Effects of Light Conditions on Tunnel Construction Workers’ Quality of Life and Work Productivity

Abstract

Higher lighting intensity promotes workers’ productivity but few studies focus on the problems caused by lower lighting intensities at tunnel construction sites without natural light. The purpose of this study is to explore the relationship between light intensity and workers’ sleep quality, alertness, vitality, and work productivity through a theoretical structural equation model based on the literature review. Data were collected through a questionnaire survey, and 5792 valid responses were adopted for the analysis. First, the results showed that greater lighting intensity promoted workers’ productivity directly and indirectly through three mediators: sleep quality, alertness, and vitality. Then, the whole sample was classified into four groups: high intensity/high comfort, moderate intensity/moderate comfort, moderate intensity/low comfort, and low intensity/low comfort. The clustered results showed that the lighting conditions of tunnel construction sites were generally poor and that lighting comfort promoted workers’ productivity to some extent. Besides, the influence of lighting intensity on productivity declined with improved lighting conditions while the impact of lighting intensity on workers’ physiological and psychological status showed differing trends as lighting conditions varied. However, the relationships between workers’ physiological and psychological status and productivity remained stable regardless of the lighting conditions. The findings could provide a reference for developing corresponding measures to promote workers’ productivity.

1. Introduction

1.1. Research Background

Duration is of great significance to construction projects in many respects [1]. On the one hand, construction projects generally require large amounts of human, material, and financial resources while project delay directly leads to an increase in the total cost including more human resources, higher construction equipment rental costs, higher material storage costs, etc. On the other hand, construction project sites may cause adverse effects on the surroundings, like traffic control and noise pollution, and delays in the construction period intensify these negative impacts. Consequently, it is of great importance to ensure the timely completion of construction projects, and workers’ productivity is one of the most important factors influencing the duration because the construction industry is a labor-intensive industry [2]. Construction lighting is a necessary condition for the normal operation of workers, as tunnel construction sites differ greatly from other project sites including residential and commercial buildings, roads, railways, and other projects above the ground in terms of lighting. Tunnel construction features poor lighting due to the special structure of the tunnels, as shown in Figure 1.

In tunnel construction sites, the light source is artificial lighting because tunnels are semi-closed, unlike other projects like buildings where sunlight can provide adequate lighting. Generally, the insufficient quantity of artificial lights due to the limited budget for construction costs leads to lower visibility inside the tunnels compared with other sites lit by sunlight. Inadequate lighting may influence workers’ productivity like wasting time searching for the operation tools, higher possibility of making mistakes, etc. Meanwhile, other risks may also cause negative impacts on the project schedule such as geological disasters, including rock bursts and unexpected accidents caused by dangerous sources like explosives and large machinery inside the tunnels [3]. The factors mentioned above are supposed to increase the uncertainty in the tunnel construction schedule, and thus tunnels are usually the key points determining the total duration of the whole project including highway or railway projects. In order to ensure timely completion, a series of management measures should be taken to improve the performance of the workers, which is the decisive factor for the project schedule.

Although the positive relationship between lighting intensity and workers’ performance has been validated in many fields [4,5,6,7], few studies focus on this problem in tunnel construction sites during the construction period. Nevertheless, the lighting conditions are generally poor inside the tunnels due to the unique structure of road tunnels and inadequate lighting installed near the tunnel excavation face. Consequently, it is necessary to investigate the actual lighting conditions across various tunnel construction sites and explore the effects of lighting intensity on the workers’ productivity to develop management measures to shorten the duration and reduce project construction costs. Except for lighting conditions, there are several other factors like sleep quality, alertness, and vitality that influence workers’ productivity [8,9,10] while the lighting intensity is also supposed to impact these factors [11,12,13], which makes these factors mediating variables between lighting conditions and workers’ productivity. In order to improve workers’ productivity, the management should take these mediators into consideration and put forward some corresponding measures to upgrade workers’ physiological and psychological status.

1.2. Research Gaps

Although a few studies have researched the relationship between the lighting environment and work performance in underground buildings [7,14], they mainly focus on this problem during the operation period when the lighting conditions during construction are much worse for underground projects. At the same time, there are distinct differences in the task content between the operation and construction periods. The workers’ jobs in the construction sites is more complex and include a series of construction instrument and machinery operations, and they may also encounter potential dangers due to improper operations or other hazards, such as geological disasters. So, it is necessary to validate whether higher lighting intensity actually promotes workers’ performance at tunnel construction sites. Further, lighting intensity may not be enough to reflect the environmental lighting conditions because some studies argued that the emotions of the occupants induced by the quality of the environment also influence the workers’ performance [15], and passive emotions lead to a degradation of task performance [16]. Consequently, lighting comfort containing light layout, luminance uniformity, and luminance contrast is supposed to be other parameters that reflect the heterogeneity of lighting conditions [7]. A comprehensive investigation into the lighting conditions across tunnel construction sites is beneficial in understanding the basic information regarding the lighting conditions and makes it possible to classify the construction sites into different groups based on lighting parameters. Meanwhile, the relationships between lighting intensity and workers’ sleep quality, alertness, vitality, and work productivity across different groups require further exploration in order to provide corresponding measures to promote the workers’ productivity in terms of lighting conditions in construction sites.

Consequently, the research gaps are as follows:

(1)

it is important to explore the relationship between the light intensity and workers’ work productivity at the tunnel construction sites, and several other factors like workers’ sleep quality, alertness, and vitality which may also influence workers’ productivity should also be taken into consideration;

(2)

a sample with adequate participants is required to grasp sufficient information across the tunnel construction sites regarding the lighting conditions including lighting intensity and comfort, and the sites could be classified into several groups based on the lighting conditions for further research;

(3)

it is of necessity to investigate the relationship heterogeneity regarding the light intensity and workers’ sleep quality, alertness, vitality, and work productivity across different groups, and thus corresponding interventions in terms of lighting conditions could be formulated to improve labor productivity and reduce the construction period.

1.3. Research Procedures and Objectives

This paper aims to study the influence of lighting conditions on workers’ sleep quality, alertness, vitality, and work productivity to fill the gaps mentioned above. A four-step approach for the research analysis was adopted, as shown in Figure 2, and the procedure is as follows:

(1)

a theoretical model was proposed based on the literature review, and the model reveals that increased lighting intensity shows a positive relationship with workers’ productivity while the relationship was mediated by three factors which are workers’ sleep quality, alertness, and vitality;

(2)

this step focused on all respondents, and the measurement model was employed to examine the convergent and discriminant validity of all constructs while structural equation modeling (SEM) was used to explore the relationship between lighting intensity and workers’ sleep quality, alertness, vitality, and productivity;

(3)

the k-means algorithm was employed to cluster the whole sample into four groups based on the lighting intensity and comfort: high intensity/high comfort, moderate intensity/moderate comfort, moderate intensity/low comfort, and low intensity/low comfort; and whether the workers’ sleep quality, alertness, vitality, and work productivity showed statistical differences across four groups was tested;

(4)

the last step emphasized the relationship heterogeneity of the lighting intensity, workers’ productivity, and three mediators across the four groups. The path coefficients and mediation effects were taken into consideration.

2. Literature Review and Research Framework

2.1. The Relationship between Lighting Intensity and Work Productivity

Light is critical for human beings as it provides people with the ability to see and perform targeted activities, and thus illuminance is a significant environmental parameter in achieving overall indoor environmental quality (IEQ) satisfaction. Various studies have validated the important role of the lighting environment in predicting the working performance of personnel [14,17].

Generally, the results of existing research are in line with the conclusion that high illumination is beneficial for workers’ performance except in the case of excessive lighting [14]. A previous study conducted a field test in a luminaire factory in Finland and found that the lighting intensity showed a significant positive correlation with productivity [4]. Meanwhile, the results also indicated that the workers had certain preferred lighting levels, and a controllable task-lighting system was beneficial for increasing their work efficiency. Another study also explored the relationship between lighting intensity and workers’ productivity [18] and found that lighting intensity and color had pronounced effects on workers’ productivity. Adequate lighting in workplaces influences workers’ comfort and satisfaction, which are facilitators of higher levels of productivity and efficiency [19]. Consequently, proper lighting intensity is an important factor when managers aim to promote their workers’ efficiency. Considering the lack of natural light near the tunnel excavation face, workers carry out their work with the help of artificial lights. In order to reduce the project costs, the number of lights is generally limited, and thus lighting conditions are relatively poor near the excavation face. It is reasonable that increasing lighting intensity properly could promote work productivity at tunnel construction sites. Based on the previous studies above, we can propose the hypothesis:

H1: 

Lighting intensity is positively associated with work productivity at tunnel construction sites.

2.2. The Mediating Variables between Lighting Intensity and Work Efficiency

Apart from the lighting intensity, there are several other factors that may affect the workers’ efficiency, like sleep quality, alertness, and vitality, and these factors are also influenced by the lighting conditions at the workplace. Consequently, there are a number of mediating variables between lighting intensity and work efficiency.

2.2.1. Sleep Quality

Light is supposed to be the dominant environmental parameter influencing the sleep-wake cycle. Many studies have explored the relationship between the amount and timing of light exposure and sleep outcome [20,21,22,23], and the positive relationship between lighting intensity and sleep has been validated. Dautovich et al. have investigated the specific conditions under which light is optimally associated with better sleep and found that bright light exposure during the day had a positive impact on objectively assessed sleep outcomes compared to dim and moderate light and also positively influenced subjectively assessed sleep outcomes compared to moderate light [20]. Lok et al. also found that bright light exposure during wakefulness promoted sleep quality in healthy men [23], and this phenomenon may result from the increased sleep pressure buildup or prevention of micro-sleep. Meanwhile, exposure to light during the night may lead to negative effects on sleep quality including shallow sleep, frequent arousal, and adverse impact on the brain oscillations related to sleep depth and stability [24]. Consequently, a 24 h lighting system with enhanced daytime brightness and restricted nocturnal light exposure is supposed to be a therapy to improve people’s sleep quality [13]. The workers at the tunnel construction sites spend most of their time near the excavation face during work, and the artificial lights are the only light source they are exposed to during work. It is reasonable to assume that the lighting intensity of artificial light is closely related to the workers’ sleep quality.

Sleep health, characterized by subjective satisfaction, appropriate timing, adequate duration, and high efficiency, plays an important role in promoting physical and mental well-being [25]. Previous studies have proved that sleep health is associated with mortality [26], metabolic syndrome [27], coronary artery disease [28], depression [29], and impaired neuro-behavioral performance [30]. Due to the close relationship between sleep health and physiological and psychological state, sleep deficiency and sleep disorders are supposed to reduce work productivity [9,31]. Swanson et al. explored the reciprocal relationships between sleep and work outcomes based on the data from the National Sleep Foundations 2008 Sleep in America telephone poll [9]. The results showed that sleep disorders substantially increase the likelihood of negative work outcomes, including accidents, absenteeism, and presenteeism. A previous study also indicates that poor sleep quality is predictive of increased chronic and acute work-related fatigue, and that low work productivity would follow [32]. Park et al. found that 79.79% of the Korean clinical nurses working in medium- and small-sized hospitals featured poor sleep quality, and poor sleep quality like sleep disturbances decreased the nurses’ productivity [33]. Providing a worksite healthy sleep program including longer intervals between shift–work cycles and a longer break time after night shift work could be the solution to this problem. Ricarda and Jana found that sleep quality was positively related to work engagement [34], which promotes workers’ efficiency. Consequently, high-quality sleep is a necessity in terms of maintaining workers’ high efficiency.

Therefore, we propose the following hypotheses:

H2: 

Sleep quality mediates the relationship between lighting intensity and work productivity.

H2a: 

Lighting intensity is positively linked with sleep quality at tunnel construction sites.

H2b: 

Sleep quality is positively associated with work productivity at tunnel construction sites.

2.2.2. Alertness

Except for its role in vision, light also influences several physiological and psychological processes, which are known as non-image-forming (NIF) effects of light, like alertness. Alertness is commonly used to denote the opposite of sleep and indeed seems almost synonymous with wakefulness [35]. In psychological studies, the term, alertness, is used to denote a state of sustained attention, in which the person is capable of responding adequately to sparse stimuli [36]. Consequently, it not only relates to the person’s arousal level but also reflects the capacity to achieve and maintain a high level of cognitive performance [37]. A number of previous studies have explored the relationship between lighting intensity and alertness [38,39,40]. A review study revealed that the higher intensity of polychromatic white light contributed to subjective alertness in a majority of studies [41]. In another study [11], the positive effects of higher-intensity white light exposure on alertness were validated, and the results also showed that white light exposure promoted cognitive functions depending on other factors like sleep drive and exposure time. Consequently, the lighting intensity at the tunnel construction site is supposed to positively influence workers’ alertness.

For many professions, sustained attention is necessary for effective and efficient work performance. For example, workers should concentrate on the tasks at hand when the job is complex, which reduces the likelihood of errors and increases the quality and efficiency of their work [42]. On the other hand, many industries involve machinery or hazardous environments, which pose a great threat to the workers, and sustained attention is crucial for maintaining safety standards to avoid accidents, which can lead to downtime and decreased productivity [43]. Lerman et al. argued that well-rested and alert employees are critical to safe and productive operations, which is beneficial in promoting productivity at the workplace [10]. For the work in tunnel construction sites, the job involves the operations of various types of machinery and tools while many potential dangers are also present around the workers, like explosives and heavy machinery. Consequently, being alert or responding adequately to stimuli around the workers is considered beneficial in promoting productivity in tunnel construction sites.

Based on the statements above, we propose the following hypotheses:

H2: 

Alertness mediates the relationship between lighting intensity and work productivity.

H2a: 

Lighting intensity is positively associated with alertness.

H2b: 

Alertness is positively related to work productivity at tunnel construction sites.

2.2.3. Vitality

In addition to activation of the visual system, photoreceptors in the human retina are capable of signaling light information to brain areas involved in the regulation of behavior, mood, and physiology [44]. Vitality, which refers to the positive feeling of having energy or resources available to the self, is among the non-image-forming (NIF) effects of light [45]. Previous studies have explored the relationship between lighting intensity and vitality and have demonstrated that bright light has beneficial effects on physiological arousal or feeling energetic [46,47]. Smolders et al. also found that exposure to white light with a higher illuminance level also had beneficial effects on vitality during daytime, even in the absence of sleep and light deprivation [12]. Partonen and Lönnqvist validated the positive effects of prolonged exposure to high levels of illuminance (2500 lx at the eye) on vitality during the darker winter months in Finland [48]. Smolders et al. investigated daily light exposure and its relationship with vitality in everyday settings, and the results showed that the impact of daily light exposure on vitality was more pronounced than those of person characteristics, time of day, activity patterns, and sleep duration for the previous night [49]. Therefore, the higher lighting intensity is supposed to promote people’s vitality.

Experiences of vitality are critical to mental well-being, health, and performance in terms of all aspects of life like work, social activities, and others. Consequently, high levels of vitality contribute to improved productivity, especially for labor-intensive industries. Jonge et al. argued that vitality at work was important for building a healthier, more engaged, sustainable, and productive workforce, which undoubtedly improves the efficiency of companies [8]. Tunnel construction workers are engaged in heavy manual work like drilling holes in rocks, installing steel support, transporting soil slag, and so on, which requires them to maintain sufficient physical strength to complete the heavy work [50]. Therefore, workers with more vitality are considered more productive in the tunnel construction sites.

Consequently, we propose the following hypotheses:

H3: 

Vitality mediates the relationship between lighting intensity and work productivity.

H3a: 

Lighting intensity is positively associated with vitality.

H3b: 

Vitality is positively related to work productivity at tunnel construction sites.

2.3. Research Framework

Based on the hypotheses above, the proposed model which aims to explore the effects of lighting intensity on workers’ sleep quality, alertness, vitality, and work productivity in this study is shown in Figure 3. It can be found that lighting intensity is the primary factor for predicting work productivity, while lighting intensity also influences productivity indirectly through three mediators (workers’ sleep quality, alertness, and vitality).

3. Methods

3.1. Sample and Data Collection

To explore the influence of lighting conditions on workers’ productivity, a questionnaire survey was conducted to collect self-reported data from tunnel construction workers, and a sample large size was required in this study to obtain comprehensive information regarding the lighting conditions of various tunnel construction sites. The internet-based survey has been proved an effective data collection approach which has been widely employed in numerous studies. For tunnel construction, which features heavy labor, most workers are male in the construction sites while female workers are generally responsible for rear service work. Consequently, a popular online questionnaire platform, Questionnaire Star, was adopted to send survey invitations to tunnel construction workers from a large number of sites with different lighting conditions in China. A total of 7765 male workers participated in this survey, which lasted from February 2023 to June 2023. All the participants signed the informed consent, and it was ensured that all their personal information was protected during the whole process and the data were only used for scientific research. In order to eliminate the unexpected effects of natural light outside the tunnels, workers who do not enter the tunnels are excluded from this study. After the removal of invalid responses (wrong selections in the trap questions and unreasonable consecutive replies) and respondents who did not meet the inclusion criteria, 5792 valid responses were employed for the next analysis. The research protocol in this study was approved by the Ethics Committee of Sichuan Normal University.

Table 1. Basic demographic parameters about participants (Date of this survey: February 2023 to June 2023).

Demographic Parameter	Item	Frequency	Percentage/%
Age
	≤25	609	10.51
	26 to 30	815	14.07
	31 to 40	1356	23.41
	41 to 50	1798	31.04
	51 to 60	1108	19.13
	>60	106	1.83
Education
	Secondary or Below	1321	22.81
	High School or Equivalent	2458	42.44
	College Diploma or Equivalent	1459	25.19
	Master’s Degree or Equivalent	498	8.60
	Doctoral or Equivalent	56	0.97
Marital Status
	Unmarried	1036	17.89
	Married without Children	408	7.04
	Married with Children	3789	65.42
	Divorced	559	9.65
Income Per Month (unit: CNY)
	≤5000	1547	26.71
	5000 to 8000	1756	30.32
	8000 to 10,000	1258	21.72
	10,000 to 15,000	789	13.62
	15,000 and above	374	6.46
	Not Applicable	68	1.17

lists the respondents’ demographic information including age, education, marital status, and income. The median age of the workers is between 41 and 50, and the respondents aged from 41 to 50 and from 31 to 40 account for 31.04% and 23.41% of the whole sample, respectively. Among the respondents, the majority (42.44%) hold a high school degree, 25.19% of the interviewed population have a college diploma, and others have lower (22.81%) or higher (9.57%) educational qualifications, which match the real-world conditions. Most of the workers (65.42%) are married with children while a proportion (17.89%) are unmarried. Besides these groups, fewer workers are married without children or divorced. The income distribution of the construction workers is in line with the industry situation. The income of the majority (30.32%) ranges from 5000 to 8000 CNY, and only a small portion of the workers (6.46%) earn over 15,000 yuan. In general, the respondents’ demographic information fits the industry statistics well.

3.2. Measurements

The questionnaire employed in this study has four parts: (1) respondents’ socio-demographic information; (2) lighting conditions of the tunnel construction sites; (3) construction workers’ physiological and psychological factors that may influence their work productivity; (4) workers’ perceived productivity.

The first part records the selected workers’ demographic information including age, education, marital status, and income. The second part focuses on the targeted construction sites’ lighting conditions including lighting intensity and comfort. As for the lighting intensity, the illuminance of the workplace is the most direct index reflecting the intensity of artificial lighting. However, it is difficult to obtain the illuminance of all the selected construction sites due to the huge size of the sample. Consequently, the perceived lighting intensity of the workers is adopted in this study and obtained by three measuring items like ‘The lights in my workplace make me see everything clearly’. Unlike lighting intensity which is an objective physical quantity, lighting comfort reflects the lighting design and arrangement related to the workers’ emotions, like the overall arrangement of the lights, lighting contrast, color temperature, uniformity of the lighting, types of the lights, and other factors. Three items adapted from previous studies [16,51] are adopted to reflect the lighting comfort with more emphasis on workers’ emotions induced by the lighting design and arrangement. The third part consists of workers’ sleep quality, alertness, and vitality which may influence their productivity to varying degrees. As for sleep quality, the schedule for tunnel construction workers is different from other occupations, and their sleep time varies according to the construction scheduling and their work shift. Consequently, three measuring items reflecting the workers’ general sleep quality were employed by referring to the Pittsburgh sleep quality index [52]. Alertness is the state of being awake, aware, attentive, and prepared to act or react, and thus alertness is defined as achieving and maintaining a state of sustained attention to respond adequately to a given stimulus [11]. This study referred to not only several previous research studies [11,53] but also two other similar self-report questionnaires targeted to reflect subjective alertness including the Karolinska Sleepiness Scale [53] and the Stanford Sleepiness Scale [54]. Finally, three items featuring feeling alert, reaction time, and being awake, which are supposed to be concise and suitable for tunnel construction workers, were included to measure the subjective alertness of the participants. The subjective trait level vitality scale [55] was adopted to reflect the workers’ vitality during work. The scale contains seven energy-related items which are seen as reflecting, from a content perspective, an adequate definition of a phenomenological sense of aliveness and energy. However, one reverse question was deleted in this study to avoid ambiguity, and the value of Cronbach’s Alpha was 0.834 for the remaining six items. As for the perceived work productivity, this study referred to a previous study [56], and three items were adopted.

Table A1. Psychological attributes in the questionnaire.

Constructs	Code	Items
Lighting intensity
	LI1	The lighting intensity in my workplace is very high
	LI2	The number of lights is sufficient in my workplace
	LI3	The lights in my workplace make me see everything clearly
Lighting comfort
	LC1	The lighting of my construction site makes me feel very comfortable
	LC2	I am very satisfied with the lighting of my construction site
	LC3	The lighting of my construction site puts me in a good mood
Sleep quality
	SQ1	My sleep quality is very high
	SQ2	My sleep duration is enough for me
	SQ3	I am very satisfied with my sleep
Alertness
	AL1	I feel alert during work
	AL2	I react quickly no matter what happens around me
	AL3	I am wide awake during work
Work productivity
	WP1	I can complete my work efficiently
	WP2	I feel productive during work
	WP3	I am very satisfied with my work efficiency

lists all the items in this study apart from measuring items reflecting vitality. Except for the first part, the remaining parts use the five-point Likert scale (completely disagree = 1, disagree = 2, neutral = 3, agree = 4, completely agree = 5).

When the questionnaire design was completed, the research team conducted two pilot survey rounds. A total of 12 professionals in tunnel construction management were invited to participate in the first round, while 35 tunnel construction workers participated in the second round. The respondents of the two rounds were asked to provide advice regarding how to revise the questionnaire. Then, the research team made the modifications to the questionnaire accordingly. In addition to the measuring items mentioned above, one question, ‘Do you usually work inside the tunnel’, was adopted to screen out the participants who do not work inside the tunnels. What is more, two trap items that require the respondents to choose a certain answer were randomly inserted into the questionnaire to eliminate invalid responses.

3.3. Data Analysis

As shown in Figure 2, a three-step approach for data analysis was adopted in this study, excluding step 1 where an overall theoretical model was built based on the literature review.

When focusing on the relationship between the lighting intensity, workers’ productivity, and their mediators of all respondents (step 2) or each cluster (step 4), structural equation modeling (SEM) was employed to test the hypotheses. SEM features the advantages of factor analysis, multiple regression analysis, path analysis, and other approaches to reveal the effect of influencing factors on the outcome factor and the interrelationship between influencing factors [57]. In the third step, the respondents were divided into different clusters according to the lighting conditions of the construction sites including lighting intensity and comfort. The number of clusters was set to be four considering the compactness and distinctiveness, and the k-means algorithm was employed for the clustering analysis [58]. Then, the study conducted a one-way analysis of variance (ANOVA) to examine whether the lighting conditions and workers’ physiological and psychological factors and productivity have statistically significant differences among the four clusters based on the cluster analysis result. Within step 4, SEM was conducted to investigate the relationship between the lighting intensity and workers’ sleep quality, alertness, vitality, and work productivity of each cluster, and the results of each cluster were compared with those of the overall model to obtain the influencing characteristics and patterns of the lighting intensity.

This study employs SmartPLS 3.3.9 (SmartPLS GmbH, Oststeinbek, Germany) for SEM and SPSS 24 (IBM, Armonk, NY, USA) for ANOVA and clustering analysis.

4. Results

4.1. Overall Analysis

This part emphasizes the influence of lighting intensity and workers’ physiological and psychological factors on their productivity in the overall analysis. The SEM comprises the analysis of measurement and structural models, and the details are as follows.

4.1.1. Measurement Modeling

This section presents the results of the convergent validity (CV) and discriminant validity (DV) tests to verify the reliability and effectiveness of the constructs.

The CV tests aim to evaluate the correlation between measuring items under the same construct. Four indices are included: (1) Cronbach’s alpha, (2) Composite Reliability (CR), (3) Average Variance Extracted (AVE), and (4) factor loading. Cronbach’s alpha and CR reflect the internal consistency and reliability of the constructs, and the values of Cronbach’s alpha and CR are expected to be greater than 0.70 [57,59,60]. Average Variance Extracted (AVE) represents the comprehensive explanatory capacity of latent variables for all measured items, and the cut-off value is 0.5 to achieve sufficient explanatory ability [57,60]. Factor loading (also called standardized outer loading) reflects the correlation coefficient between the measuring items and constructs, and it is recommended that the factor loading value of each item should be higher than 0.70 [57].

Table 2. Reliability and convergent validity test.

Construct	Items	Factor Loading	Cronbach’s Alpha	Composite Reliability	AVE
Lighting intensity
	LI1	0.675	0.750	0.792	0.560
	LI2	0.816
	LI3	0.748
Lighting comfort
	LC1	0.851	0.812	0.848	0.651
	LC2	0.765
	LC3	0.803
Sleep quality
	SQ1	0.803	0.746	0.790	0.557
	SQ2	0.745
	SQ3	0.686
Alertness
	AL1	0.768	0.796	0.856	0.665
	AL2	0.896
	AL3	0.776
Work productivity
	WP1	0.856	0.852	0.881	0.711
	WP2	0.834
	WP3	0.839

lists the results of the reliability and convergent validity tests of all constructs. From Table 2, it can be found that the values of Cronbach’s alpha range from 0.746 to 0.852 while the values of CR are from 0.790 to 0.881, which satisfy the requirements. Meanwhile, the lowest AVE value of the constructs in this study is 0.557, which is larger than the minimum threshold value. The scope of factor loading is between 0.675 and 0.896, and most of the loadings meet the requirements indicating a strong correlation between the measuring items and constructs apart from a few exceptions that are close to 0.70.

The DV test evaluates whether a construct varies from the model’s remaining constructs. In this study, the Fornell-Larcker criterion and Heterotrait-Monotrait (HTMT) ratio are adopted for the DV examination to evaluate the validity of the discrimination between constructs.

Table 3. Fornell–Larcker criteria.

Constructs	LI	LC	SQ	AL	VI	WP
LI	0.748
LC	0.654	0.807
SQ	0.578	0.379	0.746
AL	0.604	0.409	0.521	0.815
VI	0.643	0.336	0.476	0.431	0.809
WP	0.704	0.465	0.621	0.598	0.586	0.843

and

Table 4. HTMT criteria.

Constructs	LI	LC	SQ	AL	VI	WP
LI
LC	0.678
SQ	0.503	0.486
AL	0.645	0.711	0.369
VI	0.416	0.623	0.487	0.471
WP	0.378	0.521	0.547	0.653	0.539

list the results of the Fornell-Larcker and HTMT criteria. In terms of the Fornell–Larcker criterion, sufficient support is provided for the discriminant validity of the constructs because the square roots of AVE are greater than each latent variable’s inner construct correlation coefficients [61]. As for the other standard, it is recommended that the HTMT ratio should maintain a level of no more than 0.9 [62], and the results of this study satisfy the requirement.

4.1.2. Overall Structural Equation Modeling Results

The study conducts structural modeling with bootstrapping estimation at a 95% confidence interval and with 5000 samples to explore the statistical correlation between latent variables and the mediation effects between lighting intensity and workers’ productivity, which is shown in Figure 4.

Table 5. Overall structural modelling analysis results (* denotes p < 0.05; *** denotes p < 0.001).

Hypothesis		Tested Relationship	Beta	Standard Deviation	t-Value	p-Value
H1
		LI -> WP	0.316	0.029	10.89	0.000 ***
H2
	H2a	LI -> SQ	0.586	0.032	18.31	0.000 ***
	H2b	SQ -> WP	0.292	0.062	4.71	0.000 ***
H3
	H3a	LI -> AL	0.458	0.009	50.89	0.000 ***
	H3b	AL -> WP	0.186	0.083	2.249	0.025 *
H4
	H4a	LI -> VI	0.553	0.03	18.44	0.000 ***
	H4b	VI -> WP	0.242	0.036	6.72	0.000 ***

lists the details of the overall structural modeling analysis. It can be found that not only lighting intensity but also workers’ physiological and psychological factors influence their productivity. Among all the predictors, the influence of the lighting intensity (β = 0.316, p < 0.000) is the greatest, which indicates that the level of lighting intensity is the main and primary influencing factor predicting work productivity. Also, sleep quality (β = 0.292, p < 0.000) and vitality (β = 0.242, p < 0.000) play an important role. In contrast, alertness (β = 0.186, p < 0.025) has a smaller effect on workers’ productivity. It is worth noting that the lighting intensity also significantly impacts workers’ physiological and psychological factors apart from productivity. In particular, the lighting intensity has a more substantial positive impact on sleep quality (β = 0.586, p < 0.000), alertness (β = 0.458, p < 0.000), and vitality (β = 0.553, p < 0.000). Consequently, the H1, H2a, H2b, H3a, H3b, H4a, and H4b of the overall model are supported. The predictive accuracy of the constructs was evaluated by the determination coefficient, R², and the minimum acceptable value is 0.10. In this study, the R² values of the constructs are greater than 0.10 (SQ R² = 0.343; SQ R² = 0.210; SQ R² = 0.306; SQ R² = 0.583).

Table 6. Mediation results of the whole sample (** denotes p < 0.01; *** denotes p < 0.001).

Path	Indirect Effect	Standard Deviation	T-Statistics	p-Value
LI -> SQ -> WP	0.171	0.025	6.844	0.000 ***
LI -> AL -> WP	0.085	0.029	2.938	0.003 **
LI -> VI -> WP	0.134	0.019	7.043	0.000 ***

lists the results of mediation effects. The results show that the indirect effects of the three factors are statistically significant, which indicates that sleep quality, alertness, and vitality mediate the relationship between lighting intensity and workers’ productivity. Then, H2, H3, and H4 are supported for the whole sample.

4.2. Group Analysis

This section lists the lighting condition characteristics of the four clusters and the relationship heterogeneity of the lighting intensity, workers’ productivity, and three mediators across the four groups.

4.2.1. Four Groups Based on Lighting Intensity and Comfort

To balance the compactness and distinctiveness of the clustering results, the number of clusters is determined as four. Consequently, all the respondents are classified into four groups based on the lighting conditions of the construction sites where they work, and the four groups are defined as (1) high intensity/high comfort (cluster 1); (2) moderate intensity/moderate comfort (cluster 2); (3) moderate intensity/low comfort (cluster 3); (4) low intensity/low comfort (cluster 4). Figure 5 shows the distribution of the lighting conditions across the four clusters, and the lighting condition characteristics of the four groups are as follows:

(i)

it can be indicated that lighting intensity is positively related to lighting comfort because few cases feature high intensity and low comfort or the opposite; therefore, it is normal for the results of clusters 1 and 4 with the highest or lowest lighting intensity and comfort at the same time;

(ii)

as for clusters 2 and 3, the lighting intensity is comparable while a significant difference is witnessed for the two groups in terms of lighting comfort; the divergence regarding the perceived visual comfort may result from the light layout, luminance uniformity, color temperature, and other factors except for the intensity.

The details of the four clusters are listed in Figure 6. We can see that only 378 cases belong to cluster 1 and account for the smallest proportion of the whole sample (6.52%) while cluster 4 consists of 1568 cases and occupies a certain proportion of all the respondents (27.10%), which reflects that the lighting conditions are not satisfying across the tunnel construction sites. Among the four groups, the number of cases from cluster 3 is the largest, and thus 44.11% of the respondents work in construction sites where the lighting intensity is moderate but the visual comfort is rather low. The number of cases belonging to cluster 2 is slightly less than that of cluster 4, and accounts for 22.26% of the whole sample. The values of the lighting intensity and comfort decline from cluster 1 to cluster 4 and are 4.32 and 4.39 with standard deviations of 0.52 and 0.42, respectively, for cluster 1. The lighting intensity and comfort decrease to 3.35 and 3.20 with standard deviations of 0.57 and 0.42 for cluster 2, and the lighting intensity of cluster 3 is 2.80 and a bit lower than that of cluster 2 while the lighting comfort of cluster 3 is much lower than that of cluster 2. The cases belonging to cluster 4 work in the environment with the worst lighting conditions featuring the lowest lighting intensity and comfort which are 1.54 and 1.81 with standard deviations of 0.39 and 0.51, respectively.

The study compares the characteristics of lighting conditions and workers’ sleep quality, alertness, vitality, and work productivity of each cluster by ANOVA analysis, and the results are listed in

Table A2. The ANOVA results and descriptive statistics of four clusters (*** denotes p < 0.001).

		Light Intensity	Light Comfort	Sleep Quality	Alertness	Vitality	Work Productivity
	p-value (ANOVA)	0.000 ***	0.000 ***	0.000 ***	0.001 ***	0.000 ***	0.000 ***
Overall sample
	Mean	2.68	2.37	3.26	3.22	3.52	3.31
	SD	1.24	1.19	1.34	1.19	1.26	1.30
Cluster 1
	Mean	4.32	4.39	4.33	4.19	4.45	4.18
	SD	0.52	0.42	0.61	0.64	0.39	0.71
Cluster 2
	Mean	3.35	3.19	3.54	3.87	3.78	3.56
	SD	0.57	0.42	0.59	0.57	0.49	0.63
Cluster 3
	Mean	2.80	2.00	3.21	3.11	3.47	3.24
	SD	0.43	0.41	0.64	0.71	0.64	0.78
Cluster4
	Mean	1.54	1.81	2.84	2.64	3.15	3.01
	SD	0.39	0.51	0.63	0.51	0.55	0.58

.

From Figure 6 and Table A2, significant differences are witnessed for four clusters in terms of lighting conditions. As for workers’ physiological and psychological parameters, similar phenomena are also found among various clusters and the workers of cluster 1 get the highest average score (Sleep quality = 4.33, Alertness = 4.19, Vitality = 4.45, Work productivity = 4.18) across all the parameters thanks to the good lighting conditions, which indicates that the good lighting conditions promote workers’ physiological and psychological status. Cluster 2 features the second highest scores regarding the four parameters (Sleep quality = 3.54, Alertness = 3.87, Vitality = 3.78, Work productivity = 3.56) while workers’ physiological and psychological parameters of cluster 3 (Sleep quality = 3.21, Alertness = 3.11, Vitality = 3.47, Work productivity = 3.24) are lower than the average values of the whole sample and those of cluster 2 due to the poorer lighting conditions. It should be noted that the largest difference between clusters 2 and 3 exists in terms of workers’ alertness, and other parameters feature smaller differences. The workers from cluster 4 scored lowest in all parameters (Sleep quality = 2.84, Alertness = 2.64, Vitality = 3.15, Work productivity = 3.01), which are in line with the results of the lighting conditions. To summarize, the characteristics of the lighting conditions are associated with workers’ physiological and psychological parameters, and it can also be indicated that cluster 4 features the highest existing reserve of productivity improvement among the four clusters due to the lowest work productivity.

4.2.2. Results of the Structural Equation Modeling across Different Groups

This study performed structural modeling analysis on four clusters to explore the direct and indirect effects of lighting intensity on workers’ productivity across different groups. From Table 5 and Table 6, it can be found that lighting intensity plays an important role in predicting workers’ productivity while workers’ sleep quality, alertness, and vitality meditate the relationship between them. However, whether lighting intensity is still an important predictor of workers’ productivity and mediating effects of workers’ sleep quality, alertness, and vitality still hold among the four clusters is worth investigating, and the results are helpful for providing a reference for construction sites with various lighting conditions to develop measures to promote workers’ productivity.

Table 7. The hypothesis test results of the four groups (* denotes p < 0.05; *** denotes p < 0.001).

Cluster	Hypothesis	Tested Relationship	Beta	Standard Deviation	t-Value	p-Value
High intensity /high comfort
	H1	LI -> WP	0.253	0.027	9.370	0.000 ***
	H2a	LI -> SQ	0.541	0.036	15.028	0.000 ***
	H2b	SQ -> WP	0.294	0.071	4.141	0.000 ***
	H3a	LI -> AL	0.449	0.013	34.538	0.000 ***
	H3b	AL -> WP	0.187	0.081	2.309	0.021 *
	H4a	LI -> VI	0.570	0.028	20.357	0.000 ***
	H4b	VI -> WP	0.240	0.031	7.742	0.000 ***
Moderate intensity/moderate comfort
	H1	LI -> WP	0.315	0.031	10.161	0.000 ***
	H2a	LI -> SQ	0.564	0.03	18.800	0.000 ***
	H2b	SQ -> WP	0.287	0.058	4.948	0.000 ***
	H3a	LI -> AL	0.453	0.008	56.625	0.000 ***
	H3b	AL -> WP	0.180	0.086	2.093	0.037 *
	H4a	LI -> VI	0.563	0.033	17.061	0.000 ***
	H4b	VI -> WP	0.238	0.041	5.805	0.000 ***
Moderate intensity/low comfort
	H1	LI -> WP	0.335	0.036	9.306	0.000 ***
	H2a	LI -> SQ	0.581	0.044	13.205	0.000 ***
	H2b	SQ -> WP	0.289	0.055	5.255	0.000 ***
	H3a	LI -> AL	0.473	0.012	39.417	0.000 ***
	H3b	AL -> WP	0.184	0.09	2.044	0.041 *
	H4a	LI -> VI	0.521	0.036	14.472	0.000 ***
	H4b	VI -> WP	0.243	0.043	5.651	0.000 ***
Low intensity /low comfort
	H1	LI -> WP	0.423	0.029	14.586	0.000 ***
	H2a	LI -> SQ	0.612	0.048	12.750	0.000 ***
	H2b	SQ -> WP	0.296	0.052	5.692	0.000 ***
	H3a	LI -> AL	0.462	0.016	28.875	0.000 ***
	H3b	AL -> WP	0.188	0.079	2.380	0.018 *
	H4a	LI -> VI	0.492	0.039	12.615	0.000 ***
	H4b	VI -> WP	0.236	0.051	4.627	0.000 ***

and

Table 8. The mediation test results of the four groups (** denotes p < 0.01; *** denotes p < 0.001).

Cluster	Path	Indirect Effect	Standard Deviation	T-Statistics	p-Value
High intensity/high comfort
	LI -> SQ -> WP	0.159	0.026	6.117	0.000 ***
	LI -> AL -> WP	0.084	0.031	2.708	0.007 **
	LI -> VI -> WP	0.137	0.016	8.550	0.000 ***
Moderate intensity/moderate comfort
	LI -> SQ -> WP	0.162	0.031	5.222	0.000 ***
	LI -> AL -> WP	0.082	0.025	3.262	0.001 **
	LI -> VI -> WP	0.134	0.021	6.381	0.000 ***
Moderate intensity/low comfort
	LI -> SQ -> WP	0.168	0.028	5.997	0.000 ***
	LI -> AL -> WP	0.087	0.022	3.956	0.000 ***
	LI -> VI -> WP	0.127	0.014	9.043	0.000 ***
Low intensity/low comfort
	LI -> SQ -> WP	0.181	0.032	5.661	0.000 ***
	LI -> AL -> WP	0.087	0.019	4.571	0.000 ***
	LI -> VI -> WP	0.116	0.024	4.838	0.000 ***

list the results of the hypotheses and the mediation test across the four clusters (i.e., path coefficients/indirect effects, standard deviation, t-value, and p-value).

From Table 7, we can see that the impact of lighting intensity on workers’ productivity increases from cluster 1 (β = 0.253, p < 0.001) to cluster 4 (β = 0.423, p < 0.001), which reflects that the lighting intensity plays a more important role when the lighting conditions are poor. A similar phenomenon is also witnessed regarding the relationship between lighting intensity and sleep quality, and the correlation is the smallest with the highest lighting intensity (cluster 1: β = 0.541, p < 0.001; cluster 2: β = 0.564, p < 0.001; cluster 3: β = 0.581, p < 0.001; cluster 4: β = 0.612, p < 0.001). As for alertness, the impact of lighting intensity is comparable across different lighting conditions (cluster 1: β = 0.449, p < 0.001; cluster 2: β = 0.453, p < 0.001; cluster 3: β = 0.473, p < 0.001; cluster 4: β = 0.462, p < 0.001) and no huge difference is witnessed. In terms of the correlation between vitality and lighting intensity, the influence of lighting intensity is stronger (cluster 1: β = 0.570, p < 0.001; cluster 2: β = 0.563, p < 0.001) when the lighting conditions are better. Except for lighting intensity, workers’ self-reported sleep quality, alertness, and vitality all have a significant impact on their work productivity. Among the three physiological and psychological parameters, the effect of sleep quality is the highest while alertness presents the lowest performance but still shows a significant relationship with the respondents’ productivity. Unlike the effect of lighting intensity on the workers’ productivity, which varies with the change in the lighting conditions, the impact of workers’ sleep quality, alertness, and vitality all show stable influence across the four clusters regardless of the lighting conditions.

Table 8 summarizes the mediation effect results of four clusters. It can be found that the lighting intensity exerts a significant indirect effect on workers’ productivity among the four groups, which corresponds with the overall analysis results. The indirect influence of lighting intensity through sleep quality shows the highest levels among the three mediators and increases with poorer lighting conditions (cluster 1: indirect effect = 0.159, p < 0.001; cluster 2: indirect effect = 0.162, p < 0.001; cluster 3: indirect effect = 0.168, p < 0.001; cluster 4: indirect effect = 0.181, p < 0.001) because of the stronger correlation between lighting intensity and sleep quality under poorer lighting intensity and stable impact of sleep quality on the work productivity. On the contrary, the indirect influence of lighting intensity through vitality declines (cluster 1: indirect effect = 0.137, p < 0.001; cluster 2: indirect effect = 0.134, p < 0.001; cluster 3: indirect effect = 0.127, p < 0.001; cluster 4: indirect effect = 0.116, p < 0.001) when the lighting condition becomes poorer due to the weaker impact of lighting intensity on the vitality under poorer lighting conditions. The indirect effect of lighting intensity through alertness presents the lowest performance, remains stable across the four clusters, and is in the vicinity of the overall sample analysis (indirect effect = 0.085, p < 0.01).

5. Discussion

Workers’ productivity is a decisive factor in determining the duration of construction projects. Unlike other projects above the ground in terms of lighting, artificial lighting is a necessity for workers in tunnel construction sites where there is no natural light near the excavation face, and the relationship between lighting intensity and productivity requires investigation in tunnel construction sites while several factors like sleep quality, alertness, and vitality may mediate the relationship. This study explored the influence of lighting conditions on workers’ sleep quality, alertness, vitality, and productivity based on a huge sample across various construction sites.

The results of the overall structural modeling analysis show that lighting intensity and workers’ sleep quality, alertness, and vitality make a positive contribution to workers’ perceived productivity while the impact of lighting intensity is the greatest. The image-forming (IF) effects of light make it possible for people to see and perform targeted activities, and increasing lighting intensity properly could promote workers’ productivity, which has been validated by many previous studies [4,17,63]. Abdou argued that lighting intensity exerted a strong impact on worker performance in industrial facilities and pointed out that certain lighting strategies could promote productivity while reducing energy consumption [63]. However, Choi et al. concluded that the visual environment would not directly influence the workers’ task performance [16], which contradicts the results in this study, and the reason for the disagreement may be that the illuminance in the previous research was much higher than that near the excavation face of the tunnel construction sites. Another study also found that lighting intensity improved productivity when lighting conditions were relatively poor while excessive lighting might even lead to adverse effects on working efficiency [14]. Meanwhile, it is reasonable that workers’ physiological status like sleep quality, alertness, and vitality is closely related to their productivity during work. Considering that sleep health is closely related to both physiological and psychological states, sleep disorders or poor sleep quality reduces workers’ productivity while higher sleep quality is considered to promote work efficiency to some extent [9,31], and the results of the overall structural modeling analysis have proved this conclusion. Except for sleep quality, alertness is also supposed to influence the workers’ productivity in tunnel construction sites because workers should maintain a state of sustained attention to avoid potential accidents induced by dangerous sources like explosives and large construction machinery, which has been validated in this study and is in line with the previous study [10]. As a labor-intensive industry, workers need to do lots of heavy physical work in tunnel construction sites like handling materials or operating heavy machinery so they should maintain sufficient physical strength to keep work efficiency high. The results of this study have verified that workers’ efficiency could be improved when they experience high levels of vitality. Except for its image-forming (IF) effects, light also influences other physiological and psychological processes relying on photoreceptors in the retina to signal light information to the brain, known as non-image-forming (NIF) effects of light. A previous study has proved that dynamic changes in the intensity of light featuring bright light during the day and dim light during the night promote melatonin secretion and sleep initiation [64], which indicates that light intensity adjustment may be a strategy for maintaining the circadian melatonin rhythms and sleep–wake cycle to promote the sleep quality and is consistent with the conclusion of this study. Meanwhile, this study has also proved that higher lighting intensity promotes workers’ alertness and vitality, which corresponds with previous studies [41,48]. Therefore, the overall structural modeling analysis has proved that all the hypotheses in this study are valid and highlighted the dominant role of lighting intensity in predicting workers’ physiological and psychological state and productivity in tunnel construction sites.

The clustering results have provided some insights regarding lighting conditions across tunnel construction sites. First, lighting intensity is closely related to lighting comfort because few cases feature high lighting intensity and low lighting comfort or the opposite. Second, the overall lighting conditions in tunnel construction sites are not satisfactory because cluster 1 featuring high lighting intensity and comfort accounts for the smallest proportion of all the respondents (6.52%) while cluster 4 with the poorest lighting conditions still occupies a considerable proportion. Lastly, the workers’ physiological and psychological state and productivity decline from cluster 2 to 4 and from cluster 1 to 2 due to the degradation in lighting intensity. However, the significant difference between clusters 2 and 3 in terms of workers’ physiological and psychological state and productivity results from the variation in lighting comfort like the overall arrangement of the lights, lighting contrast, color temperature, uniformity of the lighting, types of the lights, and other factors because the lighting intensity is comparable between the two clusters. This finding also points out another solution to improve the workers’ productivity through upgrading the lighting comfort without enhancing the lighting intensity, which indicates a limited increase in budget and is worth promoting across tunnel construction sites. Several previous studies also validated these conclusions. For example, Shishegar et al. found that lighting parameters like color, luminance contrasts, and luminance uniformity could impact sleep quality apart from lighting intensity [65]. Lee and Jung also argued that the adoption of morning blue-enriched lighting and night blue-suppressed lighting improves sleep quality to a large extent and 480 nm blue light is effective in reducing sleep latency [66]. Therefore, increasing the lighting intensity is not the only option to improve workers’ sleep quality in terms of lighting conditions. In addition, lighting intensity or illuminance of white light plays an important role in predicting the alertness of the workers while other lighting parameters like wavelength and color temperature could also promote workers’ alertness and cause them to maintain sustained attention [38,67]. Evidence also showed that shorter wavelengths or higher color temperatures were beneficial for promoting alertness similar to higher lighting intensity [67,68]. A few studies validated the beneficial effects of prolonged exposure to blue-enriched among office workers in improving their levels of vitality and self-reported performance [69,70], which proved that alteration of light color temperature is an alternative to promoting workers’ vitality and performance. Consequently, the clustering results verified the findings from the overall structural modeling analysis further and pointed out another potential solution to productivity promotion.

The results of the structural modeling analysis on four clusters showed that all the hypotheses were valid across the four clusters, and the relationship heterogeneity of the lighting intensity, workers’ physiological and psychological status, and their productivity across the four groups has also been revealed. The influence of lighting intensity on workers’ productivity declines when the lighting conditions improve. As mentioned before, the lighting intensity promotes productivity when lighting conditions are relatively poor while excessive lighting may induce negative effects on working efficiency, which indicates the alteration of the relationship between lighting intensity and working productivity with the increase in lighting intensity [14]. The lighting intensity in tunnel construction sites with the best lighting conditions is still lower than that in other workplaces like building or bridge construction sites due to the special structure of tunnels and limited budget, and thus the impact of lighting intensity on the working efficiency only declines in this study because the lighting intensity has not reached the point where the relationship between lighting intensity and working efficiency alters. As for workers’ physiological and psychological status, the impact of lighting intensity varies with lighting conditions improving across the three parameters including sleep quality, alertness, and vitality. The influence of lighting intensity on workers’ sleep quality degrades when lighting conditions improve while lighting intensity still exerts a positive impact on the sleep quality for the sites with the best lighting conditions. Meanwhile, the lighting intensity shows a stable relationship with workers’ alertness regardless of the lighting conditions. However, the impact of lighting intensity on vitality is slightly enhanced when the lighting intensity and comfort improve. In terms of the relationships between workers’ physiological and psychological status and productivity, they remain unchanged across the four clusters, which reflects the impact of workers’ physiological and psychological status on their efficiency is positive and stable in tunnel construction sites because of the job content and industry characteristics.

5.1. Managerial and Practical Implications

This study highlights that the lighting conditions are not satisfying at various tunnel construction sites due to the small portion of sites with a high lighting intensity and comfort, while lighting intensity plays a dominant role in predicting productivity and also influences workers’ efficiency indirectly through their physiological and psychological status. Therefore, promoting the lighting intensity is of great significance in tunnel construction sites, and thus the management should invest more to install more lights near the excavation face within the budget. Meanwhile, measures to improve workers’ sleep quality, alertness, and vitality could also be adopted because workers’ physiological and psychological status also shows a positive relationship with their productivity.

In addition, the findings in this study also highlight another solution to improve workers’ productivity through upgrading the lighting comfort. Lighting comfort contains the overall arrangement of the lights, lighting contrast, color temperature, uniformity of the lighting, types of lights, and other factors, and workers could work more efficiently when they feel comfortable with the light around them. Consequently, the management could also pay attention to the selection of the light types like the color temperature, the arrangement of the lights, and other factors to maintain workers’ spirits.

5.2. Limitations

This study shares some common limitations with the studies using perceptions and self-reported data, like the subjectivity of the participants’ answers. The results of this study only show that the influence of lighting intensity on sleep quality, alertness, vitality, and productivity presents different trends with lighting conditions improving. However, the underlying mechanisms have not been uncovered, and thus further investigation, especially for experiments, is needed to explore the relationship between lighting intensity and workers’ physiological and psychological status and productivity. It should be noted that how to improve lighting comfort also requires in-depth and detailed research. In this study, the parameters contained by the lighting comfort have been highlighted, like the overall arrangement of the lights, lighting contrast, color temperature, uniformity of the lighting, types of the lights, and other factors while details regarding specific implementation methods are unknown.

6. Conclusions

Adequate lighting at construction sites is necessary for the normal operations of workers. However, tunnel construction sites feature poor lighting conditions due to the lack of natural light, and this study aims to explore the effects of light conditions on tunnel construction workers’ quality of life and work productivity. A large number of participants from various tunnel construction sites were included in this study to reveal comprehensive information regarding the lighting conditions and broaden the application scope of this research. A theoretical structural equation model was built to explore the causal relationships among lighting intensity, workers’ sleep quality, alertness, vitality, and their work productivity. The major findings are as follows:

(1)

among all the variables, lighting intensity plays the dominant role in predicting the workers’ productivity; besides, the lighting intensity also impacts the workers’ quality of life including their sleep quality, alertness, and vitality;

(2)

the whole sample is divided into four clusters based on the lighting intensity and comfort; the clustering results reveal that the lighting conditions are generally poor due to the small proportion of cluster 1 (high intensity/high comfort) and the large percentage of cluster 4 (low intensity/low comfort);

(3)

comparisons between cluster 2 (moderate intensity/moderate comfort) and 3 (moderate intensity/low comfort) in terms of lighting conditions and workers’ quality of life and work productivity show that increasing the lighting comfort is also a potential solution to productivity promotion;

(4)

the relationship between lighting intensity and workers’ quality of life and productivity may vary across the four clusters; the impact of lighting intensity on productivity declines when the lighting conditions are good while the influence of workers’ quality of life remains stable regardless of the lighting conditions.

The findings highlight the dominant role of lighting intensity on workers’ productivity and also point out another effective solution to improve the workers’ productivity by optimizing the lighting to let workers feel comfortable in terms of the overall arrangement of the lights, lighting contrast, color temperature, uniformity of the lighting, types of the lights, and other factors. Considering the sufficient sample, the findings of this study have broad application prospects. The relationship heterogeneity of the lighting intensity, workers’ physiological and psychological status, and their productivity has been revealed in this study, but the underlying mechanisms are still unknown, which require further investigation. Meanwhile, experiments can also be conducted in the future to explore how to improve lighting comfort.