Prefabrication in construction: a pathway to improved mental health and well-being for workers Open Access

The adverse impacts of poor mental health among construction workers have been widely researched. Apart from its health-related negative effects on the workers, such as burnout and suicidal tendencies (Tijani et al., 2021), poor mental health also poses a great threat to organisational, industrial and economic growth due to reduced productivity (Lingard and Turner, 2015). With constant exposure to work pressure (Pidd et al., 2017), long working hours (Lingard et al., 2012), body pains (Turner and Lingard, 2020), workplace harassment and discrimination (Sunindijo and Kamardeen, 2017) and financial difficulties (Bowers et al., 2018), resultant mental stress forces experienced construction workers into early retirement (Milner et al., 2017). Furthermore, young workers are either lost to suicide or render unproductive from addiction to harmful substances (Fagbenro et al., 2023). These construction-specific stressors, because of their psychological and physical exhaustion of workers, are referred to as job demands in the Job Demands-Resources (JD-R) model (Bakker and Demerouti, 2007).

Existing measures are workers-centric including, but not limited to, mental health education and post-distress counselling. While success has been recorded with these measures (Doran et al., 2021), their underwhelming effect is highlighted by the resilient poor mental health among construction workers. Such measures fail to affect process-driven stressors like manual handling and lengthy weather exposure (King et al., 2019; Pidd et al., 2017). However, better mental health has been linked to favourable work conditions (Lingard and Turner, 2015). Prefabrication was proposed to facilitate such a flourishing work environment through enhanced standardisation, better quality control and labour efforts efficiency (Fagbenro et al., 2023). Furthermore, recent studies have theoretically demonstrated the reducing impacts of prefabrication on stressors and poor mental health (Fagbenro et al., 2024a, b; Sunindijo et al., 2023). However, its suitability to provide working conditions that improve workers' mental health is yet to be empirically investigated. Hence, this study models the capability of prefabrication as a mental health and well-being protective job resource. This study, therefore, advances extant literature by extending the JD-R model to the influence of construction methods on workers' mental health improvement. Furthermore, the study empirically demonstrates the significance of the differences between traditional and prefabrication work environments in protecting workers' mental health.

The core assumption of the JD-R model is that every occupation/industry is characterised by two opposing factors: job demands and job resources. Job demands are the physical, psychosocial and organisational requirements of a job that must be satisfied by deploying diverse skills at the expense of the physical and mental health of the workers. Job resources are the multifaceted aspects of a job that are required to achieve the work goals, reduce the negative effects of job demands and enhance workers' development (Bakker and Demerouti, 2007). This study adopts the JD-R model because of its balanced consideration of the impacts of job demands and resources on workers' well-being. This makes it more suitable for wider industries application than its alternatives, like the effort-reward imbalance (ERI) and demand-control (DC) models. The ERI model blames occupational stress and resultant health challenges on disproportionately low rewards for high efforts to meet job demands (Siegrist, 2016). Although the DCM agrees that job demands impair workers health, it oversimplifies the solution as individual's task autonomy (Karasek, 1998). Both DCM and ERI disregard the impacts of physical and psychological work environments on workers' health challenges.

The complexity of construction and the uniqueness of every end-product have turned most of its job demands into physical and mental health stressors. Construction, in its conventional form, is mostly executed in the open with minimal workers' control over the work environment and prolonged exposure to adverse weather conditions. Furthermore, every construction project presents unique cognitive challenges and makes product and process standardisation difficult. This explains the dynamism of construction communication, which can amplify role ambiguity, unclear instruction and poor interpersonal relationships (Zhang et al., 2023). These and other job demand-turned-stressors like manual exertion, work pressure, task overload and long work hours pose greater mental health threats to construction workers than the average individual (Lingard and Turner, 2015).

In response to the prevalence of workers' poor mental health such as burnout and adjustment disorders, the construction industry has adopted various management and personnel-focused strategies. Team collaboration and work support initiatives for psychosocially isolated workers in remote areas or small workgroups in urban centres are common mental health interventions (Broadbent and Papadopoulos, 2014). Another measure is the facilitation of awareness programmes and trainings like the Australia's MATES in Construction to restore the deteriorating psychological health of distressed or suicidal workers (Lee et al., 2014). While these measures are job resources to assist the workers meet their job demands as postulated in the JD-R model, they are simply extrinsic in nature as they are meant to ensure that the work goals are achieved with little consideration for the work conditions (Bakker and Demerouti, 2007). The inadequacies of these measures lie in their overconcentration on the poor mental health symptoms against the causes (Fagbenro et al., 2024b). Workers are caught in the cycle of being stressed from the work conditions, provided with mental health support and redeploy to work under the same conditions that ruin their mental health in the first instance. However, the JD-R model argues that workers' growth and better welfare, including enhanced mental health, can be achieved with job resources initiating intrinsic motivation such as learning, role clarity, autonomy and increasing competence.

Equally important measures of mental health are emotional, social, psychological and cultural/religious well-being, which are collectively termed positive mental health (Fagbenro et al., 2024a). The subjective measure of the well-beings, divided into hedonic and eudaimonic traditions, is more acceptable because of its reliance on the self-evaluation of the individuals being assessed (Keyes, 2002, 2013). Hedonic tradition measures emotional well-being, which is the self-assessment of both negative and positive emotions to strike a balance with net positive emotions like happiness and calmness over sadness and nervousness (Keyes, 2002).Eudaimonic tradition focuses on the social and psychological well-beings. Psychological well-being assesses individuals' quality of life, health and efficient functionality with the extent of personal growth, environmental mastery, autonomy, self-acceptance and positive relations with others like their co-workers (Keyes, 2002). The quality and quantity of positive inputs of an individual in social/group tasks, like construction is social well-being. It is measured with the degree of social actualisation, coherence, integration, acceptance and contribution by an individual (Keyes, 2013). Religious/cultural well-being is important to ensure that the diverse backgrounds of the construction workforce do not strain their well-beings (Fagbenro et al., 2024a).

Prefabrication, through the off-site preassembly and subsequent installation of significantly finished building components, from non-volumetric to modular units, could potentially provide an adequate environment for workers' flourishing mental well-being (Fagbenro et al., 2024a). Although there is a dearth of empirical research on prefabrication-backed mental health improvement, its overall benefits and the need for its higher uptake are widely acknowledged. For example, Central Committee of the Communist Party of China and State Council aimed to improve construction output by mandating the procurement of 30% of new buildings in China between 2016 and 2025 with prefabrication (Luo et al., 2021). Process standardisation through prefabrication is linked to improved construction quality and productivity (Gibb, 2001). Standardisation can reduce construction errors and enhance competence through standardised and unambiguous design information. Hence, workers could be better shielded from task-related criticisms and associated inter-personal conflicts that are common stressors of mental health (Fagbenro et al., 2024b). Furthermore, prefabrication, through simultaneous site preparation and off-site components preassembly, reduces on-site congestion and symbiotic crew relationships (Court et al., 2009). Installation workers spend considerably less time on site with reduced disruption. Their productivity increases significantly, such as the 15-day completion of the 30-storey hotel building in China's Hunan Province (Boafo et al., 2016). Workers are, therefore, not exposed to weather conditions for long, and the prefabrication-powered increased efficiency could enhance their work-life balance. Furthermore, workers may be less exposed to work-related psycho-social isolation, unfavourable shift rosters, work pressure and other process-influenced poor working conditions. Additionally, prefabrication could enhance mental health through the reduction of physical safety hazards. Workers suffer less ergonomic-related injuries due to reduced risky tasks and more mechanisation of pre-assembly and installation of prefabrication units (Fagbenro et al., 2023, 2025). Despite the widely acknowledged health and safety benefits of prefabrication in the leading and emerging prefabrication markets like China and Spain, respectively (Boafo et al., 2016; Rubio-Romero et al., 2014), it still has a low adoption rate in the global construction markets. However, its safety benefits highlight its potential as a job resource to counter construction job demand and improve the overall well-being of construction workers.

Data was collected with paper and online questionnaires (see Appendix) from stratified and randomly sampled traditional and prefabricated construction workers in the Australian Capital Territory, New South Wales and Victoria. Non-identifying demographic information of the respondents, including the dominant construction method in their practices was collected. Ninety-four valid responses were received from 54 and 40 traditional and prefabricated construction workers, respectively. While large samples are generally desirable to obtain stable estimates in EFA, the adequacy of a sample is influenced by the communalities of the variables and the extent of overdetermination of the resultant factors (Fabrigar et al., 1999). Hence, the effect of small sample size is compensated by adopting a higher cut-off of 0.55 for significant factor loading, which was deemed suitable for samples between 86 and 100 (Hair et al., 2018). The sample size is also suitable for the PLS-SEM since it is larger than ten times the highest number of structural paths pointing to a construct (MH: 10 × 5 paths) in the model (Hair et al., 2021). Also, the suitability of the group samples for multi-group analysis is supported by the capability of PLS-SEM to produce reliable estimates for theory development and predictions with samples between 30 and 100 (Jannoo et al., 2014). Furthermore, the suitability of PLS-SEM over covariant-based SEM (CB-SEM) is justified because of the exploratory nature of the study. PLS-SEM is appropriate for theory development and prediction, while CB-SEM finds application in model validation (Dash and Paul, 2021). Additionally, PLS-SEM, being a non-parametric test, works well with complex models with small sample sizes even when conditions of normal distributions are not met (Jannoo et al., 2014).

Participants were provided with the objectives and project information statement, such as how the data collected would be handled, protection of privacy and eligibility criteria. Using 7-point Likert scale questions, where 1 and 7 represent “never” and “always”, respectively, the research aim of assessing the impacts of prefabrication on the mental health and well-being of construction workers was assessed.

Prior to the modelling procedures, the mental health stressors were subjected to exploratory factor analysis (EFA) to reduce them to a few and manageable factors. Low correlation coefficients and/or multicollinearity were assessed with the correlation matrix and the Bartlett's test of sphericity before the adequacy of the sample was confirmed with Kaiser–Myer–Olkin (KMO) index (Yong and Pearce, 2013). The optimal number of factors extracted was determined with maximum likelihood – a factor extraction method that tests the significance of number of factors in every model generated–and Root Mean Square Error of Approximation (RMSEA) fit index with acceptable values ranging between 0.03 and 0.08 at 95% confidence interval (Hair et al., 2018; Park et al., 2002). Furthermore, promax oblique rotation was adopted since the variables/stressors were expected to be interdependent; otherwise, their correlation statistics would be zero (Park et al., 2002). The EFA was conducted with IBM SPSS for Windows (version 28.0.1.0 of 2021)

The reduced stressors and the remaining constructs were modelled with the partial least squares-structural equation modelling (PLS-SEM). PLS-SEM simultaneously models and estimates complex relationships between independent and dependent constructs. PLS-SEM is better suited for these relationships because of its appropriateness for theory development in exploratory studies with simple or complex models, predictive capabilities and better estimation accuracy with 30–100 samples of either normally or non-normally distributed data (Dash and Paul, 2021; Hair et al., 2021). SmartPLS 4 (version 4.1.0.8, released in 2024) was used for the PLS-SEM modelling.

In addition to the complete PLS-SEM model, the responses of the prefabrication and traditional construction participants were simultaneously modelled to aid multigroup analysis and comparison. In assessing the relationships between observed variables and their constructs, variables with factor loadings below 0.50 were deleted from the model, while those between 0.50 and the recommended 0.708 were retained unless their removal increases their constructs' average variance extracted (AVE) and/or composite reliability. The internal consistency reliability was measured with Cronbach's alpha and composite reliability statistics, both with a 0.95 benchmark (Hair et al., 2019). Then, the convergent and discriminant validities of the constructs were examined with AVE and heterotrait-monotrait (HTMT) ratio, respectively. The minimum AVE and maximum HTMT ratio adopted for this study are 0.50 and 0.90, respectively. Any pair of constructs that violate the HTMT_0.90 were merged into higher-order construct (Hair et al., 2018).

The models' structural assessment involved variables' collinearities evaluation with a maximum acceptable variance inflation factor (VIF) of 5 (Hair et al., 2017), where a VIF below 3.3 indicates the absence of common method bias (CMB) in the model. CMB is any common variation observed, which is induced by either the data collection method or the instrument rather than the inter-construct relationships in the model (Kock, 2017). The model's explanatory power was assessed with the coefficient of determination (R²) and effect size (f²) with respective benchmarks of 0.25, 0.50 and 0.75 for weak, moderate and large R² (Hair et al., 2021) and 0.02, 0.15 and 0.35 for small, medium and large f², respectively (Cohen, 1988). Furthermore, the model's predictive power was assessed with Shmueli et al. (2019)'s PLS_predict before evaluating the direct and indirect path coefficients of the constructs.

The minimum factor loading of 0.55 was adopted in line with the sample size requirements. Four stressors with extraction communality values between 0.30 and 0.50 were retained because of their higher-than-0.55-benchmark loadings and no cross-loading problems. The assessment of the correlation matrix and the overall determinant values above 0.00001 shows no correlation problems. Furthermore, the higher-than-0.70 coefficients of the off-diagonal variables in the anti-image correlation matrices and the significance of the Bartlett's test of sphericity confirms the data suitability. The adequacy of the sample was confirmed with the presence of the “a” superscript on the diagonal values of the anti-image correlation matrices and the KMO indices, as shown in Table 1 (Hair et al., 2018; Yong and Pearce, 2013).

The goodness-of-fit of models with different numbers of extracted factors were compared to determine the optimal number of factors to extract. The maximum likelihood, RMSEA values and cumulative percentage of total variance explained (CPTVE) were assessed to arrive at a four-factor solution (Park et al., 2002). A one-factor solution showed an RMSEA of 0.125 and 46.47%. This indicates a poorer fit than the two-factor solution with RMSEA, change in RMSEA and CPTVE values of 0.104, 0.021 and 54.86%, respectively. The four-factor model was selected despite achieving a marginal change in RMSEA (0.008) from the 3-factor solution because the former retains more variables and has a higher CPTVE of 62.12% than the latter's 58.51%, indicating a marginal goodness of fit. The extracted factors – industry-related, management/organisational, physical health and safety and personal stressors–and their retained variables were recoded as shown in Table 2.

Sixty-two variables distributed into 10 constructs were included in the models. Four variables (IR1, MH1, PB5 and PB6) were deleted from the model due to factor loadings less than 0.50, as shown in Table 3. All the constructs satisfy the internal consistency reliability and convergent validity with AVE scores above 0.50 and Cronbach's alpha and composite reliability statistics below 0.95 (Table 4). Furthermore, the HTMT_0.90 measure of discriminant validity was fulfilled by all the inter-construct relationships except the relationship between emotional and social well-beings with an HTMT ratio of 0.938 as seen in Table 5. The two constructs were subsequently merged into a reflective-reflective higher-order construct (socioemotional well-being, SEW).

After the fusion of emotional and social well-beings into socioemotional well-being (SEW), thereby reducing the number of constructs to nine, the measurement model was reassessed, and the results satisfy all the required benchmarks as shown in Table 6. The final models have 8 lower-order and 1 higher-order constructs as shown in Figure 1.

The constructs' VIF values across the models are below the critical threshold of 5 indicating that there are no collinearity issues in the models. PB which is the models' true exogenous factor, exhibits low collinearity potentials with the intermediate and endogenous constructs in the models with its highest VIF being 1.086, 1.412 and 1.022 in its ordinary least squares regressions with MH in the traditional, prefabrication and complete models, respectively. Hence, the models showed no common method bias with VIFs less than 3.3 for all the constructs.

In the traditional construction and complete models, five endogenous constructs have R² values above the minimum of 0.25 with three (MS, PS and MH) having medium values, while the remaining two (PHS and SEW) have weak coefficients of determination. However, the number of constructs with R² above 0.25 dropped to 4 in the prefabrication model with PHS (0.563), MS (0.496), PS (0.616) and MH (0.669) having medium explanatory power.

Prefabrication benefits (PB), in the traditional construction model, have small effect size (f²) on IRS (0.075), medium effect size on CW (0.167) and large effect sizes on both SEW (0.382) and PW (0.304). In the prefabrication model, PB has negligible effect sizes on PHS (0.018) and MH (0.015). Its effect sizes on MS (0.128), PS (0.062) and CW (0.072) were weak, while it has medium effect sizes on IRS (0.158), SEW (0.192) and PW (0.173). When viewed holistically, PB has weak effect size on CW (0.184), medium effect on PW (0.311) and large effect on SEW (0.370). Furthermore, the effect size of PB on the stressors and MH are negligible, as shown in Table 7.

Although the Q²_predict of some variables of the mediating constructs are below zero, their interpretation was relegated, and the predictive power of the model was focused on the true endogenous constructs (Shmueli et al., 2019). Both traditional construction and the complete models show medium predictive power with positive Q²_predict and lesser PLS-SEM RMSE than corresponding LM-RMSE values for the variables of SEW, PW and CW. As for the prefabricated construction model, its predictive power is low, although SEW and PW have positive aggregated Q²_predict despite the presence of some negative Q²_predict among their variables. Detailed results of the PLS_predict are presented in Table 8.

To confirm the relationships among the constructs in the model, their path coefficients (β) are examined. The significance of path coefficients of the models' hypotheses was built on the bootstrapping standard errors for calculating t-values (Hair et al., 2021). The t-statistics and standardised path coefficients for all the direct relationships in the group and complete models are presented in Table 9. In the prefabrication model, PB has a statistically significant inverse causative relationship with IRS with p-value of 0.006, thereby rejecting the null hypothesis H_{0_1}. Furthermore, PB showed medium practical significance on IRS with an effect size of 0.158. The significant influence of PB on IRS is also confirmed by the positive Q²_predict of the construct and its underlining variables in the prefabrication model. However, the relationship is nonsignificant in the traditional construction and complete models, although positive but marginal Q²_predict (0.009) and weak practical significance, were observed in the latter model (f² = 0.075). Conversely, the causal relationships between PB and both SEW and PW are direct, statistically and practically significant in all the models with positive predictive powers. Practical significance was also observed for the relationship between PB and CW in the prefabrication model, although neither statistical significance nor predictive power was recorded.

The mediating roles of IRS, MS, PHS and PS on the influence of PB on poor mental health (MH) were assessed as presented in Table 10. In the traditional construction model, there is a significant and direct effect of management/organisational stressors (MS) on personal stressors (PS) through the partial mediation of physical health and safety stressors (PHS) (β = 0.236; t = 2.369; p = 0.018). The mediating effect of MS on the influence of IRS on PHS was also found to be full and positive mediation with significant indirect (β = 0.396; t = 2.541; p = 0.011) and total (β = 0.376; t = 2.736; p = 0.006) effects. The inclusion of PS in the relationship between PHS and MH as a mediator established full mediation with a significant indirect effect (β = 0.180; t = 2.311; p = 0.021). Finally, both MS and PHS serially and fully mediate the positive influence of IRS on PS (β = 0.173; t = 2.254; p = 0.024), like the full mediation of MS on IRS's impact on PS (β = 0.300; t = 2.190; p = 0.029).

The prefabrication model produced five significant mediation effects. MS fully mediates the influence of IRS on both PHS (β = −0.459; t = 4.216; p = 0.000) and MH (β = −0.297; t = 1.976; p = 0.048), with both relationships also producing significant total effects. Similarly, the impact of MS on PS (β = 0.301; t = 2.485; p = 0.013) through PHS and the serial mediation of the impact of IRS on PS (β = 0.228; t = 2.531; p = 0.011) by both MS and PHS produced partial and full mediation, respectively. Finally, PB indirectly reduces the impacts of MS (β = −0.280; t = 2.284; p = 0.022) and PHS (β = −0.170; t = 2.086; p = 0.037). The former relationship is mediated by IRS, while the latter is mediated by both IRS and MS. Furthermore, practical significance was recorded for the latter relationships (PB → IRS → MS) with a weak effect size (f² = 0.128).

In the complete model, the indirect influence of IRS on MH (β = 0.268; t = 3.168; p = 0.002) through MS and the PHS mediated influence of MS on PS (β = 0.255; t = 3.658; p = 0.000) both resulted in partial mediation. Similarly, the mediation of IRS's impact on PHS (β = 0.411; t = 4.605; p = 0.000) and PS (β = 0.275; t = 3.072; p = 0.002) through MS both resulted in full mediation. Finally, the MS and PHS serially mediated influence of IRS on PS (β = 0.182; t = 3.502; p = 0.000) produced full mediation with statistical significance recorded for both the indirect and total effects.

The results show that prefabrication directly reduces the effects of IRS, such as work pressure and work-life imbalance, because of its capability to facilitate improved working conditions. Being unable to directly influence MS, PHS and PS can be traced to the personality- and management-driven nature of these stressors. For example, neither poor workers' support mechanism (MS) nor financial difficulties (PS) can be exclusively blamed on construction. These and similar management and personal stressors are common in other industries. However, the combative nature of construction can aggravate their effects on workers. Therefore, improving construction conditions with prefabrication could lessen the effects of the general stressors on workers. Hence, the statistically non-significant but practically significant relationships of PB with MS, PHS and PS in the prefabrication model. Furthermore, tackling poor mental health with only prefabrication that can only influence IRS, will not be enough, especially when the challenge is deeply rooted in the personal and/or social life of the distressed workers. Figure 2 shows the complete structural model with the inner model parameters such as the path coefficients (β) and their corresponding p-values in brackets, and the coefficients of determination (R²) for all the endogenous constructs. A summary of the statistical and practical significance of prefabrication on mental health and well-being constructs in the models is presented in Table 11.

Before the conduct of the multigroup analysis to assess the significant differences in parameter estimates between the traditional and prefabricated construction models, the model was tested for suitability using the measurement invariance of composite models (MICOM). The first step, which is configural invariance, is conventionally established in SmartPLS 4 once the groups are created and the same number of indicators, same treatments and identical algorithm were performed to assess the groups and overall measurement models. The next step was to confirm the compositional invariance, which was done with the permutation test (10,000 samples, two-tailed test, 5% significance level) whose results are displayed in Table 12. Comparing the correlation between the composite scores of the traditional and prefabricated constructions (original correlation) with the 5% quantile shows that the quantile 5% values were lower or equal to the original correlation for all the constructs. Hence, compositional and partial measurement invariances were established.

Full measurement invariance was tested by assessing the equality of means and variances for the constructs as presented in Table 13. Only PS showed unequal mean values between the groups as indicated by its asterisked p-values above 0.05 and its mean original difference falling within the range of the lower (5%) and upper (95%) boundaries of the confidence interval. All the constructs exhibited equal variances as confirmed by their original difference values being outside the lower and upper boundaries of the confidence interval and p-value above 0.05. However, the violation of the equality of means tests means the non-suitability of the data for full measurement invariance.

The models were compared using the direct and total effects to determine the variance between the two groups (traditional and prefabrication) of construction workers. The results, as presented in Table 14, show that significant differences exist between the two groups in the direct and total effects of hypothesis H_{0_1} (PB → IRS) and H_{0_18} (PS → MH). The significant difference in H_{0_1} confirmed the capability of prefabrication to provide protective cover for the workers from the industry-related stressors. Furthermore, the extent of the adverse impact of personal stressors (PS) on the mental health of the traditional construction workers is significantly higher than the experience of the prefabrication workers. In other words, prefabrication provides work environment and conditions that provide some protective factors against the workers' experience of industry-related and personal stressors.

The Australian government reinforced the importance of prefabrication in addressing housing shortages with the recent investment of AU$54 million in the prefabrication sub-sector (Treasury Portfolio, 2025). In addition to its improvement of time, quality and cost performance of construction projects, this study identifies the positive influence of prefabrication as an effective job resource to enhance the mental health and well-being of construction workers. This could reduce the mental health gap between construction workers and the general Australian population. The direct reducing impact of prefabrication (PB) on industry-related stressors (IRS) and its indirect impact on management/organisational (MS) and physical health and safety stressors (PHS) corroborates the linking of the construction method to reducing stressors through creating favourable work conditions (Fagbenro et al., 2023). The inverse but statistically nonsignificant relationships between PB and the stressors in the traditional and complete models align with Fagbenro et al. (2024b).

Enhanced construction standardisation through prefabrication facilitates faster delivery of projects through reduced on-site time (Egege, 2018). This reduces the workers' experience of work overload, overtime, long work hours, fatigue and work-life imbalance. This is consistent with Lim and Ma's (2026) attribution of reduced project time, workload and work pressure to the psychosocial improving features of prefabrication. Efficient health risk management and reduction in dangerous tasks have been identified as catalysts for safer construction and less occurrence of anxious and traumatic site events (Ahn et al., 2016; McKay et al., 2005). Furthermore, standardisation reduces task ambiguity and indirectly reduces task-related criticism that could emanate from confusing instruction/information from relatively unstandardised practices in conventional construction. Prefabrication provides relatively better physical and psychological working conditions through less site congestion, less reliance on manual exertion and reduced exposure to adverse weather conditions (Blismas and Wakefield, 2007). These benefits make the early identification and responding to safety risks easier and help prevent psychologically stressful conditions like musculoskeletal disorders and site incidents (Fagbenro et al., 2023). Hence, the direct and indirect reducing influence of PB on the stressors, especially in the prefabrication model, is justified and agrees with past studies (Fagbenro et al., 2024b). Although this study shows that prefabrication alone cannot reduce poor mental health, the findings show its suitability to weaken stressors through conducive work conditions in agreement with past studies (Fagbenro et al., 2024a, b). Hence, its proper application with existing measures can improve the mental health protective capability of construction industry.

Contrary to its less significant impact on poor mental health, prefabrication provides an enabling environment for the positive mental health of workers. This aligns with past research that have established the difference between positive and poor mental health (Keyes, 2013). Consequently, the true mental health status of an individual is the net or the difference of the effects of both “negative” and positive mental health being exhibited and not just the absence of symptoms of poor mental health (Keyes, 2002). Furthermore, the ability of PB to reduce the effects of the mental health stressors may not result in significant reduction in mental health challenges within a short period, however, its improvement of positive mental health contributes to the overall workers' well-being which may, in the long run, overshadow the effects of the long experienced poor mental health symptoms (Keyes, 2013). Furthermore, the result confirms the previously established positive association between PB and the well-being constructs in Fagbenro et al. (2024a), which found significantly improved well-being for prefabrication workers than traditional construction workers. With prefabrication, complex and simple projects are executed in fractions of the time it would take to complete them using the conventional method (Egege, 2018; Hashemi, 2015). Workers' productivity is increased, which could have a positive effect on their psychosocial and emotional well-being. Prefabrication workers experience collegial support that is devoid of symbiotic effects and a less transient work environment, which could enhance workers' bonding, understanding and belongingness (Keyes, 2002, 2013).

The results of the multigroup analysis indicate that traditional construction workers feel the impact of stressors and their corresponding mental health effects more than the prefabrication workers as confirmed by the significantly higher impact of PB on IRS in the prefabrication model. This attests to the suitability of prefabrication, because of its better work conditions, as a job resource to address job demands-turn mental health stressors that are synonymous with the conventional construction method (Fagbenro et al., 2023). However, PB's impact on mental health and well-being is not significantly different between the groups. This aligns with past research that recommends a favourable work environment as a significant factor of improving construction workers' mental health (King et al., 2019). This shows that while properly planned and executed prefabrication is capable of providing physically and psychologically favourable work conditions for workers, other measures, whether technical or managerial, that can enhance the construction work environment would support the workers' well-being and mental stability.

The importance of providing favourable work conditions to proactively forestall poor mental health in the construction industry is acknowledged in literature. Recent studies have theoretically proposed prefabrication as a well-being, promoting construction method for workers. Based on the job demands-resources (JD-R) model, this study investigates the suitability of prefabrication to improve workers' mental health and well-being. Prefabrication benefits were modelled to investigate their influence on mental health stressors, poor mental health and positive mental health or well-being with survey responses from traditional and prefabricated construction workers in Australia. The results showed that prefabrication provides proper work conditions to support the socioemotional, psychological and cultural/religious well-beings of the workers. Although it showed no significant influence on poor mental health, the direct and significant reducing influence of prefabrication on industry-related stressors and its indirect impacts on management/organisational and physical health and safety stressors are established in the models in alignment with past theories. This study, therefore, extends the JD-R model to the construction method to prove the mental health protective benefits of prefabrication. Its better well-being promotion compared to the conventional in situ construction is also confirmed with the results of the multigroup analysis. Hence, the study recommends the adoption of prefabrication as a potent job resource and complementary measure to existing mental health improvement initiatives for better well-being and welfare for construction workers.

Certain limitations should be considered while building decisions on this study's findings. First, a cross-sectional research design was employed. Although PLS-SEM has been widely proven to produce accurate estimates, there may be variabilities in the results in a longitudinal design with different samples. Second, the model was based on combined data collected from construction workers of diverse demographic characteristics. The findings may vary if participant's information, such as age, gender and occupation, are factored into the analysis. In addition, the study broadly grouped construction methods into two traditional and prefabrication, without considering the different variants of prefabrication. The general features of these variants, such as process standardisation and off-site building preassembly may be similar, but the degree of prefabrication in each variant differs, and this may influence the extent of mental health and well-being benefits that are accrued to them. Furthermore, the restriction of the study to the Australian construction industry should be noted. A similar study in a construction sector with different economic and technological characteristics may yield different results. Although the anonymity of the research participants was guaranteed, the responses may still be prone to self-report bias because of the sensitivity of the topic. Participants may provide inaccurate responses to protect themselves from unpleasant experiences or negative public perception. Therefore, future studies should incorporate demographic categorisation to assess group sensitivity of the findings. Also, further studies can adopt objective and longitudinal research designs to reduce the impact of self-report bias and increase the validity of the findings. Finally, to establish the significance or otherwise of the degree of prefabrication on workers' mental health and well-being, future studies could explore the mental health and well-being benefits of each variant of prefabrication.

Although adopting prefabrication alone cannot eliminate workers' mental health challenges, this study exposes its importance in weakening some stressors, such as prolonged weather exposure and long work hours. However, conscious efforts must be made to properly adopt prefabrication with existing workers-focused measures to achieve improved workers' mental health and well-being.

Prefabrication adoption should not be an afterthought. Construction clients, consultants and contractors where they are involved early, should decide and plan for prefabrication from the project concept. Changing to prefabrication midways a project that is designed for in-situ construction may impose more mental demands for the workers. Furthermore, clients and consultants should ensure that prefabrication projects are administered with prefabrication-suited contract conditions. The use of contract conditions that are suited for in-situ construction may constitute contractual irregularities and a new form of mental exhaustion for workers. Furthermore, prefabrication introduces new risks that workers may not be familiar with. It is, therefore, important to ensure that workers are adequately trained before adopting prefabrication.

Governments should lead the uptake of prefabrication by complementing their investment in the subsector with regulatory policies and incentives/concessions. While policies will provide necessary guidelines to maximise health and other benefits of prefabrication, incentives and concessions could facilitate the rise of more prefabrication companies and the growth of existing ones.

This study also shows the potential of changing the culture and ways of working in construction to improve mental health and well-being. Adopting new technologies, alternative work arrangements, such as a five-day work week arrangement and contractual approaches that promote collaboration among stakeholders are strategies that may reduce the mental health burden of the industry.

The supplementary material for this article can be found online.