Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Vol. 0, No. 0, xxxx–xxxx 2021, pp. 1–18 DOI 10.1111/poms.

13523
ISSN 1059-1478|EISSN 1937-5956|21|00|0001 © 2021 Production and Operations Management Society

Emergence of Informal Safety Leadership: A Social–


Cognitive Process for Accident Prevention
Chunlin Wu
School of Economics and Management, Beihang University, Haidian District, No.37 Xueyuan Road, Beijing, 100191, China
Beijing Key Laboratory of Emergency Support Simulation Technologies for City Operations, Beihang University, Beijing, 100191, China,
wuchunlin@buaa.edu.cn

Han Yao
School of Economics and Management, Beihang University, Haidian District, No.37 Xueyuan Road, Beijing, 100191, China,
yaohan0817@buaa.edu.cn

Xin Ning*
School of Investment and Construction Management, Dongbei University of Finance and Economics, No.217 Jianshan Street, Shahekou
District, Dalian, 116025, China, ningxin@dufe.edu.cn

Lei Wang
School of Narcotics Control and Public Order Studies, Criminal Investigation Police, University of China, Huanggu District, No. 83
Tawan Street, Shenyang, 110854, China, leowang521@126.com

s an important domain of disaster management, accident prevention is an urgent and crucial subject because of the
A high frequency and severe consequences of industrial accidents in recent times. Current researchers focus on
organization-level safety antecedents, among which leadership receives the highest attention. However, formal leadership
of senior managers is limited by the level-by-level effect attenuation, such that frontline accidents cannot be controlled
substantially. Thus, this study proposes the construct of informal safety leadership (ISL), which emerges among frontline
workers through socially cognitive interactions. We develop a moderated mediating conceptual model to explore the ISL
emergence mechanism from the social-cognitive perspective. Across two studies, that is, a three-wave field study and an
agent-based modeling simulation experiment, we find that three social-cognitive elements are positively associated with
ISL emergence, which in turn predicts safety organizational citizenship behavior and perceived followership. Moreover,
this relationship is conditional on formal leadership support. We contribute to disaster management literature by articulat-
ing how informal leaders emerge from the operational frontline to possess safety leadership traits. We also give insight
into leadership emergence by clearly indicating that who becomes a leader is an inherently social process, dependent on
peers who bestow influence to a person engaging in leader-like behaviors. We thus provide a unique perspective on prac-
tical efforts to leverage benefits of the closest and most immediate leadership in frontline accident prevention.
Key words: accident prevention; frontline workers; informal leadership; operational safety; social cognitive theory
History: Received: September 2020; Accepted: June 2021 by Sushil K. Gupta, after 3 revisions.

Corresponding author.

management research, since the International Nuclear


1. Introduction Safety Advisory Group (1986) discovered that the
Accident prevention is a very important domain of Chernobyl nuclear accident stemmed from manage-
disaster management research, and needs more atten- ment and organizational errors (Hofmann et al.,
tion from the production and operations management 2017). Current studies keep revealing that although
(POM) academia (Gupta et al., 2016). Frequent occu- physical and behavioral factors are the direct antece-
pational accidents remain a severe problem in various dents of accidents, organizational factors are the root
industries, causing deaths, injuries, work-related ill- causes (De Vries et al., 2016, Wu et al., 2016), among
nesses, days away from work, and other direct and which leadership has garnered the most attention
indirect heavy losses (De Vries et al., 2016, Ramas- (Hardcopf et al., 2021, Serban et al., 2015). Effective
wamy and Mosher, 2018, Yang et al., 2017). In recent leadership is widely proven to be essential for organi-
years, accident prevention from the organizational zational success and operational performance as
behavior perspective is prevalent in disaster researchers and practitioners seek proactive

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
2 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

approaches to safety (De Vries et al., 2016). However, frontline workplace remains unexamined. The direct
the mismatch between management commitment and antecedents and immediate outcomes of ISL are
subordinate actions threatens safety conditions (Mar- unclear. Moreover, the studies ignore the essence of
tin and Lewis 2014). Specifically, senior managers are informal leadership emergence as interactively
usually the only ones concerned with safety; thus, dynamic, based on social cognition (Acton et al.,
managerial measures cannot be fully implemented on 2019); they only treat informal leadership from a
operational frontlines (Wu et al., 2016). behavioral, instead of a process, perspective. Thus,
Remarkably, as the POM academia probe deeply informal leadership is riddled with unknown elemen-
into social and psychological dynamics in complex tal properties, thereby stalling effective initiatives to
production settings (Bendoly et al., 2015), more prevent accidents. This study aims to bridge the
researchers have begun to conceptualize leadership research gap by exploring the emergence mechanism
as a dynamic social process and emergent phe- of ISL from a social–cognitive process perspective and
nomenon among group members (Hoch and Dule- clarifying what roles major variables play in the emer-
bohn 2017, Kwok et al., 2018). In other words, formal gence process.
leaders with a high level of authority are not the only We apply social cognitive theory to depict the
ones that have leadership skills; ordinary group mem- mechanisms involved in the ISL emergence process.
bers can have leadership traits too (Serban et al., Many previous studies show that informal leadership
2015). Informal leadership emerges from the process emergence is a typical social cognition process (Lynch
by which ordinary individuals gain influence via peer 2007, Uhl-Bien and Marion 2009), which actually con-
perceptions rather than formal authority or manage- tains two phases (Acton et al., 2019, Wellman 2017).
rial ranks (Cogliser et al., 2012, Taggar et al., 1999). The first phase is initial emergence, in which frontline
More than a property, it is a dynamic interaction pro- workers generate basic safety leadership attributes,
cess, driven by the deep cognitive and perceptual pro- driven by key social-cognitive factors such as attitude,
cesses of group members that form a collective self-efficacy, and competence (Chrobot-Mason et al.,
patterning of leader and follower interactions over 2016). The second phase is consolidation, in which the
time (Acton et al., 2019). In curbing operational chal- informal leaders make ISL more explicit and visible
lenges (e.g., conflicts, accidents, and calamities), infor- by exhibiting SOCB, and induce followership among
mal leadership drives frontline workers to follow the their coworkers so as to consolidate ISL emergence by
advice of one individual over another, regardless of continuous leader-follower cognition and interactions
rank, to stimulate proactive solutions rapidly and (Wellman 2017). Only through the continuous social
straightforwardly (Carnabuci et al., 2018, Pendleton cognition and interaction can ISL emerge and be con-
2011). From this perspective, leadership is emergent, solidated. Lord et al., (2016) argued that future leader-
contextual, and shared and located in the relational ship research should move beyond the static view of
processes that facilitate the collective achievement of “leadership as individuals” to a richer understanding
cooperation (Zhang et al., 2015), innovative change of leadership as a “socially constructed process.”
(Morrison and Phelps 1999), and improvement of Therefore, it is indispensable for this study to depict
other organizational behaviors, especially those lead- the complex informal leadership emergence mecha-
ing to extra-role performance or citizenship behavior nism based on social cognitive theory.
(Moideenkutty and Schmidt 2011). We first developed an initial conceptual model to
Based on the definitions of safety leadership (Bar- clarify the emergence mechanism and structure of ISL
ling et al., 2002) and informal leadership (Neubert from a social cognition and interaction perspective.
and Taggar 2004, Pan et al., 2018), we propose the We then tested the model via a three-wave multi-
construct of informal safety leadership (ISL) and sample field study (Study 1) and agent-based model-
define it as the emergent leadership of ordinary orga- ing (ABM) simulation experiment (Study 2). This
nizational members in influencing safety-related atti- study contributes to the literature in three ways. First,
tudes, behaviors, and mindsets to achieve although leadership has long been identified as a
organizational safety goals via social interactions with major leading factor of performance in disaster man-
colleagues. It begins to emerge when an organiza- agement research (Stauffer and Kumar 2021), infor-
tional member leads others with a prominent safety mal leadership among workers tends to be ignored.
attitude, safety competence, and self-efficacy and Industrial safety management cannot be strengthened
takes effect when the member with ISL (i.e., the infor- effectively because of the long fulfilling distance and
mal leader) exhibits safety organizational citizenship gradual attenuation of formal leadership. We demon-
behavior (SOCB), and others recognize and follow the strate clearly how frontline workers can develop
informal leader’s behavior (Pendleton 2011). informal leadership based on the specific need of
However, despite the importance of informal lead- workplace safety improvement, and thus provide
ership, how ISL emerges among workers in the another unique rationale by which leadership

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 3

influences operational safety more significantly. Sec- safety attitude, self-efficacy, and safety competence,
ond, although scholars often take the leadership but also on the support of formal leaders, such as
structure of an organization for granted, we consider supervisors, department managers, and senior man-
it a bottom-up emergent pattern and explain why agers (Hannah et al., 2008). Without formal leadership
informal leadership can emerge in some frontline support (FLS), average workers would have no cour-
workplaces for safety purposes, yet cannot emerge in age to be informal leaders (Zhang et al., 2020), and
other workplaces. We propose a moderated media- also no rights and resources to initiate SOCB, let alone
tion mechanism to explore the causal variable rela- recognition and followership of peers. Thus, we
tionships within the informal leadership emergence. added the FLS construct to the hypothesized model
Third, we apply the social-cognitive perspective to (Figure 1), which is essential to moderate the relation-
illustrate the overall framework of ISL emergence ship between ISL and the above three antecedents
involving both cognitive and behavioral factors, in (Newman et al., 2018).
contrast to prior leadership literature which ignores
the social process or only uses a static perspective. We 2.2. Hypotheses Development
can then establish a more holistic leadership para- 2.2.1. Safety Attitude and ISL. Attitude is a
digm on social and psychological dynamics to facili- widely concerned construct when it comes to social
tate a clearer understanding of informal leadership cognitive theory (Burke et al., 2003, Holtbr€ ugge et al.,
development. 2015), and was proven as being vital to facilitating
leadership emergence (Taggar et al., 1999). Safety atti-
2. Theoretical Foundation and tude reflects employee cognition, awareness, beliefs,
and feelings about safety policies and measures (Hen-
Hypothesis Development
ning et al., 2009). Previous studies have demonstrated
2.1. Theoretical Foundation: The Social-Cognitive that safety attitude is a useful guide to identify intra-
Theory industrial safety (Dinh et al., 2020). The active safety
The social cognitive theory first emerged from the attitude of organizational members means that they
learning theory (Bandura 1986) and became an impor- have a high awareness of the consequences of unsafe
tant research perspective in the fields of psychology, behavior, can prevent risks, and can try to avoid the
management, and leadership (Hmieleski and Baron occurrence of safety accidents (Lund and Rundmo
2009). It posits that human behavior depends on the 2009). An individual’s perceptual orientation of safety
individual’s cognition of the social context, and the behavior is better if they can maintain a higher level
mutual cognition among the members is an important of safety cognition and awareness. As Lynch (2007)
factor in the emergence of individual and organiza- notes, a positive attitude is vital in cultivating ISL and
tional behaviors (Bandura 2001). Leadership develop- the subsequent SOCB. Strong awareness and belief
ment depends on the constant interaction, mutual are also inherent attributes of leaders, and thus
cognition, and influence of leaders and followers improve people’s prototypes for leaders (Acton et al.,
(Salem et al., 2019). Social cognition is, thus, effective 2019). In other words, peers are more likely to per-
for analyzing the leadership emergence mechanism in ceive a member as an informal leader if the prospec-
social teams, which provides an important theoretical tive leader’s positive attitude is favorable enough to
model for the continuous and reciprocal interaction coincide with the followers’ expectations for ISL. As
between leadership cognition, leadership environ- such, it is posited that:
ment, and leadership behavior (Jourden 1993).
The social cognitive theory comprises various lead- Hypothesis 1. Safety attitude is positively associated
ership related constructs, and it is necessary to inte- with ISL.
grate different constructs to develop a complete
theoretical model illustrating the complex emergence 2.2.2. Safety Competence and ISL. Competence
process of ISL (Bandura 2001, Lord et al., 2016). By is a person’s basic characteristic that is causally
outlining the whole ISL emergence process as men- related to good or excellent performance in a specific
tioned in Section 1, we initially determined five key job position (Mitchell et al., 2019). It is a combination
variables which contribute to the initial emergence of skills and knowledge, including the patterns of per-
and further consolidation of ISL, and developed our sonal competency and the way they work together for
conceptual model as illustrated in Figure 1. Three achievement. As mentioned above, leadership emer-
variables, that is, safety attitude, safety competence, gence is influenced by the match between the
and self-efficacy, are antecedents of ISL, and the prospective leader’s attributes and the prospective
other two variables, that is, SOCB and perceived fol- follower’s prototype or cognition toward the leader.
lowership, are ISL outcomes. In addition, those who Previous research on social cognitive theory identified
generate ISL and initiate SOCB depend not only on universal factors that impact such cognition,

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
4 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

Figure 1 Conceptual Model of ISL Emergence among Frontline Workers

including competence, warmth, trust, and self- self-set goals, task-related efforts, and individual task
identity (or self-schema), which have garnered the performance (Chen and Chen 2014). Accident preven-
most academic attention (Acton et al., 2019, DeRue tion requires continuous and focused endeavors from
et al., 2015). In particular, as a major decisive element all personnel; they must exhibit high spirits and
in social cognition and interaction, competence signif- favorable behaviors persistently to succeed in adverse
icantly predicts great leadership emergence and situations, which is more challenging and laborious
peers’ followership (Zhang et al., 2020). To be more than operational routines. As such, it is posited that:
specific, members with high competence in occupa-
tional safety areas show high levels of ISL emergence Hypothesis 3. Self-efficacy is positively associated with
since they have the personal charisma to attract fol- ISL.
lowers. This leads us to expect:
2.2.4. SOCB and Perceived Followership:
Hypothesis 2. Safety competence is positively associ- Outcomes of ISL. As mentioned above, the process
ated with ISL. in which embedded implicit traits such as attitude,
competence, and self-efficacy induce leadership is
2.2.3. Self-efficacy and ISL. Self-efficacy is a cru- only the first phase of ISL emergence. ISL should
cial construct in Bandura’s (1986) framework of social become more explicit by informal leaders’ SOCB exhi-
cognitive theory. Bandura (2001) argued that it occu- bition and coworkers’ followership.
pied a pivotal role in the causal structure of social SOCB, a subset of organizational citizenship behav-
cognitive theory because of its salient influence on ior (OCB) (Chmiel et al., 2017), is voluntary assistance
adaptation and change of behavioral norms. Self- to other organizational members to improve safety
efficacy can be generally defined as a comprehensive (Hofmann et al., 2003). It includes activities such as
judgment about one’s capability to mobilize motiva- suggesting improvements for safety practices, report-
tion, cognitive resources, and courses of action neces- ing those who violate safety, helping others with
sary to perform a certain task (Bandura 2001). It has safety issues, and communicating safety concerns to
long been accepted as a predictor of leadership emer- coworkers (Hofmann et al., 2003). SOCB also coin-
gence (Serban et al., 2015). Smith and Foti (1998) cides with safety attitude since they are both favor-
found that higher self-efficacy is strongly correlated able safety antecedents that receive the most
to informal leadership. Paglis and Green (2002) pro- attention. The former is external and explicit, while
posed the construct of leadership self-efficacy, and the latter is internal and implicit. The transition from
showed that managers with high self-efficacy implicit attitude to explicit behavior needs courage
engaged in more leadership attempts. Across safety- and risk-taking spirits, for which leadership can pro-
related areas, self-efficacy is also a strong predictor of vide solutions (Zhang et al., 2020). Research indicates

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 5

that ISL positively predicts SOCB, such that informal emergence (DeRue et al., 2015). Under the influence
leaders with high-level safety attitude, safety compe- of high-level FLS, workers with favorable attitude,
tence, and self-efficacy have strong intention and competence, and self-efficacy will have more courage
courage to engage in SOCB to avoid accidents (Ozyil- and motivation to become informal leaders, such that
maz and Cicek 2015). In other words, ISL plays a cru- the association between ISL and its three antecedents
cial mediating role between the three ISL antecedents will be strong. In contrast, when no or only low-level
and SOCB, because it transforms implicit traits into FLS exists, workers tend to lack the courage and moti-
explicit actions. Thus, it is posited that: vation to lead within a managerial hierarchy (Luria
and Berson 2013), such that the association between
Hypothesis 4. ISL positively mediates the relationships ISL and its antecedents will be weak. Thus, it is con-
between (a) safety attitude, (b) safety competence and (c) cluded that the strength of causal relationships from
self-efficacy, and SOCB. ISL antecedents to ISL will be influenced significantly
by FLS. The external support of formal leaders can
Followership is the intentional practice by the fol- accelerate and accentuate the process of ISL emer-
lower to enhance the synergetic interchange between gence, including the exhibition of SOCB (Luria et al.,
the follower and the leader (Uhl-Bien et al., 2014). It 2019). The direct relationships embodied by Hypothe-
involves the emergence of a leadership relationship ses 1, 2, and 3 as well as the indirect relationship
that occurs when individuals in a group begin to view embodied by Hypothesis 4 differ in strength by the
themselves as members of a larger group led by a for- levels of FLS. This leads us to the following hypothe-
mal or an informal leader (Shondrick and Lord 2010). ses:
However, despite the existence of implicit leadership
theory (Gershenoff 2003), research pertaining to fol- Hypothesis 6. FLS moderates the direct relationships
lowers as a key consequence of the leadership process between ISL and (a) safety attitude, (b) safety compe-
through their enactment of followership is largely tence, and (c) self-efficacy, such that these direct relation-
missing in the leadership literature (Uhl-Bien et al., ships will be stronger at higher levels of FLS than at
2014). Thus, we incorporate followership into our lower levels of FLS.
conceptual model as a vital outcome of leadership
emergence. Compared to ordinary workers, informal
leaders have stronger awareness, firmer beliefs, and Hypothesis 7. FLS moderates the indirect relationships
higher enthusiasm in accident prevention, which typi- between SOCB via ISL and (a) safety attitude, (b) safety
cally reflect leaders’ charisma and ultimately shape competence, and (c) self-efficacy, such that these indirect
coworkers’ self-identity as followers. Informal safety relationships will be stronger at higher levels of FLS than
leaders’ favorable attitude, competence and self- at lower levels of FLS.
efficacy will lead them to pay special attention to
safety behavior of not only themselves but also others,
and thus lead to significant respect and followership
3. Study 1: Three-Wave Field Study
of peers (Conger et al., 2000). Hence, we hypothesize Since construction is noted as the most dangerous
that: industry and has the highest occupational injury
rates (Fang and Wu 2013), construction projects
Hypothesis 5. ISL positively mediates the relationships served as our field setting. The construction indus-
between (a) safety attitude, (b) safety competence and (c) try is also labor-intensive; multiple workers operate
self-efficacy, and perceived followership. on the production frontline and engage in social
cognition and interaction frequently. They face con-
2.2.5. The Moderating Role of FLS. The moder- stant threats of injuries and fatalities, and endeav-
ating effect of formal leadership in leadership emer- ors to cope with operational accidents with the
gence has been justified with sufficient evidence by help of their managers. Informal leadership is espe-
previous research (Luria et al., 2019, Newman et al., cially important for onsite safety control because
2018). For instance, Newman et al., (2018) used empir- the headquarters of construction contractors are
ical data to show that entrepreneurial leadership of usually far away. Thus, workers learn to rely on
middle-level managers exerts strong moderating themselves and their colleagues to guarantee safety,
effects on the link between employees’ creative self- which creates a favorable environment for ISL to
efficacy and their innovative behavior. Although emerge.
informal leaders are critical to accident prevention,
safety excellence cannot be achieved without the 3.1. Participants and Procedure
guidance and support of formal leaders, especially Our longitudinal questionnaire survey was con-
during the initial phase of informal leadership ducted on Chinese construction projects to test the

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
6 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

hypotheses. Invitation letters explaining the survey measurement, to measure safety attitude. Example
purpose and seeking permission to conduct site visits items are “I fully understand the potential dangers
were distributed to 23 projects in four large cities and keep an extra careful watch on them” and “I
across China: Beijing, Guangzhou, Xi’an, and rarely underestimate the risk and take shortcuts that
Chengdu. Fifteen projects agreed to participate. The involve little or no risk.”
self-reported surveys were undertaken at three time-
points (i.e., T1, T2, and T3) with a 21-day interval. The 3.2.2. Safety Competence (T1). We adapted five
interval was chosen based on previous similar studies items from Chang et al. (2012) to measure safety com-
(Olaniran 1996, Serban et al., 2015), and via sugges- petence. Example items are “I can recognize safety
tions from participant project managers, who con- and health hazards around my working environ-
firmed that it could counteract recall bias but can ments” and “I have learned to apply industrial safety
register significant fluctuations in the measures and health laws and regulations during my operating
(Tholen et al., 2013). Moreover, the internal and exter- practice.”
nal environments of construction projects change
rapidly; thus, longer time intervals may not guarantee 3.2.3. Self-Efficacy (T1). We measured self-
that all respondents will participate in the three efficacy using five items from the New General Self-
rounds of measurement. efficacy Scale developed by Chen et al. (2001). Exam-
We selected 780 new workers as our respondents so ple items include “I will be able to achieve most of the
as to track their complete course from an ordinary goals that I have set for myself” and “compared to
worker to an informal safety leader. Valid data were other people, I can do most tasks very well.”
obtained from completed questionnaires that were
not duplicates. In the first-wave (T1) survey, we 3.2.4. FLS (T2). We measured the moderating
received 645 valid responses (valid response rate of variable of FLS at T2 using three items by Amabile
82.7%). In T2 and T3, we only sent questionnaires to et al. (2004) to measure perceived leader support.
these 645 workers and controlled for valid feedbacks Amabile et al. (2004) validated that subordinates’
in the above two waves. Finally, we received three- perceived support from formal leaders improved
round valid questionnaires completed by 604 workers the overall creativity of their work, which is consis-
with a final valid response rate of 77.4%. Among tent with our hypotheses in the conceptual ISL
them, the mean age of workers was 43.49 years old model. The three items are “I constantly receive
(SD = 8.96), with an average work experience of encouragement and support from the project super-
13 years (SD = 8.44) in the construction industry. Fur- visor and other formal leaders,” “there are positive
thermore, 16.23% were women; 14.07% had senior interactions between us and the project formal lead-
high school education, and the rest had university ers,” and “the clarity of safety goals for the project
education. is high enough for us to understand and pursue.”
Respondents’ anonymity was protected rigorously. They were adapted moderately for the setting of
We assured them that there are no right or wrong this study.
answers and they should answer questions as hon-
estly as possible. We declared that this survey had 3.2.5. ISL (T2). To measure ISL, a construct incor-
nothing to do with their rewards and punishments so porating both safety leadership and leadership emer-
as to reduce their evaluation apprehension and make gence, we combined safety leadership items from Wu
them less likely to revise their responses to be more et al. (2016) and leadership emergence items from
socially desirable. Balthazard et al. (2009) to measure ISL. Eight items
were generated, all of which can fundamentally be
3.2. Measures acquired by informal leaders and, meanwhile, contain
Unless otherwise noted, survey items were assessed safety-related attributes. Example items are “I express
on a five-point scale ranging from 1 (not at all) to 5 confidence in and make coworkers confident about
(extremely), with a midpoint of 3 (moderately). The safety improvement” and “I can show a good model
basic question for each item is “to what extent does to get coworkers’ respect and make them obey safety
each item describe the work environment of rules and regulations.”
your project as you perceived it today?” Below are
the measures with their respective time points of 3.2.6. SOCB (T3). SOCB was measured at T3 via
collection. six items adapted from Lu et al. (2017) scale, designed
specifically for this construct. Three dimensions were
3.2.1. Safety Attitude (T1). Each individual covered: safety-related participation, helping, and
assessed their agreement levels using Wu et al. (2015) courtesy. Example items include “I make safety-
six-item scale, specifically designed for safety attitude related recommendations about work activities” and

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 7

“I help my coworkers learn about safe work prac- emergence cannot be probed into effectively without
tices.” addressing several obstacles via ABM. For example,
ABM can be programmed to include recursive pro-
3.2.7. Perceived Followership (T3). Perceived fol- cesses (i.e., feedback loops), which is crucial in ISL
lowership was measured at T3 via six items from Kel- emergence. ABM can detect and analyze the sequen-
ley’s (1992) Followership Style Questionnaire, a tial development of complex patterns within time-
prevalent measure adapted for many settings (Leung series data, such as cyclicity, not just linear and quad-
et al., 2018). As the original items were designed for ratic forms (Castillo and Trinh 2018, Fibich and Gibori
followers, we also made some necessary changes to 2010, Sun et al., 2021). The major recursiveness in our
the wording of items so that they can be expressed in study lies in the second stage of ISL emergence, that
the informal leaders’ tone. Examples include “under is, consolidation of ISL after a preliminary emergence
my influence, my coworkers can make decisions on of ISL. Since informal leaders exhibit more SOCB and
important matters to achieve the goal of the project” perceive more peer followership, by which their ISL
and “when I collaborate with close coworkers, they is improved cyclically until it has reached a stable and
work harder to understand and satisfy the demand desired level, we quantify this process by designing
and goal from our formal leader.” the agent behavior and interaction rules in ABM (Sec-
tion 4.2).
3.2.8. Control Variables (T1). We controlled for
age, gender, and group-member familiarity, following 4.2. Simulation Methods and Algorithms
a similar leadership emergence study by DeRue et al. 4.2.1. Simulation Scenario Design. We employed
(2015). However, in contrast, we did not control for Anylogic 8.0 software, popularly used in ABM
organizational experience because all respondents research (The Anylogic Company, 1992–2017), for the
were new members and had no prior experience of computer simulation. We selected a typical and com-
the current project. We asked each respondent to mon scenario in construction projects (i.e., informal
assess the extent to which they knew themselves leaders actively remind coworkers to wear personal
before the project for group-member familiarity (mea- safety equipment properly) to simulate the ISL emer-
sured at T1). Ratings were based on a five-point extent gence process for operational accident prevention.
scale as the above measures. Research indicated that 44% frontline workers world-
wide could not wear a safety helmet properly, thus
3.3. Analytical Approach
causing serious injuries and fatalities (Li et al., 2017).
We first checked the reliability and validity of the
Therefore, to ensure the validity and reliability of our
measurement model in Study 1 via various indexes
scenario, we interviewed three project managers and
such as Cronbach’s a reliability, content validity, con-
frontline workers for advice on the scenario, includ-
vergent validity, and discriminant validity. We then
ing rules of agent behavior and interaction (Sun et al.,
tested the hypotheses via path analysis and moder-
2019). Consequently, we confirmed that the scenario
ated mediation analysis, referring to Edwards and
could rigorously simulate ISL emergence for real-
Lambert’s (2007) analytical approach. The direct
world accident prevention. The experiment simulates
effects (H1, H2, and H3), mediating effects (H4, H5),
the process of workers’ continuous operation and
and moderating effects (H6, H7) were validated
tracks ISL emergence and evolution. It explores
sequentially via different analytical approaches (Sec-
changes in key variables including safety attitude,
tion 5). In particular, the three independent variables
safety competence, self-efficacy, SOCB, and perceived
were all grand-mean-centered to minimize the devia-
followership, and simulates 42 working days, which
tions from the multicollinearity (Aiken and West
is consistent with the total timespan of Study 1.
1991). All analytical approaches were conducted via
the Mplus 8.4 software (Muthen and Muthen, 1998–
4.2.2. Agent Definitions and Data Sources. We
2017).
constructed our simulation model within one front-
line operation team consisting of one supervisor (the
4. Study 2: ABM Simulation team leader) and 10 ordinary workers. We also
Experiment defined two other types of main agents: an onsite
safety officer and a senior project manager. The safety
4.1. Necessity of ABM Simulation officer spends the most time working on site, whereas
Many studies have noted the necessity of ABM in the senior project manager merely undertakes occa-
studying leadership emergence (Acton et al., 2019, sional onsite visits. The four agent-types are deemed
Castillo and Trinh 2018, Serban et al., 2015). They to have the greatest impact on workplace safety in the
emphasized that, as a complex cognitive and interac- frontline (Fang et al., 2015). The supervisor, onsite
tive mechanism with temporal effects, leadership safety officer, and senior project manager represented

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
8 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

the formal leaders. Based on the conceptual model improved by the same proportion, which has a value
and previous studies (Li et al., 2020, Sun et al., 2019), range of [1%, 5%].
we assigned the attributes and initial values to all Given the above scenario, attributes, initial values,
agent variables of both formal leaders and workers and rules of behavior and interactions, the simulation
(see Tables A1 and A2 in Appendix S1). The attributes program is run and tested 100 times. Subsequently, a
and initial values are the basis of the agent behavior dataset for 1000 (i.e., 100 9 10) simulated workers
rules. was generated, with values of all variables employed
Other than the variable values defining behavior in the model. Corresponding to Study 1, we also
rules, we made rules for the agents’ interactions. selected the simulation data at three time-points (i.e.,
There are four interactive variables. FLSSA, FLSSC, zero day, 21st day, and 42nd day), with a 21-day inter-
and FLSSE are the influences of FLS on the worker’s val. Data analysis was then performed using Mplus
safety attitude, safety competence, and self-efficacy, 8.4, following the same analytical approach as Study 1
respectively. Moreover, FLSON is the effect of the vis- (Section 3.3).
ibility of senior project manager on the site on work-
er’s safety attitude, safety competence, and self-
efficacy. Further, we assume that each worker has
5. Results
ISL attributes. When a worker’s ISL value exceeds 5.1. Reliability and Validity of the Measurement
0.5, the worker is considered to be an informal safety Model of Study 1
leader. Table 1 presents the correlations and descriptive
statistics of Study 1. Before the hypothesis tests, we
4.2.3. Agent Behavior and Interaction conduct reliability and validity analysis for all latent
Rules. Given the simulation scenario above, we sim- variables except for the single-item observed variable
ulated whether workers wore helmets and the subse- (i.e., group-member familiarity). Cronbach’s a relia-
quent social interactions in the operational frontline. bility coefficients for all scales appear along the diago-
A worker with a positive safety attitude, safety com- nal in Table 1. As the threshold of Cronbach’s a is 0.6,
petence, and self-efficacy who found that other all independent latent variables showed strong relia-
coworkers did not wear a helmet would develop bility. Moreover, no significant correlations exist
basic-level ISL, with the intention of reminding them among the three independent variables, thus dis-
and remedying safety violations. This informal leader pelling our anxiety about great deviations brought on
then initiates SOCB by voluntarily assisting cowork- by multicollinearity.
ers in wearing helmets and explaining the importance We then took further steps to verify the validity of
of safety compliance. Some coworkers might then the measurement model. First, as mentioned above,
develop a sense of shared identity and generate fol- we interviewed 10 workers from the sample for their
lowership. After perceiving followership, the infor- expert assessment to ensure the content validity
mal leader would develop a stronger organizational before the survey. Second, we calculated convergent
commitment and self-confidence in safety affairs, thus validity, that is, whether the scale items that purport
leading to a higher-level ISL emergence. We also sim- to measure a construct are internally correlated
ulated the moderating role of FLS, where informal (Salem et al., 2019). The average variance extracted
leaders would promote more SOCB and perceive (AVE) was selected to measure convergent validity.
higher followership if formal leaders support and The AVE values of safety attitude (0.58), safety com-
facilitate interactions between workers. The simula- petence (0.55), self-efficacy (0.51), FSL (0.56), ISL
tion algorithm with a feedback loop is shown in Fig- (0.57), SOCB (0.54), and perceived followership (0.52)
ure 2. all exceed the prescribed threshold of 0.5 (Henseler
Specifically, the emergence of ISL will go through et al., 2015), indicating a high convergent validity of
three major phases, that is, observation of safety haz- our measurement model. Third, we evaluated the dis-
ards on the workplace, initial emergence, and exhibi- criminant validity, which, together with convergent
tion of SOCB and perception of followership. In these validity, makes up the construct validity. Thus, we
links, the individual variables of the worker (e.g., tested the discriminant validity of the proposed 7-
safety attitude, safety competence, and self-efficacy) factor model (as shown in Figure 1) by contrasting it
take effects. The process by which workers improve with alternative confirmatory factor analysis (CFA)
ISL is essentially a cyclic consolidating process (as models. The fit indexes in Table 2 show that the con-
shown in Figure 2) until no safety violations can be ceptual model as the 7-factor model fits the empirical
observed, that is, ISL reaches a reasonable, high level. data significantly better than any of the alternative
In the positive feedback loop, we assume that all vari- models, confirming high discriminant validity.
ables (i.e., safety attitude, safety competence, self- Hence, all proposed variables in the model could be
efficacy, ISL, SOCB, and perceived followership) are applied to the next hypothesis tests.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 9

Figure 2 Simulation Rules of Behaviors and Interactions

Table 1 Means (M), Standard Deviations (SD), Correlations and Cronbach’s a Reliability (Study 1)

Variables M SD 1 2 3 4 5 6 7 8 9 10
1. Age 43.49
2. Gender 1.16 0.40 0.18**
3. Group-member 1.92 0.51 –0.05 –0.03
familiarity
4. Safety attitude 4.70 1.99 –0.07* 0.02 –0.07* (0.79)
5. Safety competence 4.07 0.50 –0.05 0.01 –0.03 0.06 (0.73)
6. Self-efficacy 4.76 3.96 –0.17** 0.03 –0.08* 0.03 0.05 (0.79)
7. FLS 3.82 0.51 –0.15** 0.07* –0.01 0.03 0.27* 0.02 (0.87)
8. ISL 3.45 4.97 –0.03 –0.02 –0.06* 0.42** 0.46** 0.42** 0.40** (0.89)
9. SOCB 4.14 2.46 –0.14** 0.03 –0.05 0.25** 0.29* 0.38** 0.48** 0.44** (0.77)
10. Perceived followership 3.77 0.43 –0.17** 0.06 –0.08* 0.34* 0.57** 0.45** 0.39** 0.33* 0.37** (0.77)

Notes: N = 604. Gender was coded 1 = female, 2 = male. *p < 0.05, two-tailed; **p < 0.01, two-tailed. a reliabilities appear in parentheses along the
diagonal.

We undertook several statistical remedies recom- and the first factor variance is 23.6%, indicating that
mended by Podsakoff et al. (2003) to evaluate the the single factor does not explain the vast majority of
common method variance (CMV) magnitude. We first the variance. Second, according to the CFA results in
used Harman’s single-factor test, in which all vari- Table 2, none of the fit indexes of the one-factor
ables were loaded into an unrotated exploratory fac- model reached the threshold values, meaning that the
tor analysis. Results showed that eight factors were single factor cannot account for all the variance in the
precipitated. The total cumulative variance is 57.13%, empirical data. Third, we also tested a model loading

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
10 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

Table 2 Results of Confirmatory Factor Analysis for Different Measurement Models (Study 1)

Models Variable combination approaches v2 df v2/df TLI CFI SRMR


Seven-factor model SA, SC, SE, FLS, ISL, SOCB, PF 1565.672 881 1.777 0.906 0.916 0.047
Six-factor model SA, SC, SE, FLS, ISL, SOCB+PF 1609.405 887 1.814 0.838 0.848 0.054
Five-factor model SA+SC+SE, FLS, ISL, SOCB, PF 1632.148 892 1.830 0.835 0.845 0.054
Four-factor model SA+SC+SE, FLS, ISL, SOCB+PF 1665.765 896 1.859 0.829 0.838 0.055
Three-factor model SA+SC+SE+FLS, ISL, SOCB+PF 1909.347 899 2.124 0.777 0.788 0.061
Two-factor model SA+SC+SE+FLS, ISL+SOCB+PF 1932.579 901 2.145 0.773 0.783 0.062
One-factor model SA+SC+SE+FLS+ISL+SOCB+PF 2012.313 902 2.231 0.755 0.767 0.062
Thresholds of fit indexes <2 > 0.9 > 0.9 > 0.5

Note: N = 604. SA = safety attitude, SC = safety competence, SE = self-efficacy, PF = perceived followership, TLI = Tucker-Lewis index, CFI =
comparative fit index, SRMR = standardized root mean square residual.

the items onto their latent variables as well as onto an in Figure A1 of Appendix S1. Sensitivity analysis indi-
additional method factor, which resulted in an eight- cates that all the three antecedents and the moderat-
factor model. We then compare this eight-factor ing variable (i.e., FLS) have significantly positive
model with the seven-factor model in Table 2. Results impacts on ISL, and ISL, in turn, has significantly pos-
indicated that adding an additional method factor did itive impacts on its two outcomes. Notably, the higher
not significantly improve model fit (D v2 = 42.745, the initial values of ISL antecedents, the higher the
Ddf = 44, p > 0.05). Overall, these results suggested ultimate ISL emergence level. In particular, self-
little threat of CMV and provided support for the efficacy has a noteworthy influencing mechanism,
validity of our measures. that is, the difference between the effects of lower
(value < 0.5) and higher (value > 0.5) initial values of
5.2. Sensitivity Analysis of Study 2 self-efficacy is much greater than the other variables.
Table 3 presents the descriptive statistics and correla- When entering the workplace, workers with weak
tion analysis results of the ABM simulation experi- self-efficacy have minimal ISL emergence probabili-
ment. As the major mediating variable, ISL is ties. In contrast, workers with strong self-efficacy can
significantly correlated with safety attitude, safety develop observably higher ISL because they may have
competence, self-efficacy, FLS, SOCB, and perceived higher expectations for themselves and their cowork-
followership. No significant correlations were found ers, which is a strong stimulus for their mutual infor-
among the three independent variables. Thus, there mal leadership and followership.
were no significant deviations from multicollinearity.
To prove the significant associations between ISL 5.3. Hypothesis Testing
and its antecedents or outcomes more clearly, we con- We used the data from Studies 1 and 2 to test our
ducted the sensitivity analysis using the strict variable hypotheses and realize cross-validation of all impact-
control method of experimentation, which is an ing paths. Studies 1 and 2 must both provide support-
important preparatory analysis for the hypothesis test ing results to validate one hypothesis. We only
in Section 5.3. When analyzing each association, we mentioned the results of Study 1 (in parentheses) in
changed the initial values (i.e., 0.1, 0.3, 0.5, 0.7, and this section if both studies provided supportive
0.9) of the independent variable while controlling the results.
five other variables to observe the impact of different First, we tested H1, H2, and H3 regarding the direct
initial independent-variable levels on the dependent effects of the independent variables (safety attitude,
variable. We illustrate the sensitivity analysis results safety competence, and self-efficacy) on ISL. We

Table 3 Means (M), Standard Deviations (SD), and Correlations (Study 2)

Variable M SD 1 2 3 4 5 6
1. Safety attitude 0.63 0.37
2. Safety competence 0.55 0.10 0.01
3. Self-efficacy 0.61 0.84 0.03 –0.01
4. FLS 0.61 0.10 0.21* 0.02 0.01
5. ISL 0.52 0.97 0.42** 0.31** 0.25** 0.43**
6. SOCB 0.56 0.46 0.67** 0.43** 0.56** 0.34** 0.54**
7. Perceived followership 0.56 0.08 0.24* 0.32* 0.45* 0.35** 0.22** 0.02

Notes: N = 1000 (100 groups of 10 workers). *p < 0.05, two-tailed; **p < 0.01, two-tailed.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 11

regressed ISL, respectively, onto the independent significant when the CI do not include zero. The indi-
variables, the moderator (FLS), and the interaction rect effect of safety attitude on SOCB via ISL was 0.10,
terms. We then regressed SOCB and perceived fol- with a 95% bootstrap bias-corrected CI from 0.023 to
lowership onto the independent variables, the media- 0.178. As the CI did not include zero, H4a was sup-
tor (ISL), the moderator (FLS), and interaction terms. ported. Similarly, the indirect effects of safety compe-
Table 4 shows that safety attitude (b = 0.12, p < 0.05), tence (indirect effect = 0.06, CI[0.022, 0.098]) and self-
safety competence (b = 0.25, p < 0.01), and self- efficacy (indirect effect = 0.03, CI[0.004, 0.072]) on SOCB
efficacy (b = 0.14, p < 0.05) have positive direct effects via ISL were significant. Thus, H4b and H4c were also
on ISL. Thus, H1, H2, and H3 were supported. supported. We then applied the same procedure and
H4 predicted the mediation effects of ISL on the criteria as above to validate H5, concerning the medi-
relationship between the three independent variables ating role of ISL in the relationship between the three
and SOCB. Table 4 shows that the direct effects of independent variables and perceived followership.
safety attitude (b = 0.05, n.s.), safety competence Results in Table 5 clearly indicate that the indirect
(b = 0.06, n.s.) and self-efficacy (b = 0.07, n.s.) on effects of safety attitude (indirect effect = 0.04, CI[0.010,
SOCB are not significant. However, all three indepen- 0.073]), safety competence (indirect effect = 0.04, CI
dent variables have direct and significant effects on [0.013, 0.069]), and self-efficacy (indirect effect = 0.02,
ISL (as indicated by H1, H2, and H3). Moreover, ISL CI[0.005, 0.047]) on perceived followership via ISL
is directly and significantly related with SOCB were also significant (all CIs exclude zero). Thus, H5
(b = 0.13, p < 0.01). It thus indicates that ISL can sig- was supported.
nificantly mediate the indirect effects of independent We also tested how FLS moderates the main effects
variables on SOCB. of the three independent variables on the mediating
In order to confirm the mediating role of ISL, we variable ISL (H6). First, Table 4 shows that the inter-
further applied a bootstrap mediation method with action between safety attitude and FLS (safety atti-
5000 samples with replacement and percentile boot- tude 9 FLS) is significantly related to ISL (b = 0.13,
strap confidence intervals (CI) (McClean et al., 2018). p < 0.05). Second, we tested whether the effects of
Bootstrapping requires no assumptions regarding the independent variables on ISL differed at the lower
shape of the sampling distribution when conducting ( 1 SD) and higher (+1 SD) levels of FLS. In Table 6,
mediation and moderated mediation tests (Hayes safety attitude was positively related to ISL when FLS
2013). We regressed SOCB and perceived follower- was high (effect = 0.42, CI[0.047, 0.806]), but not when
ship on ISL, and simultaneously regressed ISL on FLS was low (effect = 0.20, CI[ 0.006, 0.427]). More-
safety attitude, safety competence, and self-efficacy. over, the direct effects were significantly different
Table 5 presents the results of the indirect effect with between the two conditions (difference = 0.22, CI[0.059,
the corresponding CI. The indirect effect is considered 0.381]). This moderating pattern (see Figure 3)

Table 4 Regression Results of the Moderated Mediation Model

Study 1: Three-wave field study Study 2: ABM simulation experiment

ISL Perceived followership ISL Perceived followership


Variables (Mediator) SOCB (DV) (DV) (Mediator) SOCB (DV) (DV)
Age 0.00 (0.03) –0.02 (0.01) –0.02 (0.01) 0.00 (0.03) –0.02 –0.02 (0.01)
(0.01)
Gender –0.01 (0.05) 0.10 (0.09) 0.08 (0.03) –0.01 (0.05) 0.10 (0.09) 0.08 (0.03)
Group-member familiarity 0.00 (0.04) –0.01 (0.07) 0.05 (0.03) 0.00 (0.04) –0.01 0.05 (0.03)
(0.07)
Safety attitude 0.12* (0.07) 0.05 (0.02) 0.09* (0.10) 0.08* (0.11) 0.03 (0.07) 0.07**(0.02)
Safety competence 0.25** (0.07) 0.06 (0.03) 0.11* (0.06) 0.38** (0.24) 0.04 (0.19) 0.65**(0.18)
Self-efficacy 0.14* (0.08) 0.07 (0.06) 0.06 (0.11) 0.12* (0.05) 0.06 (0.06) 0.51*(0.14)
ISL 0.13** (0.05) 0.17** (0.14) 0.14*(0.11) 0.26*(0.17)
FLS 0.39** (0.06) 0.29** (0.06) 0.18** (0.06) 0.38**(0.29) 0.23*(0.07) 0.66**(0.21)
Safety attitude 9 FLS 0.13* (0.07) 0.31** (0.06) 0.26** (0.07) 0.42*(0.26) 0.19*(0.18) 0.69**(0.20)
Safety competence 9 0.12* (0.06) 0.77* (0.12) 0.15* (0.12) 0.16**(0.05) 0.37*(0.09) 0.55*(0.09)
FLS
Self-efficacy 9 FLS –0.07 (0.07) –0.03 (0.05) 0.04 (0.07) 0.13* (0.01) 0.12*(0.08) 0.28 (0.08)
F 29.49** 34.33** 51.24** 83.71** 85.24** 94.05**
Adjusted R2 0.57 0.56 0.54 0.73 0.71 0.65

Notes: N = 604 (Study 1) and 1000 (Study 2). DV = dependent variable. Standard errors (SE) are presented in parentheses.
*p < 0.05, two-tailed; **p < 0.01, two-tailed.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
12 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

provided support for H6a. Similarly, we successfully Thus, H7c was not supported, resulting in partial
validated the moderating effect of FLS on the direct support for H7.
relationship between safety competence and ISL In summary, five hypotheses (i.e., H1 to H5) were
(H6b), but we did not find sufficient empirical evi- completely supported by this study with the cross-
dence for H6c which proposes that FLS moderates the validation of Studies 1 and 2. In contrast, H6 and H7
main effect of self-efficacy on ISL. Table 4 shows that were only partially supported, because FLS can mod-
only the ABM simulation experiment validates the erate neither the direct effect of self-efficacy on ISL
moderating role of ISL (b = 0.13, p < 0.05), but the nor the indirect effect of self-efficacy on SOCB via ISL.
field study provided negative results (b = 0.07, n.s.).
Results demonstrated in Table 6 also gave consistent
conclusions (difference = 0.02, CI[ 0.071, 0.113]),
6. Discussion
thereby denying H6c. Thus, H6 was only partially 6.1. Theoretical Implications
supported. The primary contribution of this work is that it
To test H7, that is, the conditional indirect effects, expands the theoretical understanding of proactive
we performed a bootstrap-moderated mediation anal- accident prevention from the perspective of behav-
ysis with 5000 samples with replacement and per- ioral operations approach. Frequent accident occur-
centile bootstrap confidence intervals (CI). This rence calls for a substantial shift in the research
analysis enabled us to estimate whether indirect perspective and kernel construct of disaster manage-
effects of independent variables on SOCB via ISL dif- ment (De Vries et al., 2016). Our focus on informal
fered at the lower ( 1 SD) and higher (+1 SD) levels leadership of average workers and the mixed research
of FLS. Table 6 shows that FLS significantly moder- design provides a noteworthy boost for this shift.
ated the conditional indirect effect of safety attitude Although operational safety management is generally
and safety competence on SOCB via ISL such that associated with formal leadership (Barling et al., 2002,
safety attitude and safety competence had a positive Lu et al., 2017), informal leadership emergent among
indirect effect on SOCB when FLS was high, but not frontline workers has not received sufficient attention.
when FLS was low. Moreover, the indirect effects Safety management policies and instructions con-
were significantly different between the two condi- stantly decay or even get blocked when conveyed
tions. Thus, H7a (difference = 0.16, CI [0.013, 0.307]) from top-down; thus, formal safety leaders cannot
and H7b (difference = 0.11, CI [0.036, 0.261]) were sup- adequately address the occupational accidents sub-
ported. stantially (Wu et al., 2011). Thus, this study clarifies
However, both the empirical study and ABM how workers possess leadership traits by their attri-
simulation failed to validate that FLS significantly butes and the perceptions of others based on the
moderated the conditional indirect effect of self- specific needs of operation-level accident prevention.
efficacy on SOCB through ISL. Table 6 shows that Those with safety-related traits of positive attitude,
self-efficacy did not have a significantly positive high competence, and self-efficacy are more likely to
indirect effect on SOCB when FLS was either high produce ISL, generate citizenship behavior, and
(effect = 0.03, CI[ 0.021, 0.084]) or low (effect = 0.03, obtain positive resonance of others. They can recog-
CI[ 0.012, 0.075]). Moreover, the indirect effects nize and adapt to changing risk factors on the front-
were not significantly different between the two line and make up for the lack of “traditional safety
conditions (difference = 0.00, CI [ 0.013, 0.027]). managers.” Therefore, informal leaders are elites

Table 5 Mediating Role of ISL

Study 1: Three-wave field study Study 2: ABM simulation experiment

Indirect paths Indirect effects 95% CI Indirect effects 95% CI


Mediating role of ISL between independent variables and SOCB (H4a, H4b, and H4c)
Safety attitude ? ISL ? SOCB 0.10* [0.023, 0.178] 0.03** [0.004, 0.061]
Safety competence ? ISL ? SOCB 0.06** [0.022, 0.098] 0.06** [0.004, 0.107]
Self-efficacy ? ISL ? SOCB 0.03* [0.004, 0.072] 0.02* [0.006, 0.0034]
Mediating role of ISL between independent variables and perceived followership (H5a, H5b, and H5c)
Safety attitude ? ISL ? perceived followership 0.04* [0.010, 0.073] 0.07** [0.033, 0.107]
Safety competence ? ISL ? perceived followership 0.04** [0.013, 0.069] 0.04** [0.024, 0.056]
Self-efficacy ? ISL ? perceived followership 0.02** [0.005, 0.047] 0.05** [0.036, 0.068]

Notes: N = 604 (Study 1) and 1000 (Study 2). 95% CI = 95% Confidence intervals. Coefficient is significant if CI excludes zero. *p < 0.05, two-tailed;
**p < 0.01, two-tailed.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 13

Table 6 Moderated Mediation Analysis: Moderating Role of FLS with Confidence Intervals

Moderated Study 1: Three-wave field study Study 2: ABM simulation experiment


paths
High FLS Low FLS High FLS Low FLS
(+1 SD) ( 1 SD) Difference (+1 SD) ( 1 SD) Difference
Direct paths (corresponding to H6a, H6b, and H6c)
Safety attitude ? ISL 0.42* [0.047, 0.806] 0.20 [ 0.006, 0.427] 0.22* [0.059, 0.381] 0.39* [0.073 0.706] 0.26 [ 0.006, 0.504] 0.13* [0.021, 0.239]
Safety competence ? 0.45** [0.339, 0.571] 0.06 [ 0.238, 0.375] 0.39** [0.306, 0.474] 0.47** [0.287, 0.654] 0.12 [ 0.017, 0.256] 0.35* [0.053, 0.704]
ISL
Self-efficacy ? ISL 0.09 [ 0.142, 0.326] 0.07 [ 0.118, 0.261] 0.02 [ 0.071, 0.113] 0.21* [0.031, 0.389] –0.14 [ 0.319, 0.039] 0.35* [0.026, 0.676]
Indirect paths (corresponding to H7a, H7b, and H7c)
Safety attitude ? ISL 0.12* [0.018, 0.223] –0.04 [ 0.113, 0.033] 0.16* [0.013, 0.307] 0.66* [0.304, 1.013] 0.55 [ 0.073, 1.176] 0.11* [0.046, 0.174]
? SOCB
Safety competence ? **
0.13 [0.064, 0.198] 0.02 [ 0.063, 0.104] 0.11** [0.036, 0.261] 0.28* [0.048, 0.514] 0.14 [ 0.049, 0.327] 0.14* [0.024, 0.256]
ISL ? SOCB
Self-efficacy ? ISL ? 0.03 [ 0.021, 0.084] 0.03 [ 0.012, 0.075] 0.00 [ 0.013, 0.027] –0.03 [ 0.214, 0.154] –0.01 [ 0.205, 0.184] –0.02 [ 0.021,0.023]
SOCB

Notes: N = 604 (Study 1) and 1000 (Study 2). The square brackets contain 95% CI. Coefficient is significant if CI excludes zero. *p < 0.05, two-tailed;
**p < 0.01, two-tailed.

among workers, who can assemble collective endeav- frontline over time as a function of how people per-
ors and intelligence to conquer unfavorable condi- ceive the group and are perceived by the group. Thus,
tions and protect each other mutually. leadership no longer hinges on individual characteris-
Second, our study also extends the current behav- tics of future leaders alone but incorporates the
ioral operations research by providing new insights mutual cognition and influence between leaders and
into the emergence mechanism of leadership. Existing peers. We find that there are significant causal rela-
studies on leadership (including safety leadership) tionships among one’s values and consciousness (e.g.,
portrays leadership structure as a static property of attitude), competence, and personality (e.g., self-
the group (Eva et al., 2019). However, we emphasize a efficacy), which have complicated conjunction effects
social-psychological dynamic arrangement where the on leadership emergence. For instance, while good
leadership structure emerges in the operational performance is critical to getting ahead in an organi-
zation, who becomes a leader is an inherently social
process that also depends on peers granting influence
Figure 3 Moderating Effect of FLS on the Relationship between Safety
on a person engaging in leader-like behaviors
Attitude (Above), Safety Competence (Below), and ISL (Study 1)
(McClean et al., 2018).
Finally, we contribute to the leadership theory by
linking formal leadership based on authority and
informal leadership based on social cognition. Previ-
ous research tends to separate informal leadership
from formal leadership; it ignores the effect of formal
leadership and hierarchical authority on the emer-
gence and fulfillment of informal leadership. This
study focuses on informal leadership, but does not
ignore the impact of formal leadership. FLS is a cru-
cial element that leads to the conditional direct and
indirect effects within ISL emergence. Both empirical
and simulation data provide support for the argu-
ment that with no or very weak support from key
formal leaders, employees cannot stand out and get
ahead to become emerging leaders, even if they have
a high level of awareness and competence. Thus,
informal leadership is, no doubt, a social process
necessarily intertwined with formal leadership.
However, we have not obtained enough empirical
support from Study 1 for the significant moderating
effect of FLS on the direct relationship between self-
efficacy and ISL, or the indirect relationship via ISL
between self-efficacy and SOCB. Perhaps self-efficacy

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
14 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

is largely determined by innate traits and life experi- and habits. Constant onsite visibility should not only
ences, such as parental education and family rela- be for inspection but also show senior managers’
tionship (Trautner and Schwinger 2020). It cannot be humanistic concern, which in turn consolidates
easily changed in a short time, nor can its effect on employees’ determination for safety improvement
other constructs be significantly strengthened or and adherence to managers’ safety beliefs. Therefore,
attenuated by encouragement from formal leaders. more SOCB could be promoted, which would facili-
Frontline workers with weak self-efficacy show tate continuous consolidation and improvement of
insignificant or inadequate tendencies toward emerg- ISL. In summary, formal leaders should keep improv-
ing as informal leaders or exhibiting SOCB, even ing their FLS to enhance the positive impact of worker
when FLS is high. It is a major difference between safety attitude, safety competence, and self-efficacy
self-efficacy and the two other independent vari- and realize the constant emergence of qualified infor-
ables, which gives scope for future research and mal safety leaders.
indicates that practitioners must apply distinguish- On the other hand, non-managerial employees
ing efforts on different antecedents. should take the initiative to emerge as informal safety
leaders by improving their safety attitude and compe-
6.2. Practical Implications tence, as well as engaging in SOCB, that is, voluntary
This study provides a unique perspective on practical safety improvement activities. Although not all per-
efforts to leverage the benefits of the closest and most sonnel would become informal leaders, the improve-
convenient leadership in operational accident preven- ment in safety attitude, safety competence, and SOCB
tion and disaster management. Senior or intermediate will ultimately lead to the overall increase in the
leaders assume almost all responsibilities in prior group ISL level, as shown in our ABM simulation
organization-level accident preventive approaches, experiment. Moreover, continuous ISL improvement
given the notion that high-level policies and instruc- also depends on peers’ followership, which can be
tion could be sufficiently executed on the frontline induced by proper attributes and behaviors of lead-
(Wu et al., 2011). However, reality warrants the neces- ers. Employees should, thus, develop typical leader-
sity of close rather than distant leadership. We, thus, ship traits or rhetoric (e.g., leader-like actions, words,
provide the novel perspective of informal leadership voice, and emotion) corresponding to stereotypes for
emerging among ordinary workers, from which both qualified safety leaders and improve their ability to
managers and non-managers can learn. diagnose followers’ dynamic needs for informal lead-
On one hand, managers should realize that employ- ership to deploy the behavioral repertoire in amounts
ees are not merely followers but also possess leader- that match those needs. In turn, more colleagues can
ship potentials and should be able to shoulder more trigger loyal followership and defer to informal lead-
responsibilities on accident prevention. Thus, even ers’ influence.
though ISL tends to emerge randomly (affected by
various factors), managers should help cultivate 6.3. Limitations and Future Research Directions
favorable conditions from which outstanding infor- This study has several limitations and scope for future
mal leaders may emerge. For instance, managers research. First, CMV resulting from the single-source
should routinely assess the safety attitude and compe- design to collect field data is a limitation. Although
tence of frontline workers to determine whether statistical analysis proved that no significant threats
workers have sound values and sufficient knowledge of CMV existed, our research design can still be fur-
in accident prevention. Although self-efficacy is ther improved. Study 1 is limited by its single-
somewhat inherent and not significantly moderated country, single-industry, self-report nature and the
by leadership support, it is crucial in leadership emer- ways in which it captures the constructs of interest,
gence, especially when peers may regard initiatives as which is also a critical reason for applying the boot-
showing off, and many employees lack the self- strap analysis and designing Study 2 for cross-
confidence to lead. Thus, to improve worker self- validation. Much larger sample sizes (in thousands
efficacy, it is necessary to consolidate the motivation and across different countries) would be needed for
and support for workers in sponsoring proposals and more convincing inferences regarding the size of the
real actions for safety improvement. For example, effect at the population level. In particular, the context
managers can improve the welfare of workers, estab- of Study 1 is male-dominated (although this is repre-
lish a reasonable reward and punishment system, and sentative of many industries and work contexts), and
consolidate their belief in realizing their career objec- the setting and nature of work are somewhat unique
tives. Moreover, they should make regular onsite vis- (Joshi and Roh 2009). Although we controlled for gen-
its to disseminate safety policies, principles, der in our hypothesis testing, future studies can
strategies, attitudes, and values to frontline personnel explore these relationships in other contexts, espe-
and cultivate their advanced safety behavioral norms cially those with more female members.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 15

Second, despite great efforts in model develop- leadership theory with social cognitive theory,
ment, the constructs, interrelationships, behavioral proposing the ISL construct as well as a conceptual
and interactive rules of workers in both empirical model for its emergence mechanism, thus providing
research and agent-based simulation are still simpli- insights into how average employees might become
fied from the complex processes occurring in reality, leaders and exert substantial influence on the opera-
and interpretations are limited to the conditions and tional frontline. We applied a mixed methodology of
model properties represented by the models. For field study and ABM simulation to strengthen our
instance, informal leaders can make ISL more explicit arguments on the causal relationship between differ-
and visible by exhibiting SOCB as well as induce fol- ent variables, as ABM allows for exploring dynamic
lowership among their coworkers, which leads to the models with a large number of variables operating
consolidation of further ISL emergence. However, this under several different conditions, which are chal-
complex recursive mechanism is not covered by lenging to analyze via traditional approaches (Castillo
empirical research and simplified to meet ABM simu- and Trinh 2018).
lation requirements, which gives scope for meaning- This study deepens our understanding of the con-
ful future research. notation of leadership in behavioral operations and
Finally, the causal relationship between ISL and disaster management studies through the exploration
safety performance, which is a special case of the rela- of the ISL emergence mechanism. We hope it serves
tionship between informal leadership and organiza- as motivation for future disaster management
tional performance, also remains unanswered. research on the social-psychological and organiza-
Although we articulate the advantage of ISL in acci- tional behavioral foundations of informal leadership
dent prevention and depict the possible impacting emergence in groups. These insights will be particu-
approaches in this study, we neither established a larly important as organizations become more com-
structured conceptual model nor found enough empir- plex and dynamic. Thus, more adaptive leadership
ical support for the ISL effect as a significant predictor should be identified and cultivated to cope with
of safety performance and accident reduction. Thus increasingly serious operational safety conditions.
far, research on the effect of leadership emergence on
work performance with rigorous methodologies such
as longitudinal field study or intervention experiment Acknowledgments
remains scarce (Acton et al., 2019, Balthazard et al., The authors are grateful to the department editor Prof. Sus-
2009). Thus, future studies should build upon the find- hil K. Gupta, the senior editor, and the two anonymous
ings by identifying ways that informal leaders with reviewers for their insightful suggestions and comments
favorable safety attitudes and competence can improve that significantly improved this study. This work is sup-
safety performance, incorporating constructs such as ported financially by National Natural Science Foundation
task complexity, team task interdependence, the span of China (Grant Nos. 71801007 and 71971045), MOE (Min-
of control, network density, and network centralization istry of Education of China) Project of Humanities and
(Hoch and Dulebohn 2017). Social Sciences (Grant No. 18YJCZH188), and Fundamental
Research Funds for the Central Universities (Grant No.
YWF-21-BJ-W-225).
7. Conclusions
A large number of research works suggested more References
efforts and new perspectives to study safety within Acton, B. P., R. J. Foti, R. G. Lord, J. A. Gladfelter. 2019. Putting
the POM domain (Gupta et al., 2020, Gupta et al., emergence back in leadership emergence: A dynamic, multi-
2016, Paciarotti and Valiakhmetova 2021, Starr 2001). level, process-oriented framework. Leadership Q. 30: 145–164.
For example, based on their disaster management Aiken, L. S., S. G. West 1991. Multiple Regression: Testing and Inter-
preting Interactions, Sage, Newbury Park, CA.
review work, Gupta et al. (2016) argued that “pre-
Amabile, T. M., E. A. Schatzel, G. B. Moneta, S. J. Kramer. 2004.
venting accidents is not generating the kind of interest Leader behaviors and the work environment for creativity:
it should create,” and called on POM leaders “to pro- Perceived leader support. Leadership Q. 15(1): 5–32.
vide some guidance to incentivize research along pre- The Anylogic Company. 1992–2017. AnyLogic 8.0 User’s Guide.
vention and mitigation lines.” We hope this study Available at https://help.anylogic.com/index.jsp (accessed
serves as a beneficial attempt which can respond to date September 27, 2020).
the above call. A frequently asked question across dis- Balthazard, P. A., D. A. Waldman, J. E. Warren. 2009. Predictors
of the emergence of transformational leadership in virtual
ciplines pertaining to disaster management is “how decision teams. Leadership Q. 20(5): 651–663.
can accidents and calamities be prevented and con- Bandura, A. 1986. Social Foundations of Thought and Action: A Social
trolled if major leaders and executives are far from Cognitive Theory, Prentice-Hall, Englewood Cliffs, NJ.
the frontline and cannot ensure attendance all the Bandura, A. 2001. Social cognitive theory: An agentic perspective.
time?” We address this question by linking the classic Annu. Rev. Psychol. 52(1): 1–26.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
16 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

Barling, J., C. Loughlin, E. K. Kelloway. 2002. Development and Gershenoff, A. B. 2003. Individual differences and leader emergence in
test of a model linking safety-specific transformational leader- a transformational context: An examination of person and process.
ship and occupational safety. J. Appl. Psychol. 87(3): 488–496. Unpublished doctoral dissertation. Virginia Polytechnic Insti-
Bendoly, E., W. van Wezel, D. G. Bachrach 2015. The Handbook of tute and State University, Blacksburg, VA.
Behavioral Operations Management: Social and Psychological Gupta, S., M. K. Starr, R. Z. Farahani, N. Matinrad. 2016. Disaster
Dynamics in Production and Service Settings, Oxford University management from a POM perspective: Mapping a new
Press, New York, NY. domain. Prod. Oper. Manag. 25(10): 1611–1637.
Burke, C. S., S. M. Fiore, E. Salas. 2003. The role of shared cognition Gupta, S., M. K. Starr, R. Zanjirani Farahani, M. M. Ghodsi.
in enabling share leadership and team adaptability. In C. L. Pearce 2020. Prevention of terrorism–an assessment of prior POM
Conger, J. A. (Eds.), Shared Leadership: Reframing the Hows work and future potentials. Prod. Oper. Manag. 29(7): 1789–
and Whys of Leadership. Sage, Thousand Oaks, CA, pp. 103– 1815.
122. Hannah, S. T., B. J. Avolio, F. Luthans, P. D. Harms. 2008. Leader-
Carnabuci, G., C. Emery, D. Brinberg. 2018. Emergent leadership ship efficacy: Review and future directions. Leadership Q. 19
structures in informal groups: A dynamic, cognitively (6): 669–692.
informed network model. Organ. Sci. 29(1): 118–133. Hardcopf, R., R. Shah, S. Dhanorkar. 2021. The impact of a spill
Castillo, E. A., M. P. Trinh. 2018. In search of missing time: A or pollution accident on firm environmental activity: An
review of the study of time in leadership research. Leadership empirical investigation. Prod. Oper. Manag. Forthcoming.
Q. 29(1): 165–178. https://doi.org/10.1111/poms.13392
Chang, S., D. Chen, T. Wu. 2012. Developing a competency model Hayes, A. F. 2013. Introduction to Mediation, Moderation, and Condi-
for safety professionals: Correlations between competency tional Process Analysis: A Regression-Based Approach, Guilford
and safety functions. J. Saf. Res. 43(5–6): 339–350. Press, New York, NY.
Chen, C. F., S. C. Chen. 2014. Measuring the effects of safety man- Henning, J. B., C. J. Stufft, S. C. Payne, M. E. Bergman, M. S.
agement system practices, morality leadership and self- Mannan, N. Keren. 2009. The influence of individual differ-
efficacy on pilots’ safety behaviors: Safety motivation as a ences on organizational safety attitudes. Saf. Sci. 47(3): 337–
mediator. Saf. Sci. 62: 376–385. 345.
Chen, G., S. M. Gully, D. Eden. 2001. Validation of a new general Henseler, J., C. M. Ringle, M. Sarstedt. 2015. A new criterion for
self-efficacy scale. Organiz. Res. Methods 4(1): 62–83. assessing discriminant validity in variance-based structural
Chmiel, N., J. Laurent, I. Hansez. 2017. Employee perspectives on equation modeling. J. Acad. Mark. Sci. 43(1): 115–135.
safety citizenship behaviors and safety violations. Saf. Sci. 93: Hmieleski, K. M., R. A. Baron. 2009. Entrepreneurs’ optimism and
96–107. new venture performance: A social cognitive perspective.
Chrobot-Mason, D., A. Gerbasi, K. L. Cullen-Lester. 2016. Predict- Acad. Manag. J. 52(3): 473–488.
ing leadership relationships: The importance of collective Hoch, J. E., J. H. Dulebohn. 2017. Team personality composition,
identity. Leadership Q. 27(2): 298–311. emergent leadership and shared leadership in virtual teams:
Cogliser, C. C., W. L. Garnder, M. B. Gavin, J. C. Broberg. 2012. A theoretical framework. Hum. Resou. Manag. Rev. 27(4): 678–
Big five personality factors and leader emergence in virtual 693.
teams: Relationships with team trust worthiness, member per- Hofmann, D. A., M. J. Burke, D. Zohar. 2017. 100 years of occupa-
formance contributions, and team performance. Group Orga- tional safety research: From basic protections and work analy-
niz. Manag. 37(6): 752–784. sis to a multilevel view of workplace safety and risk. J. Appl.
Conger, J. A., R. N. Kanungo, S. T. Menon. 2000. Charismatic Psychol. 102(3): 375–388.
leadership and follower effects. J. Organiz. Behav. 21(7): 747– Hofmann, D. A., F. P. Morgeson, S. J. Gerras. 2003. Climate as a
767. moderator of the relationship between leader-member
De Vries, J., R. De Koster, D. Stam. 2016. Safety does not happen exchange and content specific citizenship: Safety climate as an
by accident: Antecedents to a safer warehouse. Prod. Oper. exemplar. J. Appl. Psychol. 88(1): 170–178.
Manag. 25(8): 1377–1390. Holtbr€
ugge, D., A. Baron, C. B. Friedmann. 2015. Personal attri-
DeRue, D. S., J. D. Nahrgang, S. J. Ashford. 2015. Interpersonal butes, organizational conditions, and ethical attitudes: A
perceptions and the emergence of leadership structures in social cognitive approach. Busin. Ethics 24(3): 264–281.
groups: A network perspective. Organ. Sci. 26(4): 1192–1209. International Nuclear Safety Advisory Group. 1986. Summary report
Dinh, D. D., N. H. Vu, R. C. McIlroy, K. A. Plant, N. A. Stanton. on the post-accident review meeting on the chernobyl accident.
2020. Examining the roles of multidimensional fatalism on Report, International Atomic Energy Agency, Vienna, Austria.
traffic safety attitudes and pedestrian behaviour. Saf. Sci. 124, Joshi, A., H. Roh. 2009. The role of context in work team diversity
104587. research: A meta-analytic review. Acad. Manag. J. 52(3): 599–
Edwards, J. R., L. S. Lambert. 2007. Methods for integrating mod- 627.
eration and mediation: A general analytical framework using Jourden, F. J. 1993. The new leadership paradigm: Social learning
moderated path analysis. Psychol. Methods 12(1): 1–22. and cognition in organizations. Acad. Manag. Perspect. 2(7):
Eva, N., J. Cox, H. Tse, K. Lowe. 2019. From competency to con- 93–94.
versation: A multi-perspective approach to collective leader- Kelley, R. 1992. The Power of Followership, Knopf Doubleday Pub-
ship development. Leadership Q. Forthcoming. lishing Group, New York, NY.
Fang, D., C. Wu, H. Wu. 2015. Impact of the supervisor on Kwok, N., S. Hanig, D. J. Brown, W. Shen. 2018. How leader role
worker safety behavior in construction projects. J. Manag. Eng. identity influences the process of leader emergence: A social
31(6): 04015001. network analysis. Leadership Q. 29(6): 648–662.
Fang, D., H. Wu. 2013. Development of a safety climate interaction Leung, C., A. Lucas, P. Brindley, S. Anderson, J. Park, A. Vergis,
(SCI) model for construction projects. Saf. Sci. 57: 138–149. L. M. Gillman. 2018. Followership: A review of the literature
Fibich, G., R. I. Gibori. 2010. Aggregate diffusion dynamics in in healthcare and beyond. J. Crit. Care 46: 99–104.
agent-based models with a spatial structure. Oper. Res. 58(5): Li, F., Z. Li, H. Chen, Z. Chen, M. Li. 2020. An agent-based
1450–1468. learning-embedded model (ABM-learning) for urban land use

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society 17

planning: A case study of residential land growth simulation Pendleton, G. B. 2011. The informal leader’s role on construction sites:
in Shenzhen, China. Land Use Policy 95, 104620. A comparative analysis of formal and informal leadership structures
Li, H., X. Li, X. Luo, J. Siebert. 2017. Investigation of the causality within the construction industry. Unpublished doctoral disserta-
patterns of non-helmet use behavior of construction workers. tion, Virginia Polytechnic Institute and State University,
Automation Constru. 80: 95–103. Blacksburg, VA.
Lord, R. G., P. Gatti, S. L. M. Chui. 2016. Social-cognitive, rela- Podsakoff, P. M., S. B. MacKenzie, J. Y. Lee, N. P. Podsakoff.
tional, and identity-based approaches to leadership. Organ. 2003. Common method biases in behavioral research: A criti-
Behav. Hum. Decis. Process. 136: 119–134. cal review of the literature and recommended remedies. J.
Appl. Psychol. 88(5): 879–903.
Lu, C., H. Weng, C. Lee. 2017. Leader-member exchange, safety
climate and employees’ safety organizational citizenship Ramaswamy, S. K., G. A. Mosher. 2018. Using workers’ compen-
behaviors in container terminal operators. Maritime Bus. Rev. sation claims data to characterize occupational injuries in the
2(4): 331–348. biofuels industry. Saf. Sci. 103: 352–360.
Lund, I. O., T. Rundmo. 2009. Cross-cultural comparisons of traf- Salem, M., N. V. Quaquebeke, M. Besiou, L. Meyer. 2019. Inter-
fic safety, risk perception, attitudes and behaviour. Saf. Sci. 47 group leadership: How leaders can enhance performance of
(4): 547–553. humanitarian operations. Prod. Oper. Manag. 28(11): 2877–
2897.
Luria, G., Y. Berson. 2013. How do leadership motives affect
informal and formal leadership emergence? J. Organiz. Behav. Serban, A., F. J. Yammarino, S. D. Dionne, S. S. Kahai, C. Hao, K.
34(7): 995–1015. A. McHughd, K. L. Sotak, L. Mushore, T. L. Friedrich, D. R.
Peterson. 2015. Leadership emergence in face-to-face and vir-
Luria, G., A. Kahana, J. Goldenberg, Y. Noam. 2019. Leadership
tual teams: A multi-level model with agent-based simulations,
development: Leadership emergence to leadership effective-
quasi-experimental and experimental tests. Leadership Q. 26(3):
ness. Small Group Res. 50(5): 571–592.
402–418.
Lynch, K. D. 2007. Modeling role enactment: Linking role theory
Shondrick, S. J., R. G. Lord. 2010. Implicit leadership and follow-
and social cognition. J. Theory Soc. Behav. 37(4): 379–399.
ership theories: Dynamic structures for leadership percep-
Martin, H., T. M. Lewis. 2014. Pinpointing safety leadership fac- tions, memory, and leader-follower processes. Intern. Rev.
tors for safe construction sites in Trinidad and Tobago. J. Con- Indus. Organiz. Psychol. 25: 1–33.
stru. Eng. Manag. 140(2): 04013046.
Smith, J. A., R. J. Foti. 1998. A pattern approach to the study of
McClean, E. J., S. R. Martin, K. J. Emich, C. T. Woodruff. 2018. leader emergence. Leadership Q. 9(2): 147–160.
The social consequences of voice: An examination of voice
Starr, M. K. 2001. Safety and security: Critical qualities call for
type and gender on status and subsequent leader emergence.
refocusing POM. Prod. Oper. Manag. 10(4): 361–362.
Acad. Manag. J. 61(5): 1869–1891.
Stauffer, J. M., S. Kumar. 2021. Impact of incorporating returns
Mitchell, P., J. Nightingale, P. Reeves. 2019. Competence to capa-
into pre-disaster deployments for rapid-onset predictable dis-
bility: An integrated career framework for sonographers.
asters. Prod. Oper. Manag. 30(2): 451–474.
Radiography 25(4): 378–384.
Sun, J., M. Zheng, M. Skitmore, B. Xia, X. Wang. 2019. Industry
Moideenkutty, U., S. M. Schmidt. 2011. Leadership tactics:
effect of job hopping: An agent-based simulation of Chinese
Enabling quality social exchange and organizational citizen-
construction workers. Front. Eng. Manag. 6(2): 249–261.
ship behavior. Organiz. Manag. J. 8(4): 229–241.
Sun, Z., C. Zhang, P. Tang. 2021. Modeling and simulating the
Morrison, E. W., C. C. Phelps. 1999. Taking charge at work: Extra-
impact of forgetting and communication errors on delays in
role efforts to initiate workplace change. Acad. Manag. J. 42(4):
civil infrastructure shutdowns. Front. Eng. Manag. 8(1): 109–
403–419.
121.
Muthen, L. K., B. O. Muthen. 1998–2017. Mplus User’s Guide, 8th
Taggar, S., R. Hackett, S. Saha. 1999. Leadership emergence in
edn. Los Angeles, CA.
autonomous work teams: Antecedents and outcomes. Pers.
Neubert, M. J., S. Taggar. 2004. Pathways to informal leadership: Psychol. 52(4): 899–926.
The moderating role of gender on the relationship of individ-
Tholen, S. L., A. Pousette, M. T€ orner. 2013. Causal relations
ual differences and team member network centrality to infor-
between psychosocial conditions, safety climate and safety
mal leadership emergence. Leadership Q. 15(2): 175–194.
behavior: A multi-level investigation. Saf. Sci. 55: 62–69.
Newman, A., H. Tse, G. Schwarz, I. Nielsen. 2018. The effects of
Trautner, M., M. Schwinger. 2020. Integrating the concepts self-
employees’ creative self-efficacy on innovative behavior: The
efficacy and motivation regulation: How do self-efficacy
role of entrepreneurial leadership. J. Bus. Res. 89: 1–9.
beliefs for motivation regulation influence self-regulatory suc-
Olaniran, B. A. 1996. A model of group satisfaction in computer- cess? Learn. Individ. Differen. 80, 101890.
mediated communication and face-to-face meetings. Behav.
Uhl-Bien, M., R. Marion. 2009. Complexity leadership in bureau-
Inf. Technol. 15(1): 24–36.
cratic forms of organizing: A meso model. Leadership Q. 20(4):
Ozyilmaz, A., S. S. Cicek. 2015. How does servant leadership 631–650.
affect employee attitudes, behaviors, and psychological cli-
Uhl-Bien, M., R. E. Riggio, K. B. Lowe, M. K. Carsten. 2014. Fol-
mates in a for-profit organizational context? J. Manag. Organiz.
lowership theory: A review and research agenda. Leadership
21(3): 1–28.
Q. 25(1): 83–104.
Paciarotti, C., I. Valiakhmetova. 2021. Evaluating disaster opera-
Wellman, N. 2017. Authority or community? A relational models
tions management: An outcome-process integrated approach.
theory of group-level leadership emergence. Acad. Manag.
Prod. Oper. Manag. 30(2): 543–562.
Rev. 42(4): 1–61.
Paglis, L. L., S. G. Green. 2002. Leadership self-efficacy and man-
Wu, C., F. Wang, P. X. W. Zou, D. Fang. 2016. How safety
agers’ motivation for leading change. J. Organiz. Behav. 23(2):
leadership works among owners, contractors and subcontrac-
215–235.
tors in construction projects. Int. J. Project Manag. 34(5): 789–
Pan, J., S. Liu, B. Ma, Z. Qu. 2018. How does proactive personality 805.
promote creativity? A multilevel examination of the interplay
Wu, T. C., S. H. Chang, C. M. Shu, C. T. Chen, C. P. Wang. 2011.
between formal and informal leadership. J. Occupa. Organiz.
Safety leadership and safety performance in petrochemical
Psychol. 91(4): 852–874.

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523
Wu, Yao, Ning, and Wang: Informal Safety Leadership
18 Production and Operations Management 0(0), pp. 1–18, © 2021 Production and Operations Management Society

industries: The mediating role of safety climate. J. Loss Prev. Zhang, X. A., N. Li, T. Brad Harris. 2015. Putting non-work ties to
Process Ind. 24(6): 716–721. work: The case of guanxi in supervisor–subordinate relation-
Wu, X., Q. Liu, L. Zhang, M. J. Skibniewski, Y. Wang. 2015. ships. Leadership Q. 26(1): 37–54.
Prospective safety performance evaluation on construction
sites. Accid. Anal. Prev. 78: 58–72. Supporting Information
Yang, X., Y. Yu, H. Li, X. Luo, F. Wang. 2017. Motion-based anal- Additional supporting information may be found online
ysis for construction workers using biomechanical methods.
in the Supporting Information section at the end of the
Frontiers Eng. Manag. 4(1): 84–91.
article.
Zhang, C., J. D. Nahrgang, S. J. Ashford, D. S. DeRue. 2020. The
risky side of leadership: Conceptualizing risk perceptions in
informal leadership and investigating the effects of their over- Appendix S1: Attributes and Initial Values of Formal
time changes in teams. Organ. Sci. 31(5): 1138–1158. Leaders in Anylogic 8.0

Please Cite this article in press as: Wu, C., H. Yao, et al. Emergence of Informal Safety Leadership: A Social–Cognitive Process for Acci-
dent Prevention. Production and Operations Management (2021), https://doi.org/10.1111/poms.13523

You might also like