Psychosocial hazards in residential and commercial construction: a mixed-methods comparison of demands, resources and distress

Residential construction presents distinct psychosocial risk conditions, stemming from structural and organisational features that differ sharply from commercial settings. The job demands-resources (JD-R) framework provides a platform for understanding how sector-specific conditions shape worker wellbeing. Psychological strain becomes more likely when high job demands (e.g. time pressure, role ambiguity and interpersonal conflict) co-occur with constrained job resources (e.g. control, justice, recognition and support; Bakker et al., 2023; Hon et al., 2024). Although psychosocial hazards are now formally recognised as controllable Workplace Health and Safety (WHS) risks in Australia (Safe Work Australia, 2022), most research applying the JD-R framework, measurement tools, programs and policies have been developed in commercial or civil environments. Direct empirical comparisons between residential and commercial settings remain rare, though growing evidence suggests residential small and medium-sized enterprises (SMEs) operate under different organisational constraints, supervision patterns and workforce compositions (Loudoun et al., 2023a; Sun et al., 2022; Thompson and Doran, 2025; Waqar et al., 2024).

Given these gaps, sector-specific evidence is needed to determine whether psychosocial hazards manifest differently in residential construction, and if the pathways linking these hazards to psychological distress operate similarly across contexts. This distinction is important for the adaptation of industry-wide prevention efforts, including programs such as MATES in Construction in Australia, which have largely been implemented and evaluated within commercial settings with existing WHS systems (Gullestrup et al., 2023). Understanding whether residential workers experience different psychosocial conditions – and whether those conditions carry different implications for mental health – provides a foundation for tailoring prevention strategies to an under-researched segment of the industry.

This study uses a mixed-methods design to compare psychosocial demands, resources, and mental-health indicators across residential and commercial construction workers. Quantitatively, we draw on the People at Work – Construction (PAW-CON) instrument (Loudoun et al., 2024) to examine sector differences and test whether industry context conditions the relationship between specific job demands and psychological distress. Qualitatively, semi-structured interviews provide insight into how workers interpret supervisory relationships, job pressures and external stressors, highlighting the mechanisms of sectoral differences observed in the survey data. The next section reviews relevant literature and positions these aims within the broader evidence base.

Psychosocial hazards refer to the social and psychological conditions of work that have potential to harm workers' mental and physical health, safety and wellbeing. As Quinlan (2025) notes, “hazard” refers to the harmful work characteristic (e.g. overwork, bullying and insecure employment), whereas “risk” denotes the likelihood and magnitude of exposure. These hazards span work characteristics such as workload and pace, role clarity, hours and scheduling, payment regimes, work–life interface and the social-organisational environment, including supervisory relationships, support, justice, ill-treatment and career mobility, with porous boundaries between these categories. In this study, psychosocial hazards are conceptualised as a broad category of work-related factors with the potential to cause psychological harm, encompassing both job demands and deficits in job resources, consistent with the JD-R framework.

The JD-R framework provides a widely used theoretical basis for understanding how these hazards translate into worker strain. Within JD-R theory, psychological harm becomes more likely when high job demands (e.g. time pressure, role overload and interpersonal conflict) co-occur with constrained job resources (e.g. job control, recognition, procedural justice and social support; Bakker et al., 2023). JD-R theory emphasises that these conditions exhaust workers' emotional and physical resources, increasing the likelihood of stress, psychological distress and reduced wellbeing.

Evidence from the construction industry supports these pathways. Quantitative studies consistently show that ongoing exposure to high job demands such as work overload and excessive working hours elevates the probability of mental ill-health, stress and reduced wellbeing (Thompson et al., 2024). Job control has been repeatedly identified as a primary stressor in construction (Hon et al., 2024; Loudoun et al., 2023b; Sun et al., 2022), and research across occupations demonstrates that insufficient resources combined with high demands undermines mental health and performance (Lee et al., 2020). Qualitative studies complement this work by highlighting how localised work roles, identity-based factors, safety culture, exposure to critical incidents and traumatic events contribute to workers' mental health experiences within construction environments (Carvajal-Arango et al., 2021; Biggs et al., 2024; Hon et al., 2024; Thompson et al., 2025).

Together, this evidence establishes construction as an environment where psychosocial hazards are both prevalent and consequential. However, despite the increasing policy emphasis on psychosocial risk management in Australia (Safe Work Australia, 2022), most empirical knowledge is derived from large commercial or civil construction settings. The construction industry is heterogeneous, and work organisation differs substantially across its segments. The next section therefore examines the emerging evidence on the residential sector, where organisational structures, subcontracting patterns and work practices may result in distinct psychosocial conditions not captured in existing studies.

While the construction mental-health literature is dominated by large commercial and civil contexts, the Australian residential sector exhibits a distinct risk architecture characterised by short project cycles, dispersed worksites, extensive subcontracting and a predominance of SMEs. These organisational features are associated with more informal HR and WHS practices and weaker capacity for systematic risk control (Lingard et al., 2015; Thorpe et al., 2009; O'Neill et al., 2025; Thompson and Doran, 2025; Furci and Sunindijo, 2020). Recent Australian market conditions have intensified these pressures, with persistent demand, supply chain frictions, labour constraints and input price volatility driving schedule pressure, rework risk and cost uncertainty at the site level (Construction Skills Queensland, 2024; Master Builders Australia, 2024). According to JD-R theory, these features likely amplify job demands (e.g. time pressure and coordination ambiguity) while reducing job resources such as induction, supervisor development and worker consultation (Bakker et al., 2023).

Although limited, evidence specific to residential settings underscores these vulnerabilities. In the United States (US), a survey of 200 residential construction workers found 30% reported no WHS practices or induction activities that could address mental health concerns, signalling weak foundational controls (Choi and Carlson, 2014). Safety climate research also suggests perception gaps in residential construction, with managers typically rating conditions more positively than frontline workers, indicating potential under-recognition of psychosocial exposures (Dale et al., 2021). Comparative work indicates apprentices in residential settings report poorer work organisation factors and more adverse outcomes than apprentices in commercial settings (Dale et al., 2021). Cunningham et al. (2018) similarly highlight the critical role of supervisors in small residential teams and note lower access to induction and safety training, particularly for immigrant workers.

Subcontracting is another key mechanism shaping psychosocial hazard exposure in residential construction. Subcontractors tend to be small, specialised and subject to payment practices, coordination pressures and diffuse responsibility for WHS, each with implications for role clarity, procedural justice and voice (Arditi and Chotibhongs, 2005; Al-Bayati et al., 2023). Field studies describe limited health and safety meetings, variable supervisory behaviours and permissive attitudes to risky work, contributing to weaker everyday controls (Hung et al., 2011).

Overall, the residential sector's organisational structure and market conditions depict an environment where worker resources (induction, consultation and supervisor support) are relatively thin while demands (time pressure, coordination, payment stress and ambiguity) are amplified. This alignment is consistent with JD-R pathways to strain observed in construction more broadly (Sun et al., 2022). Given the sparse direct comparisons with commercial work, an exploratory sector analysis is required to determine whether meaningful differences in psychosocial hazards and mental-health outcomes exist and to guide adaptation of prevention approaches for SME-dominated residential settings.

Despite the strong indications outlined above, there remains limited empirical evidence directly comparing psychosocial hazards across residential and commercial construction settings. Existing measurement tools, programs and interventions have been developed and validated almost exclusively in large commercial or civil environments, and program evaluations such as MATES in Construction (MATES) have predominantly been conducted in settings where a more formalised WHS system is already in place (Gullestrup et al., 2023). As a result, it is unclear whether the psychosocial conditions experienced by residential workers align with those observed in the broader industry or whether key hazards manifest differently in SME-dominated residential environments.

A further gap concerns the mechanisms linking psychosocial hazards to psychological distress. Although JD-R theory suggests that sector context may shape how demands and resources translate into strain, the comparative evidence directly contrasting psychosocial hazards between residential and commercial construction remains limited, and no quantitative studies have tested whether industry type moderates these relationships. Understanding whether the hazard-distress pathway operates differently across segments of the construction industry is critical for targeting prevention strategies and adapting established programs to diverse work environments.

Qualitative insights are also needed to contextualise these patterns. Workers' interpretations of supervisory relationships, coordination demands and external pressures such as finances and family life can illuminate why certain hazards may be more salient in one sector than another. Integrating qualitative perspectives with survey data enables a more comprehensive understanding of sector-specific risk profiles and the mechanisms underpinning psychosocial harm in residential construction.

Guided by these gaps, the present study undertakes an exploratory sector comparison using a mixed-methods design. Survey data from the People at Work – Construction (PAW-CON) instrument are used to assess differences in psychosocial demands and resources and to test whether sector context conditions the relationship between job demands and psychological distress. Semi-structured interviews are then used to enrich and explain these patterns.

The study addresses the following research questions:

1. RQ1. How do job demands and job resources differ between residential and commercial construction workers?

2. RQ2. Does industry sector moderate the relationship between job demands and psychological distress?

3. RQ3. How do qualitative accounts from workers in each sector help explain sectoral differences in psychosocial hazards and their consequences?

A mixed-methods design was adopted to enable a richer examination of sector differences. Quantitative PAW-CON data provided a comparative profile of job demands, resources and mental-health indicators across residential and commercial workers. Qualitative interviews were used to contextualise these patterns and examine how workers experience and interpret psychosocial conditions within their work environments. This sequential explanatory approach is well-suited to JD-R research, whereby understanding both the presence of hazards and the mechanisms through which they are perceived and enacted is essential.

Survey data were collected from 1,450 building and construction workers (1,333 broader building and construction industry and 117 residential construction workers) across 21 construction sites within Queensland and Northern New South Wales between January and December 2024. This imbalance reflects the practical difficulty of accessing residential workers, who typically operate across smaller, dispersed and less centrally coordinated work environments. A total of 30 qualitative interviews were conducted: 15 with residential workers and 15 with commercial workers, which is within the sample size recommended by Saunders and Townsend (2016) for interview-based research.

Data on psychosocial hazards were collected using the 36-item People at Work-Construction Survey (PAW-CON; Loudoun et al., 2024). The PAW-CON is an empirically validated measure of the work environment underpinned by JD-R theory and has been used previously in research investigating psychosocial hazards within the building and construction industry (Loudoun et al., 2023b; Thompson et al., 2024). The PAW-CON asks participants to report on 11 job demands and job resources relevant to the construction industry those being role overload, role ambiguity, role conflict, job control, co-worker support, supervisor support, supervisor task conflict, supervisor relationship conflict, praise and recognition, procedural justice and change consultation. All job resource items were scored on a 7-point scale, ranging from 1 (Never) to 7 (Always), with higher scores indicating more resources. Job demand items were also scored on a similar 7-point scale, with higher scores indicating greater job demands, except for role ambiguity, where items were subsequently reverse-coded such that higher scores indicated greater role ambiguity. Internal consistency in the present sample across the 11 job demands and resources was acceptable, ranging from α = 0.82 to α = 0.96 (refer to Table 1 for further details).

Psychological distress was measured using an adapted version of the 12-item General Health Questionnaire (GHQ; Goldberg, 1972). Psychological distress was conceptualised as a state of psychological strain characterised by symptoms of anxiety, depression and reduced functional capacity. The GHQ is a commonly used measure for detecting minor psychiatric disorders and psychological distress among participants in community and organisational settings (Banks et al., 1980). In line with previous research evaluating psychosocial hazards and general mental health in the construction industry (Loudoun et al., 2024), we adapted the standard 4-point Likert response scale of the GHQ (0-1-2-3) to a continuous 7-point scale, ranging from 1 (Never) to 7 (Always). This adaptation was made primarily to reduce participant burden and minimise careless responding in a field-based construction survey (Ward and Meade, 2023). Construction workers often complete surveys under significant time pressure and in operational environments that are busy, interrupted, and not conducive to cognitively demanding response formats. A uniform 7-point response continuum was therefore used across the majority of the survey to support faster and more consistent responding, including alignment with the PAW-CON. All positively worded items were subsequently reverse-coded such that higher average scores across the 12 items indicated increased psychological distress and the likelihood of poor mental health. Finally, to identify and analyse cases of psychological distress within our multivariate analysis of variance (MANOVA), survey responses were transformed into the original 4-point Likert scale prior to analyses, with Never = 0, Rarely/Once in a While = 1; Some of the Time/Fairly Often = 2; Often/Always = 3. Once transformed, a participant's sum score of the GHQ was computed out of a maximum of 36, where a score exceeding 12 was used as a threshold for identifying cases of “psychological distress” versus “typical” general mental health. This clinical threshold is commonly used to detect minor psychiatric disorders and instances of poor mental health among participants (Kuipers et al., 2019; Rathore et al., 2022). This recoding was used only where threshold-based GHQ interpretation was required; moderation analyses retained the original 7-point Likert scale. Internal consistency in the present sample for the 12-items was acceptable (α = 0.90).

Past and future suicidal ideation was measured using two questions adapted from the Suicide Behaviours Questionnaire-Revised (SBQ-R; Osman et al., 2001). Participants were asked to evaluate feelings of past suicidal ideation by responding to the question, “How often have you thought about taking your own life in the past 6 months?” The language in this question was adapted to suit the Australian building and construction context by using the wording “taking your own life”, instead of “killing yourself” that is used in the SBQ-R. Feelings of future suicidal ideation were evaluated using the question, “How likely is it that you will attempt suicide someday?” Both questions were reported using the original response scales as noted in the SBQ-R (e.g. Never-Very Often; Never-Very Likely). To ease interpretation of results and help identify at-risk individuals, responses were transformed prior to MANOVA into two states, whereby past suicidal ideation was recoded such that Never/Rarely = Rarely, and Sometimes/Often/Very Often = Often. Conversely, future suicidal ideation was recoded such that Never/No Chance at All/Rather Unlikely/Unlikely = Unlikely, and Likely/Rather Likely/Very Likely = Likely.

Participants for questionnaires were recruited from among various principal contractors and subcontractors across the industry, including commercial, residential, civil construction and interior fit-out organisations. These organisations represent many of Australia’s most common construction works (IBISWorld, 2022). The survey was administered with support from Mates in Construction.

Multiple recruitment methods were used for residential construction workers because organisations in this sector typically operate in small trade-based work environments with limited interaction beyond their immediate workgroup (Wadick, 2010). Specifically, quantitative data were collected using two methods, with agreement between the principal contractors, industry-based barbeques were held in larger residential developments and estates, allowing small trade teams to be grouped into larger cohorts for survey participation. The second method involved supportive principal contractors, enabling data collection across multiple individual small housing projects, where work groups typically consisted of three to six tradespeople.

In total, 1,450 completed questionnaires were returned. Owing to the dynamic nature of project work at construction sites, it is not possible to determine the exact response rate. However, daily site reports for each of the 21 construction sites suggest an overall response rate of approximately 85%. It was as low as 50% and as high as 100% in some workplace settings.

Analyses were performed in several steps. First, the data were screened for accuracy and missing values, and descriptive statistics were computed. Next, a one-way between-group MANOVA was conducted to examine group differences with significant effects, followed by univariate analyses and pairwise comparisons. Finally, moderated regression analyses were conducted using the PROCESS macro (Version 4.2) in SPSS to test the moderating effect of industry type on the relationship between job demands and psychological distress. To account for a potential imbalance between the residential and commercial samples, all reported PROCESS analyses were performed using 5,000 re-sampling iterations.

The PAW-CON data were supplemented with 30 semi-structured interviews to provide deeper insights into workers' perspectives on psychosocial hazards (Carvajal-Arango et al., 2021). Qualitative data collection involved pooled interviews with mirrored questioning, with two distinct groups within the existing quantitative sample: residential and commercial workers (Thompson and Doran, 2025). Semi-structured interviews used consistent core questions across both cohorts with flexible thematic prompts (Bryant, 2020; Saldaña, 2020). Questions examined psychosocial demands and resources, safety culture, supervision, subcontracting conditions, mental-health and suicide-prevention training, and experiences of suicide exposure (Hon et al., 2024). Interviews were conducted by a construction mental-health practitioner, supporting rapport and data credibility. Interviews were transcribed and thematically analysed using open coding to identify key patterns and sector differences.

Demographics: The overall sample comprised 1,450 commercial and residential construction workers. The commercial group comprised 1,333 workers, of whom 90% were male. Among those who provided age data, the average age was 40.3 years, and 61.8% reported having 10 or more years of industry experience. Employment in this group was primarily subcontractors (62.1%), with 34% employed by the principal contractor. The most common role descriptions among commercial workers were tradespersons (36.6%), labourers (15.1%) and other professional staff (11.1%). Regarding direct supervision, the leading hand was the most frequently reported (32.6%), followed by the foreman (29.6%) and the site manager (17.5%). The residential group consisted of 117 workers, of whom 94% were male. Their average age was 36.6 years, and 53.5% had 10 or more years of industry experience. In terms of employment, 43.6% of the residential workers were subcontractors, and 39.3% were employed by the principal contractor. The most common role descriptions in the residential sample were tradespersons (36.8%), labourers (18.8%) and leading hands/foremen (12%). The predominant reported direct supervisor among residential workers was the site manager (34.2%), followed by the foreman (32.5%) and the leading hand (19.7%). Table 1 outlines the mean, standard deviation, reliability and intercorrelations across all study variables.

MANOVA was conducted to assess the differences between commercial and residential construction workers on 14 dependent variables (11 psychosocial hazards, psychological distress and measures of past and future suicidal ideation). The normality of all variables was assessed using Shapiro–Wilk tests and visual inspections of Q–Q plots in SPSS. As expected with a large sample, all variables yielded significant Shapiro–Wilk test results (p < 0.001), reflecting sensitivity to even minor deviations from normality (Field, 2024). Q–Q plots for each variable showed that the observed values closely followed the expected normal distribution, with only minor positive skewness and no substantial outliers or extreme departures from normality. Given these findings, and the known robustness of MANOVA to moderate non-normality in large samples (Tabachnick and Fidell, 2019), all variables were retained for analysis.

The overall MANOVA was statistically significant, Wilks' Lambda = 0.965, F(14, 1269) = 3.317, p < 0.001, with a partial eta squared of 0.035, indicating that the combined dependent variables differed by industry type. Follow-up univariate analyses revealed that residential workers reported significantly higher levels of supervisor relationship conflict (M = 1.719, SE = 0.098) than commercial workers (M = 1.491, SE = 0.029), yielding a mean difference of 0.23 (p = 0.026, Adjusted Cohen's d = 0.23). Similarly, residential workers scored higher on job control (M = 4.845, SE = 0.147) than commercial workers (M = 4.299, SE = 0.043), with a mean difference of 0.55 (p < 0.001, Adjusted Cohen's d = 0.37). Additionally, change consultation was significantly greater among residential workers (M = 5.549, SE = 0.144) than among commercial workers (M = 5.109, SE = 0.042), yielding a mean difference of 0.44 (p = 0.003, Adjusted Cohen's d = 0.30). However, no significant differences were reported for role overload, role ambiguity, role conflict, supervisor task conflict, coworker support, supervisor support, praise and recognition, or procedural justice. In terms of key mental health outcomes, no significant differences were observed between residential and commercial workers' psychological distress, past suicide ideation or future suicide ideation. Overall, these findings indicate that residential construction workers experience distinct psychosocial work conditions, particularly with elevated supervisor relationship conflict, increased job control and greater involvement in the change consultation processes (Table 2).

Moderated regression analyses with 5000 re-sampling iterations were conducted to determine whether industry type moderated the relationship between job demands and psychological distress. Five job demands (role overload, role ambiguity, role conflict, supervisor task conflict and supervisor relationship conflict), representing a subset of psychosocial hazards, were entered as independent variables, with psychological distress as the dependent variable and industry type (commercial = 0 or residential worker = 1) as the moderator.

In the full sample, direct effects indicated that higher levels of role overload (β = 0.131, SE = 0.021, p = 0.001), role ambiguity (β = 0.249, SE = 0.025, p = 0.001), role conflict (β = 0.113, SE = 0.021, p = 0.001) and supervisor relationship conflict (β = 0.120, SE = 0.027, p = 0.001) were significantly associated with increased psychological distress, whereas the direct effect for supervisor task conflict was not significant (β = 0.036, SE = 0.027, p = 0.185). The moderation analyses revealed that industry type did not significantly moderate the relationships between worker psychological distress and role overload (β = −0.005, SE = 0.066, p = 0.946, 95% CI [−0.1288, 0.1313]), role ambiguity (β = 0.013, SE = 0.090, p = 0.883, 95% CI [−0.1389, 0.2090]), role conflict (β = 0.013, SE = 0.057, p = 0.833, 95% CI [−0.0941, 0.1308]) or supervisor task conflict (β = 0.027, SE = 0.079, p = 0.719, 95% CI [−0.1162, 0.1959]). By contrast, a significant interaction was observed for supervisor relationship conflict (β = −0.287, SE = 0.080, p < 0.001, 95% CI [−0.4106, −0.1055]), indicating that the association between supervisor relationship conflict and psychological distress differed as a function of industry type. To clarify this relationship, conditional effects were examined, which indicated that for commercial workers, the relationship between supervisor relationship conflict and psychological distress was positive and significant (β = 0.3637, SE = 0.0258, t = 14.11, p < 0.001, 95% CI [0.3131, 0.4193]). Conversely, among residential workers, this relationship was not statistically significant (β = 0.0772, SE = 0.0595, t = 1.30, p = 0.195, 95% CI [−0.0396, 0.1939]).

These results suggest that while higher supervisor relationship conflict is associated with greater psychological distress among commercial workers, this association does not hold for residential workers. Consequently, industry type moderates the impact of supervisor relationship conflict on psychological distress, highlighting a differential risk profile between commercial and residential construction workers. Overall, the moderation analyses indicate that industry sector does not broadly condition the relationship between job demands and psychological distress, but instead operates selectively, with a significant effect observed only for supervisor relationship conflict. The qualitative findings presented below provide further insight into this pattern, particularly in relation to how supervisory relationships are structured and experienced across sectors. Table 3 outlines the interaction results for commercial and residential construction workers.

The qualitative interviews were used to contextualise the sector differences observed in the PAW-CON data and to explore how workers describe supervisory dynamics, job pressures and broader stressors in their daily work. The residential interview cohort included a higher proportion of younger workers and apprentices, whereas the commercial cohort contained more experienced workers. Both groups were overwhelmingly male. Several participants across both sectors reported having lost a colleague to suicide (eight residential workers and 13 commercial workers), reflecting the broader context of suicide exposure in the industry. Full demographic details for interviewees are provided in Table 4. Across the dataset, responses were broadly consistent within each sector, enabling identification of clear cross-cutting patterns rather than isolated individual experiences. The qualitative analysis highlighted four themes that illustrate workers' experiences and help contextualise the sector differences observed in the quantitative data.

• Theme 1: Supervisory relationships and conflict

Across the residential cohort, supervisory relationships were described as interpersonal, variable and dependent on individual supervisor behaviour rather than structured systems. Participants consistently reported limited communication, unpredictable moods and reduced access to supportive feedback. These patterns were particularly pronounced among younger workers and apprentices, suggesting supervisory conflict in residential settings is experienced as intermittent and relational rather than embedded in formal organisational processes. Several participants reported avoidance-based supervisory styles, abrupt mood changes and a lack of supportive feedback:

In contrast, commercial workers described supervision as more structured and systematised, with clearer communication channels, defined roles and more stable interpersonal dynamics. This suggests that supervisory relationships in commercial settings are more firmly embedded in formal site processes rather than depending heavily on individual supervisors

These accounts indicate that although residential workers report higher levels of supervisory conflict, the nature of that conflict differs across sectors. In residential settings, conflict is episodic and relational, whereas in commercial environments it is more structural and ongoing. This distinction provides a plausible explanation for the survey finding that supervisory relationship conflict is associated with psychological distress among commercial workers but not residential workers.

• Theme 2: Time pressure and production demands

Time pressure was consistently identified across both sectors as a routine feature of construction work, though it manifested through different organisational dynamics. Residential workers described continuous task-switching, coordination pressure across multiple trades and limited buffers, reflecting more informal and fragmented work processes:

In contrast, commercial workers described time pressure in relation to formalised productivity targets and measurable outputs, indicating tighter performance structures:

These accounts suggest that while time pressure is pervasive across construction, it arises from coordination ambiguity in residential settings and from production demands in commercial environments. These differences may help to explain the absence of significant sector differences in quantitative measures of workload.

• Theme 3: External stressors influencing mental health

Across sectors, workers highlighted external stressors – particularly finances, family responsibilities and relationship difficulties – as primary influences on mental health. These external pressures were described as intersecting with, and often compounding workplace demands, rather than operating as separate influences. In many accounts, personal stressors were positioned as more immediate or consequential than work-related factors, particularly during periods of financial strain or family disruption.

Participants also described how these external pressures could intensify the impact of workplace demands, particularly where long hours, unstable income or geographic mobility limited their ability to manage responsibilities outside of work. These accounts indicate that while psychosocial hazards differ across sectors, external stressors represent a shared and persistent influence on mental health. This may help explain the lack of significant sector differences in overall psychological distress observed in the quantitative results.

• Theme 4: Access to WHS and mental health systems

Access to WHS and mental health systems differed markedly across sectors, with residential workers frequently noting the absence of formal structures such as WHS committees, safety inductions and mental health or suicide prevention training. These patterns were widely reported across the residential cohort and reflect the constraints of the sector's SME-dominated work environments. These findings align with those of previous studies (Arcury et al., 2012; Choi and Carlson, 2014; Ormiston et al., 2020).

Residential workers described how the absence of formal systems contributed to uncertainty, reduced access to support and increased reliance on informal peer networks. Commercial workers described more established systems and processes, including regular site meetings, structured induction processes and greater exposure to safety and mental-health initiatives.

These differences highlight an uneven distribution of organisational resources across residential and commercial sectors. The interview findings align with the quantitative findings, showing higher autonomy but fewer formal supports in residential settings, which means psychosocial risks are less likely to be formally identified or managed.

The qualitative findings indicate that sector differences are not defined by the presence of psychosocial hazards, but by how those hazards are structured and experienced in different organisational contexts. Supervisory conflict appears more pervasive and emotionally loaded in residential settings, while commercial workers operate within more structured and continuous supervisory systems. Time pressure is common across both sectors but arises through coordination ambiguity in residential work and formalised production demands in commercial environments. External stressors – particularly finances and family pressures – are salient for all workers, providing a shared influence on mental health across sectors.

These patterns extend the quantitative findings by demonstrating that the impact of psychosocial hazards is context dependent. In particular, the qualitative data provide a plausible explanation for the sector-specific effect of supervisory conflict, showing how differences in supervision structure may shape whether conflict translates into psychological distress. Differences in the structure and continuity of supervisory relationships across sectors provide a clear qualitative basis for the sector-specific effect of supervisory conflict observed in the quantitative analysis.

Analysis of questionnaire and interview data provides an integrated account of psychosocial hazards across residential and commercial construction settings and directly addresses the study's research questions. Quantitatively, sector differences were modest: residential workers reported higher supervisor relationship conflict, greater job control and more involvement in change, and only supervisor relationship conflict showed a sector-specific association with psychological distress. The qualitative findings provide contextual depth to these patterns; residential workers experience supervisory conflict as brief and relational, whereas commercial workers encounter more continuous, workflow-embedded oversight. Across both sectors, workers described time pressure and external stressors as central influences on wellbeing. Together, these findings help to clarify how psychosocial conditions differ across sectors (RQ1), suggest that supervisory conflict operates differently by industry type (RQ2), and illustrate how workers' interpretations of supervision and work organisation may explain these differences (RQ3).

Existing applications of JD-R theory in construction have largely treated the relationship between job demands and psychological strain as consistent across work contexts. The present findings suggest that this assumption may benefit from greater contextual nuance, particularly in regard to supervisory relationship conflict which varied according to sector-specific conditions. These findings should be interpreted as indicative rather than definitive evidence of context-conditioned demand–strain relationships, given the exploratory design and the reliance on a single observed moderation effect. The qualitative findings are therefore used to provide a theoretically informed interpretation of this pattern, rather than to confirm the underlying mechanism.

Traditionally, JD-R theory assumes that the strain associated with job demands depends on their intensity, duration and the level of control available to workers (Bakker et al., 2023). Our findings provide preliminary evidence that while some demands (e.g. role overload and role ambiguity) operate similarly across settings, others are contingent on how work is organised. Supervisory relationship conflict appears particularly sensitive to differences in supervision structure and interaction patterns across sectors. As such, the contribution is best understood as identifying a potential boundary condition within JD-R applications in construction, rather than establishing a generalisable theoretical refinement.

Although prior research has identified supervisory relationships as an important psychosocial factor in construction (e.g. Hon et al., 2024), studies have typically examined these relationships within single construction contexts. By demonstrating a sector-conditioned relationship between supervisory conflict and distress, this study contributes to existing JD-R applications and highlights the need for more context-sensitive models of psychosocial risk in construction. More broadly, this finding suggests that JD-R assumptions may benefit from greater attention to organisational context. In particular, demand–strain relationships may not operate uniformly across settings but could be contingent on organisational structures, especially where demands are shaped by interpersonal dynamics such as supervisory relationships.

JD-R would typically suggest that higher levels of interpersonal conflict increase psychological distress through resource depletion and/or emotional exhaustion. However, our analysis indicates that although residential workers reported higher levels of supervisory conflict, qualitatively, these interactions appear to be episodic and relational. Workers described brief but emotionally charged exchanges and inconsistent supervisory availability, patterns consistent with residential construction's structure in which supervisors commonly oversee multiple small projects and are not continuously present on site. This arrangement, together with the higher reported job control, may help explain why cumulative strain typically associated with interpersonal conflict is not observed in the residential sample.

By contrast, commercial sites are characterised by continuous supervisory presence, clearer hierarchical structures and direct oversight of task allocation, sequencing and productivity expectations. Participants described supervisory behaviour as “dripping down” through the chain of command and closely tied to daily workflow, making conflict both harder to avoid and more consequential for task progress. These organisational conditions align with research showing continuous interpersonal demands, particularly when combined with lower autonomy, have a stronger impact on wellbeing (Hon et al., 2024; Sun et al., 2022).

Prior work also indicates supervisors in small residential teams have fewer resources and less formal training to manage psychosocial risks (Cunningham et al., 2018), whereas commercial environments offer more consistent supervisory systems despite higher productivity pressures. This interpretation is consistent with sectoral features outlined earlier, where residential SMEs operate with lean supervisory capacity and fewer embedded WHS and mental-health supports (Thorpe et al., 2009). When considered together, these findings provide a plausible explanation for why supervisor conflict predicted psychological distress among commercial workers but not residential workers, suggesting sector context may act as a “boundary condition” (Bakker et al., 2023) shaping when and how specific demands influence outcomes. While this interpretation is supported by the qualitative data, it remains a plausible explanatory account rather than a confirmed causal mechanism.

The qualitative findings help explain the sector differences identified in the survey data and clarify how workers experience psychosocial hazards in their day-to-day work. Residential workers described supervisory interactions as relational and intermittent, whereas commercial workers experienced continuous oversight embedded in production and coordination processes, supporting the quantitative moderation result. The interviews also reinforced points of convergence: time pressure was perceived as a universal feature of construction work, and financial strain, family responsibilities and relationship difficulties were common across both sectors. These shared stressors help explain why many psychosocial hazards did not differ statistically between residential and commercial workers despite differences in organisational structure. Finally, qualitative accounts highlighted limited access to structured WHS and mental health supports in residential construction, complementing quantitative findings of higher autonomy but fewer formal resources. Overall, the integration of methods strengthens confidence in the interpretation of sector differences and clarifies the organisational mechanisms shaping how psychosocial hazards translate into distress.

The findings suggest psychosocial risk management in construction should be tailored to sector context. The JD-R framework suggests interventions should target both demand reduction and resource enhancement. Importantly, the results indicate that not all psychosocial hazards operate uniformly across sectors, with supervisory relationship conflict emerging as a context-dependent risk factor. This indicates that intervention strategies must focus not only on reducing exposure to hazards, but on how specific hazards are experienced and enacted within different organisational structures.

In residential settings, where supervisory contact is intermittent, brief “toolbox talk” interventions, conflict-management supports and targeted supervisor capability development may be more effective than formal programs that require sustained engagement which are more efficacious in larger commercial sites. Supervisor capability development in these contexts may include short, task-integrated training focused on (1) clear and consistent communication across dispersed sites, (2) managing interpersonal conflict in small teams and (3) recognising early signs of distress among apprentices and younger workers. These approaches align with the resource constraints of SME-dominated environments and can be embedded within existing work practices. Strengthening quality and consistency of supervisory interactions may be particularly important for younger workers and apprentices.

In commercial settings, the strong association between supervisory conflict and distress highlights supervisory behaviour as a key intervention point. Practices that support predictable communication, role clarity and emotionally regulated supervision may help reduce the cumulative strain associated with continuous oversight. At the micro-level (team), this may involve structured daily pre-start briefings, clearer task sequencing and expectations around respectful communication. At the meso-level (organisation), organisations may implement formal supervisor training, incorporate psychosocial risk indicators into performance management systems and establish clear escalation pathways for interpersonal conflict. At the macro-level (industry and policy), these findings support the development of guidance on supervisory standards within WHS frameworks, including clearer expectations regarding psychosocial risk management and leadership behaviour.

These findings should also be considered in light of emerging regulatory expectations in Australia. Under the Model Code of Practice: Managing Psychosocial Hazards at Work (Safe Work Australia, 2022), employers must identify, assess and control psychosocial hazards as WHS risks. In residential SME contexts like Australia, limited formal systems mean obligations will rely on proportionate mechanisms such as basic consultation and supervisor capability building. This may include simplified psychosocial risk assessment processes, integration of mental health considerations into routine safety discussions and informal but consistent consultation practices with workers. In commercial settings, stronger governance structures support regulatory implementation, but the stronger conflict-distress pathway also increases legal and organisational exposure where supervisory behaviours remain unregulated. Accordingly, organisations may need to explicitly incorporate supervisory behaviour and interpersonal risk into WHS audits, compliance processes and leadership accountability structures.

This study has several limitations. Although the overall survey sample was large, the residential cohort was comparatively small and substantially underrepresented relative to the commercial sample, reflecting the fragmented nature of the sector and limiting generalisability. To further examine this issue, post-hoc power analyses were conducted in SPSS 31 using the Univariate Linear Regression Power Analysis procedure, specified as a Type III F-test of R² increase comparing the nested main-effects model with the full interaction model. The significant supervisor relationship conflict × industry interaction was well powered in the full sample (1 − β = 0.983), whereas the remaining interaction models showed negligible incremental variance explained by the interaction term. However, these estimates apply to the overall interaction tests and do not remove the greater uncertainty associated with conditional estimates for the smaller residential subgroup. Accordingly, non-significant interaction effects should not be interpreted as definitive evidence that sector does not condition the relationship between job demands and psychological distress. Rather, the findings should be interpreted as indicative rather than definitive, particularly in relation to the residential sample and the broader heterogeneity of residential construction. Replication with larger residential cohorts is needed to determine whether additional sector-specific interaction effects are present.

The structural characteristics of residential construction described in this study – particularly SME dominance and fragmented subcontracting – reflect the Australian context and may not generalise to residential construction systems in other regions where large developer-led models are more common. The cross-sectional design prevents causal inference, and the qualitative sample was male dominated. Self-report measures also introduce potential recall and social-desirability biases. A further limitation concerns the measurement of psychological distress. Although the adapted GHQ-12 demonstrated acceptable internal consistency in this sample, the original 4-point GHQ response format was modified to a 7-point scale to reduce participant burden and minimise careless responding in a time-pressured construction survey context. The responses were subsequently collapsed back into four ordered categories to perform our MANOVA analyses. This approach was pragmatic and allowed comparison against the established GHQ threshold literature, but the psychometric equivalence of the adapted format to the original GHQ-12 was not formally established in the present study. Accordingly, threshold-based distress classifications should be interpreted cautiously, as some measurement imprecision may have been introduced through recoding. Future research should directly validate this adapted format in construction samples or retain the original GHQ response structure where conventional threshold interpretation is central to the study design. Nonetheless, the mixed-methods design and multi-site data collection strengthen confidence in the identification of sector-specific supervisory dynamics and contextual mechanisms shaping psychosocial risk.

This study provides indicative evidence of how psychosocial hazards manifest across residential and commercial construction environments. Quantitatively, sector differences were modest, and only supervisor conflict showed a sector-specific association with distress. Qualitative findings illuminate why this pattern emerges, highlighting how organisational structures shape the frequency, intensity and consequences of supervisory interactions. The results suggest the need for sector-appropriate approaches to psychosocial risk management, particularly regarding supervisory practices and the constraints of SME-dominated residential work. More broadly, the findings contribute to JD-R applications in construction by providing initial evidence that demand–strain relationships may be sensitive to sector-specific organisational conditions, particularly in relation to supervisory dynamics.

N.T.: conceptualisation, investigation, project administration, data collection, qualitative interviews, writing—original draft, and review and editing.

A.K.: thematic analysis, writing—original draft, and review and editing.

A.R.: investigation, formal analysis, project administration, supervision, methodology, validation, visualisation, writing—original draft, and review and editing.

A.B.: investigation, writing—original draft, and review and editing.

J.M.: investigation, thematic analysis, writing—original draft, and review and editing.

K.T.: investigation, writing— original draft, review and editing.

R.L.: conceptualisation, funding acquisition, investigation, project administration, writing—original draft, and review and editing.

Ethics approval was provided by the Griffith University Human Research Ethics Committee (application reference 2024/818).

Informed consent was obtained from all participants involved in the study in accordance with the ethics approval provided by the Griffith University Human Research Ethics Committee.

The authors declare there are no competing interests to declare.

We gratefully acknowledge all the project stakeholders and particularly the workers in the Building and Construction industry who gave their time freely to participate and complete surveys and interviews.