From a SET perspective, TIS is a core antecedent of OP, as it shapes employees’ willingness to reciprocate through enhanced effort and citizenship behaviour (Blau, 2017). Trust strengthens the quality of workplace interactions and fosters a climate conducive to collaboration, innovation and productivity. Recent empirical findings confirm that supervisory trust significantly enhances organisational citizenship behaviours, creative problem-solving and employee commitment, all of which positively influence performance outcomes (Ertosun & Aşçı, 2021). Moreover, trust not only promotes cooperation but also motivates employees to exceed formal job expectations, thereby reinforcing organisational effectiveness (Yusnita, 2025). Even in cases where trust does not directly translate into measurable performance gains, it often acts as a foundational mechanism that enables critical intermediate behaviours – such as extra-role efforts and discretionary work practices – that contribute to long-term performance success (Lay et al., 2020). A multilevel study by Salanova et al. (2021) further demonstrated that trust serves as a psychosocial resource, linking supportive organisational practices and team-level resources to improved performance, underscoring its influence across multiple organisational levels. Consistent with SET, these findings suggest that when employees perceive their supervisors as fair, competent and benevolent, they are more inclined to reciprocate with loyalty, psychological engagement and performance-enhancing behaviour. This evidence underscores the strategic value of building supervisory trust as a foundation for sustainable, high-performing organisational systems.

Drawing on the JD-R model, EE acts as a psychological resource that transforms organisational support into sustained energy, productivity and commitment (Bakker & Demerouti, 2007). It serves as a motivational driver, helping employees to cope with high job demands and use available resources – such as supervisory trust, autonomy and recognition – to optimise their performance. Engagement is widely recognised as a strategic asset that influences both attitudinal and behavioural outcomes in the workplace. Recent empirical evidence indicates that engaged employees exhibit greater commitment, enhanced productivity and a greater willingness to contribute beyond formal job expectations, thereby reinforcing organisational success (Wahyutomo et al., 2025). For instance, a study on public universities in Ethiopia revealed that vigour, dedication and absorption significantly improved OP, confirming the positive role of engagement in high-demand contexts (Gede & Huluka, 2024). Similarly, research within industrial organisations found that engagement directly enhanced job performance and mediated the influence of job satisfaction and communication satisfaction (Rizky et al., 2023). Engagement has also been linked to greater organisational citizenship behaviour, which contributes to long-term sustainability and service quality in public-sector organisations (Hazarika, 2024). In the corporate sector, case analyses such as Alibaba Group demonstrate how engagement initiatives drive innovation, customer satisfaction and competitive advantage (Ouhammou & Manar, 2024). Moreover, human resource strategies that prioritise participative decision-making, recognition and training have been shown to significantly strengthen the engagement–performance link (Arshad & Ming, 2024; Chatterjee, 2022). Taken together, these findings reinforce the JD-R model’s proposition that EE is a pivotal mechanism through which both structural and relational workplace resources are channelled into sustainable performance outcomes.

According to LMX theory, the quality of the supervisor–employee relationship determines how effectively trust is transformed into performance-enhancing behaviours (Graen & Uhl-Bien, 1995). This theory emphasises relational differentiation, suggesting that employees in high-quality exchanges experience more mutual trust, support and respect, which makes them more responsive to organisational cues – such as engagement opportunities and performance expectations. In this context, EE serves as a critical moderator, shaping the strength and direction of the trust–performance relationship.

Nabilla et al. (2023) demonstrated that while trust fosters organisational citizenship behaviours, its effect on job performance was significantly more substantial when employees exhibited high engagement. Similarly, Yusnita (2025) found that engagement amplified the influence of supervisory trust on performance by reinforcing motivation and perceptions of organisational support. Studies further highlight that engaged employees are more likely to transform supervisory trust into proactive behaviours and sustained contributions to organisational success (Alam et al., 2024). Moreover, Honnamane et al. (2024) emphasised that work engagement moderates the impact of tenure on both organisational citizenship behaviour and task performance, underscoring its contextual role in enhancing supervisor–employee dynamics. Relatedly, Darmawan and Alawiyah (2024) showed that flexible work arrangements and perceived supervisor support improved employee performance through engagement, suggesting that engagement acts as a critical mediator and moderator in trust-based workplace outcomes. Hossain et al. (2024) further argued that supervisor support strengthens the association between person–organisation fit and engagement, which ultimately boosts performance. These findings collectively reinforce the LMX perspective by illustrating how the depth of the supervisor–subordinate relationship influences the extent to which engagement can magnify the performance outcomes of trust. High engagement thus functions not only as a motivational resource but also as a relational amplifier within trust-driven performance dynamics. Based on the literature, the following hypotheses are proposed:

The study sample comprised 401 officials drawn through simple random sampling from 31 local government units in a southern province of Thailand, including a provincial administrative organisation, two city municipalities, five town municipalities and 23 sub-district administrative organisations. To ensure effective participation, a team of 10 trained research assistants facilitated data collection by explaining the research objectives, emphasising confidentiality and voluntary participation, and guiding respondents through the survey process. Data were collected over 3 months (July–September 2023). The sample demonstrated diverse demographic characteristics, with 34.66% identifying as male, 43.89% as female and 21.45% as lesbian, gay, bisexual, transgender, queer, intersex, asexual and other. In terms of marital status, 52.37% were single, 30.92% were married, 7.98% were divorced, 7.73% were separated, and 1% were categorised as other. Educationally, the majority held a bachelor’s degree (55.86%), followed by 33.92% with higher qualifications and 1% with a lower qualification. The participants’ mean age was 33.63 years (standard deviation [SD] = 6.97), with an average tenure of 6.62 years (SD = 5.27) and a mean monthly income of 22541.06 Thai Baht (approximately United States Dollar [USD] 623.16, SD = 8557.64). This demographic distribution provides a comprehensive foundation for interpreting the study’s findings within the diverse local government context.

Trust in supervisor was measured using the organisational trust instrument developed by Mayer and Davis (1999), which was selected for its strong validity in assessing TIS. The instrument evaluates three key dimensions – ability, benevolence and integrity – through 17 items rated on a 5-point Likert scale (1 = ‘strongly disagree’ to 5 = ‘strongly agree’). Example items include: ‘My supervisor acts justly’, ‘My supervisor reliably keeps their commitments’, ‘My supervisor has the necessary expertise and abilities for their responsibilities’ and ‘My supervisor considers and recognises my welfare needs’. The instrument demonstrated excellent internal consistency in this study, with a Cronbach’s alpha of 0.95. Employee engagement was assessed using the ISA engagement scale developed by Soane et al. (2012), which captures engagement across three dimensions: intellectual, social and affective. The instrument comprises nine items, each rated on a 5-point Likert scale (1 = ‘strongly disagree’ to 5 = ‘strongly agree’). Illustrative items include: ‘I am invested in performing my tasks diligently’, ‘My career goals resonate with those of my colleagues’ and ‘I feel a positive sentiment toward my profession’. This scale also showed strong reliability, with a Cronbach’s alpha of 0.91. Organisational performance was measured using a researcher-developed instrument comprising seven items, designed to capture perceptions of organisational effectiveness and efficiency. Participants responded on a 5-point Likert scale (1 = ‘strongly disagree’ to 5 = ‘strongly agree’). Example items include: ‘My organisation fully utilises my knowledge and skills to enhance work efficiency’, ‘My organisation strives to reduce administrative and operational costs’ and ‘Over the past two years, my organisation has performed well’. This instrument demonstrated excellent internal consistency, with a Cronbach’s alpha of 0.95.

Data were analysed using partial least squares structural equation modelling (PLS-SEM) with SmartPLS 4, suitable for testing complex models with moderation effects. Reliability and validity were confirmed through Cronbach’s alpha, composite reliability and average variance extracted (AVE), while discriminant validity was assessed via cross-loadings, the Fornell–Larcker criterion and the heterotrait–monotrait (HTMT) ratio. Model fit was evaluated using standardised root mean square residual (SRMR), normed fit index (NFI), d_ULS and d_G. These procedures align with recent structural equation modelling (SEM) practices, including applications in digital health and behavioural research (Alduosari et al., 2025). Hypotheses, including the moderating effect of EE, were tested through bootstrapping (5000 resamples) to obtain path coefficients, t-statistics and p-values, consistent with recent SEM research (Alam et al., 2024; Amrina et al., 2024; Ghani et al., 2023).

Ethical clearance to conduct this study was obtained from the Suan Sunandha Rajabhat University Ethics Committee (No. COA.2-042/2023).

Before testing structural relationships, the measurement model was evaluated for reliability and validity using Cronbach’s alpha, composite reliability, AVE, cross-loadings, the Fornell–Larcker criterion and the HTMT ratio, ensuring the constructs were suitable for SEM.

The results in Table 1 show that the constructs of EE, OP and TIS exhibit strong reliability and validity. Cronbach’s alpha values (0.771 for EE, 0.924 for OP and 0.906 for TIS) exceed the commonly accepted threshold of 0.70, indicating adequate to excellent internal consistency (Kilic, 2016). Both composite reliability coefficients (ρₐ and ρ_c) exceed 0.77, aligning with recommendations that values above 0.70 indicate measurement reliability (Peterson & Kim, 2013). The AVE values – 0.536 for EE, 0.601 for OP and 0.763 for TIS – are all above the 0.50 threshold, signifying good convergent validity, meaning that each construct explains more variance than error (Ulbegi, 2019). Overall, the findings demonstrate that the measurement instruments used for these constructs are psychometrically sound and suitable for further SEM (Coffie et al., 2023).

The cross-loadings in Table 2 provide evidence of discriminant validity by showing how strongly each indicator relates to its intended construct relative to other indicators. Items AE, IE and SE load more strongly on EE (0.800, 0.581, 0.794) than on OP or TIS, confirming their alignment with EE. Similarly, OP1–OP9 demonstrate higher loadings on OP (ranging from 0.740 to 0.842) than on other constructs, supporting construct validity. Trust in supervisor items ABI, BEN and INT display very high loadings (0.885, 0.865, 0.871) on TIS, surpassing cross-loadings with EE and OP, indicating strong construct specificity. Notably, the interaction term (EE × TIS) loads exclusively on its designated construct (1.000) and shows negative correlations with the others, consistent with interaction modelling. This pattern demonstrates adequate discriminant validity, meaning the constructs – EE, OP and TIS – are empirically distinct, which is crucial for valid SEM.

The Fornell–Larcker criterion in Table 3 evaluates discriminant validity by comparing the square root of the AVE for each construct (on the diagonal) with its correlations (off-diagonal). The diagonal values – 0.732 for EE, 0.775 for OP and 0.874 for TIS – are all higher than their corresponding interconstruct correlations, such as EE–OP (0.640) and OP–TIS (0.748). This indicates that each construct accounts for more variance in its measures than in those of other constructs, supporting discriminant validity. Such findings suggest that EE, OP and TIS are empirically distinct, which is critical in ensuring reliable SEM results (Ab Hamid et al., 2017; Tensay, 2018).

The HTMT ratio results in Table 4 assess discriminant validity by comparing correlations across constructs. All reported HTMT values fall below the conservative threshold of 0.85 (e.g. EE–OP = 0.636, EE–TIS = 0.662, OP–TIS = 0.747), suggesting adequate discriminant validity, because the constructs are more strongly related to their indicators than to others. The interaction term (EE × TIS) also shows moderate HTMT values with other constructs (0.263–0.483), reinforcing construct distinctiveness. These findings indicate that EE, OP and TIS are empirically distinguishable, supporting the robustness of the measurement model in SEM (Ab Hamid et al., 2017; Dirgiatmo, 2023).

Following confirmation of the measurement model’s reliability and validity, the structural model was assessed to test the hypothesised relationships among EE, TIS and OP. Model fit was evaluated using SRMR, NFI and discrepancy measures (d_ULS and d_G). Path coefficients were analysed with bootstrapping (5000 resamples) to assess the significance of direct and moderating effects. A simple slope analysis further clarified the moderating role of EE on the TIS–OP relationship.

Table 5 presents the model fit indices for both the saturated and estimated models in an SEM framework. The SRMR values (0.045 and 0.047) are below the recommended cutoff of 0.08, indicating a good fit between the hypothesised model and observed data (Shi & Maydeu-Olivares, 2020). The d_ULS and d_G values (0.211–0.235 and 0.164–0.166, respectively) reflect discrepancies between empirical and model-implied correlations, with lower values indicating greater model adequacy (Kono & Sato, 2022). The Chi-square statistics (366.530 for saturated; 378.558 for estimated) are significant, which is common in SEM with larger samples, but emphasis is placed on approximate fit indices rather than Chi-square alone (Sathyanarayana & Mohanasundaram, 2024). Finally, the NFI values (0.903 and 0.900) exceed the 0.90 threshold, confirming acceptable model fit. Overall, the indices indicate that the estimated model has a strong fit, validating its structural adequacy.

Table 6 and Figure 1 present the path coefficients explaining the structural relationships among EE, TIS, their interaction (EE × TIS) and OP. The results indicate that EE has a significant positive effect on OP (β = 0.316, p = 0.001), suggesting that higher EE is associated with better OP. Trust in supervisor demonstrates an even more substantial positive effect on OP (β = 0.595, p < 0.001), highlighting that employees’ trust in their supervisors is a key driver of organisational success. Additionally, the interaction term (EE × TIS) is significant (β = 0.080, p = 0.010), showing that TIS enhances the positive impact of engagement on performance. Together, these findings suggest that both EE and supervisory trust, individually and jointly, play crucial roles in improving organisational outcomes.

The simple slope analysis provides evidence for the moderating role of EE in the relationship between TIS and OP. As depicted in Figure 2, the positive association between TIS and OP varies with EE level. When EE is low (−1 SD; red line), the slope is positive but relatively shallow, suggesting that while trust in a supervisor contributes to better performance, its effect is limited in low-engagement conditions. At average levels of EE (blue line), the slope becomes steeper, indicating a more substantial influence of TIS on OP when employees exhibit moderate engagement. Importantly, at high levels of EE (+1 SD; green line), the slope is steepest, demonstrating that the impact of TIS on OP is substantially amplified when employees are highly engaged. This interaction pattern supports the hypothesised moderation, showing that EE enhances the extent to which TIS translates into OP. Conceptually, this implies that TIS relationships provides the foundation for improved performance. However, the magnitude of its effect depends on the degree of EE, with the most favourable outcomes achieved when both trust and engagement are high.