable
with well-established risk factors for mortality, including
those identified by the U.S. Department of Health and
Human Services (physical activity, obesity, substance
abuse, responsible sexual behavior, mental health, injury
and violence, environmental quality, immunization, and
access to health care; see www.hhs.gov/safety/index). A
substantial body of research has also elucidated the psy-
chological, behavioral, and biological pathways by which
social isolation and loneliness lead to poorer health and
decreased longevity (for reviews, see Cacioppo, Cacioppo,
Capitanio, & Cole, 2015, this issue; Shankar et al., 2011;
Thoits, 2011; see also Cacioppo et al., 2015; Hawkley &
Cacioppo, 2003, 2010). In light of mounting evidence that
social isolation and loneliness are increasing in society
(McPherson & Smith-Lovin, 2006; Perissinotto, Stijacic
Cenzer, & Covinsky, 2012; Victor & Yang, 2012; Wilson &
Moulton, 2010), it seems prudent to add social isolation
and loneliness to lists of public health concerns. The pro-
fessional literature and public health initiatives can accord
social isolation and loneliness greater recognition.
To draw a parallel, several decades ago scientists who
observed widespread dietary and behavior changes
(increasing consumption of processed and calorie-rich
foods and increasingly sedentary lifestyles) raised warn-
ings about obesity and related health problems (e.g.,
Brewster & Jacobson, 1978; Dietz & Gortmaker, 1985).
The present obesity epidemic (Wang & Beydoun, 2007)
had been predicted. Obesity now receives constant cov-
erage in the media and in public health policy and
236
initiatives. The current status of research on the risks of
loneliness and social isolation is similar to that of research
on obesity 3 decades ago—although further research on
causal pathways is needed, researchers now know both
the level of risk and the social trends suggestive of even
greater risk in the future. Current evidence indicates that
heightened risk for mortality from a lack of social rela-
tionships is greater than that from obesity (Flegal, Kit,
Orpana, & Graubard, 2013; Holt-Lunstad et al., 2010),
with the risk from social isolation and loneliness (control-
ling for multiple other factors) being equivalent to the
risk associated with Grades 2 and 3 obesity. Affluent
nations have the highest rates of individuals living alone
since census data collection began and also likely have
the highest rates in human history, with those rates pro-
jected to increase (e.g., Euromonitor International, 2014).
In a recent report, researchers have predicted that loneli-
ness will reach epidemic proportions by 2030 unless
action is taken (Linehan et al., 2014). Although living
alone can offer conveniences and advantages for an indi-
vidual (Klinenberg, 2012), this meta-analysis indicates
that physical health is not among them, particularly for
adults younger than 65 years of age. Further research is
needed to address the complexities of social interactions,
interdependence, and isolation (Parigi & Henson, 2014;
Perissinotto & Covinsky, 2014), but current evidence cer-
tainly justifies raising a warning.
Acknowledgments
We express deep appreciation to Spencer Ford, Devin Peterson,
Lauren Brown, Naomi Worden, Breydan Wright, Eric R. Smith,
Chris Badger, and Jae Cho for their work with database searches
and coding.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to the authorship or the publication of this article.
Funding
This research was generously supported by grants from Brigham
Young University awarded to Timothy B. Smith and Julianne
Holt-Lunstad.
Supplemental Material
Additional supporting information may be found at
http://pps.sagepub.com/content/by/supplemental-data
References
Bekhet, A. K., & Zauszniewski, J. A. (2012). Mental health of
elders in retirement communities: Is loneliness a key factor?
Archives of Psychiatric Nursing, 26, 214–224. doi:10.1016/j.
apnu.2011.09.007
Berkman, L. F., & Syme, S. L. (1979). Social networks, host
resistance, and mortality: A nine-year follow-up study
Holt-Lunstad et al.
of Alameda County residents. American Journal of
Epidemiology, 109, 186–204.
Booth, R. (2000). Loneliness as a component of psychiatric dis-
orders. Medscape General Medicine, 2, 1–7.
Brewster, L., & Jacobson, M. F. (1978). The changing American
diet. Washington, DC: Center for Science in the Public
Interest.
Cacioppo, J. T., Cacioppo, S., & Boomsma, D. I. (2014).
Evolutionary mechanisms for loneliness. Cognition &
Emotion, 28, 3–21. doi:10.1080/02699931.2013.837379
Cacioppo, J. T., Cacioppo, S., Capitanio, J. P., & Cole, S. W.
(2015). The neuroendocrinology of social isolation. Annual
Review of Psychology, 66, 9.1–9.35. doi:10.1146/annurev-
psych-010814-015240
Cacioppo, J. T., Cacioppo, S., Cole, W. W., Capitanio, J. P.,
Goossens, L., & Boomsma, D. I. (2015). Loneliness across
phylogeny and a call for animal models. Perspectives on
Psychological Science, 10, 202–212.
Cacioppo, J. T., Hawkley, L. C., Ernst, J. M., Burleson, M.,
Berntson, G. G., Nouriani, B., & Spiegel, D. (2006).
Loneliness within a nomological net: An evolutionary per-
spective. Journal of Research in Personality, 40, 1054–1085.
Cacioppo, J. T., Hawkley, L. C., Crawford, L. E., Ernst, J. M.,
Burleson, M. H., Kowalewski, R. B., . . . Berntson, G. G.
(2002). Loneliness and health: Potential mechanisms.
Psychosomatic Medicine, 64, 407–417.
Cacioppo, J. T., & Patrick, W. (2008). Loneliness: Human nature
and the need for social connection. New York, NY: W. W.
Norton.
Cohen, S., Doyle, W. J., Skoner, D. P., Rabin, B. S., & Gwaltney,
J. M. (1997). Social ties and susceptibility to the common
cold. Journal of the American Medical Association, 277,
1940–1944.
Cooper, H. (1998). Synthesizing research: A guide for literature
reviews (3rd ed.). Thousand Oaks, CA: Sage.
Coyle, C. E., & Dugan, E. (2012). Social isolation, loneliness and
health among older adults. Journal of Aging and Health,
24, 1346–1363. doi:10.1177/0898264312460275
De Jong-Gierveld, J., & Kamphuis, F. (1985). The development
of a Rasch-type loneliness scale. Applied Psychological
Measurement, 9, 289–299.
Dietz, W. H., & Gortmaker, S. L. (1985). Do we fatten our chil-
dren at the television set? Obesity and television viewing in
children and adolescents. Pediatrics, 75, 807–812.
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-
plot-based method of testing and adjusting for publication
bias in meta-analysis. Biometrics, 56, 455–463.
Dykstra, P. A., van Tilburg, T. G., & de Jong Gierveld, J. (2005).
Changes in older adult loneliness: Results from a seven-
year longitudinal study. Research on Aging, 27, 725–747.
doi:10.1177/0164027505279712
Euromonitor International. (2014). The rising importance of single
person households globally: Proportion of single person house-
holds worldwide. Available from http://blog.euromonitor.com
Flegal, K. M., Kit, B. K., Orpana, H., & Graubard, B. I. (2013).
Association of all-cause mortality with overweight and obe-
sity using standard body mass index categories: A systematic
review and meta-analysis. Journal of the American Medical
Association, 309, 71–82. doi:10.1001/jama.2012.113905
Loneliness and Isolation as Mortality Risk Factors
Grant, N., Hamer, M., & Steptoe, A. (2009). Social isolation and
stress-related cardiovascular, lipid, and cortisol responses.
Annals of Behavioral Medicine, 37, 29–37. doi:10.1007/
s12160-009-9081-z
Greenfield, T. K., Rehm, J., & Rogers, J. D. (2002). Effects of
depression and social integration on the relationship
between alcohol consumption and all-cause mortality.
Addiction, 97, 29–38.
Hawkley, L. C., & Cacioppo, J. T. (2003). Loneliness and path-
ways to disease. Brain, Behavior, and Immunity, 17(Suppl.
1), S98–105.
Hawkley, L. C., & Cacioppo, J. T. (2010). Loneliness matters:
A theoretical and empirical review of consequences and
mechanisms. Annals of Behavioral Medicine, 40, 218–227.
doi:10.1007/s12160-010-9210-8
Hawkley, L. C., Thisted, R. A., & Cacioppo, J. T. (2009).
Loneliness predicts reduced physical activity: Cross-
sectional & longitudinal analyses. Health Psychology, 28,
354–363. doi:10.1037/a0014400
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social
relationships and mortality risk: A meta-analytic review.
PLoS Medicine, 7(7), e1000316. doi:10.1371/journal.
pmed.1000316
Holwerda, T. J., Beekman, A. T., Deeg, D. J., Stek, M. L., van
Tilburg, T. G., Visser, P. J., . . . Schoevers, R. A. (2012).
Increased risk of mortality associated with social isolation
in older men: Only when feeling lonely? Results from the
Amsterdam Study of the Elderly (AMSTEL). Psychological
Medicine, 42, 843–853. doi:10.1017/S0033291711001772
House, J. S., Landis, K. R., & Umberson, D. (1988). Social rela-
tionships and health. Science, 241, 540–545.
Hughes, M. E., Waite, L. J., Hawkley, L. C., & Cacioppo, J. T.
(2004). A short scale for measuring loneliness in large sur-
veys: Results from two population-based studies. Research
on Aging, 26, 655–672.
Klinenberg, E. (2012). Going solo: The extraordinary rise and
surprising appeal of living alone. New York, NY: Penguin.
Linehan, T., Bottery, S., Kaye, A., Millar, L., Sinclair, D., &
Watson, J. (2014). 2030 vision: The best and worst futures
for older people in the UK. London, England: Independent
Age and International Longevity Centre-UK.
McPherson, M., & Smith-Lovin, L. (2006). Social isolation in
America: Changes in core discussion networks over two
decades. American Sociological Review, 71, 353–375.
Parigi, P., & Henson, W. (2014). Social isolation in America.
Annual Review of Sociology. Advance online publication.
Peplau, L. A., & Perlman, D. (Eds.). (1982). Loneliness: A source-
book of current theory, research, and therapy. New York,
NY: Wiley.
Perissinotto, C. M., & Covinsky, K. E. (2014). Living alone,
socially isolated or lonely—What are we measuring? Journal
of General Internal Medicine, 11, 1429–1431. doi:10.1007/
s11606-014-2977-8
Perissinotto, C. M., Stijacic Cenzer, I., & Covinsky, K. E. (2012).
Loneliness in older persons: A predictor of functional
decline and death. Archives of Internal Medicine, 172,
1078–1083. doi:10.1001/archinternmed.2012.1993
Peters, J., Sutton, A., Jones, D. R, Abrams, K. R., & Rushton, L.
(2006). Comparison of two methods to detect publication
237
bias in meta-analysis. Journal of the American Medical
Association, 295, 676−680.
Pressman, S. D., Cohen, S., Miller, G. E., Barkin, A., Rabin, B. S.,
& Treanor, J. J. (2005). Loneliness, social network size, and
immune response to influenza vaccination in college fresh-
men. Health Psychology, 24, 297–306. doi:10.1037/0278-
6133.24.3.297
Qualter, P., Vanhalst, J., Harris, R., Van Roekel, E., Lodder, G.,
Bangee, M., . . . Verhagen, M. (2015). Loneliness across
ontogeny. Perspective on Psychological Science, 10, 250–
264.
Roelfs, D., Shor, E., Kalish, R., & Yogev, T. (2011). The rising
relative risk of mortality among singles: Meta-analysis and
meta-regression. American Journal of Epidemiology, 74,
379–389.
Russell, D., Peplau, L. A., & Cutrona, C. E. (1980). The revised
UCLA Loneliness Scale: Concurrent and discriminant valid-
ity evidence. Journal of Personality & Social Psychology, 39,
472–480.
Shankar, A., McMunn, A., Banks, J., & Steptoe, A. (2011).
Loneliness, social isolation, and behavioral and biological
health indicators in older adults. Health Psychology, 30,
377–385. doi:10.1037/a0022826
Shor, E., Roelfs, D. J., & Yogev, T. (2013). The strength of family
ties: A meta-analysis and meta-regression of self-reported
social support and mortality. Social Networks, 35, 626–638.
doi:10.1016/j.socnet.2013.08.004
Steptoe, A., Shankar, A., Demakakos, P., & Wardle, J. (2013).
Social isolation, loneliness, and all-cause mortality in
older men and women. Proceedings of the National
Academy of Sciences, USA, 110, 5797–5801. doi:10.1073/
pnas.1219686110
Theeke, L. A. (2010). Sociodemographic and health-related risks
for loneliness and outcome differences by loneliness status
in a sample of U.S. older adults. Research in Gerontological
Nursing, 3, 113–125. doi:10.3928/19404921-20091103-99
Thoits, P. A. (2011). Mechanisms linking social ties and support
to physical and mental health. Journal of Health and Social
Behavior, 52, 145–161.
Uchino, B. N. (2006). Social support and health: A review of
physiological processes potentially underlying links to dis-
ease outcomes. Journal of Behavioral Medicine, 29, 377–
387. doi:10.1007/s10865-006-9056-5
Victor, C. R., & Bowling, A. (2012). A longitudinal analysis of
loneliness among older adults in Great Britain. The Journal
of Psychology, 146, 313–331.
Victor, C. R., & Yang, K. (2012). The prevalence of loneliness
among adults: A case study of the United Kingdom. The
Journal of Psychology, 146, 85–104.
Wang, Y., & Beydoun, M. A. (2007). The obesity epidemic in
the United States—Gender, age, socioeconomic, racial/eth-
nic, and geographic characteristics: A systematic review and
meta-regression analysis. Epidemiologic Reviews, 29, 6–28.
Wilson, C., & Moulton, B. (2010). Loneliness among older
adults: A national survey of adults 45+. Washington, DC:
AARP Inc.
Yildirim, Y., & Kocabiyik, S. (2010). The relationship between
social support and loneliness in Turkish patients with can-
cer. Journal of Clinical Nursing, 19, 832–839.
 Journal of Personality and Social Psychology Copyright 1987 by the American Psychological Association. Inc.
1987. Vol. 52. No. 6 , ' 1 1 2 2 - 1 1 3 1 0022-3514/87/500.75

Positive Affect Facilitates Creative Problem Solving

Alice M. Isen
University of Maryland Baltimore County

Kimberly A. Daubman Gary P. Nowicki

University of Maryland (College Park) University of Maryland Baltimore County

Four experiments indicated that positive affect, induced by means of seeing a few minutes of a
comedy film or by means of receiving a small bag of candy, improved performance on two tasks
that are generally regarded as requiring creative ingenuity: Duncker's (1945) candle task and M. T.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Mednick, S. A. Mednick, and E. V. Mednick's (1964) Remote Associates Test. One condition in
This document is copyrighted by the American Psychological Association or one of its allied publishers.

which negative affect was induced and two in which subjects engaged in physical exercise (intended
to represent atfectless arousal) failed to produce comparable improvements in creative performance.
The influence of positive affect on creativity was discussed in terms of a broader theory of the impact
of positive affect on cognitive organization.

Recent research has suggested that positive affect can influ-
ence the way cognitive material is organized and thus may in-
fluence creativity. Studies using three types of tasks (typicality
rating, sorting, and word association) indicated that persons in
whom positive affect had been induced differed from those in
control conditions in the associations that they gave to com-
mon, neutral words (Isen, Johnson, Mertz, & Robinson, 1985)
and in the pattern and degree of relatedness that they depicted
among stimulus elements (Isen & Daubman, 1984). It has been
suggested that these differences are due to differences between
the groups in the tendency to relate and integrate divergent ma-
terial. This process of bringing together apparently disparate
material in a useful or reasonable but unaccustomed way is cen-
tral to most current conceptualizations of the creative process
(e.g., Koestler, 1964; S. A. Mednick, 1962). Thus, it seems likely
that positive affect may promote creativity.
In one of these series of studies, it was found that persons in
whom positive affect had been induced (in any of three ways)
tended to categorize stimuli more inclusively than did persons
in the control conditions (Isen & Daubman, 1984). This ten-
dency was reflected by performance on both a rating task and a
sorting task. On the sorting task, positive-affect subjects tended
to group more stimuli together than control subjects did, thus
indicating that, for them, more of the items could be seen as
related. On the rating task, persons in whom positive affect had
been induced tended to rate nontypical exemplars of a category
more as members of the category than control subjects did. For
example, in a task similar to that used by Rosch (1975) in as-
sessing the prototypicality of category exemplars, persons in
whom positive affect had been induced—by refreshments at the
experimental session, receipt of a small gift, or viewing 5 min
of a comedy film—gave higher ratings to the atypical exemplars
This research was supported in part by Research Grant 37452 from
the National Institute of Mental Health.
Correspondence concerning this article should be addressed to Alice
M. Isen, Department of Psychology, University of Maryland Baltimore
County, Catonsville, Maryland 21228.
elevator, camel, and feet as members of the category vehicle than
did subjects in control conditions. That is, they gave evidence
of being better able than control subjects to see the relatedness
between these exemplars and the category or to see aspects of
these exemplars that would make them like members of the cat-
egory.
In a second series of studies, persons in whom positive affect
had been induced—by having refreshments (juice and cookies)
at the experimental session, giving word associations to positive
words, receiving a small gift, or viewing 5 min of a comedy
film—gave more unusual first associates to neutral words, ac-
cording to the Palermo and Jenkins (1964) norms, than did per-
sons in the control conditions (Isen et al., 1985).
Both of these effects have been interpreted as indicating an
influence of positive affect on cognitive organization because
they reflect the relatedness that people see among ideas or cog-
nitive elements. One shows how persons organize stimuli set be-
fore them when they set out to do so; the other indicates the
concepts that are cued for people by given stimulus words. In
each type, there is evidence of greater integration or perception
of interrelatedness of stimuli among people who are feeling
happy.
Either of the effects of positive feelings on cognitive organiza-
tion that has been observed thus far (atypical categorization and
word association) might also be seen as reflective of an influence
of affect on creativity. The categorization task involves either
seeing nontypical yet plausible ways of relating items, or seeing
aspects of the items that are real and useful but not usually focal
in people's attention; these processes are central to creativity.
Likewise, responding with related but nontypical word associa-
tions can be seen as creative, and in fact word association tasks
have often been explicitly linked with creativity (e.g., Freed-
ntan, 1965; Maltzman, Simon, Raskin, & Licht, I960; M. T.
Mednick, S. A. Mednick, & E. V. Mednick, 1964; S. A. Med-
nick, 1962).
Moreover, S. A. Mednick's (1962) theory of creativity spe-
cifically relates word associations to cognitive representation
and defines creativity in terms of the formation of new associa-

1122
 POSITIVE AFFECT PROMOTES CREATIVITY
tions or combinations of cognitive elements that are in some
way useful. Thus, this theory of creativity is also compatible
with the process suggested to result from positive affect, a pro-
cess involving making new associations and combining cogni-
tive elements in new ways.
Our studies are designed to test more directly the proposition
that positive affect promotes creativity, and to extend the evi-
dence beyond that available in the word association and catego-
rization studies. In these experiments, we investigate whether
the creativity promoted by positive affect includes problem-
solving innovation.
The task that we used in the first two studies was the candle
task used by Karl Duncker (1945) in his demonstrations of cre-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ative problem solving (actually, his demonstration of what he
This document is copyrighted by the American Psychological Association or one of its allied publishers.
termed functional fixedness). In this task, the subject is pre-
sented with a box of tacks, a candle, and a book of matches and
is asked to attach the candle to the wall (a corkboard) in such a
way that it will burn without dripping wax on the table or floor.
The task in Studies 3 and 4 was based on the Remote Associates
Test used by M. T. Mednick et al. (1964) as a test of creativity.
This task requires subjects to think of a word that is related to
each of three other words presented.
Experiment 1
Method
Subjects Subjects were 65 male and female students who partici-
pated in exchange for extra credit in their introductory psychology
classes.
Design. Subjects were assigned randomly to one of four conditions.
Two of the four involved manipulation of aifect (positive and neutral)
by means of films; the other two involved differences in the way materi-
als for the task were displayed.
Affect manipulation. Positive affect was induced in Condition 1 of
this study by having subjects view 5 min of a comedy film. Gag Reel.
used previously in research and found to induce the desired state (e.g.,
Isen & Gorgoglione, 1983). The segment shown consists of blooperc
(i.e., errors that were cut from the film and not shown to the public)
from two old television westerns (Have Gun Will Travel and Gunsmoke]
and one from The Red Skelton Show. In Condition 2, a neutral-film
control condition, subjects were shown a 5-min segment from a math
film. Area Under a Curve, also used previously. In each case, subjects
were told that the film was being pretested for use in another study to
be held the following term and that our interest was in people's general
reactions to it. They were instructed not to try to memorize anything
from the film, but just to watch it.
Display manipulation. In the remaining conditions, no films were
shown, and affect was not induced. Instead, differences between these
two conditions were created by differences in the ways in which the
items for the candle task were displayed. In Condition 3 (as in Condi-
tions I and 2), the usual display was presented: a box filled with tacks, a
candle, and a book of matches. In Condition 4, the same items were
presented but in a facilitative display: The tacks were placed in a pile
next to the empty box. Adamson (1952) found that arranging the items
in this way facilitated performance on the task.
Procedure. Subjects were admitted to the laboratory in groups of 2
to 4. They were seated, each several seats away from the next person,
and given a minute of introduction to the study. During this introduc-
tion they were asked not to speak to each other, interact with each other
in any other way, or call out to the experimenter during the session.
Then, in Conditions 1 and 2, subjects were shown one of the 5-min film
1123
Table 1
Study I: Number and Percentage of Subjects Obtaining
Correct Solution in Each Condition
Condition n %
Positive film 9/12 7S
Neutral film 3/15 20
Facilitative display 19/23 83
No manipulation 2/15 13
segments described earlier. Subjects in Conditions 3 and 4 proceeded
directly to the next phase of the experiment.
After the films were shown but before the problem-solving task was
presented, as a check on the manipulation of affect, subjects completed
a word-rating scale in which they rated the pleasantness of unfamiliar
words. Previous studies have shown effects of induced feelings on such
measures (e.g., ratings of ambiguous slides) and in some cases have used
them as indirect assessments of affect (e.g.. Isen el al.. [985; Isen &
Shalker, 1982).
Next, subjects were seated at individual tables m the four corners of
the room, approximately 25 ft (7 m) apart, facing the corner so that
they were unable to see each other. The materials for the candle task
were on the tabl-j, in the appropriate display, but under a cover until the
task was explained by the experimenter. The following instructions were
read to subjects in Conditions 1, 2, and 3:
On the table are a book of matches, a box of tacks, and a candle.
Above the table on the wall is a corkboard. Your task is to affix the
candle to the corkboard in such a way that it will burn without
dripping wax onto the table or the floor beneath. You will be given
10 minutes to work on the problem.
Subjects in Condition 4 heard the same instructions except that the box
and the tacks were listed separately as items on the table. The problem
as usually presented can be solved if the box is emptied, tacked to the
wall, and used as a platform (candle holder) for the upright candle.
At the end of the session the purpose of the study was explained to
the subjects, they were shown the solution to the problem, and the ex-
perimenter thanked the subjects for participating.
Results and Discussion
Results of the manipulation check indicated that unfamiliar
words were rated more positively by subjects in the positive-
affect condition than by subjects in the neutral-film condition,
((25) = 2.00, p< .05. This suggests that the appropriate affective
state was induced.
Table 1 presents the data showing the number and percentage
of subjects obtaining the solution in each condition. Chi-square
tests indicated that, as predicted, subjects in the positive-affect
condition produced significantly more solutions than did sub-
jects in the neutral-affect control condition, x 2 ( l . A' = 27) =
8.19, p < .01, and subjects in the facilitative-display condition
produced significantly more solutions than did subjects in the
control-display condition x 2 ( l , A' = 36) = 17.62, p < .01. The
control conditions did not differ from each other on number of
solutions produced, suggesting that the neutral-film condition
may be considered comparable to a no-manipulation control in
effect.
These results suggest that positive affect can facilitate creative
problem solving. Duncker (1945) spoke of people's inability to
 1124 A. ISEN, K. DAUBMAN, AND G. NOWICKI
solve this task as stemming from functional fixedness, or their
inability to consider alternative uses for the box. Adamson
(1952) showed, and we have replicated his results, that display-
ing the items separately facilitates performance on this task. He
reasoned that highlighting the independence of the box and
tacks might allow each object to be utilized in its own right.
Likewise, Glucksberg and Weisberg (1966) found that providing
labels of all of the items in the display, including the box, also
facilitated performance; and Higgins and Chaires (1980) found
that giving subjects experience with unaccustomed linguistic
structures that tended to emphasize the separateness of mem-
bers of common pairs ("tray and tomatoes" instead of "tray of
tomatoes") also facilitated performance on the candle task. In
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
addition, Glucksberg and Danks (1967. 1968), in a series of
This document is copyrighted by the American Psychological Association or one of its allied publishers.
studies using the electrical-circuit problem (another task re-
quiring innovation to break set), demonstrated that procedures
that increased the accessibility of semantic categories relevant
to the solution of the problem facilitated seeing the required
novel function and solving the problem.
All of these findings (except the present ones, that positive
affect can facilitate performance on this task) can be interpreted
as suggesting that calling attention to the box as an item in its
own right, or calling attention specifically to the aspect of the
box that might prove helpful, can facilitate seeing the crucial
alternative function of the box. Subjects in our positive-affect
condition were also able to break set and see the additional fea-
tures of the box that would allow them to realize its potential
for solving the problem, even though there was nothing in our
procedure that specifically called their attention to the box, to
the potential solution, or to the aspects of the box that might
prove especialK helpful in solving the task. Thus, there may be
something about positive affect itself that facilitates either sec-
ing more aspects of objects or in some other way seeing objects
more fully; including their potential for combination with other
objects and with the problem under consideration.
Possibly it is through creating a complex cognitive context
that positive affect promotes creative problem solving. The im-
portance of context in determining interpretation has been dis-
cussed recently by several cognitive psychologists (e.g., Brans-
ford. 1979;Jenkins, 1974). Moreover, complexity of context has
been specifically related to originality of word associations
(Cramer, 1968). In addition, complexity of context is likely to
result in a relaxing or broadening of focus of attention, a pro-
cess that has also been related to the creative process (Martin-
dale, 1981). Thus, it seems that increasing complexity of cogni-
tive context might influence interpretation and organization of
stimuli and promote creative responding.
For persons who are feeling happy, the complex context arises
from the fact that positive feelings cue and facilitate access to
positive material in memory (e.g., Isen, Shalker, Clark, & Karp,
1978; Teasdale & Fogarty, 1979) combined with the fact that
positive material is more extensive and diverse than other mate-
rial (e.g., Boucher & Osgood, 1969). Thus, a person who is feel-
ing happy has more ready access to a large and diverse set of
cognitive material, and this constitutes a complex cognitive
context.
This elaborated and enriched context may influence cogni-
tive organization by increasing the number of ways in which
ideas or objects can be integrated or related to one another. This
may occur because of an increase in the number of aspects of
ideas that are noticed, but there are other possible means by
which context may influence organization as well, and the exact
mechanism of the effect is as yet unknown. For example, as
noted, the easy accessibility of many ideas simultaneously likely
results in defocused attention, and this effect itself may influ-
ence cognitive organization. Thus, the presence of a complex
cognitive context may cause many features of items and prob-
lems to become salient, so that more functions of the objects
and more possibilities for solution can be seen. This interpreta-
tion is only speculative, however, and will have to be addressed
specifically in future research. The studies that follow in this
article focus on establishing the empirical link between positive
feelings and creative problem solving.
Experiment 2
Experiment 2 was conducted to replicate and extend the
finding that one's affective state can influence creative problem
solving. First, we wanted to explore the conditions of affect that
might produce the same result, in order to learn more about the
specific aspects of feeling states that might be responsible for
the effect. We examined whether positive affect other than feel-
ings of amusement might have the same effect, we included a
negative-affect condition, to begin investigation of the impact of
negative affect on creativity, and we included an arousal-control
group, to get some idea about whether arousal, independent of
affect, might influence performance on this task. The predic-
tion that arousal alone might improve performance on tasks
such as these has an intuitive appeal to many people, and it
may also follow from a theory such as spreading activation, as
described by, for example, Anderson (1983).
We predicted that the second means of positive-affect induc-
tion (a small gift) would facilitate solution of the candle prob-
lem, as would the humorous film. Because we conceptualized
this effect as attributable to cognitive processes resulting from
positive affect, rather than to a process of general activation, we
expected the arousal condition to have no effect. We did not
predict the deficit in performance due to arousal that some
might have expected on this complex task requiring innovative
responding, in part because success rates were already so low in
the control conditions.
We did not expect negative affect to improve creative prob-
lem solving, because it has often been associated with con-
stricted thinking and reduced cue utilization (e.g., Bruner. Mat-
ter, & Papanek, 1955; Easterbrook, 1959). At the same time, we
did not expect impaired performance, because of the relatively
low success rates in the control conditions. Moreover, recent
research on the impact of negative affect on social behavior,
memory, other cognitive processes, and performance has indi-
cated complex effects, with negative affect sometimes facilitat-
ing, sometimes impairing, and sometimes leaving unaffected
behaviors of interest (see, e.g., Isen, 1984, for discussion). Thus,
the prediction regarding the impact of negative affect relative to
the control group was difficult to make; we did. however, expect
a difference between the negative-affect condition and the com-
parable positive-affect group.
 POSITIVE AFFECT PROMOTES CREATIVITY
Table 2
Study 2: Manipulation Check and Mean Affect Ratings
(Positive-negative Scale) in Each Film Condition
Condition M MS, n film separately). Two separate analyses of variance (ANOVAS),
Comedv 3.11 1.29 u
Neutral 3.90 1.04 10
Negative 5.52 2.12 12
Ratings were made on 7-point scales (1 = positive, 1 = negative).
Method
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Subjects. Subjects were 33 male and 83 female students who partici-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
pated in this experiment for extra credit for their introductory psychol-
ogy classes. Male-female composition was roughly equivalent in each
of the conditions (between 26% and 32% men).
Manipulations. In one positive-affect condition, subjects each re-
ceived a junior candy bar as an expression of thanks for their participa-
tion. In the other positive-affect condition, participants viewed the same
segment of Gag Reel that was shown in Experiment 1. The negative-
affect manipulation consisted of viewing 5 min of Night and Fog, a doc-
umentary film depicting Nazi concentration camps. To control for the
effects of watching a film on performance, one control group watched
the same segment of the math film (Area Under a Curve) that was shown
in Experiment 1. Again, subjects were told that the films were being
pretested for use in another study the following term and that our inter-
est was in people's general reactions to them.
To control for the effect of simple arousal on performance, another
control group exercised for 2 min by stepping up on and down from a
cement block. This exercise is similar to that known as the Step Test
and results in elevated heart rate. The third control group received no
manipulation.
Procedure. Subjects assigned randomly to one of the conditions de-
scribed (two positive-affect, one negative-affect, and three affect-control
conditions) were admitted to the laboratory in groups of 1 to 3. The
same instructions and procedures used in Experiment 1, including pre-
cautions against interaction among subjects, were used in this study.
After undergoing the assigned manipulation, some randomly selected
subjects from each condition indicated their feelings on five 7-point Lik-
en scales representing five affective dimensions. In the film conditions,
subjects were asked to indicate how the film had made them feel; in the
other affect-induction conditions, subjects were asked just to indicate
their feelings. Thus, the manipulation check was slightly but meaning-
fully different in the two types of conditions (film vs. no film).
Four of the scales in the questionnaire were intended only as filler
items (refreshed vs. tired, calm vs. anxious, alert vs. unaware, and
amused vs. sober). The other scale (positive vs. negative) was included
for the purpose of checking whether the appropriate affective states had
been induced. We expected that subjects in the comedy-film condition
would feel more positive, and subjects in the negative-film condition
more negative, than subjects in the neutral-film condition and that sub-
jects in the candy condition would feel more positive than subjects in
the no-manipulation control condition; the exercise condition was not
expected to differ from the control condition in rated affect.
Subjects, seated at individual tables approximately 20 ft (6 m) apart,
with dividers positioned between them so that they could not see each
other, were instructed about the task in a manner identical to that used
in Conditions 1, 2. and 3 in Experiment 1. They were then given 10 min
to complete the task. Whether they solved the problem successfully and,
if so, time to solution were recorded. At the end of the session, subjects
were debriefed as to the purpose of the study and were shown the solu-
tion to the problem if they had not solved it.
1125
Results and Discussion
Manipulation check. Tables 2 and 3 present the mean affect
ratings (manipulation check) in each condition (film and no
one including the film conditions and the other including the
candy, exercise, and no-manipulation conditions, were per-
formed on the data. Separate analyses were performed because
the manipulation check question (as described in the Proce-
dure) was different for the two types of conditions. The analysis
contrasting the three film conditions indicated a significant
effect, t\2, 30) = 11.44, p< .01. Subsequent / tests revealed that
subjects in the comedy-film condition reported that they felt
more positive, /(19) = 1.68, p = .056, one-tailed, and those in
the negative-film condition more negative ((20) = 3.05. p < .01.
one-tailed, than did subjects in the neutral-film condition.
However, the analysis contrasting the candy, exercise, and no-
manipulation conditions indicated no significant differences
among these three groups, F(2, 45) = .60. Thus, contrary to
expectation, subjects in the candy condition did not report that
they felt more positive than subjects in their comparison group,
the no-manipulation condition, ((34) = 1 . 1 7 , / j > .10. Subjects
in these no-film conditions were not compared with subjects
in any of the film conditions on the affect manipulation check
because, as noted, the affect-rating tasks were differently fo-
cused in these two distinct contexts and therefore should be dis-
tinguished from one another.
The manipulation check data confirm the expected positive-
affect and negative-affect induction in the film conditions but
not in the gift (candy bar) condition. This suggests that our gift,
a small (junior) candy bar in nothing but its commercial wrap-
per, was not successful in inducing pleasant feelings in subjects.
Perhaps it was too small a gift or inappropriately packaged.
Moreover, in retrospect, it is clear to us that this type of ma-
nipulation check (self-reported mood) is especially inappropri-
ate for persons who have received treatments such as a gift of
candy. The measure is simply too reactive and may even cause
subjects to be suspicious of the experimenter's intent in giving
them the candy and resentful rather than elated in the remain-
der of the session. It is probably inappropriate also for treat-
ments such as the exercise condition, in which, again, there is
no apparent reason for the question and it is therefore too reac-
tive. (In the film conditions, in contrast, the affect question is
an integral part of the task of pretesting the film.)
Problem solving. Table 4 presents the percentage of subjects
in each condition who solved the problem. We predicted that a
higher proportion of subjects in the positive-affect conditions
Table 3
Manipulation Check and Mean Affect Ratings (Positive-
Negative Scales) in Each Nonfilm Condition
Condition M MS, n
Candy bar 2.69 1.27 IS
No manipulation 3.18 207 18
Exercise 3.10 2.88 12
Note. Ratings were made on 7-point scales ( 1 - pasinve, 1 - negative).
 1126 A. ISEN, K. DAUBMAN, AND G. NOWICKI
than in their respective control conditions (neutral film and no
manipulation) would solve the problem correctly. As expected,
a x2 test indicated that a higher percentage of subjects who had
viewed the comedy film solved the problem than of subjects
who had viewed the control film, x2( 1, A' = 38) = 9.46, p < .01;'
however, subjects who had received a candy bar did not perform
better than subjects in the no-manipulation condition. x 2 U,
N = 39) < 1. We did not expect any of the other conditions to
facilitate solution of the problem, and in fact chi-squares com-
paring the exercise condition with its appropriate control (no
manipulation) and the negative-affect condition with its control
group (neutral film) revealed no significant difference, x 2 0,
N= 38) < 1 and x 2 (l, N = 39) = 2.27, p > .1, respectively.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Moreover, subjects in the positive-affect film condition per-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
formed significantly better than subjects in all of these compari-
son conditions, combined, x 2 U, A' = 96) = 10.41, p < .01, or
singly, x 2 0, A' = 38) = 3.89, p < .05; x 2 0, N = 39) = 3.09, p <
.05;2 and x 2 U, If = 38) = 7.24, p < .01, for exercise, negative
affect, and no-manipulation, respectively.
Although no specific prediction was made regarding time to
reach a solution, we thought that positive affect might reduce
solution time (among those solving the problem) as well as facil-
itate problem solution, as had been observed. Individual / tests
revealed, however, that there were no significant differences
among the conditions in the amount of time taken to solve the
problem correctly (see Table 5 for means and variances). These
results should be interpreted with caution, however, because of
the extremely small sample sizes, especially in some of the con-
ditions where only two or three persons solved the problem.
The results of this experiment support and extend those of
the preceding one. Subjects in whom positive affect had been
induced through viewing a comedy film were more likely to find
a creative solution than were subjects who viewed a negative
film or subjects in any one of three neutral-affect conditions
(neutral film, no manipulation, and no-film arousal). Further-
more, this experiment suggests that the superior performance
of subjects in the positive-affect condition is not due to a rela-
tively high arousal level, as subjects in the exercise condition
performed no better than subjects in the other control condi-
tions and significantly worse than subjects in the comedy-film
condition. We chose to represent arousal by means of exercise,
even though this may seem questionable on some counts, be-
cause there is a growing body of literature that conceptualizes
arousal in this way (e.g., Zillmann. 1979). If the meaning of
arousal is taken to be better represented by the negative-affect
condition than the exercise condition, then again there is no
Table 4
Study 2: Number and Percentage of Subjects Obtaining
Correct Solution in Each Condition
Condition n %
Positive film 11/19 58
Neutral film 2/19 11
Negative film 6/20 30
Candy bar 5/20 25
No manipulation 3/19 16
Exercise 5/19 26
Table 5
Study 2: Mean Amount of Time in Minuter to Solve
Problem in Each Condition
Condition M MS, /;
Comedy film 4.30 11.29 8'
Neutral film 4.20 8.14 2
Negative film 4.29 7.27 6
Candy bar 5.81 14.44 5
No manipulation 3.06 9.01 3
5
Exercise 6.71 11.68
a
Three data points were lost because of a malfunction of timing devices.
evidence that the effect of positive affect in our study is attribut-
able only to arousal.
Contrary to expectation, subjects in the gift condition did not
show improved performance as those in the comedy-film condi-
tion did. Thus, it may be specifically humor, and not positive
affect more generally, that gives rise to improved creative prob-
lem solving. On the other hand, the manipulation check data
suggest that positive affect may not have been induced in that
condition. Moreover, as we have speculated, the rather heavy-
handed technique of inexplicably inquiring about subjects'
moods after giving them a gift may have ruined the lighthcarted
affect induction that was intended. Previous research has indi-
cated that some affect inductions cannot withstand a manipula-
tion check questionnaire (e.g.. Frost & Greene, 1982; [sen &
Gorgoglione, 1983), and this may be another instance of a sim-
ilar phenomenon, if for a slightly different reason.
Thus, in subsequent studies, we retained the hypothesis that
positive affect, induced in ways other than via humor, could fa-
cilitate creativity. However, we attempted to be more sophisti-
cated in the way we went about giving the gift. First, we made
the gift more believable and more charming by wrapping it, and
second, we did not follow the gift with an obvious question
about mood. (Later, in Study 4, we also considered again the
effect of affectless arousal.)
Study 3, then, again examined the impact of a small gift on
creativity. This time, however, a different measure of creative-
1
Because some of the subjects in this experiment were run in pairs
or in triads, it is important to address the issue of independence of sub-
jects. To the extent that subjects who participate in the same session are
allowed to interact with each other, independence of subjects is jeopard-
ized. As indicated in the Procedure sections of the studies, we took great
pains to make sure subjects did not interact with each other, and there-
fore the issue of independence is not likely to call the conclusions of this
particular set of studies into question. One indication that this conclu-
sion is correct is available in the data of 38 subjects who were run with
no other subject present. Under this circumstance, subjects in the com-
edy-film condition were more likely to solve the problem than were sub-
jects in the control conditions combined, x 2 C- N = 20) - 5.06. p <
.05, whereas the comparable subjects in the negative-affect and exercise
conditions were not more likely than control subjects 10 solve the prob-
lem, x 2 (l, N = 21) < 1.
2
This probability level reflects a one-tailed test, for which the proba-
bility levels of critical values in a two-tailed, standard chi-squarc table
are halved (Siegel, 1956, p. 110).
 POSITIVE AFFECT PROMOTES CREATIVITY
problem-solving performance was used: items from the Remote
Associates Test.3 This test is an individual difference measure
created by the Mednicks (M. T. Mednick et al., 1964) in accord
with S. A. Mednick's (1962) theory of creativity. We used it as
a dependent measure, assuming that performance on it reflects
cognitive organization (which, in turn, we hypothesize to be
malleable and responsive to affective state, despite the fact that
people may differ in their typical style of cognitive organiza-
tion). Before proceeding with Studies 3 and 4, we pretested the
items for difficulty because we assumed that, given the relatively
small sample sizes with which we had to work, differences be-
tween affect conditions might not be apparent on items of very
great or very low difficulty.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Pretest for Experiment 3
Method
Subjects. Fifty-one students from an introductory psychology class,
fulfilling a class requirement, served as subjects in this pretest.
Materials. Seventy-eight Remote Associates Test items of differing
difficulty levels were assembled into 78-page booklets containing one
item per page. Each item consisted of three words followed by a blank
space. Subjects were instructed to provide, in the blank space, a word
that related to each of the three words given in the item. An example of
a Remote Associates Test item (of moderate difficulty) is this:
MOWER ATOMIC FOREIGN .4
Procedure. Subjects were seated individually al tables and given a
78-page remote-associates booklet. Each page of the booklet contained
three words and a blank, as illustrated in the display.
Subjects were told the instructions for the task and advised that no
more than I or 2 min should be spent on any one item, as any additional
time was not likely to produce an answer. In actuality, subjects spent
about a half minute per item. After completion of the booklet, subjects
were debriefed and given credit for their participation.
Results
For each item, the percentage of subjects obtaining the cor-
rect answer was calculated and used as the item's score. Scores
ranged from 0 to 88.2. Items with scores between 0 and 22 (up
to 22% of the sample answered them correctly) were designated
difficult items ( 1 2 items). Those with scores between 66 and 88
were designated as easy items (11 items), and those with scores
between 23 and 65 were considered of moderate difficulty (55
items).
Seven items were selected from each pool (difficult, easy, and
moderate items) to create a test containing 21 items, of three
difficulty levels, for use in our study. We attempted to select
items representative of their range. The average difficulty scores
of the three groups of items thus obtained were 4.2 (difficult
items), 37.8 (moderate items), and 72.1 (easy items).
Experiment 3
Method
Subject*. Subjects were 24 male and 22 female students enrolled in
introductory psychology classes who participated as part of a course
requirement. Approximately half the subjects in each condition were
1127
Table 6
Study 3: Mean Number of Remote Associates Test
Items Correct in Each Condition
High Moderate
difficulty difficulty Low difficulty
Condition M MS, M MS, M MS,
Candy .50 .66 4.38 2.57 5 38 3.45
Control .60 .36 3.45 3.00 5.10 3.04
Note. For the candy condition, n - 26. For the control condition, n --
20.
male (12 women and 14 men in the positive-affect condition. 10 women
and 10 men in the control).
Procedure. Subjects were assigned randomly to either a positive-affect
or a neutral-affect condition. In the positive-affect condition, the experi-
menter thanked the subjects for coming and gave them a decorated hag
of candy (10 pieces of wrapped hard candy in a Glad Funtime sandwich
bag), which has been used in other studies to induce affect (e.g.. Isen et
al., 1985). In the neutral-affect control condition, there was no attempt
to manipulate affect.
Subjects, working independently at separate desks, were given a short-
ened version (the 21 items described) of the Remote Associates Test. On
the basis of the pretest results, subjects were presented with seven easy,
seven moderately difficult, and seven very difficult items. The items
were presented in random order in a booklet, with one on each page,
and subjects were given the usual instructions for the Remote Associates
Test. Each subject was allowed to work on the test until he or she had
finished or felt that additional time would be of no use in solving the
problems. At the end of the session, subjects were debriefed and given
credit for participation, as always.
Results and Discussion
Table 6 presents the mean number of each type of item cor-
rect, by condition. A mixed-design ANOVA, with affect as a be-
tween-subjects variable and item difficulty level as a within-sub-
jects variable, was performed on the data. This analysis revealed
a main effect of difficulty level, F(2. 88) = 174.48. p < .0001.
Planned comparisons showed that, as might be expected, sub-
jects performed significantly better on the moderately difficult
items than on the very difficult items, r(88) = 13.52. p < .01, and
significantly better on the easy ilems than on the moderately
difficult items, ((88) = 5.05,/)<.01.
We predicted that positive affect should facilitate perfor-
mance on items of moderate difficulty. It was less clear, however,
if positive affect would influence performance on easy or diffi-
cult items, because of restricted ranges of performance in these
conditions (e.g., something like a ceiling effect with easy items
and a door effect with difficult items). An examination of the
means indicates that a floor effect did occur on difficult items:
Very few subjects could answer any of these items correctly.
Also, scores on the easy items in every condition were fairly
high, indicating that most subjects, regardless of affective state.
3
We wish lo thank Martha T. Mednick for making thL- Remote Asso-
ciates Test items available to us.
4
The correct answer to this item is power.
 1128 A. ISEN, K. DAUBMAN, AND G. NOWICKI
found these items quite easy. Therefore, it is not surprising that
the main effect for affect was not significant, F(\, 44) = 1.32,
p < .26. Although the interaction between affect and difficulty
level only approached but did not reach the customary signifi-
cance level, F(2, 88) = 2.09, p = .13, a planned comparison
indicated that positive affect resulted in improved performance
on the moderately difficult items, as predicted, /(110) = 2.13,
p < .025.5
Thus, Study 3 indicated that positive affect induced by means
of a small gift could facilitate creativity as indicated by perfor-
mance on the Remote Associates Test. This extends the previ-
ous findings by showing that affect-induction procedures other
than humor can give rise to creative responding, and it demon-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
strates the impact of affect on creativity as measured by a task
This document is copyrighted by the American Psychological Association or one of its allied publishers.
other than Duncker's candle task.
Experiment 4
Experiment 4 was conducted to replicate conceptually the
results of Experiment 3. In this experiment, the comedy film
described earlier was used to induce affect and, as in Experi-
ment 2, a control for the effect of arousal was added.
Method
Subject Ninety-eight male and female college students participated
in this experiment in exchange for extra credit toward their introduc-
tory psychology grade. A maximum of 5 subjects participated in any
given session.
Procedure Subjects who had been assigned randomly to one of three
conditions—comedy film, exercise, or no manipulation—were told that
they would be helping to develop procedures and materials for experi-
ments to be run at a later time.
Subjects who had been assigned to the exercise condition were asked
to perform the step exercise, described in conjunction with Study 2,
for 2 min. All subjects complied, and pre-exercise and postexercise
heart rate measures were taken. The average subject's heart rate in-
creased 66%.
Subjects in the comedy-film condition were shown the same segment
of Gag Reel described in Experiments 1 and 2. As before, they were
instructed not to try to memorize anything from the film but just to
watch it and get a general impression of it.
Next, all subjects were told that norms for an association task needed
to be established. The seven Remote Associates Test items (each consist-
ing of a set of three words) that had been determined to be of moderate
difficulty (Study 3) were then read to subjects, who were asked to write
them down as read. They were then given 15 min to complete the task.
Results
The data from 6 subjects in the comedy-film condition, 5 in
the no-manipulation condition, and 1 in the exercise condition
were discarded because of failure of subjects to follow instruc-
tions or to hear or transcribe the items correctly.
Table 7 presents the number of items correct in each of the
three conditions. A one-way ANOVA with three levels of the
affect variable was performed. There was a significant main
effect for affect, F(2, 83) = 5.96, p < .01. Planned comparisons
revealed a difference between the comedy-film and no-manipu-
lation conditions, /(83) = 3.40, p < .01, whereas the difference
between the exercise and no-manipulation conditions was not
Table 7
Study 4: Mean Number of Remote Associates Test
Items Correct in Each Condition
Condition M MS, ,'J
Comedy film 5.00 3.69 33
No manipulation 3.04 6.35 26
27
Exercise 3.81 4.84
significant, /(83) = 1.28, p > . 1. As in Experiment 2. a planned
comparison revealed that the positive-affect and exercise condi-
tions differed significantly also, ((83) = 2.08, p < .025.
General Discussion
Results of these four studies taken together show that positive
affect, induced by a comedy film or a small gift of candy, can
facilitate creative responding on tasks usually thought to reflect
creativity. At the same time, a manipulation designed to induce
negative affect (negative film) and one promotive of arousal de-
void of any particular affective tone (exercise) had no effect on
these measures. Thus, it appears that elation, if it involves
arousal, is unlike some other aroused states in that it seems to
lead to the kinds of thinking that enable people to solve prob-
lems thai require ingenuity or innovation.
It has been proposed that a creative-problem-solving task is
one involving the ability to see relatedness in diverse stimuli
that normally seem unrelated. This is the essence of the defini-
tion of creativity provided by a number of theorists: S. A. Med-
nick (1962), for example, has proposed that creativity involves
the combination of elements that are remotely associated;
Koestler (1964) spoke of bisociation, the association and com-
bination of two different frames of reference; and the mathema-
tician Poincare (cited in Martindale, 1981) suggested that cre-
ativity involves useful new combinations of associative ele-
ments. At the same time, a similar process—one of seeing
relatedness in stimuli that are not usually seen as related to one
another—has been proposed and demonstrated to result from
the induction of a happy affective state (e.g., Isen, 1984, in press;
Isen & Daubman, 1984; Isen etal., 1985). Thus, it makes sense
that a person who is feeling good might be more creative than
others (or than she or he might be at another time).
These results indicate that creativity, an important skill that
is often thought of as a stable characteristic of persons, can be
facilitated by a transient pleasant affective state. Moreover, the
affective state sufficient to do this can be induced subtly, by
small everyday events. This suggests that creativity can be fos-
tered by appropriate modification of the physical or interper-
sonal environment.
Our findings have potential application to several domains.
' This contrast for a mixed design requires a combined error term
based on both the between-subjects error term and the within-subjects
error term. The degrees of freedom were computed by use of the for-
mula found in Winer (1971, p. 545). The critical value was computed
according to the formula found in Kirk (1982, p. 508) and described
there as a conservative test. A value of 2.01 is needed forp = .025.
 POSITIVE AFFECT PROMOTES CREATIVITY
Their implications for the educational enterprise are clear:
Teachers (and the students themselves) should regard everyone
as potentially creative, and an effort should be made to provide
the conditions that are conducive to creativity. One of those
conditions is a happy feeling state. Although we induced this
state by means of a small gift of candy and a comedy film in
these experiments, it is likely that other inductions of good feel-
ing might also be effective in facilitating creativity. We would
suggest, for the educational context, that an atmosphere of in-
terpersonal respect conducive to good self-esteem might be the
kind of condition that would promote creativity.
These results are also potentially relevant to other settings,
including other organizational settings such as businesses, in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
which leaders seek to promote creativity or innovative problem
This document is copyrighted by the American Psychological Association or one of its allied publishers.
solving. Once again, the most important way of inducing good
feelings may be by allowing workers to achieve a sense of compe-
tence, self-worth, and respect. Pleasant surprises may also be
effective, but if used too frequently, they will become expected
and may lose their ability to induce happy feelings.
Finally, our findings may bear on interpersonal problem solv-
ing more generally. A recent study has reported that persons in
whom positive affect was induced negotiated more effectively
and obtained higher joint benefits on an integrative bargaining
task that required finding an innovative solution (Carnevale &
Isen, 1986). The creation and maintenance of good interper-
sonal relationships often involves finding ways of resolving dis-
putes or negotiating arrangements of various kinds. When these
would benefit from a creative approach, positive feelings might
facilitate the interpersonal process. Thus, it may be possible to
extend our findings regarding the facilitative effect of good feel-
ings on creative problem solving to many organizational and
interpersonal domains.
Arousal
In these studies we have attempted to address a possible alter-
native interpretation of the effect observed in Study 1, namely
that positive affect constitutes an aroused state and therefore
the effects on creativity are attributable to arousal rather than
to positive affect. For this reason, we included groups engaging
in exercise (to represent arousal with no affective tone), and the
negative-affect induction can be seen as representing the kind
of arousal associated with upset, anxiety, or anger. Neither of
these types of conditions produced any significant increase in
creativity over that displayed by control groups; moreover, each
was found to be significantly below comparison positive-affect
conditions. Thus, it does not appear that arousal alone is suffi-
cient to result in creativity, even though such a hypothesis might
follow from spreading-activation theory (e.g., Anderson, 1983).
Moreover, it does not seem appropriate to attribute the ob-
served effect of positive feelings on creativity to increased
arousal.
Indeed, it is not clear whether positive affect does in fact cre-
ate an aroused state, although many people assume so. (See Isen
et al., 1985, for a fuller discussion of this issue.) Recently Man-
dler, long a proponent of the role of arousal in emotion, sug-
gested that arousal is not a central component in the type of
affect state (pervasive affect, or mood) involved in these studies
(Mandler, 1984, p. 277).
1129
Although it is not possible to say for certain that positive
affect of the kind used in these studies (mild elation) involves
no arousal, it does seem clear that feeling good has effects
different from those commonly associated with or produced by
arousal. For example, it is not clear that arousal should be ex-
pected to facilitate creativity. Arousal of at least some types has
been shown to increase a person's likelihood of giving dominant
or usual responses (e.g., Matlin & Zajonc, 1 %X) and. compati-
bly with that finding, to improve performance on simple tasks
but to impair performance on complex tasks or tasks requiring
innovation (e.g., Martindale, 1981; Yerkes&Dodson, 1908). In
contrast with effects that have been observed for arousal, posi-
tive affect has been shown to facilitate u n u s u a l responding
rather than typical responding in several kinds of tasks (e.g..
Isen & Daubman, 1984; Isen etal., 1985). This is in addition to
the results obtained in the present studies that demonstrate a
facilitative effect of positive feelings, but none of arousal, on
creative problem solving.
Thus, in sum, there seems to be reason that we should not
think of elation as involving arousal. Yet the mere use of the
term elation, rather than positive affect, in the preceding sen-
tence makes clear how difficult it will be to t h i n k of positive
affect as devoid of arousal. Thus, let us conclude by saying that
even if positive affect does have an arousal component or does
result in increased arousal, it also has other effects; it has been
demonstrated to have influences on cognition and behavior that
mere arousal does not.
Negative Affect and Creativity
Study 2 included one group in which negative affect was in-
duced; this group failed to differ from the relevant control group
and scored significantly lower than those in the comparable pos-
itive-affect condition. We interpreted this result as suggesting
that negative affect neither facilitates nor impairs creativity.
However, a few words of caution are warranted here.
First, negative affect was induced in only one study, by only
one means. The film that we showed for the purpose (Ntghl and
Fog) undoubtedly does induce unpleasant feelings in most per-
sons, and our manipulation check confirmed this expectation.
However, there may be something specific to this particular
means of affect induction that produced the observed outcome.
In particular, this film is about a very disturbing chapter in the
history of the world, and it depicts suffering with great impact.
Therefore, its intensity level may be incomparable to those of
the other affect inductions used. Moreover, conversations with
persons who have seen the film suggest that it may induce more
than one negative affect at once (disgust, sadness, anger, shame,
regret, fear, and despair have been reported by students viewing
the film in classes). This complexity could complicate interpre-
tation of the effects of the film. Finally, as noted, the film is
about something relatively specific, and this specificity of con-
tent may influence cognitive processes in ways independent of
the affect induced. Thus, before conclusions can be drawn re-
garding the impact of negative affect on creativity, studies in-
ducing negative affect in more than one way (and perhaps
differentiating among negative affects such as sadness, fear, and
anger) should be conducted. It is still possible that negative
 1130 A. ISEN, K. DAUBMAN, AND G. NOWICKI
affect may be found to influence creative problem solving in
some way.
More specifically regarding negative affect, it should be noted
that our negative-affect induction did not result in impaired
creativity. Sometimes it is inviting to think of positive and nega-
tive affect as, if not the same thing (arousal, as discussed earlier;
Duffy, 1934, 1941), then opposite poles of a single dimension
and therefore likely to produce opposite effects. Thus, some
might interpret our finding regarding the impact of positive
affect on creativity as suggestive that negative affect should im-
pair creativity. This may yet be true, when appropriate levels
and types of negative affect are investigated; however, on the
basis of the information now available, such a conclusion or
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
prediction would seem unwarranted. Rather, positive and nega-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
tive affect may be seen as distinct states, influencing behavior
in accord with the thoughts that each brings to mind and the
processes that each fosters.
Moreover, this view would suggest that individuals may differ
in the impact that affect states may have on them, depending on
the cognitive patterns available to them and the coping strate-
gies that they typically use. One may expect commonalities
among people, and indeed results of our studies do indicate pre-
dictable effects of positive-affect inductions in general; however,
individuals may differ, possibly in identifiable ways having to do
with cognitive structure. Thus, although negative affect may not
typically facilitate creativity for most people, the case of Edgar
Allan Poe or Vincent van Gogh can be understood as excep-
tions.
Mention of these two creative geniuses not only points out
the possibility of individual differences but also underscores the
potential distinction with regard to the creative process between
normal persons and those clinically depressed or otherwise
emotionally disturbed. It may be that for clinical depressives,
compared with normal persons, more cognitive material is ac-
cessed by sadness; and it may also be that for such persons sad-
ness cues more material than other affective states do. If so, the
process proposed to occur with happiness for normal persons
(i.e., increased access to a large range and amount of material)
may occur in response to sadness for these persons. The asym-
metry between the cuing effects of happiness and sadness in
normal persons has been discussed recently (e.g., Isen, 1984,
1985), and this kind of cognitive difference between depressives
and normals has been hypothesized. However, at present it re-
mains to be investigated.
Cognitive Capacity
The results of the four studies presented in this article suggest
that it would be unwise to think of positive affect as reducing
cognitive capacity or as, for some other reason, leading to lazy
and inefficient problem solving. Earlier research reported
effects—such as increased use of heuristics, more rapid deci-
sion making, and broadened categorization among positive-
affect subjects (e.g., Isen & Daubman, 1984; Isen & Means,
1983; Isen, Means, Patrick, & Nowicki, 1982)—that might have
been seen as evidence of lowered cognitive capacity. Most of
those effects provided stronger evidence for changes in process
than for changes in available cognitive capacity as a result of
positive affect; however, the alternative interpretation remained
possible. Our findings suggest that that alternative is not appro-
priate, because they show improved performance on the part of
persons in positive-affect states and, in fact, improved perfor-
mance on tasks usually considered very difficult and demanding
of cognitive capacity. Thus, it seems that positive affect should
be viewed as influencing the way in which material is processed,
rather than just the amount of capacity present.
In conclusion, the interpretation that we have suggested for
the impact of positive affect on creative problem solving is thai
good feelings increase the tendency to combine material in new
ways and to see relatedness between divergent stimuli. We hy-
pothesize that this occurs because the large amount of cognitive
material cued by the positive affective state results in defocused
attention, and the more complex cognitive context thus experi-
enced by persons who are feeling happy allows them a greater
number and range of interpretations. This increased range of
interpretations results in awareness of more aspects of stimuli
and more possible ways of relating and combining them.
One question that remains regarding this process is whether
persons in positive states notice more features of stimuli be-
cause of the increased number of interpretations of them (thus
rendering more of their features useful, functional, and there-
fore salient). Another question arising from the formulation
presented here is whether these processes also promote sche-
matic functioning and whether, therefore, persons who are feel-
ing happy may also be likely to structure material i n schematic
or functional ways. These two kinds of processes, although
different, would not seem to be mutually exclusive. Moreover,
together they would account for the very distinct kinds of cogni-
tive processing shown by persons in whom positive affect has
been induced. Thus, these seem particularly good questions for
further study.
References
Adamson, R. E. (1952). Functional fixedness as related to problem solv-
ing. Journal of Experimental Psychology, 44, 288-291.
Anderson, J. R. (1983). The architecture of cognition Cambridge: Har-
vard University Press.
Boucher, J., & Osgood, C. E. (1969). The Pollyanna hypothesis. Journal
of Verbal Learning and Verbal Behavior, 8, 1-8.
Bransford, J. D. (1979). Human cognition Learning, understanding.
and remembering. Belmont, CA: Wadsworth.
Bruner, J. S., Matter, J., & Papanek, M. L. (1955). Breadth of learning
as a function of drive level and mechanization. Psychological Review.
62, 1-10.
Carnevale, P. J. D., & Isen, A. M. (1986). The influence of positive affect
and visual access on the discovery of integrative solutions in bilateral
negotiation. Organizational Behavior and Human Decision Pro-
cesses, 37, 1-13.
Cramer, P. (1968). Word association. New York: Academic Press.
Duify, E. (1934). Emotion: An example of the need for reorientation in
psychology. Psychological Review, 41, 184-198.
Duffy, E. (1941). An explanation of "emotional" phenomena without
the use of the concept of "emotion." Journal of General Psychology
25,282-293.
Duncker, K. (1945). On problem solving. Psychological Monographs.
58(5, Whole No. 270).
Easterbrook, J. A. (1959). The effect of emotion on cue utilization and
the organization of behavior. Psychological Review. 66, 183-20!
Freedman, J. L. (1965). Increasing creativity by free-association train-
ing. Journal of Experimental Psychology. 69. 89-91.
 POSITIVE AFFECT PROMOTES CREATIVITY
Frost, R. O., & Greene, M. L, (1982). Duration and post-experimental
removal of Velten Mood Induction Procedure effects. Personality and
Social Psychology Bulletin, 8, 341-347.
Glucksberg, S., & Danks, J. H. (1967). Functional fixedness: Stimulus
equivalence mediated by semantic-acoustic similarity. Journal of Ex-
perimental Psychology, 74, 400-405.
Glucksberg, S., & Danks, J. H. (1968). Effects of discriminative labels
and of nonsense labels upon availability of novel function. Journal of
Verbal Learning and Verbal Behavior, 7, 72-76.
Glucksberg, S., & Weisberg, R. W. (1966). Verbal behavior and problem
solving: Some effects of labeling in a functional fixedness problem.
Journal oj Experimental Psychology, 71, 659-664.
Higgins, E. T., & Chaires, W. M. (1980). Accessibility of interrelational
constructs: Implications for stimulus encoding and creativity. Journal
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
of Experimental Social Psychology, 16, 348-361.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Isen, A. M. (1984). Toward understanding the role of affect in cognition.
In R. Wyer & T. Srull (Eds.), Handbook of social cognition (pp. 174-
236). Hillsdale, NJ: Erlbaum.
Isen, A. M. (1985). The asymmetry of happiness and sadness in effects
on memory in normal college students. Journal of Experimental Psy-
chology: General, 114, 388-391.
Isen, A. M. (in press). Positive affect, cognitive processes, and social
behavior. In L. Berkowitz (Ed.), Advances in experimental social psy-
chology New York: Academic Press.
Isen, A. M., & Daubman, K. A. (1984). The influence of affect on cate-
gorization. Journal of Personality and Social Psychology, 47, 1206-
1217.
Isen, A. M.. & Gorgoglione, J. M. (1983). Some specific effects of four
affect-induction procedures. Personality and Social Psychology Bul-
letin, 9, 136-143.
Isen, A. M.. Johnson, M. M. S., Mertz, E., & Robinson, G. F. (1985).
The influence of positive affect on the unusualness of word associa-
tions. Journal oj Personality and Social Psychology, 48, 1-14.
Isen, A. M., & Means, B. (1983). The influence of positive affect on
decision-making strategy. Social Cognition. 2, 18-31.
Isen, A. M., Means, B., Patrick, R., & Nowicki, G. P. (1982). Some
factors influencing decision-making strategy and risk taking. In M. S.
Clark & S. T. Fiske (Eds.), Affect and cognition: The 17th Annual
Carnegie Symposium on Cognition (pp. 243-261). Hillsdale, NJ: Erl-
baum.
Isen, A. M.. & Shalker, T. E. (1982). Do you "accentuate the positive,
1131
eliminate the negative" when you are in a good mood? Social Psy-
chology Quarterly, 45, 58-63.
Isen, A. M., Shalker, T. E., Clark, M., & Karp, L. ( 1 9 7 8 ) . Affect, accessi-
bility of material in memory, and behavior: A cognitive loop? Journal
of Personality and Social Psychology, 36, 1-12.
Jenkins, J. J. (1974). Remember that old theory of memory? Well, for-
get it! American Psychologist, 29, 785-795.
Kirk, R. E. (1982). Experimental design (2nd ed.). Uelmont, CA:
Brooks/Cole.
Koestler, A. (1964). The act of creation. New York: Macmillan.
Maltzman, 1., Simon, S., Raskin. D., & Licht, L. (1960). Experimental
studies in the training of originality. Psychological Monographs. 74,
1-23.
Mandler, G. (1984). Mind and body: The psychology of emotion and
stress. New York: Norton.
Martindale, C. (1981). Cognition and consciousness. Homewood, II,:
Dorsey Press.
Matlin, M. W., & Zajonc, R. B. (1968). Social facilitation of word asso-
ciations. Journal of Personality and Social Psychology 10. 455-460.
Mednick, M. T, Mednick, S. A., & Mednick, E. V. (1964) Incubation
of creative performance and specific associative priming. Journal of
Abnormal and Social Psychology, 69, 84-88.
Mednick, S. A. (1962). The associative basis of the creative process.
Psychological Review, 69, 220-232.
Palermo, D. S., & Jenkins, J. J. (1964). Word association norms. Minne-
apolis: University of Minnesota Press.
Rosch, E. (1975). Cognitive representations of semantic categories.
Journal of Experimental Psychology: General. 104, 192-233.
Siegel, S. (1956). Nonparametric statistics for the behavioral sciences
New York: McGraw-Hill.
Teasdale, J. D., & Fogarty, S. J. (1979). Differential effects of induced
mood on the recall of pleasant and unpleasant events from episodic
memory. Journal of Abnormal Psychology, 88, 248-257.
Winer, B. J. (1971). Statistical principles in experimental design (2nd
ed.). New York: McGraw-Hill.
Yerkes, R. M., & Dodson, J. D. (1908). The relationship of strength of
stimulus to rapidity of habit formation. Journal of Comparative and
Neurological Psychology, IS, 459-482.
Zillmann, D. (1979). Hostility and aggression Hillsdale, NJ: Erlbaum.
Received March 24, 1986
Revision received September 2, 1986 •