You are on page 1of 12

Mediating Effect of Risk Propensity between

Personality Traits and Unsafe Behavioral


Intention of Construction Workers
Jian Zhang 1; Pengcheng Xiang 2; Rong Zhang 3; Dan Chen 4; and Yitian Ren 5
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The construction industry has the highest number of casualties per accident, and workers at construction sites are vulnerable to
accidents. Current research on unsafe behavior mainly focuses on behavior-based safety at the individual level and safety culture at the
organization level; these two safety management rubrics enjoy respective advantages and are essentially complementary. This study takes
a psychological approach and examines how personality traits affect unsafe behavior of construction workers. Based on the Big Five person-
ality model and the theory of planned behavior, it is postulated that risk propensity mediates the relationships between personality traits and
unsafe behavioral intention of construction workers. Structural equation modeling and bootstrapping techniques were used to analyze data
collected from 533 construction workers in China. Results show that (1) personality traits, risk propensity, and unsafe behavioral intention
of construction workers are related to each other; and (2) risk propensity plays a partial mediating role on the relationships between person-
ality traits (extraversion, openness to experience, and conscientiousness) and unsafe behavioral intention. The findings provide a deeper
understanding of how the perceptions of construction workers affect their behaviors. DOI: 10.1061/(ASCE)CO.1943-7862.0001792.
© 2020 American Society of Civil Engineers.
Author keywords: Construction workers; Big Five personality traits; Risk propensity; Unsafe behavioral intention.

Introduction transportation industry (Zhang et al. 2015); it rose and ranked second
only to the transportation industry in 2016 (Nie 2017).
Construction is a high-risk industry, and safety accidents seriously Construction accidents are often major accidents, such as the 42
interfere with the construction period and affect site management major accidents in the construction industry in 2016, which caused
(Kines et al. 2010). Especially, the complex environment means an 423 casualties (Zhang 2018). Major accidents are often caused by
increased risk of accidents and injuries to workers in the construc- minor and premonitory ones and hidden dangers, and once a major
tion industry (Aksorn and Hadikusumo 2007). The casualty rate of accident occurs, there are at least 29 minor accidents, 300 premoni-
the construction industry is in the forefront of almost all industries tory accidents, and 1,000 hidden dangers behind it (Wang 2014).
in the world (Ju et al. 2013; Hyoung et al. 2009). Since 2010, the Workers are the major participants and direct victims of safety ac-
number of construction accidents and casualties in China has been at cidents on construction sites. Most accidents involve human factors,
a high level, and it has continued to rise since 2015 (Shen et al. 2019). and approximately 80% are caused by workers’ unsafe behaviors
It ranks third in all industries, second only to the mining industry and (Haslam et al. 2005; Mitropoulos et al. 2009; Li et al. 2015).
There are two main research areas for the study of unsafe
1
Ph.D. Candidate, School of Management Science and Real Estate, behavior. The first is called behavior-based safety, which aims to
Chongqing Univ., Chongqing 400030, China. Email: jason2016.cqu@ eliminate unsafe behavior by intervening through certain measures.
foxmail.com These types of studies focus directly on the workers themselves,
2
Professor, School of Management Science and Real Estate, Chongqing such as their material award, goal setting, and performance feed-
Univ., Chongqing 400030, China; Senior Research Fellow, Construction back (Austin et al. 1996; Cooper et al. 1994). Another research area
Economics and Management Research Center, Chongqing Univ., Chongqing focuses on safety culture. These studies argue that individuals’
400030, China (corresponding author). Email: pcxiang@cqu.edu.cn behavior is influenced by organizational culture, and the unsafe
3
Lecturer, School of Management Science and Real Estate, Chongqing
behavior of workers should be eliminated from the level of the
Univ., Chongqing 400030, China; Senior Research Fellow, International
Research Center for Sustainable Built Environment, Chongqing Univ., organization (Choudhry et al. 2007; Nordlöf et al. 2015). However,
Chongqing 400030, China. Email: zhang.rong@cqu.edu.cn each of these research foci provide insufficient understanding of the
4
Ph.D. Candidate, School of Management Science and Real Estate, causes of unsafe behavior. Behavior-based safety research cannot
Chongqing Univ., Chongqing 400030, China. Email: 20180301001@cqu fundamentally eliminate the causes of unsafe behavior (Fang et al.
.edu.cn 2015); meanwhile, safety culture research neglects to consider indi-
5
Ph.D. Candidate, Dept. of Planning and Environmental Management, vidual factors related to workers, resulting in a lack of explanation
Univ. of Manchester, Oxford Rd., Manchester M13 9PL; Research Assis- of the micromechanisms through which organizational factors
tant, International Research Center for Sustainable Built Environment, affect workers’ behavior (Wen Lim et al. 2018; Zhang 2012).
Chongqing Univ., Shazheng St., Chongqing 400045, China. Email:
Several scholars have introduced psychological theory into their
rrtyijia@163.com
Note. This manuscript was submitted on February 8, 2019; approved on
research and have focused on the psychological state of workers
September 10, 2019; published online on February 11, 2020. Discussion at work; these studies have established understanding of the mi-
period open until July 11, 2020; separate discussions must be submitted croscopic relationships between safety culture and behavior-based
for individual papers. This paper is part of the Journal of Construction safety (Fogarty and Shaw 2010; Hu and Wu 2009). Several scholars
Engineering and Management, © ASCE, ISSN 0733-9364. have suggested that the risk propensity of individuals is strongly

© ASCE 04020023-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


related to personality traits (Nicholson et al. 2005); other research- however, a short item scale can improve measurement efficiency
ers argue that risk propensity is a consistent predictor of risk behav- and has certain necessity and value (Gosling et al. 2003).
iors of individuals (Zuckerman and Kuhlman 2010). Based on the Subtle differences in scale translation also affect the reliability
related theoretical foundations and through introducing psycho- and validity of the questionnaire (Li 2013). In the analysis of adult
logical variables—specifically, risk propensity, personality traits, samples over 21 years old, personality characteristics were also
and behavior—this research explores the psychological state of significantly correlated with age (Li et al. 2003). The influence
construction workers and aims to determine how they influence of gender factors often exceeds age and education level (Dai and
unsafe behavior. Wu 2005). Even whether the personality traits in the personality
This study has three main objectives. First, this study investi- model are five or more or fewer remains controversial (Howard
gates how construction workers with diverse personalities differ 2011). But these discussions about the Big Five are not the main
in the way they perceive unsafe behavior; the second is to determine objectives of this study.
whether there is a link between personality traits and unsafe behav-
ior through risk propensity; the third is to clarify what personality
Risk Propensity
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

traits increase unsafe behavior through risk propensity at construc-


tion sites. This research will not only contribute to psychological Risk propensity refers to a person’s attitude toward risk, which
knowledge with respect to the construction industry, but the results represents a generic orientation toward taking or avoiding risks
may also be used to predict unsafe behavior or accidents to some (Rohrmann 2002). Research on the risk propensity of individuals
extent, which has certain theoretical and practical significance. can be classified into two viewpoints. The traditional viewpoint
indicates that risk propensity is a stable and constant dispositional
trait (Ghosh and Ray 1997). Surveys have demonstrated the stability
Literature Review of risk propensity across contexts, gender, age, and backgrounds and
have indicated that it remains constant regardless of the presence of
other traits (Dohmen et al. 2011). Risk-takers can be classified into
Personality Traits
three nonexclusive types: stimulation seekers are seen as truly risk
The Big Five personality model, referred to the Five-Factor model seeking, whereas the other two, goal achievers and risk adapters, are
of personality, has emerged as an influential model for understand- more correctly viewed as risk bearers (Nicholson et al. 2005). The
ing the relationship between individuals’ personality and behaviors consideration of risk attitude as a personality trait were initially
(Jensen and Patel 2011). The Five-Factor model has received defined as stable personality characteristics (Blais and Weber
considerable support and is the most widely accepted and compre- 2006), and through the questionnaire method to measure risk atti-
hensive model of core personality across adulthood (John and tudes, it has shown high test–retest stability (Lönnqvist et al. 2015).
Srivastava 1999; Lang et al. 2011). The model includes five dimen- The second viewpoint argues that risk propensity varies with de-
sions: neuroticism, extraversion, openness to experience, agree- cision making and is changeable over time (Sitkin and Weingart
ableness, and conscientiousness. Neuroticism refers to degree of 1995; Weber et al. 2002). The level of risk taking is relatively incon-
emotional stability and is a measure of whether an individual easily sistent across situations (Kahneman and Tversky 1979; Nicholson
experiences psychological distress, unrealistic thoughts, and exces- et al. 2005). The study of risk attitude can be extended from risk-
sive expectations (Costa and McCrea 1992). Extraversion repre- seeking to risk aversion, and there is no convincing evidence the
sents the propensity to which a person is assertive, energetic, and dimension is a general trait that is consistent across domains
enthusiastic (Costa and McCrea 1992). Openness to experience (Rohrmann 2002). In light of these two viewpoints, this research
measures the inclination to seek a diverse and broad range of new views risk propensity as stable but changeable, taking a balanced
experiences (Zhao and Seibert 2006). Agreeableness examines approach to the investigation of risk propensity (Wang et al. 2016).
individuals’ attitude toward others; those who score high on it are
generally amenable and easy to get along with (Hough 1992). Con-
scientiousness relates to individuals’ degree of organization and Unsafe Behavior
durability, and their dynamic nature of the target orientation and Although there is a large body of literature on unsafe behavior,
behavior (Zhao and Seibert 2006). there remains no clear definition. Unsafe acts (Heinrich et al. 1980;
In recent years, the description of personality traits of Big Five Jones et al. 1999), human error (Reason 1990; Rigby 1970), and
has been accepted by personality psychologists (Luo et al. 2016; Li violation (Shappell and Wiegmann 2001; Wiegmann and Shappell
2013; John et al. 2008). The NEO (Neuroticism Extraversion 2003), among others, are common themes in the research on behav-
Openness Personality Inventory) Personality Inventory-Revised ioral safety. Other scholars classify unsafe behaviors according to
(NEO-PI-R) has received considerable support and is the most engineering classifications (Sanmiquel et al. 2010; Smith et al.
widely used instrument for measuring personality (Gosling et al. 2013), cognitive classifications (Reason 1990), or behavioral perfor-
2003; Li 2013). The scale contains more than 200 questions, con- mance classifications (Moore et al. 2009; Shappell and Wiegmann
sisting of 5 dimensions, 30 levels, and 240 items (Gosling et al. 2001). This study uses the cognitive classifications according to
2003). Several scholars have developed shorter measurement its root causes: unintentional behavior and intentional behavior
scales, such as the 60 item NEO-Five-Factor Inventor (NEO-FFI) (Reason 1990). This research focuses on intentional unsafe behav-
scale, the 50 pairs of adjectives scale (Goldberg 1990), the 10 item ior. The theory of planned behavior (TPB) is the most famous theory
TIPI (Ten Item Personality Inventory) (John and Srivastava 1999), in behavioral psychology for the study of mechanisms underlying
and the 5 item FIPI (Five Item Personality Inventory) scale (Saucier individual behavior. In TPB, behavioral intention directly predicts
1994). These scales have been tested and are considered to be re- individual behavior, and it is determined by three major factors:
liable and effective tools for measuring personality. The NEO-PI-R attitude toward the behavior, subjective norms, and perceived
takes more than 40 min to complete; completion of such a long behavioral control (Ajzen 1991). TPB is depicted in Fig. 1.
scale is unable to meet the requirement of measuring the five factors Attitude toward the behavior refers to the positive or negative
of personality rapidly and accurately. Theoretically speaking, the value assessment of potential actions. Subjective norm refers to
accuracy of a short scale is generally lower than that of long scale; the social pressure that an individual perceives when deciding

© ASCE 04020023-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Attitude towards and so on (Costa and McCrea 1992). Individuals with a high degree
the behavior
of agreeableness may prefer safer solutions or choices so as to
reduce their uncomfortable feelings (Lauriola and Levin 2001).
Conscientiousness relates to “control of impulses” (McCrae and
Subjective norm Intention Behavior
Costa 1991), and empirical studies have suggested that there is a
negative correlation between conscientiousness and the occurrence
of accidents (Cellar et al. 2001). Individuals scoring high on this
tend to correlate well across criterion measures of safety perfor-
Perceived mance (Arthur and Graziano 1996); they are less likely to produce
behavioral control unsafe behavior and have a relatively low risk tendency (Hogan and
Ones 1997). Based on these theories, the following hypothesis is
Fig. 1. Model of theory of planned behavior. proposed:
Hypothesis 1: Neuroticism, extraversion, and openness to ex-
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

perience are positively related to unsafe behavioral intention among


construction works, whereas agreeableness and consciousness are
whether to perform a particular behavior. Perceived behavioral negatively related to unsafe behavioral intention.
control refers to the perceived difficulty in performing a particular
behavior. Besides, other factors may have indirect effects on behav-
ioral intention by affecting the three major elements of TPB (Ajzen Big Five Personality and Risk Propensity
1991). Construction workers’ unsafe behavior originates from their Personality traits represent individuals’ psychological and behav-
deliberate intentions, and the unsafe behavioral intention refers to ioral systems; an individual’s personality can lead to consistent risk
an individual who still wants to do the behavior based on the prem- tendency or risk aversion. Data have indicated risk propensity to be
ise that they knew that this behavior violated relevant safety regu- strongly rooted in personality (Nicholson et al. 2005). The relevant
lations, so the analysis of the influencing factors of unsafe behavior studies are discussed next. Neuroticism reflects differences in emo-
intention is also the research focus. This research uses TPB to sys- tional stability and adjustment ability (Costa and McCrea 1992).
tematically analyze the unsafe behavioral intention of construction People who score low on neuroticism usually show more self-
workers. confidence, mildness, and may even be more relaxed (Zhao and
Seibert 2006). Because neuroticism is a “defensive response” to
threat in relevant motivation theory, neurotic individuals may ex-
Analysis of Hypotheses and Theoretical Model perience more negative events in daily life than other individuals
(Deyoung 2015).
Because extraverts are predisposed to positive emotions and
Big Five Personality Traits and Unsafe Behavioral
external stimulation, they are more susceptible to risk taking
Intention
(McCarthy 2003); they usually attach importance to peripheral feel-
According to social cognitive theory, individuals’ behaviors are af- ings and needs and have a need for stimulation. They are prone to
fected by psychological factors (Bandura 1991; Kahn and Byosiere adventurous ideas, and high extraversion supplies the motivational
1992). Behavioral genetics research has indicated that heritable dif- force for specific risk taking, such as financial or career risk taking
ferences in temperament and dispositions can influence the choices (Nicholson et al. 2005).
people make in life (Soane and Nicholson 2009). In the Five-Factor Openness to experience is characterized by a tendency to seek
model of personality, neuroticism is related to the trait of anxiety; novel experiences and new ideas (Zhao and Seibert 2006). Individ-
there is a positive correlation between it and unsafe behavior inten- uals high on openness to experience are curious and tend toward
tion (Lauriola and Levin 2001), and neurotic individuals generally novelty-seeking (John and Srivastava 1999). In contrast, individ-
have negative perceptions (McCrae and Costa 1991). Generally, uals low on it are regarded as conventional and traditional; they
individuals who score high on the degree of emotional stability are may tend to avoid risk (Kowert and Hermann 1997).
less likely to act impulsively and pursue risk; correspondingly, indi- High levels of agreeableness may inhibit one’s willingness to
viduals who score high on unsafe behavior intention also score drive hard bargains, and these individuals may obey rules and prin-
lower in emotional stability (Klein and Kunda 1994). ciples in order to reduce self-conflict and restlessness (Lauriola and
Extraversion is closely related to excitement seeking, social Levin 2001). Individuals who are low in agreeableness tend to be
attention seeking, and other positive characteristics such as soci- more mistrustful and skeptical (McCrae and Costa 1989) and even
ability and talkativeness (Costa and McCrea 1992). Extraversion apply insulation against guilt or anxiety about negative results
is associated with behavioral exploration; highly extraverted indi- (Nicholson et al. 2005).
viduals have a tendency to control things, which may thereby trig- Related research has viewed conscientiousness as the degree
ger aggressive behavior (Nielsen and Knardahl 2015). Research to which an individual has the ability to perform better at work
showed that high extraversion supplies the motivational force for (Barrick and Mount 1991). The dependable and responsible facets
unsafe behavior (Nicholson et al. 2005). of conscientiousness reflect the extent to which one can be relied
Openness to experience is related to one’s perception of the upon (Zhao and Seibert 2006). Relevant characteristics, such as
world. Because of characteristics of imagination, curiosity, and being dutiful or industrious, have strong relationships with ration-
unconventionality, they may be more prone to rule violations, par- ality, implying that conscientious individuals are able to control
ticularly when working in a routinized environment (Clarke and themselves (McCrae and Costa 1989). This research proposes the
Robertson 2005). Those who score high on Openness to experience following hypothesis:
ignore risks and show risky behaviors for different reasons, pos- Hypothesis 2: Neuroticism, extraversion, and openness to ex-
sibly just because of their impulsiveness and willfulness. perience are positively related to risk propensity among construc-
Agreeableness reflects one’s interpersonal orientation, including tion workers. On the contrary, agreeableness and consciousness are
empathetic traits such as trust, understanding, sympathy, altruism, negatively related to risk propensity.

© ASCE 04020023-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Risk Propensity and Unsafe Behavioral Intention H4
The Big Five
Risk propensity refers to an individual’s attitude toward risk Risk propensity
(Rohrmann 2002). Related studies have demonstrated the impor- Neuroticism
tance of risk propensity as a consistent predictor of various risk H2 H3
behaviors (Zuckerman and Kuhlman 2010). Risk attitude is seen Extraversion
as an influential factor in the evaluation of whether an individual
chooses to take risks (MacCrimmon and Wehrung 1990). According Openness to Unsafe behavioral
to TPB, behavioral intention refers to the subjective probability that a Experience H1 intention
person will take some practical action, where the first influencing
factor is attitude toward the behavior. The attitude toward the behav- Agreeableness
ior reflects the extent to which a person evaluates whether an action
is beneficial (Ajzen 1991); the criterion for judging this is whether
Conscientiousness
the behavior meets one’s motivation or intention (Zhang and Fang
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

2013). For example, a study found that the potential gains and losses
determined by risk propensity affect an individual’s decision making.
Fig. 2. Theoretical model.
Based on these arguments, the following hypothesis is proposed:
Hypothesis 3: Construction workers’ risk propensity is posi-
tively associated with unsafe behavioral intention.
workload (Brust et al. 2016). A pilot study was performed to ensure
Mediating Effect of Risk Propensity the items in the questionnaire were scientific and practical for con-
struction workers, including observation, expert interviews, and
The Big Five personality traits have a huge impact on risk propensity some tips to reduce biases. Especially, questionnaires were catego-
(Nicholson et al. 2005), as well as behavior and motivation (Major rized as invalid if more than 15% of the responses failed to express
et al. 2006). Risk propensity is closely related to one’s behavioral the respondent’s views clearly in the distribution process.
intention or motivation (Zuckerman and Kuhlman 2010). Research
frameworks and models indicated that factors related to personality
traits are important predictors of unsafe behaviors through risk Construct Measurement
propensity (Williams et al. 2008). Measuring risk attitudes through
the questionnaire method has shown correlations with the Big Five Personal Particulars
personality traits and is correlated with risky behaviors (Lönnqvist Other personal particulars besides the three major factors, such as
et al. 2015). Prior research has established that personality traits, risk age, work experience, and company type, may have indirect effects
propensity, and unsafe behavioral intention are associated with each on risk propensity or behavioral intention (Ajzen 1991). Two types
other. However, it is unclear what role risk propensity plays between of personal particulars were considered in this study. The first type
the personality traits and unsafe behavioral intention of construction is demographic information, including gender, age, marital status,
workers. Based on the current research and the three aforementioned education, and family size. The second is occupational information,
hypotheses, the authors sought to examine the relationships between namely work experience, work type, company type, average daily
them, and posit the following hypothesis: working hours, and injury experience. Respondent characteristics
Hypothesis 4: Construction workers’ risk propensity has a are provided in Table 1.
mediating effect between the Big Five personality traits and unsafe
Big Five Personality Traits
behavioral intention.
The Big Five framework represents interindividual differences in
personality dimensions. Research has indicated that long question-
Theoretical Model naires are not always better than their diminutive versions, short
As discussed in the preceding sections, this study aims to (1) explore scales take less time and less labor, which is advantageous (Gosling
how construction workers’ personality traits affect their unsafe et al. 2003). Typically, three items per dimension is the minimum
behavioral intention, and (2) verify whether risk propensity medi- required for identification of the Big Five personality trait domains
ates the relationship between personality traits and unsafe behav- (Gagne and Hancock 2006; Marsh et al. 1998). Considering the
ioral intention. The proposed theoretical model is shown in Fig. 2. applicability for construction workers, this research adopted the
short 15-item version of the Big Five Inventory (BFI-S); there are
four reverse coded items.
Research Method
Risk Propensity
Self-administered questionnaires were distributed at construction Many scales have been developed to measure individuals’ risk
sites for data collection. propensity in different situations, including the Four-Domain
Scale (Rohrmann 2002), the Five-Domain Specific Risk-Taking
(Adult) Scale (Blais and Weber 2006), and the Six-Domain Scale
Questionnaire Structure
(Nicholson et al. 2005). This research referred these scales and
The self-administered questionnaire consisted of four parts: (1) per- adopted the Chinese version of the General Risk Propensity (GRP)
sonal particulars, (2) Big Five personality traits, (3) risk propensity, Scale (Hung et al. 2012) to measure risk propensity across multiple
and (4) unsafe behavioral intention. The personal particulars part decisions. In addition to measuring general risk propensity, the
contained 10 questions, including demographic and occupational GRP scale can also measure openness, problem-specific ambiguity,
information. The remainder of the questionnaire consisted of 39 general ambiguity tolerance, and social risk propensity, among
questions, and respondents were required to tick using a five-point others; this allows the scale to address different facets of the indi-
Likert scale (1 = strongly disagree to 5 = strongly agree); this con- vidual. The scale has four reverse coded items among the eight
densed questionnaire may have considerably reduced the mental items.

© ASCE 04020023-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Table 1. Personal particulars the individual level and safety culture at the organizational level.
Personal In the questionnaire, attitude toward the behavior were calculated
particulars Category Frequency Percentage by the pursuit of energy saving and risk perception. Subjective
norms were measured with respect to two aspects: influence of the
Gender Male 470 88.18
Female 63 11.82 manager and workmates. Perceived behavioral control was mea-
sured with respect to both self-efficacy and external conditions
Age group <20 75 14.07 (Wang 2014). The improved questionnaire contained 16 questions,
20–29 142 26.64 including six reverse scored questions. Specially, the statements
30–39 103 19.32 were improved instead of a five-point Likert scale (1 = strongly
40–49 76 14.26
disagree to 5 = strongly agree) in order to diminish social desirabil-
≥50 137 25.70
ity bias and acquiescence bias.
Marital status Single 128 24.02
Married 405 75.98
Data Collection and Analysis Method
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Education Primary school graduate 243 45.59


Junior middle school graduate 167 31.33
High school graduate 52 9.76 Questionnaire Distribution
Bachelor’s degree 48 9.01 A pilot study was conducted to optimize the version before formal
Master’s degree 23 4.32 distribution. The investigation was conducted at two specific con-
Family size 1 59 11.07 struction sites in Chongqing, China. According to observations and
2 66 12.38 interviews for 1 week, workers came from different backgrounds,
3 115 21.58 ranged in age mainly from 20 to 50 years old, and their attitudes
4 95 17.82 and usage about labor protection articles varied. However, different
5 107 20.08 types’ workers should wear homologous labor protection articles
≥6 91 17.07 (Wang 2014). These items included scaffolder’ antiskid shoes, car-
Work experience ≤3 109 20.45 penter’s safety belts, metal worker’ dust masks, and thus, “not prop-
(years) 4–10 124 23.26 erly wearing labor protection articles” was used as the content
11–15 92 17.26 according to six dimensions (part: unsafe behavioral intention).
16–20 69 12.95 In order to reduce social desirability bias, acquiescence bias, and
>20 139 26.08 increase the accuracy, improvements were made to make it appli-
Work type Metal worker 73 13.70 cable for the participants. For example, a self-administered ques-
Concreter 70 13.13 tionnaire was used to diminish social desirability bias (Rammstedt
Scaffolder 46 8.63 et al. 2004), and 14 reverse scored items were included to diminish
Plumber and electrician 56 10.51 acquiescence bias. Then, 256 questionnaires were distributed, 242
General worker 70 13.13 were returned, 37 were removed, and finally 205 valid ones were
Carpenter 55 10.32 used for the pilot study. Based on the feedback, structure and con-
Mason 52 9.76 tent were improved and the final version was developed. Factor
Painter 61 11.44
analysis in SPSS software requires the ratio between variables and
Other workers 50 9.38
samples to preferably be less than 1∶5. For the variables’ number
Company type State-owned enterprise 101 18.95 (39), estimated recovery rate (90%), and effective rate (80%) in the
Private enterprise 332 62.29 study, at least 271 sets of questionnaires should be issued. After a
Others 100 18.76 nationwide distribution over a 6 month period, including 18 con-
Average daily ≤5 27 5.07 struction sites, a total of 867 self-administered questionnaires were
working hours 6–9 31 5.82 collected in the form of centralized distribution and anonymous
10–13 240 45.03 collection through managers. Finally, 533 valid ones were obtained,
14–17 188 35.27 with an effective response rate of 61.5%. The main reasons that the
≥18 47 8.82 effective response rate did not reach a higher level may because of
Injury Yes 133 24.95 the education level, too many reverse questions, or the strict validity
experience No 400 75.05 conditions of the questionnaire.
Sum — 533 100 Data Analysis Method
SPSS version 22.0 was used to perform exploratory factor analysis
(EFA) in order to explore the proper structure and reliability of
the questionnaire in Chinese sample. AMOS software version 21.0
Unsafe Behavioral Intention was used for structural equation modeling (SEM) to run the hypoth-
In this study, the behavior of not properly wearing labor protection esized structural model, and confirmatory factor analysis (CFA)
articles was used to assess the workers’ unsafe behavioral intention. was performed to check the convergent validity and discriminant
Usually, it is achieved by asking several questions, like “In the next validity. Ten fitness indices were used to indicate the model fit
2 weeks, I will not wear labor protection articles because of my own degree. The relationships among personality traits, risk propensity,
habits or the requirement of the managers.” However, it is difficult and unsafe behavioral intention were identified using AMOS.
to obtain an accurate answer to a scenario that has not yet hap- A bootstrapping procedure was used to provide a robust test of
pened. This study revised and improved the questionnaire based the mediating effect hypothesis. The independent sample t-test and
on the influential factors of a Chinese scale from the cognitive per- one-way ANOVA method were used to test correlation between
spective through TPB (Wang 2014; Zhang 2012). The factors con- personal particulars with risk propensity and unsafe behavioral
sist of six dimensions that consider both behavior-based safety at intention.

© ASCE 04020023-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Results The measurement items included in each factor were homogenized
after promax rotation (Table 2), indicating that the scale had appro-
Preliminary Analysis priate construct validity.
First, EFA was performed using SPSS version 22.0. Principal com-
ponent analysis (PCA) with promax rotation was used to assess the Reliability and Validity
factorial structure of the 39 items of the questionnaire. In Table 2, All Cronbach’s alpha values given in Table 3 were above 0.8,
terms with a factor loading less than 0.6 and a cross loading greater which is greater than the recommended value of 0.70 (Peterson
than 0.35 should be deleted (Kolar and Zabkar 2010). The results 1994). Thus, the items had good internal consistency. In Table 3,
indicated that the 39 items could be classified into seven factors, the standardized loadings λ all exceeded 0.7, and sampling adequacy
which accounted for 75.336% of the total variance. Items loaded was greater than the minimum level of 0.60, with a significant t-value
on the first five factors corresponded to personality traits, the sixth (p < 0.001). The stricter acceptable standards of the composite reli-
factor corresponded to risk propensity, and the seventh factor ability (CR) test is 0.7 (Hair et al. 1998). Ideally, acceptable average
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

corresponded to unsafe behavioral intention. The Kaiser-Meyer- variance extracted (AVE) should exceed 0.5 (Hair et al. 1998). Table 3
Olkin (KMO) test indicated that all values of the independent and provides the results of the CR and AVE tests, which indicate that the
dependent variables achieved the criterion, being greater than model had acceptable internal consistency and convergent validity.
the minimum level of 0.60 (KMO ¼ 0.954). Further, other indices In addition, reasonable discriminant validity can be assumed
(χ2 ≈ 17; 861.088 and Significance ¼ 0.000) indicated that the when the root-mean square of the AVE value is greater than its cor-
data were appropriate to continue with the CFA procedure. relation coefficient with other latent variables (Hair et al. 1998).
Table 4 shows that the square root of the AVE of the construct was
greater than its estimates of correlation. The model had acceptable
discriminant validity.
Table 2. Results of exploratory factor analysis
Indicator I P O A C N E
Overall Model Fit Estimate
N1 −0.096 −0.022 −0.002 0.115 0.069 0.859 −0.037
N2 −0.091 −0.059 −0.032 0.280 0.092 0.843 −0.032 AMOS version 21.0 was applied to run the model; the impact
N3 −0.049 −0.076 0.000 0.359 0.129 0.796 −0.081 coefficients were estimated using the maximum likelihood esti-
E1 0.340 0.159 0.114 −0.007 −0.045 −0.077 0.801 mates method. Ten commonly used indices for goodness of fit were
E2 0.313 0.241 0.185 −0.090 0.017 0.042 0.761 employed to assess the overall model fit. The results are presented
E3 0.247 0.177 0.126 −0.091 −0.177 −0.135 0.776
O1 0.297 0.143 0.824 −0.035 −0.042 −0.022 0.123
in Table 5. The overall model fit was adequate, all indicators met
O2 0.283 0.128 0.831 −0.093 −0.047 −0.008 0.153 the standards, and the final 39-item questionnaire could be used for
O3 0.314 0.177 0.819 −0.026 −0.037 −0.005 0.105 further analysis.
A1 −0.139 −0.067 −0.054 0.795 0.098 0.330 −0.072
A2 −0.095 −0.093 −0.050 0.851 0.174 0.177 −0.035
A3 −0.106 −0.025 −0.048 0.849 0.155 0.227 −0.054 Hypotheses and Mediating Effect Test
C1 −0.197 −0.258 −0.055 0.204 0.830 0.060 −0.047 Based on good model fit, the hypotheses were analyzed. In Fig. 3
C2 −0.286 −0.276 −0.053 0.127 0.807 0.163 −0.104
and Table 6, it can be seen that extraversion and openness to
C3 −0.215 −0.260 −0.034 0.190 0.827 0.139 −0.064
P1 0.242 0.778 0.055 0.012 −0.062 −0.071 0.094 experience are positively related to unsafe behavioral intention
P2 0.270 0.810 0.116 −0.060 −0.170 0.002 0.130 (β ¼ 0.295 and p < 0.001; β ¼ 0.279 and p < 0.001), and consci-
P3 0.275 0.842 0.042 −0.054 −0.034 −0.069 0.113 entiousness is negatively related to unsafe behavioral intention
P4 0.272 0.774 0.138 −0.053 −0.231 −0.007 0.076 (β ¼ −0.177 and p < 0.001). The other two dimensions, neuroti-
P5 0.278 0.826 0.128 −0.029 −0.174 −0.012 0.108 cism and agreeableness, had no significant relationships with un-
P6 0.223 0.850 0.074 −0.059 −0.071 −0.053 0.108 safe behavioral intention, and thus H1 was partially supported.
P7 0.209 0.805 0.063 −0.042 −0.074 −0.049 0.029 Further, extraversion and openness to experience had positive ef-
P8 0.218 0.804 0.076 −0.039 −0.176 0.010 0.104 fects on risk propensity (β ¼ 0.264 and p < 0.001; β ¼ 0.171
I111 0.710 0.240 0.275 −0.048 −0.124 −0.079 0.151
I112 0.727 0.256 0.225 −0.073 −0.100 −0.041 0.163
and p < 0.001), whereas conscientiousness had a negative effect
I113 0.777 0.210 0.068 −0.071 −0.167 −0.015 0.203 (β ¼ −0.430 and p < 0.001). The other two dimensions, neuroti-
I114 0.765 0.140 0.173 −0.035 −0.100 −0.035 0.201 cism and agreeableness, had no significant relationship with risk
I121 0.723 0.340 0.233 −0.085 −0.181 −0.037 0.221 propensity, and thus H2 was partially supported. Moreover, risk
I122 0.735 0.260 0.219 −0.055 −0.064 −0.124 0.221 propensity was positively associated with unsafe behavioral inten-
I211 0.743 0.226 0.162 −0.075 −0.036 −0.035 0.108 tion (β ¼ 0.232 and p < 0.001); thus, H3 was supported.
I212 0.833 0.129 0.073 −0.067 −0.028 −0.026 0.025 In addition, bias-corrected and percentile bootstrapping methods
I213 0.825 0.141 0.050 −0.088 −0.101 −0.015 0.050 were used to test the mediating effect of risk propensity. Relevant
I221 0.805 0.148 0.079 0.026 −0.153 −0.075 0.025
scholars have found that bootstrapping provides a robust test of
I222 0.825 0.138 0.071 −0.048 −0.043 −0.028 0.084
I223 0.782 0.169 0.079 −0.048 −0.037 −0.073 0.113
mediating hypotheses (Cheung and Lau 2008). Subsequently, the
I311 0.763 0.209 0.104 −0.037 −0.047 −0.073 0.063 significant effect of personality traits on unsafe behavioral intention
I312 0.762 0.217 0.091 −0.108 −0.140 −0.001 0.149 was through risk propensity utilizing a SEM with bootstrapping of
I321 0.787 0.192 0.039 −0.043 −0.105 0.001 0.086 5,000 subsamples. In Table 7, because there was no zero between
I322 0.743 0.223 0.154 −0.050 −0.097 −0.059 0.086 the lower and upper limits of the 95% bias-corrected and percentile
Note: Maximum factor load in each dimension is denoted in bold font; I = bootstrap confidence intervals, the risk propensity mediates the
unsafe behavioral intention; P = risk propensity; O = openness to positive effects of extraversion and openness to experience on un-
experience; A = agreeableness; C = consciousness; N = neuroticism; and safe behavioral intention, and mediates the negative effect of con-
E = extraversion. scientiousness. Thus, H4 was partially supported.

© ASCE 04020023-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Table 3. Reliability and convergent validity
Unstandardized Standardized Standard Composite Cronbach’s Average variance
Construct Indicator loading loading error t-value reliability α extracted
Neuroticism N1 0.837 0.725 0.045 18.62*** 0.862 0.861 0.678
N2 0.968 0.861 0.043 22.727***
N3 1 0.875 — —
Extraversion E1 1.107 0.844 0.058 19.039*** 0.853 0.852 0.660
E2 1.131 0.821 0.061 18.662***
E3 1 0.771 — —
Openness to O1 1.043 0.837 0.045 23.214*** 0.890 0.889 0.729
experience O2 0.991 0.858 0.041 23.911***
O3 1 0.866 — —
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Agreeableness A1 1.081 0.835 0.049 21.894*** 0.874 0.872 0.698


A2 1.083 0.824 0.05 21.591***
A3 1 0.847 — —
Conscientiousness C1 0.973 0.863 0.036 27.009*** 0.917 0.916 0.785
C2 1.081 0.909 0.037 29.328***
C3 1 0.886 — —
Risk propensity P1 1 0.793 — — 0.954 0.953 0.721
P2 1.198 0.88 0.05 23.945***
P3 1.119 0.871 0.047 23.589***
P4 1.186 0.845 0.053 22.586***
P5 1.18 0.897 0.048 24.612***
P6 1.293 0.868 0.055 23.455***
P7 1.128 0.797 0.054 20.856***
P8 1.123 0.837 0.05 22.29***
Unsafe behavioral I111 1 0.808 — — 0.968 0.968 0.657
intention I112 1.038 0.819 0.046 22.422***
I113 1.034 0.839 0.044 23.239***
I114 1.038 0.814 0.047 22.235***
I121 0.952 0.865 0.039 24.355***
I122 0.826 0.831 0.036 22.916***
I211 0.773 0.788 0.036 21.216***
I212 0.871 0.807 0.04 21.968***
I213 0.921 0.815 0.041 22.294***
I221 0.819 0.803 0.038 21.823***
I222 0.877 0.813 0.04 22.2***
I223 0.927 0.79 0.043 21.322***
I311 0.762 0.779 0.036 20.896***
I312 0.868 0.816 0.039 22.328***
I321 0.763 0.792 0.036 21.399***
I322 0.846 0.789 0.04 21.268***
Note: t-value is obtained by bootstrapping (5000 times), and *** means p < 0.001.

Table 4. Discriminant validity


Construct 1 2 3 4 5 6 7
Unsafe behavioral intention 0.974 — — — — — —
Neuroticism −0.191** 0.823 — — — — —
Extraversion 0.547** −0.183** 0.812 — — — —
Conscientiousness −0.448** 0.316** −0.318** 0.866 — — —
Agreeableness −0.251** 0.554** −0.217** 0.410** 0.836 — —
Openness to experience 0.523** −0.093** 0.432** −0.244** −0.177** 0.854 —
Risk propensity 0.549** −0.150** 0.433** −0.492** −0.191** 0.365** 0.849
Mean value 2.227** 3.9149 2.1664 3.8762 3.9206 2.2908 2.7491
Standard deviation 0.89336 1.01238 0.86774 1.12654 1.09384 1.02142 1.24725
Note: *p < 0.05; **p < 0.01; and ***p < 0.001; the nondiagonal elements are the latent correlations and the diagonals (in bold) represent the square root of
average variance extracted (AVE).

Correlation of Personal Particulars p ¼ 0.001 < 0.01), and there is no gender difference in unsafe
SPSS was used to make correlation analysis of personal particulars, behavioral intention (t ¼ −1,940 and p ¼ 0.056 > 0.05). How-
and some interesting results were found, such as that male workers ever, there is no difference in risk propensity between single
have a higher risk propensity than female workers (t ¼ 3.291 and and married workers (t ¼ 1.720 and p ¼ 0.086 > 0.05), and single

© ASCE 04020023-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Table 5. Estimated values for goodness-of-fit indices adventurous, especially in the face of new circumstances. Likewise,
Fit index Recommended value Estimate it is comparatively easier for these workers to generate risky ideas,
and they tend to have the desire to be the “first to eat crab” when
X 2 =d:f: ≤0.3 2.151
facing an emergency on the construction site. Conscientious workers
GFI ≥0.8 0.874
AGFI ≥0.8 0.853
may have the opposite thoughts before action due to self-discipline
RMR ≤0.05 0.05 and rationality. Direct observations made on the construction site also
RMSEA ≤0.1 0.047 support these findings. In the distribution process, the authors saw an
CFI ≥0.9 0.956 extraverted scaffolder forgot to wear his safety belt as he talked to
NFI ≥0.9 0.921 others; a carpenter who scored high on openness to experience wore
TLI ≥0.9 0.952 slippers as he entered the worksite; and a conscientious concreter
PNFI ≥0.05 0.835 reminded his peers to put on their crash helmets before work.
PGFI ≥0.05 0.753 However, the results did not reveal significant relationships
Note: GFI = goodness-of-fit index; AGFI = adjusted goodness-of-fit index; among neuroticism, agreeableness, and unsafe behavioral inten-
tion. Some other studies have found correlations between them.
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

RMR = root mean square residual; RMSEA = root mean square error of
approximation; CFI = comparative fit index; NFI = norm fit index; TLI = A positive relationship has been found between neuroticism and
tacker-Lewis index; NNFI = non-normed fit index; PNFI = parsimony- risk behaviors; this may be due to its association with the anxiety
adjusted norm fit index; PGFI = parsimony goodness-of-fit index; and (Lauriola and Levin 2001). Those who score high on agreeableness
d.f. = degree of freedom. respect others’ opinions and prefer safer solutions to reduce their
uncomfortable feelings (Lauriola and Levin 2001). Individuals with
workers possess higher unsafe behavioral intention than married these two personality traits are easily affected by external factors
workers (t ¼ 2,901 and p ¼ 0.007 < 0.01). In addition, the con- when making decisions, and their relationships can be further stud-
struction workers’ risk propensity declines with the increase of ied in future in the construction industry.
age (except 20–29 and 40–49), their unsafe behavioral intention
declines with the increase of age group (except 20–29). Influence of Personality on Risk Propensity
H2 was also partially supported. As Nicholson et al. (2005) stated,
Discussion risk propensity is strongly rooted in personality. Extraverted work-
ers live in groups and are social; they have the need for excitement
and derive pleasure from it. Workers who score high on openness
Influence of Personality on Unsafe Behavioral Intention
to experience like to break traditions and pursue curiosity; these
H1 was partially supported. Construction workers with high levels of characteristics are closely associated with risk-seeking. However,
extraversion and openness to experience tend to be stimulating and construction workers with a higher level of conscientiousness

Adoption
Neuroticism
Rejection

Risk propensity
Extraversion
R2=0.405

Openness to
0.232***
Experience

-0.024 Unsafe behavioral


Agreeableness intention
R2=0.567

Conscientiousness

Note: *sig. at p 0.05; **sig. at p 0.01; **sig. at p 0.001

Fig. 3. Tests of hypotheses.

© ASCE 04020023-8 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Table 6. Analysis of hypotheses (n ¼ 533)
To risk propensity To unsafe behavioral intention
From Beta p-value Standard error Beta p-value Standard error
Neuroticism 0.011 0.841 0.058 0.004 0.935 0.047
Extraversion 0.264*** <0.001 0.071 0.295*** <0.001 0.061
Openness to experience 0.171*** <0.001 0.051 0.279*** <0.001 0.043
Agreeableness 0.077 0.187 0.064 −0.024 0.642 0.052
Conscientiousness −0.430*** <0.001 0.048 −0.177*** <0.001 0.041
Risk propensity — — — 0.232 <0.001 0.041
Note: *p < 0.05; **p < 0.01; and ***p < 0.001.
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Table 7. Mediating effect


Bootstrapping
Bias-corrected 95% Percentile 95%
Point Standard
Path Effects estimation error Lower Upper Lower Upper
Neuroticism to unsafe behavioral intention Total 0.007 0.054 −0.097 0.112 −0.095 0.113
Indirect 0.003 0.013 −0.024 0.030 −0.024 0.030
Direct 0.004 0.052 −0.097 0.106 −0.094 0.108
Extraversion to unsafe behavioral intention Total 0.356 0.056 0.249 0.466 0.249 0.466
Indirect 0.061 0.018 0.031 0.102 0.029 0.100
Direct 0.295 0.053 0.194 0.398 0.193 0.398
Openness to experience to unsafe behavioral intention Total 0.319 0.051 0.222 0.425 0.217 0.420
Indirect 0.04 0.015 0.016 0.076 0.015 0.073
Direct 0.279 0.049 0.190 0.387 0.182 0.378
Agreeableness to unsafe behavioral intention Total −0.006 0.06 −0.120 0.116 −0.124 0.112
Indirect 0.018 0.014 −0.008 0.050 −0.012 0.046
Direct −0.024 0.059 −0.138 0.096 −0.138 0.095
Conscientiousness to unsafe behavioral intention Total −0.277 0.06 −0.392 −0.154 −0.395 −0.157
Indirect −0.100 0.023 −0.151 −0.059 −0.147 −0.057
Direct −0.177 0.062 −0.295 −0.056 −0.298 −0.059

usually have strong self-management or control. They show a Mediating Effect of Risk Propensity
strong sense of responsibility and are more trustworthy, and it
H4 was also partially supported. Workers’ risk propensity acts as a
is reasonable that they generally avoid risks when making choices.
bridge and bond between personality traits and unsafe behavioral
There were no significant associations among neuroticism, agree-
intention. Risk propensity is rooted in personality traits, which are
ableness, and risk propensity found. Because neurotic individuals determined by individual genes and early life experience; it is likely
have a poor ability to control emotions, they tend to be nervous, one of the strongest predictors of unsafe behavior (Nicholson et al.
anxious, and lack security. Agreeable individuals can be either 2005). Risk propensity is also a determinant of risky behavior
sympathetic and trustworthy or ruthless and hostile; thus, there (Sitkin and Weingart 1995). Attitude has been found to moderate
is a duality in the traits of agreeableness. In a complex environment intention–behavior relations, and these moderating variables pro-
such as a construction site, the risk propensity of construction duce a “strong” intention (Sheeran and Abraham 2003). Thus, the
workers may depend on the actual situations. Further, with an personality traits of extraversion and openness to experience in-
increased sample size, the relationships between variables may crease unsafe behavioral intention through risk propensity, whereas
be more likely to be significant. conscientiousness decreases unsafe behavioral intention through
risk propensity. To assess the unsafe behavioral intention of con-
Influence of Risk Propensity on Unsafe Behavioral struction workers, one not only needs to pay attention to their per-
Intention sonality traits, but also their risk propensity.
However, no evidence of significant relationships were found for
This study’s findings support H3. Construction workers’ risk pro- neuroticism and agreeableness on risk propensity or unsafe behav-
pensity had a significant positive influence on their unsafe behav- ioral intention; there is no mediating effect between them. This may
ioral intention. Risk propensity refers to an individual’s willingness due to the size or the intrinsic characteristics of the sample, which
to take risks, and it has a large influence on decision making requires further study.
(Nicholson et al. 2005). Individuals’ unsafe behavior results from
conscious thought, which is influenced by attitudes; it can be mea-
sured by unsafe behavioral intention (Reason 1990), which is a Conclusions
direct predictor of behavior (Ajzen 1991). Thus, construction
workers’ risk propensity and unsafe behavioral intention are con- This research examined the relationships among personality traits,
sistently related. risk propensity, and unsafe behavioral intention of construction

© ASCE 04020023-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


workers. Incorporating the Big Five personality model and TPB, a Aksorn, T., and B. Hadikusumo. 2007. “The unsafe acts and the decision-
theoretical model was built among the constructs. Results showed to-err factors of Thai construction workers.” J. Constr. Developing
construction workers’ personality traits, risk propensity, and unsafe Countries 12 (1): 1–25.
behavioral intention were related to some extent. Extraversion and Arthur, W., and W. G. Graziano. 1996. “The five-factor model, conscien-
openness to experience were positively associated with unsafe tiousness, and driving accident involvement.” J. Personality 64 (3):
593–618. https://doi.org/10.1111/j.1467-6494.1996.tb00523.x.
behavioral intention and risk propensity. Conscientiousness was
Austin, J., M. L. Kessler, J. E. Riccobono, and J. Bailey. 1996. “Using feed-
negatively associated with unsafe behavioral intention and risk pro-
back and reinforcement to improve the performance and safety of a
pensity. There were no significant effect of neuroticism or agree- roofing crew.” J. Organizational Behav. Manage. 16 (2): 49–75.
ableness on unsafe behavioral intention and risk propensity. Risk https://doi.org/10.1300/J075v16n02_04.
propensity had a positive effect on unsafe behavioral intention. Bandura, A. 1991. “Social cognitive theory of self-regulation.” Organiza-
Thus, risk propensity had a partial mediating effect on the relation- tional Behav. Hum. Decis. Processes 50 (2): 248–287. https://doi.org
ship between personality traits (extraversion, openness to experi- /10.1016/0749-5978(91)90022-L.
ence, and conscientiousness) and unsafe behavioral intention. Barrick, M. R., and M. K. Mount. 1991. “The Big Five personality dimen-
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

These findings will not only contribute to the literature on sions and job performance: A meta-analysis.” Personnel Psychol.
behavior-based safety and safety culture but will also provide some 44 (1): 1–26. https://doi.org/10.1111/j.1744-6570.1991.tb00688.x.
practical significance for scientifically predicting the occurrence of Blais, A. R., and E. U. Weber. 2006. “A domain-specific risk-taking
unsafe behavior or accidents. In terms of theory, this study enriches (DOSPERT) scale for adult populations.” Judgment Decis. Making
theory on risk management and safety management in the construc- 1 (1): 33–47.
tion industry and provides new research insight into construction Brust, O. A., S. Häder, and M. Häder. 2016. “Is the short version of the
safety, promoting multidisciplinary research. On the practical level, big five inventory (BFI-S) applicable for use in telephone surveys?”
construction workers can improve their safety awareness and safety J Off. Stat. 32 (3): 601–618. https://doi.org/10.1515/jos-2016-0031.
Cellar, D. F., Z. C. Nelson, C. M. Yorke, and C. Bauer. 2001. “The five-
behavior by knowing themselves better, and managers can improve
factor model and safety in the workplace: Investigating the relationships
safety management performance by effectively understanding their
between personality and accident involvement.” J. Prev. Intervention
subordinates. Community 22 (1): 43–52. https://doi.org/10.1080/108523501095
This research also has several limitations. First, personality traits 11210.
were significantly correlated with age, gender, education, and other Cheung, G. W., and R. S. Lau. 2008. “Testing mediation and suppression
traits. This study used BFI-S to measure personality traits without effects of latent variables: Bootstrapping with structural equation mod-
further analysis of how these factors affect the selection of question- els.” Organizational Res. Methods 11 (2): 296–325. https://doi.org/10
naire version. Second, the GRP scale was originally designed to .1177/1094428107300343.
measure the risk propensity of managers’ decisions, but this study Choudhry, R. M., D. Fang, and S. Mohamed. 2007. “Developing a model of
used the complete scale to measure the general risk propensity of construction safety culture.” J. Manage. Eng. 23 (4): 207–212. https://
construction workers. Although all the items have also obtained rea- doi.org/10.1061/(ASCE)0742-597X(2007)23:4(207).
sonable validity, more surveys are required to test the differences Clarke, S., and I. T. Robertson. 2005. “A meta-analytic review of the Big
between the general (Item 1) and other categories (Items 2–8) of Five personality factors and accident involvement in occupational and
risk propensity items in BFI-S. In addition, this research examined non-occupational settings.” J. Occup. Organizational Psychol. 78 (3):
relationships among personal particulars, risk propensity, and unsafe 355–376. https://doi.org/10.1348/096317905X26183.
Cooper, M. D., R. A. Phillips, V. J. Sutherland, and P. J. Makin. 1994.
behavioral intention. Future studies should examine these relation-
“Reducing accidents using goal setting and feedback: A field study.”
ships and causes in greater depth.
J. Occup. Organizational Psychol. 67 (3): 219–240. https://doi.org/10
.1111/j.2044-8325.1994.tb00564.x.
Costa, P. T., and R. R. McCrea. 1992. Revised neo personality inventory
Data Availability Statement (NEO PI-R) and neo five-factor inventory (NEO-FFI). Odessa, FL:
Psychological Assessment Resources.
Some or all data, models, or code generated or used during the Dai, X. Y., and Y. Q. Wu. 2005. “A study on NEO-PI-R used in 16∼20 years
study are available from the corresponding author by request. old people.” Chin. J. Clin. Psychol. 13 (1): 14–18. https://doi.org/10
.16128/j.cnki.1005-3611.2005.01.004.
Deyoung, C. G. 2015. “Cybernetic Big Five theory.” J. Res. Personality
Acknowledgments 56 (Jun): 33–58. https://doi.org/10.1016/j.jrp.2014.07.004.
Dohmen, T., A. Falk, D. Huffman, U. Sunde, J. Schupp, and G. G. Wagner.
This research was supported by the Fundamental Research Funds 2011. “Individual risk attitudes: Measurement, determinants, and
for the Central Universities (Grant Nos. 2019CDSKXYJSG0041, behavioral consequences.” J. Eur. Econ. Assoc. 9 (3): 522–550. https://
2017CDJSK03XK19, and 2019CDJSK03XK24). doi.org/10.1111/j.1542-4774.2011.01015.x.
Fang, D., C. Wu, and H. Wu. 2015. “Impact of the supervisor on worker
safety behavior in construction projects.” J. Manage. Eng. 31 (6):
04015001. https://doi.org/10.1061/(ASCE)ME.1943-5479.0000355.
Supplemental Data Fogarty, G. J., and A. Shaw. 2010. “Safety climate and the theory of
planned behavior: Towards the prediction of unsafe behavior.” Accid.
Survey data are available online in the ASCE Library (www Anal. Prev. 42 (5): 1455–1459. https://doi.org/10.1016/j.aap.2009.08
.ascelibrary.org). .008.
Gagne, P., and G. R. Hancock. 2006. “Measurement model quality,
sample size, and solution propriety in confirmatory factor models.”
References Multivariate Behav. Res. 41 (1): 65–83. https://doi.org/10.1207/s15327
906mbr4101_5.
Ajzen, I. 1991. “The theory of planned behavior.” Organizational Behav. Ghosh, D., and M. R. Ray. 1997. “Risk, ambiguity, and decision choice:
Hum. Decis. Processes 50 (2): 179–211. https://doi.org/10.1016/0749 Some additional evidence.” Decis. Sci. 28 (1): 81–104. https://doi.org
-5978(91)90020-T. /10.1111/j.1540-5915.1997.tb01303.x.

© ASCE 04020023-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Goldberg, L. R. 1990. “An alternative “description of personality”: The Kowert, P. A., and M. G. Hermann. 1997. “Who takes risks? Daring and
Big-Five factor structure.” J. Personality Social Psychol. 59 (6): 1216. caution in foreign policy making.” J. Conflict Resolut. 41 (5): 611–637.
https://doi.org/10.1037/0022-3514.59.6.1216. https://doi.org/10.1177/0022002797041005001.
Gosling, S. D., P. J. Rentfrow, and W. B. Swann. 2003. “A very brief mea- Lang, F. R., D. John, O. Lüdtke, J. Schupp, and G. G. Wagner. 2011. “Short
sure of the Big-Five personality domains.” J. Res. Personality 37 (6): assessment of the Big Five: Robust across survey methods except
504–528. https://doi.org/10.1016/S0092-6566(03)00046-1. telephone interviewing.” Behav. Res. Methods 43 (2): 548–567. https://
Hair, J. F., R. E. Anderson, R. L. Tatham, and W. C. Black. 1998. Multi- doi.org/10.3758/s13428-011-0066-z.
variate data analysis. Upper Saddle River, NJ: Prentice Hall. Lauriola, M., and I. P. Levin. 2001. “Personality traits and risky decision-
Haslam, R. A., S. A. Hide, A. G. Gibb, D. E. Gyi, T. Pavitt, S. Atkinson, making in a controlled experimental task: An exploratory study.” Per-
and A. R. Duff. 2005. “Contributing factors in construction accidents.” sonality Individual Differences 31 (2): 215–226. https://doi.org/10.1016
Appl. Ergon. 36 (4): 401–415. https://doi.org/10.1016/j.apergo.2004.12 /S0191-8869(00)00130-6.
.002. Li, H., L. Chen, H. Pei, F. G. Zhang, B. Zhang, and W. Y. Ma. 2003. “A
Heinrich, H. W., D. C. Petersen, N. R. Roos, and S. Hazlett. 1980. Indus- study on correlation of personality traits to employees’ stress evaluation.”
trial accident prevention: A safety management approach. New York: Chin. J. Clin. Psychol. 11 (4): 270–272. https://doi.org/10.16128/j.cnki
McGraw-Hill. .1005-3611.2003.04.009.
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Hogan, J., and D. S. Ones. 1997. “Conscientiousness and integrity at Li, H., M. Lu, S. C. Hsu, M. Gray, and T. Huang. 2015. “Proactive behavior-
work.” Chap. 32 in Handbook of personality psychology, 849–870. based safety management for construction safety improvement.” Saf. Sci.
Amsterdam, Netherlands: Elsevier. 75: 107–117. https://doi.org/10.1016/j.ssci.2015.01.013.
Hough, L. M. 1992. “The “Big Five” personality variables–construct Li, J. D. 2013. “Psychometric properties of ten-item personality inventory
confusion: Description versus prediction.” Hum. Perform. 5 (1–2): in China.” China J. Health Psychol. 21 (11): 1688–1692. https://doi.org
139–155. https://doi.org/10.1080/08959285.1992.9667929. /10.13342/j.cnki.cjhp.2013.11.008.
Howard, S. F. 2011. Personality: Classic theories and modern research. Lönnqvist, J. E., M. Verkasalo, G. Walkowitz, and P. C. Wichardt. 2015.
Beijing: China Machine Press. “Measuring individual risk attitudes in the lab: Task or ask? An empiri-
Hu, X. J., and C. Wu. 2009. “Review on the studying methodologies cal comparison.” J. Econ. Behav. Organ. 119 (Nov): 254–266. https://
of human safety psychological characteristics.” China Saf. Sci. J. doi.org/10.1016/j.jebo.2015.08.003.
19 (7): 5–13. https://doi.org/10.16265/j.cnki.issn1003-3033.2009.07 Luo, J., Y. Zhou, W. Chen, Y. Pan, and S. Zhao. 2016. “A reliability gen-
.003. eralization of the Big-Five factor personality tests in China.” Int. J.
Hung, K., C. Tangpong, J. Li, and Y. Li. 2012. “Robustness of general risk Psychol. 32 (1): 121–128. https://doi.org/10.16187/j.cnki.issn1001
propensity scale in cross-cultural settings.” J. Managerial Issues 24 (1): -4918. 2016. 01.16.
78–96. MacCrimmon, K. R., and D. A. Wehrung. 1990. “Characteristics of risk
Hyoung, J. I., Y. Kwon, S. G. Kim, Y. S. Ju, and H. P. Lee. 2009. “The taking executives.” Manage. Sci. 36 (4): 422–435. https://doi.org/10
characteristics of fatal occupational injuries in Korea’s construction .1287/mnsc.36.4.422.
industry, 1997–2004.” Saf. Sci. 47 (8): 1159–1162. https://doi.org/10 Major, D. A., J. E. Turner, and T. D. Fletcher. 2006. “Linking proactive
.1016/j.ssci.2008.11.008. personality and the Big Five to motivation to learn and development
Jensen, J. M., and P. C. Patel. 2011. “Predicting counterproductive work activity.” J. Appl. Psychol. 91 (4): 927. https://doi.org/10.1037/0021
behavior from the interaction of personality traits.” Personality Individ- -9010.91.4.927.
ual Differences 51 (4): 466–471. https://doi.org/10.1016/j.paid.2011.04 Marsh, H. W., K. Hau, J. R. Balla, and D. Grayson. 1998. “Is more ever
.016. too much? The number of indicators per factor in confirmatory factor
John, O. P., L. P. Naumann, and C. J. Soto. 2008. “Paradigm shift to the analysis.” Multivariate Behav. Res. 33 (2): 181–220. https://doi.org/10
integrative Big Five trait taxonomy.” In Vol. 3 of Handbook of person- .1207/s15327906mbr3302_1.
ality: Theory and research, 114–158. New York: Guilford Press. McCarthy, B. 2003. “Strategy is personality-driven, strategy is crisis-
John, O. P., and S. Srivastava. 1999. “The Big Five trait taxonomy: History, driven: Insights from entrepreneurial firms.” Manage. Decis. 41 (4):
measurement, and theoretical perspectives.” In Vol. 2 of Handbook of 327–339. https://doi.org/10.1108/00251740310468081.
personality: Theory and research, 102–138. New York: Guilford Press. McCrae, R. R., and P. T. Costa. 1989. “More reasons to adopt the five-factor
Jones, S., C. Kirchsteiger, and W. Bjerke. 1999. “The importance of near model.” Am. Psychologist 44 (2): 451–452. https://doi.org/10.1037
miss reporting to further improve safety performance.” J. Loss Prev. /0003-066X.44.2.451.
Process Ind. 12 (1): 59–67. https://doi.org/10.1016/S0950-4230(98) McCrae, R. R., and P. T. Costa. 1991. “Adding Liebe und Arbeit: The full
00038-2. five-factor model and well-being.” Personality Social Psychol. Bull.
Ju, J., G. S. Yang, and P. Yang. 2013. “Study on influence factor and control 17 (2): 227–232. https://doi.org/10.1177/014616729101700217.
measures for unsafe behavior of construction workers.” J. Saf. Sci. Mitropoulos, P., G. Cupido, and M. Namboodiri. 2009. “Cognitive approach
Technol. 9 (11): 179–184. https://doi.org/10.11731/j.issn.1673-193x to construction safety: Task demand-capability model.” J. Constr. Eng.
.2013.11.032. Manage. 135 (9): 881–889. https://doi.org/10.1061/(ASCE)CO.1943
Kahn, R. L., and P. Byosiere. 1992. “Stress in organizations.” In Handbook -7862.0000060.
of industrial and organizational psychology, edited by M. D. Dunnette Moore, S. M., W. L. Porter, and P. G. Dempsey. 2009. “Fall from equipment
and L. M. Hough, 571–650. Palo Alto, CA: Consulting Psychologists injuries in U.S. mining: Identification of specific research areas for fu-
Press. ture investigation.” J. Saf. Res. 40 (6): 455–460. https://doi.org/10.1016
Kahneman, D., and A. Tversky. 1979. “Prospect theory: An analysis of /j.jsr.2009.10.002.
decision under risk.” Econometrica 47 (2): 263–291. https://doi.org/10 Nicholson, N., E. Soane, M. Fenton-O’Creevy, and P. Willman. 2005.
.2307/1914185. “Domain specific risk taking and personality.” J. Risk Res. 8 (2):
Kines, P., L. P. Andersen, S. Spangenberg, K. L. Mikkelsen, J. Dyreborg, 157–176. https://doi.org/10.1080/1366987032000123856.
and D. Zohar. 2010. “Improving construction site safety through leader- Nie, L. Q. 2017. Study on emergency management level evaluation of safety
based verbal safety communication.” J. Saf. Res. 41 (5): 399–406. emergency in construction project. Chongqing, China: Chongqing
https://doi.org/10.1016/j.jsr.2010.06.005. Univ.
Klein, W. M., and Z. Kunda. 1994. “Exaggerated self-assessments and the Nielsen, M. B., and S. Knardahl. 2015. “Is workplace bullying related to the
preference for controllable risks.” Organizational Behav. Hum. Decis. personality traits of victims? A two-year prospective study.” Work Stress
Processes 59 (3): 410–427. https://doi.org/10.1006/obhd.1994.1067. 29 (2): 128–149. https://doi.org/10.1080/02678373.2015.1032383.
Kolar, T., and V. Zabkar. 2010. “A consumer-based model of authenticity: Nordlöf, H., B. Wiitavaara, U. Winblad, K. Wijk, and R. Westerling. 2015.
An oxymoron or the foundation of cultural heritage marketing?” “Safety culture and reasons for risk-taking at a large steel-manufacturing
Tourism Manage. 31 (5): 652–664. https://doi.org/10.1016/j.tourman company: Investigating the worker perspective.” Saf. Sci. 73 (Mar):
.2009.07.010. 126–135. https://doi.org/10.1016/j.ssci.2014.11.020.

© ASCE 04020023-11 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023


Peterson, R. A. 1994. “A meta-analysis of Cronbach’s coefficient alpha.” Wang, C. M., B. B. Xu, S. J. Zhang, and Y. Q. Chen. 2016. “Influence of
J. Consum. Res. 21 (2): 381–391. https://doi.org/10.1086/209405. personality and risk propensity on risk perception of Chinese construc-
Rammstedt, B., B. Holzinger, and T. Rammsayer. 2004. “Zur Äquivalenz tion project managers.” Int. J. Project Manage. 34 (7): 1294–1304.
der Papier-Bleistift- und einercomputergestützten Version des NEO- https://doi.org/10.1016/j.ijproman.2016.07.004.
Fünf-Faktoren-Inventars (NEO-FFI).” Diagnostica 50 (2): 88–97. Wang, S. Q. 2014. Study on construction workers’ unsafe behavior on
https://doi.org/10.1026/0012-1924.50.2.88. construction site. Wuhan, China: Huazhong Univ. of Science and
Reason, J. 1990. Human error. Cambridge, UK: Cambridge University Technology.
Press. Weber, E. U., A. R. Blais, and N. E. Betz. 2002. “A domain-specific
Rigby, L. V. 1970. “Nature of human error.” Ann. Surg. 244 (5): 642–648. risk-attitude scale: Measuring risk perceptions and risk behaviors.”
Rohrmann, B. 2002. Risk attitude scales: Concepts and questionnaires, 12. J. Behav. Decis. Making 15 (4): 263–290. https://doi.org/10.1002
Melbourne, Australia: Univ. of Melbourne. /bdm.414.
Sanmiquel, L., M. Freijo, J. Edo, and J. M. Rossell. 2010. “Analysis of work Wen Lim, H., N. Li, D. Fang, and C. Wu. 2018. “Impact of safety climate
related accidents in the Spanish mining sector from 1982-2006.” J. Saf. on types of safety motivation and performance: Multigroup invariance
Res. 41 (1): 1–7. https://doi.org/10.1016/j.jsr.2009.09.008. analysis.” J. Manage. Eng. 34 (3): 04018002. https://doi.org/10.1061
Saucier, G. 1994. “Mini-markers: A brief version of Goldberg’s unipolar /(ASCE)ME.1943-5479.0000595.
Downloaded from ascelibrary.org by University of Liverpool on 02/16/20. Copyright ASCE. For personal use only; all rights reserved.

Big-Five markers.” J. Personality Assess. 63 (3): 506. https://doi.org/10 Wiegmann, D., and S. Shappell. 2003. A human error approach to aviation
.1207/s15327752jpa6303_8.
accident analysis. London: Taylor and Francis Group.
Shappell, S. A., and D. A. Wiegmann. 2001. “Applying reason: The human
Williams, S., M. Zainuba, and R. Jackson. 2008. “Determinants of mana-
factors analysis and classification system (HFACS).” Gastroenterol.
gerial risk perceptions and intentions.” J. Manage. Res. 8 (2): 59–75.
Res. 1 (5): 207–212.
Zhang, B. 2018. “Brief talk on common collapse accidents in construction
Sheeran, P., and C. Abraham. 2003. “Mediator of moderators: Temporal
and solutions.” Constr. Saf. 33 (1): 4–8.
stability of intention and the intention-behavior relation.” Personality
Social Psychol. Bull. 29 (2): 205–215. https://doi.org/10.1177 Zhang, J. S., Y. Xu, and Q. Wang. 2015. “Analysis of the difference
/0146167202239046. between the actual and legal industrial injury death compensation in
Shen, Y. Y., X. H. Xu, K. L. Xu, and X. L. Ju. 2019. “Statistical analysis of construction industry.” Ind. Saf. Environ. Prot. 41 (8): 67–69.
construction accidents based on cluster analysis.” Constr. Saf. 34 (5): Zhang, M., and D. Fang. 2013. “A cognitive analysis of why Chinese
66–69. scaffolders do not use safety harnesses in construction.” Constr.
Sitkin, S. B., and L. R. Weingart. 1995. “Determinants of risky decision- Manage. Econ. 31 (3): 207–222. https://doi.org/10.1080/01446193.2013
making behavior: A test of the mediating role of risk perceptions and .764000.
propensity.” Acad. Manage. J. 38 (6): 1573–1592. https://doi.org/10 Zhang, M. C. 2012. Cognitive mechanism of construction worker’s unsafe
.5465/256844. behaviors and its application. Beijing: Tsinghua Univ.
Smith, P., H. Kincannon, R. Lehnert, Q. Wang, and M. D. Larrañaga. 2013. Zhao, H., and S. E. Seibert. 2006. “The Big Five personality dimensions
“Human error analysis of the Macondo well blowout.” Process Saf. and entrepreneurial status: A meta-analytical review.” J. Appl. Psychol.
Prog. 32 (2): 217–221. https://doi.org/10.1002/prs.11604. 91 (2): 259–271. https://doi.org/10.1037/0021-9010.91.2.259.
Soane, E., and N. Nicholson. 2009. “Individual differences and decision Zuckerman, M., and D. M. Kuhlman. 2010. “Personality and risk-taking:
making.” In The Oxford handbook of organizational decision making. Common bisocial factors.” J. Personality 68 (6): 999–1029. https://doi
Oxford, UK: Oxford University Press. .org/10.1111/1467-6494.00124.

© ASCE 04020023-12 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2020, 146(4): 04020023

You might also like