You are on page 1of 11

JSR 1614 No.

of Pages 11, Model 5G


13 November 2019

Journal of Safety Research xxx (xxxx) xxx


1

Contents lists available at ScienceDirect

Journal of Safety Research


journal homepage: www.elsevier.com/locate/jsr

5
6

3 Impact of psychological capital on construction worker safety behavior:


4 Communication competence as a mediator
7 Changquan He a,b,⇑, Guangshe Jia a, Brenda McCabe b, Yuting Chen c, Jide Sun a
8 a
School of Economics and Management, Tongji University, Shanghai, China
9 b
Construction Engineering & Management Group, Department of Civil Engineering, University of Toronto, Toronto, Canada
10 c
Department of Engineering Technology and Construction Management, University of North Carolina at Charlotte, Charlotte, North Carolina, USA

12
11
a r t i c l e i n f o a b s t r a c t
1
2 4
7
15 Article history: Introduction: Construction workers face a work environment of high risk and mental stress. Psychological 28
16 Received 14 February 2019 capital (PsyCap) could influence employee’s mental health and work performance. It would be helpful to 29
17 Received in revised form 17 May 2019 determine whether PsyCap affects worker safety behavior. However, few studies empirically examined 30
18 Accepted 28 September 2019
the impacts of the sub-dimensions of PsyCap on the safety behavior in construction settings, reducing 31
19 Available online xxxx
the potential practicability of PsyCap to improve workplace safety performance. Thus, this study tested 32
the relationship between sub-dimensions of PsyCap (self-efficacy, hope, resilience, optimism) and safety 33
20 Keywords:
behaviors (safety compliance, safety participation), while the mediating role of communication compe- 34
21 Psychological capital
22 Safety behavior
tence was also explored. 35
23 Communication competence Method: Data were collected from 655 construction workers in China using a psychological capital ques- 36
24 Construction industry tionnaire (PCQ). The theoretical model were tested with confirmatory factor analyses (CFA) and structural 37
25 Structural equation modeling equation modeling (SEM) techniques. 38
26 Results: Results show that: (a) the self-efficacy dimension of PsyCap positively affected safety compliance 39
and safety participation, while the resilience dimension positively impacted safety participation; (b) the 40
hope dimension was not directly related to safety behaviors, while the optimism dimension negatively 41
associated with safety participation; and (c) communication competence mediated the relationships 42
between the hope and optimism dimensions of PsyCap and safety participation. 43
Conclusions: A multidimensional perspective on PsyCap should be taken while examining its effects on 44
safety behavior and the individual communication competence helps to enhance construction safety. 45
Practical applications: Findings of this study shed lights on safety behavior promotion practices based on 46
the multidimensional model. Initiating flexible psychological capital training and intervention, and 47
strengthening communication skills of construction employees are suggested to improve safety perfor- 48
mance in the construction industry. 49
Ó 2019 Published by Elsevier Ltd. 50
51

52
53
54 1. Introduction world, construction safety performance seems to be approaching 62
a plateau (Chen, McCabe, & Hyatt, 2017a; Lingard, Cooke, & 63
55 The construction industry is one of the most dangerous indus- Blismas, 2010; Statistics New Zealand, 2014) with little improve- 64
56 tries as evidenced by its high rate of safety accidents and injuries ment over the last decade. These situations call for broadening 65
57 (Fang & Wu, 2013; U.S. Bureau of Labor Statistics. (2012), 2012). the boundary of safety research to include interdisciplinary per- 66
58 In 2017, there were 807 fatalities deriving from 692 accidents in spectives and focus on worker safety behavior, because unsafe 67
59 the Chinese construction industry, representing an increase of worker behavior contributes to more than 80% of accidents (Choi, 68
60 9.15% and 9.80%, respectively, over the same period in 2016 Ahn, & Lee, 2017; Willamson & Feyer, 1990). 69
61 (MOHURD, 2018). This situation is not unique. Throughout the A set of organizational, physical and psychological factors may 70
influence the cognitive mechanism of safety behavior. As a core 71

⇑ Corresponding author at: School of Economics and Management, Tongji concept of positive organizational behavior, psychological capital 72

University, 1 Zhangwu Road, Shanghai, China. (PsyCap) is attracting attention in safety science (Brunetto et al., 73
E-mail addresses: changquan.he@mail.utoronto.ca (C. He), jiagsh803@tongji. 2016; Wang, Wang, & Xia, 2018). It has been widely applied to 74
edu.cn (G. Jia), brenda.mccabe@utoronto.ca (B. McCabe), ychen106@uncc.edu study its relationship with employee behavior, work engagement, 75
(Y. Chen), jdsun126@126.com (J. Sun).

https://doi.org/10.1016/j.jsr.2019.09.007
0022-4375/Ó 2019 Published by Elsevier Ltd.

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

2 C. He et al. / Journal of Safety Research xxx (xxxx) xxx

76 and work performance (Avey et al., 2010; Chen, 2018; Madrid, of focusing on the asset, risk, and process. Optimism is an explana- 140
77 Diaz, Leka, Leiva, & Barros, 2018). While PsyCap was treated as a tory style for someone attributing positive things to inner, ever- 141
78 multidimensional construct in many studies (Avey et al., 2010; lasting, and common factors. It can be enhanced by forgiving the 142
79 Hodges, 2010; Luthans, Luthans, & Luthans, 2004), few had empir- past, correctly evaluating the present, looking for opportunities 143
80 ically confirmed the impacts of its sub-dimensions on worker in the future, maintaining a pragmatic viewpoint, and having flex- 144
81 behavior in the safety domain, especially in construction settings. ible cognition. In this study, we use the Luthans et al. (2004) four- 145
82 Further, construction workers face a work environment of high risk dimensional concept of PsyCap. 146
83 and pressure, which might affect their psychological state. Thus, it
84 would be helpful to determine whether PsyCap influences their 1.2. Safety behavior 147
85 safety behavior and how this might be used to improve safety
86 performance. Safety behavior is the safety related actions performed by indi- 148
87 Using the meta-analysis approach, an integrative model linking viduals in an organization. As a leading indicator of safety perfor- 149
88 situation-related factors (e.g., safety climate) and person-related mance (Hinze, Thurman, & Wehle, 2013), safety behavior has 150
89 factors (e.g., personality characteristics) with safety behavior indi- some advantages over lagging indicators, such as injuries and fatal- 151
90 cated that the consciousness dimension of personality characteris- ities (Chen, McCabe, & Hyatt, 2017b; Cooper, 2009; McCabe, 152
91 tics indirectly impacted safety behavior via safety motivation and Alderman, Chen, Hyatt, & Shahi, 2017). Safe behavior data are more 153
92 safety knowledge (Christian, Bradley, Wallace, & Burke, 2009). likely to have a normal distribution, thus the relationships with its 154
93 However, more empirical research is needed to confirm the effect antecedents are easier to analyze (Christian et al., 2009; Zohar, 155
94 of sub-dimensions of PsyCap other than consciousness on safety 2000), easier to predict, and more suitable to be used for safety 156
95 behavior. Another theoretical model included the relationship evaluation and intervention (Guo & Yiu, 2015). Safety behavior 157
96 between authentic leadership, safety climate, PsyCap and safety has two dimensions (Griffin & Neal, 2000): safety compliance 158
97 outcomes, but this model was not empirically tested (Eid, and safety participation. Safety compliance refers to in-role safety 159
98 Mearns, Larsson, Laberg, & Johnsen, 2012). As such, the impact related actions, such as observing safety regulations and complying 160
99 mechanism of PsyCap on safety behavior remains to be further with safety instructions. Safety participation belongs to out-role 161
100 explored from the multidimensional perspective. safety related behaviors, which are voluntary. It includes the will- 162
101 To help fill this gap, this study examined the impacts of sub- ingness of employees to help coworkers, actively participate in 163
102 dimensions of PsyCap on safety behavior with the data collected safety training activities, and put forward safety suggestions 164
103 from Chinese construction workers. The mediating role of individ- (Clarke & Ward, 2006; Neal, Griffin, & Hart, 2000). 165
104 ual communication competence was also checked, which may Both safety compliance and safety participation could effec- 166
105 jointly enhance the safety performance of the construction tively reduce negative safety outcomes. For example, the risk of 167
106 industry. accidents can be reduced by improving construction workers’ 168
safety behaviors (Aryee & Hsiung, 2016; Leung, Liang, & 169
107 1.1. Psychological capital Olomolaiye, 2016). Safety compliance behaviors (e.g., wearing per- 170
sonal protective equipment) play a mediating role between safety 171
108 Psychological capital (PsyCap) is a positive state or capacity that climate and injuries (Liu et al., 2015) and has a direct impact on 172
109 cultivates during one’s growth and development. This capacity can safety outcomes (Fernández-Muñiz, Montes-Peón, & Vázquez- 173
110 be assessed, developed, and leveraged for performance improve- Ordás, 2014). Research to date has clearly shown the critical func- 174
111 ment (Luthans, Avolio, Avey, & Norman, 2007b). There are three tion of safety behavior in preventing accidents and injuries. This 175
112 perspectives on the concept of PsyCap. The first focuses on individ- paper explores the antecedents of safety behavior based on PsyCap 176
113 ual characteristics that result from the interaction between one’s theory to better understand the psychological interaction mecha- 177
114 environment and heredity. It has been described using a five- nism of construction safety. 178
115 factor model that includes neuroticism, extraversion, openness,
116 agreeableness, and conscientiousness (Letcher & Niehoff, 2004). 1.3. Psychological capital and safety behavior 179
117 The second perspective points out that PsyCap is a psychological
118 state that could be used to predict and improve personal perfor- PsyCap has positive impacts on employee performance and sat- 180
119 mance (Avolio, 2004; Luthans et al., 2007b). The third perspective isfaction (Luthans, Norman, Avolio, & Avey, 2008; Youssef & 181
120 proposes that PsyCap is an integrated psychological ability consist- Luthans, 2007) and organizational citizenship behaviors (Avey 182
121 ing of both individual characteristics and psychological state. Fur- et al., 2010), and they can be predicted using PsyCap constructs 183
122 ther, it could be enhanced by intervention measures and are (Luthans et al., 2007b). For example, hope has stronger ties than 184
123 relatively stable (Avolio & Walumbwa, 2006). optimism and resilience to work-related outcomes, such as perfor- 185
124 Given the different perspectives, it is not surprising that experts mance, job satisfaction, work happiness, and organizational com- 186
125 have not agreed upon the dimensions of PsyCap. Structures vary mitment (Youssef & Luthans, 2007). It has been suggested that 187
126 from two to five dimensions (Avey et al., 2010; Goldsmith, Veum, PsyCap might be used to enhance job satisfaction and outcomes 188
127 & Darity, 1997; Hodges, 2010; Luthans, 2002; Luthans et al., (Avey et al., 2010). Although it is commonly accepted that PsyCap 189
128 2004; West, Patera, & Carsten, 2009). Among them, the most has a positive effect on job performance, the effect of PsyCap on 190
129 widely adopted construct includes four elements: self-efficacy/ safety performance may show another pattern because of differ- 191
130 self-confidence, hope, resilience, and optimism (Luthans et al., ences related with the task, its context, or applicability (Christian 192
131 2004). Self-efficacy/self-confidence is a personal belief that one can et al., 2009). Thus, it is necessary to test the impact of PsyCap on 193
132 use known resources to get their desired results. This belief could safety performance as a separate domain of job performance. 194
133 be obtained from mastered experience, individual study, social Besides, there are differing opinions about the exact influence 195
134 faith, positive feedback, and psychological enforcement. Hope pathways between PsyCap and job performance. One camp sug- 196
135 refers to someone having the motivation and approach for achiev- gests that PsyCap has direct influences on employee performance 197
136 ing goals. Hope can be improved by goal-setting, participation, (Luthans et al., 2008). The other believes that PsyCap has indirect 198
137 preparation, flexibility, mental rehearsals, and goal-resetting. Resi- influences on worker behavior via mediating variables, such as 199
138 lience reflects one’s ability to recover from adversity, failure, or safety motivation (Christian et al., 2009). In the middle, some pro- 200
139 irreversible change. Resilience can be developed using strategies pose that PsyCap has both direct and indirect correlations with 201

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

C. He et al. / Journal of Safety Research xxx (xxxx) xxx 3

202 worker behavior (Newman, Ucbasaran, Zhu, & Hirst, 2014; Seo, Lee, H3d. Optimism is positively associated with communication 266
203 Kim, & Jee, 2015). Further, most previous studies treated PsyCap as competence. 267
204 a unidimensional construct and neglected its sub-dimensional H4. Communication competence is positively associated with 268
205 effect on safety performance. To contribute to this debate and safety compliance. 269
206 gap, Hypothesis 1 and 2 are presented. H5. Communication competence is positively associated with 270
207 H1. Psychological capital is positively associated with safety safety participation. 271
208 compliance behavior. For the reader’s benefit, Fig. 1 provides a graphical summary of 272
209 H1a. Self-efficacy is positively associated with safety the five hypotheses presented in this study. This model is based on 273
210 compliance. the theoretical model proposed by Eid et al. (2012), the integrative 274
211 H1b. Hope is positively associated with safety compliance. safety model developed by Christian et al. (2009) and the PsyCap 275
212 H1c. Resilience is positively associated with safety compliance. framework introduced by Newman et al. (2014). 276
213 H1d. Optimism is positively associated with safety compliance.
214 H2. Psychological capital is positively associated with safety
215 participation behavior. 2. Methodology 277
216 H2a. Self-efficacy is positively associated with safety
217 participation. 2.1. Questionnaire design 278
218 H2b. Hope is positively associated with safety participation.
219 H2c. Resilience is positively associated with safety participation. The questionnaire comprised three sections, namely demo- 279
220 H2d. Optimism is positively associated with safety participation. graphic information, PsyCap and communication competence, 280
and, safety behaviors. The format, items, and semantics were dis- 281
221 1.4. Communication competence as a mediator cussed with two safety experts and five senior construction engi- 282
neers, each with more than 15 years of work experience. The 283
222 Communication is an employee’s overall perception of sharing wording and expression of each question was carefully reviewed 284
223 information, ideas, and emotion among individuals, team leaders to accord with the habits and idioms of site workers. Two workers 285
224 and team members, to effectively and creatively integrate profes- were interviewed to evaluate the content and intelligibility of the 286
225 sional knowledge and information between teams (Patrashkova- questionnaire. Six workers were invited to fill out the question- 287
226 Volzdoska, Mccomb, Green, & Compton, 2003). Communication is naire to test the average answering time and appropriate survey 288
227 an important factor affecting organizational safety climate, indi- procedure. Three reverse questions were randomly set in the ques- 289
228 vidual psychological state, and safety behavior (Liao, Lei, Xue, & tionnaire to reduce the influence of social desirable responding 290
229 Fang, 2015; Niu, Leicht, & Rowlinson, 2016; Shen, Koh, (Bergheim, Nielsen, Mearns, & Eid, 2015). Items about PsyCap 291
230 Rowlinson, & Bridge, 2015; Wu, Liu, Zhang, Skibniewski, & Wang, and communication competence adopted a 5-point Likert scale, 292
231 2015; Zohar & Luria, 2003). More frequent communication where 1 translated to ‘‘strongly disagree” while 5 meant ‘‘strongly 293
232 between safety leaders and workers, and more communication agree.” For safety behaviors, workers were asked the frequency 294
233 channels amongst individuals help to improve organizational with which they performed safety compliance and participation 295
234 safety climate and personal psychological safety climate (Liao, using discrete values of 1 for ‘‘never” to 5 for ‘‘quite often.” 296
235 Lei, Fang, & Liu, 2014; Shen et al., 2015). Conversely, a supportive
236 safety climate can significantly impact safety communication
237 (Kath, Marks, & Ranney, 2010). Further, safety communication 2.2. Measures 297
238 affects the safety behavior of construction workers (Cigularov,
239 Chen, & Rosecrance, 2010). As the safety-related interaction PsyCap was measured with Psychological Capital Questionnaire 298
240 between supervisors and workers increases, the safety behavior (PCQ) (Luthans et al., 2007b). As a widely used tool, PCQ has shown 299
241 of workers is also enhanced (Zohar & Luria, 2003). its applicability in many studies (Avey et al., 2010; Luthans et al., 300
242 The impact of communication depends on one’s communication 2008; West et al., 2009) and its suitability for Chinese workers 301
243 competence, which is ‘‘the ability to communicate in a personally (Luthans, Avolio, Walumbwa, & Li, 2005). Twenty-four statements 302
244 effective and socially appropriate manner” (Trenholm & Jensen, with six assessing each of the four dimensions of PsyCap were 303
245 2013). Communication competence is therefore key for safety per- used. Example statements include ‘‘I feel confident helping to set 304
246 formance (Marin & Roelofs, 2017). It might be used to interpret targets/goals in my work area,” ‘‘At the present time, I am energet- 305
247 individual behavior variance, resulting from cognition, emotion, ically pursuing my work goals,” ‘‘I can get through difficult times at 306
248 motivation internally and circumstances, culture, interpersonal work because I have experienced difficulty before,” and ‘‘I am opti- 307
249 relationships externally. A person with high communication com- mistic about what will happen to me in the future as it pertains to 308
250 petence may also be more likely to keep a healthy psychological work.” Responses to three of the 24 statements were deleted from 309
251 state through effective communication with his or her supervisor the original questionnaire because of low factor loadings (0.339, 310
252 and coworkers. As discussed above, this healthy psychological
253 state would be beneficial to further develop safety performance.
254 However, few studies examined the correlations of communication
255 competence with PsyCap and safety behavior. To systematically
256 examine the role of communication competence in PsyCap and
257 safety behavior, the authors suggest Hypotheses 3 to 5.
258 H3. PsyCap is positively associated with communication
259 competence.
260 H3a. Self-efficacy is positively associated with communication
261 competence.
262 H3b. Hope is positively associated with communication
263 competence.
264 H3c. Resilience is positively associated with communication
265 competence. Fig. 1. The hypothesis model.

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

4 C. He et al. / Journal of Safety Research xxx (xxxx) xxx

311 0.468, 0.521 respectively), leaving 21 statements of PsyCap in the scores from six. Second, SPSS Statistic22.0 was utilized to examine 373
312 final data analysis. the correlation, collinearity and normality of the variables. The 374
313 Communication competence can be defined from various per- thresholds of absolute skewness, absolute kurtosis, and variance 375
314 spectives, such as situation, state, and characteristics (Adler & inflation factor (VIF) are all less than or equal to 2, 7, 5, respectively 376
315 Rodman, 2003; Readon, 1987; Spano, 1988). In this study, we (Kim, 2013; O’Brien, 2007; Pektasß, 2015). Third, reliability and 377
316 adopted the perspective of Trenholm and Jensen (2013), which validity analyses were applied to test the internal consistency, con- 378
317 viewed communication competence as an interactive five-step vergent validity, and discriminant validity of the questionnaire. 379
318 process. This process includes five sub-processes: planning, per- Finally, as an appropriate technique for multivariate analysis that 380
319 ceiving, adapting, coding and decoding, and self-presentation, cor- integrates factor analysis, path analysis and multiple regression 381
320 responding to the five types of process competence: goal analysis and a mature approach for safety research (Fang, Wu, & 382
321 competence, interpretive competence, role competence, message Wu, 2015; Hox, 2007; Lyu, Ckh, Chan, Fkw, & Javed, 2018; Seo 383
322 competence, and self competence, thus giving more functional et al., 2015), the technique of structural equation modeling (SEM) 384
323 and comprehensive interpretation of this concept. Five items was used to create the conceptual model, perform confirmatory 385
324 assessed the process-perspective communication competence. factor analyses (CFA) of the measurement models, and test the the- 386
325 Example statements included ‘‘I know how to achieve communica- oretical hypotheses of the structural models with the software 387
326 tion goals” (i.e., the planning process), and ‘‘I know to send and SPSS AMOS22.0. Bootstrapping could be used to create multiple 388
327 receive messages effectively” (i.e., the coding and decoding sub-samples from an original database, enabling the researcher 389
328 process). to evaluate the stability of parameter estimates and thus report 390
329 Safety behaviors (compliance and participation) were evaluated their values with a higher level of accuracy (Byrne, 2010), thus 391
330 by six statements adopted from Neal and Griffin (2006). Three the bootstrapping technique was adopted to provide robustness 392
331 were for safety compliance: (a) I use all the necessary safety equip- estimation (10,000 bootstrap samples and 95% confidence 393
332 ment to do my job, (b) I use the correct safety procedures for car- intervals). 394
333 rying out my job, and (c) I ensure the highest levels of safety when I
334 carry out my job. The other three statements assessed safety par-
2.5. Reliability and validity analyses 395
335 ticipation: (a) I promote the safety program within the organiza-
336 tion, (b) I put in extra effort to improve the safety of the
Reliability tests were done to examine the internal consistency 396
337 workplace, and (c) I voluntarily carry out tasks or activities that
of each construct in the questionnaire. For good reliability, the 397
338 help to improve workplace safety.
coefficient of Cronbach’s alpha must be higher than 0.7 398
(Nunnally & Bernstein, 1994). Cronbach values of the seven con- 399
339 2.3. Data collection procedures
structs varied from 0.82 to 0.93 (shown in Table 2), which denotes 400
a high robustness and stability of the questionnaire. 401
340 Data were collected by questionnaire from Chinese construction
Convergent validity means the interpretation power of observ- 402
341 workers. The questionnaires were printed and distributed to front-
able variables exerting on latent variables. It could be assessed 403
342 line workers during safety training meetings or breaks. The respon-
with three commonly used indicators: standardized factor loadings 404
343 dents answered the questions with paper and pencil and
(SFL), construct reliability (CR), and average variance extracted 405
344 immediately returned the questionnaires to the coordinators. It
(AVE). The criteria of these indicators for acceptance are: 406
345 took 10–20 minutes to complete the questionnaire. Further, the
SFL > 0.6, CR > 0.7, and AVE > 0.5 (Bagozzi & Yi, 1988; Shen et al., 407
346 questionnaire was anonymous and the workers were told that
2015). All constructs met these criteria. The second-order indica- 408
347 the information provided by them was completely confidential
tors of SFL, CR, and AVE for PsyCap and safety behavior were also 409
348 and that it was only used for academic research. They were encour-
calculated (shown in Table 2). All values meet the criteria, which 410
349 aged to fill out the questionnaire according to their own thoughts
demonstrated acceptable convergent validity of each construct. 411
350 and were required to answered the questions independently. All
Discriminant validity signifies the distinguishing power of a 412
351 the questionnaires were collected by the researchers once the sur-
latent variable that shows different features or meanings from 413
352 vey was finished at a site.
other variables. The method of measuring discriminant validity is 414
353 One thousand questionnaires were sent to 22 construction pro-
comparing some variable’s square root of AVE with its values of 415
354 jects in 10 Chinese provinces, autonomous regions, or municipali-
correlation coefficients (Bagozzi, 1981). If a variable’s square root 416
355 ties. Of those, 830 questionnaires were returned. After examining
of AVE is bigger than its correlation coefficients, then it shows a 417
356 the returned questionnaires, 175 were considered invalid because
good discriminant validity. For example, self-efficacy (EFF)’s square 418
357 (a) more than 10% of data were missing, (b) there existed identical
root of AVE was 0.777, which was higher than its biggest correla- 419
358 answers in every response, or (c) there were obvious patterns in
tion coefficient 0.741 (shown in Table 3), illustrating an appropri- 420
359 the responses. The final data contained 655 questionnaires, with
ate discriminant validity. The other six constructs also passed this 421
360 a valid rate of 65.5%. The details of the respondents are shown in
test. 422
361 Table 1. Generally, the invalid respondents are elder, with more
362 work experience, lower education level, and more weekly work
363 hours than valid respondents. For instance, the percentage for 3. Results 423
364 the invalid respondents with equal to or more than 15 years work
365 experience is 23.4%, while this percentage for the valid respon- 3.1. Measurement model 424
366 dents is 14.8%. Thus the invalid respondents may have less willing-
367 ness or ability to complete a questionnaire, which constrains the To evaluate the quality of the four-factor structure of PsyCap 425
368 valid rate. and the one-factor structure of communication competence, the 426
overall model fit test was applied. If the test results are acceptable, 427
369 2.4. Data analysis procedures the quality of the two measurement models can be confirmed 428
(Fang et al., 2015). Three main indexes of the overall model fit were 429
370 To ensure consistency across the Likert scale statements, the adopted: the relative v2 (v2/degree of freedom), root-mean-square 430
371 scores of the negative items were reversed to be consistent with error of approximate (RMSEA) and comparative fit index (CFI). The 431
372 the scores of the positive statements by subtracting the negative upper thresholds of the relative v2 and RMSEA are 5 and 0.08, 432

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

C. He et al. / Journal of Safety Research xxx (xxxx) xxx 5

Table 1
Demographic information of the respondents.

Category Items Valid respondents (N = 655) Invalid respondents (N = 175)


Frequency Percentage (%) Frequency Percentage (%)
Gender Male 600 91.6 166 94.9
Female 55 8.4 9 5.1
Age <30 137 20.9 39 22.3
<40 260 39.7 54 30.9
<50 186 28.4 59 33.7
50 72 11.0 23 13.1
Work experience <5 177 27.0 39 22.3
<10 236 36.0 59 33.7
<15 145 22.2 36 20.6
<20 42 6.4 18 10.3
20 55 8.4 23 13.1
Educational level Primary school 117 17.9 26 14.9
Middle school 290 44.3 92 52.6
High school 118 18.0 35 20.0
Junior College 59 9.0 12 6.9
College or higher 71 10.8 10 5.7
Average weekly work 40 38 5.8 11 6.3
hours <50 111 16.9 14 8.0
<60 190 29.0 64 36.6
<70 155 23.7 43 24.6
70 161 24.6 43 24.6
Region Fujian 117 17.9 30 17.1
Shanxi 115 17.6 20 11.4
Anhui 88 13.4 26 14.9
Guizhou 83 12.7 8 4.6
Hebei 59 9.0 44 25.1
Heilongjiang 50 7.6 10 5.7
Inner Mongolia 40 6.1 10 5.7
Liaoning 38 5.8 11 6.3
Shanghai 36 5.5 3 1.7
Hubei 29 4.4 13 7.4

Table 2
Convergent validity test.

Constructs Items SFL CR AVE a Second-order SFL Second-order CR Second-order AVE


Self-efficacy (EFF) EFF1 0.768 0.90 0.60 0.91 0.846 0.89 0.67
EFF2 0.804
EFF3 0.796
EFF4 0.790
EFF5 0.767
EFF6 0.733
Hope (HOP) HOP1 0.693 0.88 0.56 0.89 0.854
HOP2 0.772
HOP3 0.677
HOP4 0.747
HOP5 0.822
HOP6 0.768
Resilience (RES) RES1 0.639 0.85 0.53 0.85 0.822
RES2 0.750
RES3 0.729
RES4 0.736
RES5 0.772
Optimism (OPT) OPT1 0.665 0.82 0.53 0.82 0.736
OPT2 0.763
OPT3 0.741
OPT4 0.732
Communication Competence (CC) CC1 0.747 0.87 0.57 0.87 —— —— ——
CC2 0.740
CC3 0.738
CC4 0.770
CC5 0.777
Safety Compliance (SC) SC1 0.888 0.92 0.78 0.93 0.835 0.86 0.76
SC2 0.895
SC3 0.873
Safety Participation (SP) SP1 0.853 0.92 0.79 0.93 0.905
SP2 0.914
SP3 0.893

Note: SFL = standardized factor loadings; CR = construct reliability; AVE = average variance extracted; a = Cronbach value.

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

6 C. He et al. / Journal of Safety Research xxx (xxxx) xxx

Table 3
Discriminant validity test.

Construct Mean SD EFF HOP RES OPT CC SC SP


EFF 3.758 0.655 0.777
HOP 3.822 0.599 0.741** 0.748
RES 3.829 0.554 0.619** 0.671** 0.727
OPT 3.846 0.588 0.446** 0.442** 0.472** 0.726
CC 3.893 0.533 0.555** 0.585** 0.513** 0.496** 0.755
SC 4.185 0.726 0.369** 0.314** 0.304** 0.233** 0.271** 0.885
SP 3.760 0.939 0.509** 0.499** 0.459** 0.261** 0.391** 0.521** 0.887

Note: **p < 0.01; SD = standard deviation; diagonal values are the square roots of AVE.

433 respectively, while the lower threshold of CFI is 0.90 (Hooper, 3.3. Hypotheses test 476
434 Coughlan, & Mullen, 2008; Marsh & Hau, 1996; Xiong, Skitmore,
435 & Xia, 2015). Table 4 shows the overall model fit statistics of the A hypothesis was supported if the P value was less than 0.05 477
436 confirmatory factor analyses, which indicate favorable fitness for and the estimate was positive. For the hypothesis with several 478
437 measurement models of PsyCap and communication competence. sub-hypotheses (e.g., H1 in Table 6), if some sub-hypotheses were 479
supported, then the hypothesis was considered partially sup- 480
ported. The results of hypotheses tests showed that H1, H2, and 481
438 3.2. Structural model
H3 were partially supported. H4 (communication competence to 482
safety compliance) was rejected while H5 (communication compe- 483
439 The aforementioned hypothesized model (Fig. 1) was tested
tence to safety participation) was supported (shown in Fig. 2 and 484
440 with SEM technique. Goodness-of-fit was analyzed to check
Table 6). Self-efficacy was positively related with safety compli- 485
441 whether the surveyed data fit the structural model. Firstly, prelim-
ance and safety participation, while resilience was only associated 486
442 inary data fit was examined to make sure there were no abnormal
with the latter. There was no significant relationship between hope 487
443 variables (Bagozzi & Yi, 1988). All error variances were significant
and safety behaviors. Although the relationship between optimism 488
444 and greater than zero; all standard errors were less than 0.04; all
and safety participation is significant, the standardized regression 489
445 SFL were significant and located between 0.50 and 0.95. This is
coefficient is negative. However, both hope and optimism have sig- 490
446 strong evidence of a good preliminary fit of the data. Secondly,
nificant and positive links with communication competence, which 491
447 the overall model fit was evaluated with three index classes: abso-
means that these two constructs might affect safety behavior via 492
448 lute, incremental, and parsimonious. The criteria (Hooper et al.,
communication competence. Communication competence was 493
449 2008; Marsh & Hau, 1996; Xiong et al., 2015) and values of these
not significantly associated with safety compliance, but was signif- 494
450 indexes for the original and competing models were shown in
icantly and positively related with safety participation. 495
451 Table 5. All the indexes met the criteria, demonstrating an accept-
The path coefficients among PsyCap, communication compe- 496
452 able overall model fit.
tence and safety behaviors are shown in Fig. 2. Firstly, self- 497
453 Three models were developed. Model 1 was the original model,
efficacy positively influenced both safety behaviors, which sup- 498
454 and displayed full connections between the four dimensions of
ported H1a and H2a. Hope had no significant direct effect on safety 499
455 PsyCap, communication competence, and two dimensions of safety
behavior, which led to the rejection of H1b, H2b. Resilience signifi- 500
456 behavior. Model 2 was based on the hypotheses test results of
cantly and positively influenced safety participation, which sup- 501
457 model 1, in which five insignificant connections (EFF ? CC;
ported H2c. Optimism negatively influenced safety participation, 502
458 HOP ? SP; RES ? CC; OPT ? SC; CC ? SC) were removed and the
which caused H1d, H2d to be rejected. Secondly, both hope and opti- 503
459 correlation between the variances of SC and SP was added based
mism could positively affect safety participation via communica- 504
460 on modification indices. Another two insignificant relationships
tion competence, which meant that H3 was partially supported 505
461 from model 2 (HOP ? SC; RES ? SC) were taken out, resulting in
and H5 was supported. Finally, according to the values of squared 506
462 model 3. The v2 and degrees of freedom were 1648.56 and 444
multiple correlations, 39.5% of the variance for safety participation 507
463 for model 1, and 1428.17 and 442 for model 2. The v2 difference
was explained by PsyCap and communication competence, while 508
464 also follows a v2 distribution (Chen et al., 2017a, 2017b). For exam-
58.0% and 19.9% of the variance for communication competence 509
465 ple, the v2 difference between model 1 and model 2 was 220.39
and safety compliance were explained by PsyCap, respectively. 510
466 with 2 degrees of freedom. As 220.39 > v2 0.05(2) = 5.99, a signifi-
These results also demonstrated the obvious effects of PsyCap on 511
467 cant difference was evident between these two models. This sug-
safety behaviors and communication competence. 512
468 gests that model 2 was better than model 1. There were no
The effects of PsyCap (with 4 dimensions) and communication 513
469 significant differences for the values of model fit test between
competence (CC) on safety compliance (SC) and safety participa- 514
470 model 2 and 3 (Ho, 2013; Williams & Holahan, 1994). However,
tion (SP) are shown in Table 7. Both direct and indirect effects were 515
471 model 3 was more parsimonious than model 2. For instance, PNFI
assessed with SEM, of which the latter could not be calculated with 516
472 for model 3 was higher than that for model 2. Thus model 3 was
regression due to measurement error (Brown, 1997). The results 517
473 better than model 2. As a result, model 3 was adopted as the best
show that hope (HOP) and optimism (OPT) had positive and signif- 518
474 option (shown in Fig. 2). For model 3, all of the indexes meet the
icant effects on communication competence (CC), while self- 519
475 criteria, which illustrated favorable overall model fitness.

Table 4
Fit indexes for the measurement models.

Model v2 D.f. v2/D.f. RMSEA CFI


Model of PsyCap 645.228 179 3.605 0.063 0.921
Model of Communication Competence 7.507 3 2.502 0.048 0.995

Note: D.f. = degree of freedom.

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

C. He et al. / Journal of Safety Research xxx (xxxx) xxx 7

Table 5
Fit indexes of the structural models.

Type Index Criteria Values


Model 1 Model 2 Model 3
Absolute indexes v2 —— 1648.56 1428.17 1428.86
Degree of freedom —— 444 442 444
RMSEA <0.08 0.065 0.058 0.058
AGFI >0.80 0.811 0.833 0.833
Incremental indexes CFI >0.90 0.877 0.903 0.903
IFI >0.90 0.878 0.903 0.903
Parsimonious indexes PGFI >0.50 0.707 0.720 0.723
PNFI >0.50 0.753 0.771 0.775
NC <5 3.794 3.231 3.218

Note: RMSEA = root-mean-square error of approximation; AGFI = adjusted goodness-of-fit index; CFI = comparative fit index; IFI = incremental fit index; PGFI = parsimony
goodness-of-fit index; PNFI = parsimony normed-fit index; NC = normed v2 (i.e., v2/degree of freedom).

Fig. 2. The final structural model and impact paths. Note:*p < 0.05; **p < 0.01; ns = not significant; SMC = squared multiple correlation.

Table 6 efficacy (EFF) and resilience (RES) did not. However, self-efficacy 520
Path coefficients of the final model. (EFF) and resilience (RES) had positive and significant effects on 521
Path Estimate S.E. C.R. p safety behavior. Meanwhile, optimism (OPT) had negative and sig- 522

H1. PsyCap ? SC —— —— —— ——
nificant effects on safety participation (SP). Communication com- 523
H1a. EFF ? SC 0.446 0.053 9.660 *** petence (CC) played a mediating role between hope (HOP), 524
H2. PsyCap ? SP —— —— —— —— optimism (OPT) and safety participation (SP), which could reduce 525
H2a. EFF ? SP 0.420 0.097 5.594 *** the negative effects. 526
H2c. RES ? SP 0.248 0.173 3.109 0.002
H2d. OPT ? SP –0.186 0.102 –3.088 0.002
H3. PsyCap ? CC —— —— —— —— 4. Discussion 527
H3b. HOP ? CC 0.510 0.049 8.681 ***
H3d. OPT ? CC 0.351 0.052 6.003 ***
H5. CC ? SP 0.141 0.111 2.420 0.016
This study focuses on the effects of PsyCap on construction 528
workers’ safety behaviors while incorporating communication 529
Note: ***p < 0.001; Estimate = standardized regression coefficients; S.E. = standard- competence as a mediator. The relationships between PsyCap 530
ized error; C.R. = critical ratio (>[1.96]).
(self-efficacy, hope, resilience, and optimism), communication 531
competence and safety behaviors (safety compliance, safety partic- 532
ipation) were examined. The results show that self-efficacy and 533
resilience positively impacted safety behavior. It is interesting to 534
point out that optimism negatively affected safety participation, 535
Table 7
which denotes that a deepened and specific perspective should 536
Standard direct and indirect effects for the final model.
be drawn on PsyCap. The partially mediating role of communica- 537
Endogenous Exogenous Direct Indirect 10,000 tion competence to reduce aforementioned negative influence also 538
variables variables effects effects bootstrapping 95%
emerged. 539
CI
SC EFF 0.446** —— [0.349, 0.536]
SP EFF 0.420** —— [0.272, 0.571]
4.1. Main effects of PsyCap on construction workers’ safety behavior 540
SP RES 0.248** —— [0.079, 0.431]
SP OPT –0.186** —— [–0.333, –0.070] Self-efficacy significantly and positively affected both safety 541
SP CC 0.141* —— [0.026, 0.268] compliance and safety participation, which is consistent with pre- 542
CC HOP 0.510** —— [0.368, 0.628]
vious studies (Bandura, 1978; Chen & Chen, 2014; Prinzel, 2002; 543
CC OPT 0.351** —— [0.212, 0.500]
SP HOP —— 0.072* [0.013, 0.138] Wang et al., 2018). Assuming that feeling and self-motivating is 544
SP OPT —— 0.049* [0.008, 0.111] impacted by self-efficacy (Bandura, 1978), construction workers 545
with higher states of self-efficacy may have better feelings of work 546
Note: *p < 0.05; **p < 0.01.
control and motivation to perform safely. According to the Theory 547

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

8 C. He et al. / Journal of Safety Research xxx (xxxx) xxx

548 of Planned Behavior (Ajzen, 1991), all of these factors foster safe opposing relationships. This seemingly conflicting result also 614
549 behavior. So self-efficacy could be used as a stable predictor of appears in other safety behavior research (Clarke & Ward, 2006; 615
550 safety behavior and behavioral change (Graham, 1995). Further, Leung et al., 2016; Seo et al., 2015), indicating the complex rela- 616
551 self-efficacy may encourage people to pursue new challenging tionships between safety behavior and its antecedents. 617
552 goals (Bandura & Cervone, 1983), and continuous reinforcement In summary, the impacts of sub-dimensions of PsyCap on safety 618
553 of self-efficacy (Gist, Schwoerer, & Rosen, 1989) for construction behaviors varied by significance, direction and coefficients. PsyCap 619
554 workers could be effective in improving safe behaviors. might act like a double-edged sword for safety behaviors, and 620
555 Hope was not related to safety participation and safety compli- should be treated with caution. 621
556 ance, which did not meet common conceptions (Luthans, Youssef,
557 & Avolio, 2007a). One of the reasons may be the social and cultural
4.2. Mediating effects of communication competence 622
558 context (Luthans & Youssef, 2007). Chinese construction workers
559 face many challenges including high work-related pressures
Communication competence partially mediates the relationship 623
560 because of tight schedules, unstable employment, and high mobil-
between PsyCap and safety behavior. Self-efficacy and resilience, 624
561 ity (Fang & Wu, 2013; Lei, Xiang, Song, & Wang, 2010). This may
which directly affect safety behavior, do not have significant rela- 625
562 lead them to take risks or use unsanctioned methods to complete
tionships with communication competence, while hope and opti- 626
563 their tasks more quickly. These actions might increase false hope,
mism are positively associated with it. Although optimism has a 627
564 which could further increase the likelihood that they will violate
negative effect on safety participation, this effect could be reduced 628
565 safety regulations and avoid safety participation. False hope and
via communication competence. This result, which is often 629
566 the complex relationship between hope and success could cause
ignored, indicates that the communication competence of con- 630
567 a waste of energy and resources (Luthans et al., 2007b). In partic-
struction workers is a necessary instrument for improving safety 631
568 ular, there are two basic mechanisms for someone using hope:
performance. This capability was related more strongly with safety 632
569 willpower to achieve their goals, and, waypower to choose path-
participation than with safety compliance. The industry experts 633
570 ways to get to the goals (Luthans & Youssef, 2007; Snyder et al.,
attribute it in part to the difficulties for evaluating nonverbal com- 634
571 1991). Thus higher levels of hope may not necessarily be associ-
munication skills (Friedman, Prince, Riggio, & Dimatteo, 1980; 635
572 ated with better safety behavior because of strong willpower and
Riggio, 1986), which could reveal stronger connections with safety 636
573 waypower focus on tasks rather than safety.
compliance. Questions related to communication competence in 637
574 Resilience was significantly and positively related to safety par-
the questionnaire focused more on properties of verbal communi- 638
575 ticipation, which echoes other studies (Luthans et al., 2007a; Wang
cation skills, and therefore reflected a closer relationship with 639
576 et al., 2018). In China, 88.9% construction workers in 2017 come
safety participation. Some other individual factors, such as the 640
577 from the agricultural industry as temporary or seasonal workers
level of angry hostility (Slaughter, Christian, Podsakoff, Sinar, & 641
578 (NBSC, 2018). They often face unforeseen problems, such as skill
Lievens, 2014), could also influence the relationship between com- 642
579 learning, communication with others, and stressful isolation from
munication skills and performance. Thus, adding communication 643
580 home. The resilient workers more easily adapt to the new environ-
channels and frequencies, and launching verbal and nonverbal 644
581 ment and solve those problems effectively. This capability further
communication skills training programs are suggested. 645
582 encourages them to participate in safety training and speak out
As with all research, this study has limitations. First, due to typ- 646
583 to help their coworkers, which may also cultivate safety participa-
ical time and resource constrains of this kind of investigation, 647
584 tion. However, resilient construction workers may also tend to rely
cross-sectional data were utilized. Although cross-sectional data 648
585 on their own safety knowledge and experiences rather than learn
are commonly used to examine relationships in safety science, lon- 649
586 new safety regulations and procedures, even when they encounter
gitudinal data would be better for exploring inner causations. Sec- 650
587 previously unseen safety situations or incidents. This self-reliance
ond, it is understood that PsyCap is heavily influenced by cultural 651
588 may have led to the lack of relationship between resilience and
and social contexts. The results in this paper should be used cau- 652
589 safety compliance.
tiously, because of the backgrounds of Chinese respondents (e.g., 653
590 It is interesting that optimism was negatively associated with
relatively low education level and high average weekly work hours 654
591 safety participation, while it had no significant relationship with
shown in Table 1). Global research on PsyCap and safety perfor- 655
592 safety compliance. While this differs from some studies (Strutton
mance would be interesting. Finally, similar to the aforementioned 656
593 & Lumpkin, 1992), it is consistent with the findings of Tuten and
relationships between stress and performance, the connections 657
594 Neidermeyer (2004), which showed that employee pessimism in
between dimensions of PsyCap and safety performance may also 658
595 call centers resulted in higher levels of performance and satisfac-
show nonlinear patterns. An exploration of suitable levels of Psy- 659
596 tion. Industry safety experts attribute the negative correlation
Cap to foster optimal safety performance might be necessary. 660
597 between optimism and safety behavior to the situational factors.
598 In some industries, such as accounting, finance, quality control,
599 and security management, a relatively pessimistic explanatory 5. Conclusions 661
600 style may be suitable (Luthans et al., 2007a), which enables the
601 practitioners developing preventive measures in serious situations. The research reported in this paper enriches the current body of 662
602 Another possible reason is the complex relationship between stress knowledge by examining the impacts of PsyCap (self-efficacy, 663
603 and performance, which may demonstrate different shapes, such hope, resilience, and optimism) on construction workers’ safety 664
604 as positive, negative or inverted-U (Sullivan & Bhagat, 1992). The behaviors (safety compliance and safety participation), and the 665
605 inverted-U model describes situations where pressure or stress mediating role of communication competence. This research 666
606 increase performance but only to a certain extent, after which it makes several contributions: (a) the antecedent role of PsyCap on 667
607 results in a reduction of performance. Working in the same safety worker safety behaviors helps to understand the psychological 668
608 climate, pessimists may feel higher level of stress, which urges mechanism of individual safety performance, (b) differential 669
609 them to take more responsibility and accountability, while opti- impacts of the sub-dimensions of PsyCap on worker safety behav- 670
610 mists may perceive lower levels of stress, which lets them avoid iors indicate the complexity of PsyCap-safety interrelationship and 671
611 from safety activities. Thus realistic and flexible optimism is the necessity of multidimensional PsyCap intervention, and (c) the 672
612 needed (Luthans et al., 2007a). It should be pointed out that the partially mediating role of communication competence between 673
613 direct and indirect effects of optimism on safety participation show PsyCap and safety behaviors offers a potential for better construing 674

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

C. He et al. / Journal of Safety Research xxx (xxxx) xxx 9

675 psychological-behavioral processes to reduce workplace safety Avey, J. B., Luthans, F., & Youssef, C. M. (2010). The additive value of positive 737
psychological capital in predicting work attitudes and behaviors. Journal of 738
676 incidents. 739
Management, 36(2), 430–452.
677 The results show that self-efficacy significantly and positively Avolio, B. J. (2004). Transformational leadership and organizational commitment: 740
678 affected safety behaviors while resilience had significant positive Mediating role of psychological empowerment and moderating role of 741
structural distance. Journal of Organizational Behavior, 25(8), 951–968. 742
679 effect on safety participation. Hope had no significant relationship 743
Avolio, B. J., & Walumbwa, F. O. (2006). Authentic leadership: Moving hr leaders to a
680 with safety behaviors while optimism negatively influenced safety higher level. Research in Personnel & Human Resources Management, 25(25), 744
681 participation. The good news is that communication competence 273–304. 745
Bagozzi, R. P. (1981). Evaluating structural equation models with unobservable 746
682 could reduce this negative effect. Both hope and optimism were
variables and measurement error: A comment. Journal of Marketing Research, 18 747
683 significantly and positively related to communication competence, (3), 375–381. 748
684 which was positively associated with safety participation. Thus Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. 749
685 communication competence demonstrated a partial mediating role Journal of the Academy of Marketing Science, 16(1), 74–94. 750
Bandura, A. (1978). Self-efficacy: Toward a unifying theory of behavioral change. 751
686 between PsyCap and safety behavior. Advances in Behaviour Research & Therapy, 1(4), 139–161. 752
687 These findings indicate that PsyCap may be a double-edged Bandura, A., & Cervone, D. (1983). Self-evaluative and self-efficacy mechanisms 753
688 sword to safety behavior. The differential impacts of sub- governing the motivational effects of goal systems. Journal of Personality & Social 754
Psychology, 45(5), 1017–1028. 755
689 dimensions of PsyCap on safety behavior have important practical Bergheim, K., Nielsen, M. B., Mearns, K., & Eid, J. (2015). The relationship between 756
690 implications for both construction workers and safety managers. psychological capital, job satisfaction, and safety perceptions in the maritime 757
691 For construction workers, false hope and blind optimism could industry. Safety Science, 74, 27–36. 758
Brown, R. L. (1997). Assessing specific mediational effects in complex theoretical 759
692 waste their energy for safety and encourage them neglect the 760
models. Structural Equation Modeling, 4(2), 142–156.
693 potential dangers and difficulties in construction, which may jeop- Brunetto, Y., Xerri, M., Farr-Wharton, B., Shacklock, K., Farr-Wharton, R., & 761
694 ardize their safety performance. Appropriate levels of hope and Trinchero, E. (2016). Nurse safety outcomes: Old problem, new solution-the 762
differentiating roles of nurses’ psychological capital and managerial support. 763
695 optimism might be more useful. For safety managers, taking a con- 764
Journal of Advanced Nursing, 72(11), 2794.
696 tingent view of PsyCap may help to improve the effectiveness of Byrne, B. M. (2010). Structural equation modeling with AMOS: Basic concepts, 765
697 safety training and safety intervention. It suggests that safety man- applications, and programming ((2nd ed.),). New York: Routledge. 766
Chen, S. L. (2018). Cross-level effects of high-commitment work systems on work 767
698 agers can plan PsyCap intervention programs based on frontline
engagement: The mediating role of psychological capital. Asia Pacific Journal of 768
699 workers’ sub-dimensional and overall PsyCap levels. It also sug- Human Resources, 56, 384–401. 769
700 gests that more attention should be paid to developing and main- Chen, C. F., & Chen, S. C. (2014). Measuring the effects of safety management system 770
701 taining construction workers’ communication competence, which practices, morality leadership and self-efficacy on pilots’ safety behaviors: 771
Safety motivation as a mediator. Safety Science, 62, 376–385. 772
702 could reduce the unexpected effects on safety performance. Chen, Y., McCabe, B., & Hyatt, D. (2017b). Relationship between individual 773
703 The results are primarily based on the survey of construction resilience, interpersonal conflicts at work, and safety outcomes of 774
704 workers. However, the approach can be extended for other occupa- construction workers. Journal of Construction Engineering & Management, 143 775
(8), 04017042. 776
705 tional employees, such as occupational drivers, firefighters and Chen, Y., McCabe, B., & Hyatt, D. (2017a). Impact of individual resilience and safety 777
706 hospital doctors. They face immense mental pressures at the work- climate on safety performance and psychological stress of construction 778
707 places. The PsyCap-safety relationships may vary between differ- workers: A case study of the Ontario construction industry. Journal of Safety 779
Research, 61, 167–176. 780
708 ent occupational groups. It is also extended that safety 781
Choi, B., Ahn, S., & Lee, S. H. (2017). Construction workers’ group norms and
709 researchers can compare subgroup PsyCap-safety relationships personal standards regarding safety behavior: Social identity theory 782
710 between organization levels and regions in a same occupational perspective. Journal of Management in Engineering, 33(4), 04017001. 783
Christian, M. S., Bradley, J. C., Wallace, J. C., & Burke, M. J. (2009). Workplace safety: 784
711 group or an organization to explore different psychological impact 785
A meta-analysis of the roles of person and situation factors. Journal of Applied
712 mechanisms, especially in large cross-regional or international Psychology, 94(5), 1103–1127. 786
713 organizations. Furthermore, future studies may focus on how the Cigularov, K. P., Chen, P. Y., & Rosecrance, J. (2010). The effects of error management 787
climate and safety communication on safety: A multi-level study. Accident 788
714 PsyCap changes over time and how it affects the performance of
Analysis & Prevention, 42(5), 1498–1506. 789
715 safety behavior. Safety managers could promote flexible safety ini- Clarke, S., & Ward, K. (2006). The role of leader influence tactics and safety climate 790
716 tiatives based on this dynamic relationship to elevate workplace in engaging employees’ safety participation. Risk Analysis, 26(5), 1175–1185. 791
717 safety performance. Cooper, M. D. (2009). Behavioral safety interventions a review of process design 792
factors. Professional Safety, 54(2), 36–45. 793
Eid, J., Mearns, K., Larsson, G., Laberg, J. C., & Johnsen, B. H. (2012). Leadership, 794
718 Funding psychological capital and safety research: Conceptual issues and future research 795
questions. Safety Science, 50(1), 55–61. 796
Fang, D., & Wu, H. (2013). Development of a safety culture interaction (SCI) model 797
719 This research was supported by the National Natural Science for construction projects. Safety Science, 57(8), 138–149. 798
720 Foundation of China (Grant No. 71472139) and the International Fang, D., Wu, C., & Wu, H. (2015). Impact of the supervisor on worker safety 799
behavior in construction projects. Journal of Management in Engineering, 31(6), 800
721 Cooperation Training Program for Graduate Students at Tongji 801
04015001.
722 University (Grant No. 2018XKJC-004). Fernández-Muñiz, B., Montes-Peón, J. M., & Vázquez-Ordás, C. J. (2014). Safety 802
leadership, risk management and safety performance in Spanish firms. Safety 803
Science, 70, 295–307. 804
723 Uncited references Friedman, H. S., Prince, L. M., Riggio, R. E., & Dimatteo, M. R. (1980). Understanding 805
and assessing nonverbal expressiveness: The affective communication test. 806
724 Avey, Luthans, Smith, and Palmer (2010); Avey, Luthans, and Journal of Personality & Social Psychology, 39(2), 333–351. 807
Gist, M. E., Schwoerer, C., & Rosen, B. (1989). Effects of alternative training methods 808
725 Youssef (2010). on self-efficacy and performance in computer software training. Journal of 809
Applied Psychology, 74(6), 884–891. 810
726 References Goldsmith, A. H., Veum, J. R., & Darity, W. J. R. (1997). The impact of psychological 811
and human capital on wages. Economic Inquiry, 35(4), 815–829. 812
Graham, S., & Weiner, B. (1995). Handbook of Educational Psychology. New York: 813
727 Adler, R. B., & Rodman, G. R. (2003). Understanding human communication. New
Simon & Schuster Macmillan. 814
728 York: Oxford University Press.
Griffin, M. A., & Neal, A. (2000). Perceptions of safety at work: A framework for 815
729 Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior & Human
linking safety climate to safety performance, knowledge, and motivation. 816
730 Decision Processes, 50(2), 179–211.
Journal of Occupational Health Psychology, 5(3), 347–358. 817
731 Aryee, S., & Hsiung, H. H. (2016). Regulatory focus and safety outcomes: An
Guo, B. H. W., & Yiu, T. W. (2015). Developing leading indicators to monitor the 818
732 examination of the mediating influence of safety behavior. Safety Science, 86,
safety conditions of construction projects. Journal of Management in Engineering, 819
733 27–35.
32, 04015016. 820
734 Avey, J. B., Luthans, F., Smith, R. M., & Palmer, N. F. (2010). Impact of positive
Hinze, J., Thurman, S., & Wehle, A. (2013). Leading indicators of construction safety 821
735 psychological capital on employee well-being over time. Journal of Occupational
performance. Safety Science, 51(1), 23–28. 822
736 Health Psychology, 15(1), 17–28.

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

10 C. He et al. / Journal of Safety Research xxx (xxxx) xxx

823 Ho, R. (2013). Handbook of univariate and multivariate data analysis and Niu, M., Leicht, R. M., & Rowlinson, S. (2016). Overview and analysis of safety 906
824 interpretation with SPSS. New York: CRC Press. climate studies in the construction industry. Construction Research Congress, 907
825 Hodges, T. D. (2010). An experimental study of the impact of psychological capital 2926–2935. 908
826 on performance, engagement, and the contagion effect. Dissertations & Theses- Nunnally, J., & Bernstein, I. (1994). Psychometric theory (3rd ed.). New York: 909
827 Gradworks. McGraw-Hill. 910
828 Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural equation modeling: O’Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation 911
829 Guidelines for determining model fit. Electronic Journal on Business Research factors. Quality & Quantity, 41(5), 673–690. 912
830 Methods, 6(1), 141–146. Patrashkova-Volzdoska, R. R., Mccomb, S. A., Green, S. G., & Compton, W. D. (2003). 913
831 Hox, J. J. (2007). An introduction to structural equation modeling. Computers in Examining a curvilinear relationship between communication frequency and 914
832 Human Behavior, 27(1), 63–68. team performance in cross-functional project teams. IEEE Transactions on 915
833 Kath, L. M., Marks, K. M., & Ranney, J. (2010). Safety climate dimensions, leader- Engineering Management, 50(3), 262–269. 916
834 member exchange, and organizational support as predictors of upward safety Pektasß, A. O. (2015). Determining the essential parameters of bed load and 917
835 communication in a sample of rail industry workers. Safety Science, 48(5), suspended sediment load. International Journal of Global Warming, 8(3), 918
836 643–650. 335–359. 919
837 Kim, H. Y. (2013). Statistical notes for clinical researchers: Assessing normal Prinzel, L.J. (2002). The relationship of self-efficacy and complacency in pilot- 920
838 distribution (2) using skewness and kurtosis. Restorative Dentistry & automation interaction. NASA/TM-2002-211925. 921
839 Endodontics, 38(1), 52–54. Readon, K. K. (1987). Interpersonal Communication: Where Minds Meet. Belmont: 922
840 Lei, Z., Xiang, H. Y., Song, X. Q., & Wang, Z. Z. (2010). Nonfatal unintentional injuries Wadsworth Publishing Company. 923
841 and related factors among male construction workers in central china. American Riggio, R. E. (1986). Assessment of basic social skills. Journal of Personality & Social 924
842 Journal of Industrial Medicine, 53(6), 588–595. Psychology, 51(3), 649–660. 925
843 Letcher, L., & Niehoff, B. (2004). Psychological capital and wages: A behavioral Seo, H. C., Lee, Y. S., Kim, J. J., & Jee, N. Y. (2015). Analyzing safety behaviors of 926
844 economic approach. Midwest Academy of Management, May 2004, temporary construction workers using structural equation modeling. Safety 927
845 Minneapolis, MN. Science, 77, 160–168. 928
846 Leung, M. Y., Liang, Q., & Olomolaiye, P. (2016). Impact of job stressors and stress on Shen, Y., Koh, T. Y., Rowlinson, S., & Bridge, A. J. (2015). Empirical investigation of 929
847 the safety behavior and accidents of construction workers. Journal of factors contributing to the psychological safety climate on construction sites. 930
848 Management in Engineering, 32(1), 04015019. Journal of Construction Engineering & Management, 141(11), 04015038. 931
849 Liao, P. C., Lei, G., Fang, D., & Liu, W. (2014). The relationship between Slaughter, J. E., Christian, M. S., Podsakoff, N. P., Sinar, E. F., & Lievens, F. (2014). On 932
850 communication and construction safety climate in china. KSCE Journal of Civil the limitations of using situational judgment tests to measure interpersonal 933
851 Engineering, 18(4), 887–897. skills: The moderating influence of employee anger. Personnel Psychology, 67(4), 934
852 Liao, P. C., Lei, G., Xue, J. W., & Fang, D. (2015). Influence of person-organizational fit 847–885. 935
853 on construction safety climate. Journal of Management in Engineering, 31(4), Snyder, C. R., Harris, C., Anderson, J. R., Holleran, S. A., Irving, L. M., Sigmon, S. T., ... 936
854 04014049. Harney, P. (1991). The will and the ways: Development and validation of an 937
855 Lingard, H. C., Cooke, T., & Blismas, N. (2010). Safety climate in conditions of individual-differences measure of hope. Journal of Personality and Social 938
856 construction subcontracting: A multi-level analysis. Construction Management Psychology, 60, 570–585. 939
857 and Economics, 28(8), 813–825. Spano, S. J. (1988). Assessing communication competence: The development of a 940
858 Liu, X., Huang, G., Huang, H., Wang, S., Xiao, Y., & Chen, W. (2015). Safety climate, self-report measure of communication flexibility. Doctoral dissertation, Indiana 941
859 safety behavior, and worker injuries in the Chinese manufacturing industry. University. 942
860 Safety Science, 78(6), 173–178. SPSS Amos 22 [Computer software]. IBM, Armonk, NY. 943
861 Luthans, F. (2002). Positive organizational behavior: Developing and managing SPSS Statistics 22 [Computer software]. IBM, Armonk, NY. 944
862 psychological strengths. Academy of Management Executive, 16(1), 57–75. Statistics New Zealand, 2014. Injury Statistics-Work-related Claims: 2014- 945
863 Luthans, F., Avolio, B. J., Avey, J. B., & Norman, S. M. (2007b). Positive psychological trendstables for 2002–13. In: Zealand, S.N. (Ed.). 946
864 capital: Measurement and relationship with performance and satisfaction. Strutton, D., & Lumpkin, J. (1992). Relationship between optimism and coping 947
865 Personnel Psychology, 60(3), 541–572. strategies in the work environment. Psychological Reports, 71(3 Pt 2), 948
866 Luthans, F., Avolio, B. J., Walumbwa, F. O., & Li, W. (2005). The psychological capital 1179–1186. 949
867 of Chinese workers: Exploring the relationship with performance. Management Sullivan, S. E., & Bhagat, R. S. (1992). Organizational stress, job satisfaction and job 950
868 & Organization Review, 1(2), 249–271. performance: Where do we go from here? Journal of Management, 18(2), 951
869 Luthans, F., Luthans, K. W., & Luthans, B. C. (2004). Positive psychological capital: 3535–13374. 952
870 Beyond human and social capital. Business Horizons, 47(1), 45–50. Trenholm, S., & Jensen, A. (2013). Interpersonal Communication (7th ed.). Oxford 953
871 Luthans, F., Norman, S. M., Avolio, B. J., & Avey, J. B. (2008). The mediating role of University Press. 954
872 psychological capital in the supportive organizational climate: Employee Tuten, T. L., & Neidermeyer, P. E. (2004). Performance, satisfaction and turnover in 955
873 performance relationship. Journal of Organizational Behavior, 29(2), 219–238. call centers: The effects of stress and optimism. Journal of Business Research, 57 956
874 Luthans, F., & Youssef, C. M. (2007). Emerging positive organizational behavior. (1), 26–34. 957
875 Journal of Management, 33(3), 321–349. U.S. Bureau of Labor Statistics. (2012). Fatal occupational injuries by industry and 958
876 Luthans, F., Youssef, C. M., & Avolio, B. J. (2007a). Psychological capital: Developing the event or exposure, http://www.bls.gov/iif/oshwc/cfoi/cfch0011.pdf, (Aug. 20, 959
877 human competitive edge. Oxford, UK: Oxford University Press. 2014). 960
878 Lyu, S., Ckh, H., Chan, A., Fkw, W., & Javed, A. A. (2018). Relationships among safety Wang, D., Wang, X., & Xia, N. (2018). How safety-related stress affects workers’ 961
879 climate, safety behavior, and safety outcomes for ethnic minority construction safety behavior: The moderating role of psychological capital. Safety Science, 962
880 workers. International Journal of Environmental Research & Public Health, 15(3), 103, 247–259. 963
881 484. West, B. J., Patera, J. L., & Carsten, M. K. (2009). Team level positivity: Investigating 964
882 Madrid, H. P., Diaz, M. T., Leka, S., Leiva, P. I., & Barros, E. (2018). A finer grained positive psychological capacities and team level outcomes. Journal of 965
883 approach to psychological capital and work performance. Journal of Business & Organizational Behavior, 30(2), 249–267. 966
884 Psychology, 33, 461–477. Willamson, A., & Feyer, A. M. (1990). Behavioural epidemiology as a tool for 967
885 Marin, L. S., & Roelofs, C. (2017). Promoting construction supervisors’ safety-efficacy accident research. Journal of Occupational Accidents, 12(1), 207–222. 968
886 to improve safety climate: Training intervention trial. Journal of Construction Williams, L. J., & Holahan, P. J. (1994). Parsimony-based fit indices for multiple- 969
887 Engineering & Management, 143(8), 04017037. indicator models: Do they work? Structural Equation Modeling: A 970
888 Marsh, H. W., & Hau, K. T. (1996). Assessing goodness of fit: Is parsimony always Multidisciplinary Journal, 1(2), 161–189. 971
889 desirable? Journal of Experimental Education, 64(4), 364–390. Wu, X., Liu, Q., Zhang, L., Skibniewski, M. J., & Wang, Y. (2015). Prospective safety 972
890 McCabe, B. Y., Alderman, E., Chen, Y., Hyatt, D. E., & Shahi, A. (2017). Safety performance evaluation on construction sites. Accident Analysis & Prevention, 78, 973
891 performance in the construction industry: Quasi-longitudinal study. Journal of 58–72. 974
892 Construction Engineering & Management, 143(4), 04016113. Xiong, B., Skitmore, M., & Xia, B. (2015). A critical review of structural equation 975
893 Ministry of Housing and Urban-Rural Development of China (MOHURD). (2018). modeling applications in construction research. Automation in Construction, 49, 976
894 Report on safety accidents in housing and municipal construction projects in 59–70. 977
895 December 2017, http://www.mohurd.gov.cn/wjfb/201801/t20180130_234983. Youssef, C. M., & Luthans, F. (2007). Positive organizational behavior in the 978
896 html. workplace: The impact of hope, optimism, and resilience. Journal of 979
897 National Bureau of Statistics of China (NBSC). (2018). China Statistical Yearbook Management, 33(5), 774–800. 980
898 2018, http://www.stats.gov.cn/tjsj/ndsj/2018/indexch.htm. Zohar, D. (2000). A group-level model of safety climate: Testing the effect of group 981
899 Neal, A., & Griffin, M. A. (2006). A study of the lagged relationships among safety climate on microaccidents in manufacturing jobs. Journal of Applied Psychology, 982
900 climate, safety motivation, safety behavior, and accidents at the individual and 85(4), 587–596. 983
901 group levels. Journal of Applied Psychology, 91(4), 946–953. Zohar, D., & Luria, G. (2003). The use of supervisory practices as leverage to improve 984
902 Neal, A., Griffin, M. A., & Hart, P. M. (2000). The impact of organizational climate on safety behavior: A cross-level intervention model. Journal of Safety Research, 34 985
903 safety climate and individual behavior. Safety Science, 34(1), 99–109. (5), 567–577. 986
904 Newman, A., Ucbasaran, D., Zhu, F., & Hirst, G. (2014). Psychological capital: A 987
905 review and synthesis. Journal of Organizational Behavior, 35(S1), S120–S138. 988
Changquan He is a PhD Candidate of management science and engineering at the
Department of Construction Management and Real Estate, School of Economics and 989

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007
JSR 1614 No. of Pages 11, Model 5G
13 November 2019

C. He et al. / Journal of Safety Research xxx (xxxx) xxx 11

990 Management, Tongji University. His research interests include worker safety the construction of tall buildings. 1004
991 behavior, safety climate in construction sites, construction safety performance, etc. 1005
992 He has published in peer reviewed journals within the field of safety research. Yuting Chen, PhD, is an Assistant Professor at Department of Engineering Tech- 1006
993 nology and Construction Management, University of North Carolina at Charlotte. 1007
994 Guangshe Jia, PhD, is a Professor at the Department of Construction Management She obtained her doctoral degree from University of Toronto. Her research area 1008
995 and Real Estate, School of Economics and Management, Tongji University. His focuses on safety climate, individual resilience, interpersonal conflicts at work, etc. 1009
996 research area focuses on airport construction management, project management She has published several papers in peer reviewed journals within the field of safety 1010
997 maturity model, mega project controlling, etc. research. 1011
998 1012
999 Brenda McCabe, PhD, PEng, is a Professor at the Department of Civil Engineering, Jide Sun, PhD, is an Associate Professor at the Department of Construction Man- 1013
1000 University of Toronto and has been elected a Fellow of the Engineering Institute of agement and Real Estate, School of Economics and Management, Tongji University. 1014
1001 Canada (EIC). She is currently the inaugural director of the Building Tall research His research area focuses on construction project management, occupational 1015
1002 center. Her research interests are focused on construction engineering and include mobility of construction workers, value engineering of construction projects. 1016
1003 risk analysis, probabilistic modeling, worker safety, automated data collection, and 1017

Please cite this article as: C. He, G. Jia, B. McCabe et al., Impact of psychological capital on construction worker safety behavior: Communication compe-
tence as a mediator, Journal of Safety Research, https://doi.org/10.1016/j.jsr.2019.09.007

You might also like