This study employed a cross-sectional survey design to examine the conceptual framework of audit data governance in public service auditing for people with disabilities in Thailand. Using multi-group structural equation modeling (MG-SEM), we compared the measurement and structural relationships of the proposed model between two stakeholder groups: government officials involved in policy formulation, implementation, and auditing (service providers), and people with disabilities who access public services (service recipients). The hypothesised relationships and latent constructs are presented in Figure 1.

As shown in Figure 1, the detailed parameter values and item-level information are provided in Supplementary Table S1.

Based on the theoretical framework and previous empirical findings, this study formulates the following hypothesis: Hypothesis (H1):

There are variations in the causal factors influencing audit data governance for public service auditing between government officials (service providers) and people with disabilities (service recipients).

Data were collected from 711 respondents in two stakeholder groups: (1) government officials involved in policy formulation, implementation, and auditing of public services, and (2) people with disabilities holding official disability identification cards. Government officials completed an online questionnaire, while people with disabilities (including individuals with visual and physical impairments) were surveyed via face-to-face fieldwork in eight provinces representing Thailand’s major regions: Chiang Mai and Phitsanulok (North); Khon Kaen and Nakhon Ratchasima (Northeast); Nonthaburi (Central); Chonburi (East); and Surat Thani and Songkhla (South).

The structured questionnaire comprised four parts: (i) participant characteristics; (ii) measures of public service delivery for people with disabilities (S) and audit data governance (A); (iii) items assessing the perceived importance of audit data governance factors (A); and (iv) items assessing data-driven culture (C) and its relationship with audit data governance (A). All items used a 5-point Likert response scale (1 = not very important to 5 = very important).

Prior to the main survey, the instrument was pilot-tested separately in each stakeholder group (n = 30 per group). Internal consistency was high in both groups (Cronbach’s alpha = 0.95), exceeding commonly used thresholds for acceptable reliability ( Nunnally & Bernstein, 1994).

Multi-group structural equation modeling (MG-SEM) was implemented in IBM SPSS AMOS 24.0 ( IBM Corp., 2016; https://www.ibm.com/products/structural-equation-modeling-sem), with IBM SPSS Statistics 25.0 ( IBM Corp., 2017; https://www.ibm.com/products/spss-statistics) used for data preparation. MG-SEM was selected to evaluate the proposed latent-variable model and to examine whether key measurement and structural parameters were comparable between government officials (service providers) and people with disabilities (service recipients) ( Byrne, 2010; Hair et al., 2019).

The analysis proceeded in a stepwise sequence of nested models. First, an unconstrained baseline model was estimated. We then imposed increasingly restrictive equality constraints across groups: (1) measurement weights (factor loadings), (2) structural weights (regression paths among latent variables), (3) structural covariances, (4) structural residuals, and (5) measurement residuals. This sequence allows assessment of cross-group comparability at the measurement and structural levels while preserving the same model form across groups.

Model adequacy and between-model comparisons followed the decision criteria reported in Table 2, using global fit indices and model-comparison statistics (Δχ ² and ΔCFI) to evaluate whether additional constraints materially degraded model fit.

The confirmatory factor analysis (CFA) results indicate an acceptable fit of the measurement model. Specifically, the Relative Chi-Square (χ ²/df), Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR) met the standard thresholds (see Table 2 and Table 4; Hair et al., 2019; Kline, 2015). Item-parameter estimates are reported in Supplementary Table S2.

Factor loadings for all observed variables exceeded 0.50, indicating sufficient convergent validity. Cronbach’s alpha coefficients ranged from 0.75 to 0.95, indicating acceptable to high internal consistency.

The MG-SEM analysis was conducted to examine the differences between government officials and people with disabilities regarding factors influencing audit data governance. The results demonstrate differences between these groups in several structural paths. The unconstrained baseline model indicated a good fit to the data (see Table 4), and the standardized path coefficients by group are summarized in Table 3.

Further testing of structural invariance using model comparison indices (Δχ ² and ΔCFI) confirmed differences in certain relationships between the two groups. Specifically, differences were notable in relationships involving data-driven organizational culture and audit data governance factors (see Table 4).

Hypothesis H1, stating that there are significant variations in the causal factors influencing audit data governance between government officials and people with disabilities, was supported. Standardized path coefficients are presented in Table 3, and the model comparisons used for invariance testing are reported in Table 4.

This study identified between-group differences between government officials and people with disabilities. This variation reflects differences in the social context and data-driven organizational culture between service providers and service recipients, particularly among People with disabilities, who often face limitations in accessing information. The Measurement Weights Model, which compares the differences in the model between the two groups, is shown in Figure 2. Between-group differences are summarized in Table 3, with supporting model-comparison evidence reported in Table 4.

Based on the multi-group path coefficient invariance testing, it was found that the path relationships between Data-Driven Organization and Public Service Delivery, as well as Audit Data Governance, vary between the group of government officials, who provide services, and the group of People with disabilities, who receive public services. This reflects that a Data-Driven Organization can enhance decision-making and data management more effectively within the Government Officials group, which has sufficient resources and access to information. Meanwhile, the People with disabilities group often faces challenges in accessing information and participating in policy decision-making processes ( Hofstede, 2010). This results in variations along these two paths, aligning with the study by Chen & Lee (2017), which indicates that People with disabilities faces barriers in accessing the information and services necessary to improve their quality of life. These variations highlight the lack of government support in bridging the information access gap for vulnerable groups.

However, although the service recipients may not have in-depth knowledge of organizational culture or data-driven decision-making processes, their perceptions are shaped by their experiences with public service delivery, particularly in terms of service quality and access to information. High-quality and efficient public service delivery can lead to greater satisfaction among People with disabilities, even though they may not have the same level of information access as government officials ( Nunnally & Bernstein, 1994; Parasuraman et al., 1988).

When testing measurement invariance, the constrained and unconstrained models were compared; the Measurement Residuals Model did not meet the model-fit criteria ( Table 4). This clearly indicates measurement variance or error between groups, possibly reflecting technical issues or environmental factors that cause inconsistencies in variable measurement between groups. For instance, limited access to public services may contribute to a lower quality of life for People with disabilities, consistent with research by the World Bank (2019), which highlights the connection between disability and poverty in developing countries. The report indicates that People with disabilities often has fewer opportunities for education or employment, and the lack of health resources and public service support in rural areas may create barriers to traveling to urban centers. Access to basic healthcare services for People with disabilities is therefore limited compared to the general population ( GHRP, 2017).

Additionally, statistics from this research indicate that 46.90% of People with disabilities are aged 50 or older, reinforcing the relevance of age-related issues. For example, a case study in Malawi found that elderly individuals with multiple chronic illnesses often prefer public hospitals due to the reliability of services, but still face geographical and social barriers that hinder access ( BMC Public Health, 2022). This similarity highlights the lack of public services in rural areas. Although technology has been introduced to public service delivery to reduce geographical barriers, challenges in accessing information through such technology remain, particularly for the visually impaired. This is reflected in the research by Henriquez-Camacho et al. (2014), which discusses the use of eHealth and mHealth technologies to assist the elderly in accessing healthcare information in rural areas. It was found that elderly individuals with higher education and those using technology to search for medical information had better access to services compared to those with lower education or who did not use technology.

While this study provides valuable insights into audit data governance and its impact on public service delivery for people with disabilities, several limitations should be noted.

First, the use of survey data allows for broad coverage and statistical robustness, capturing perceptions from a diverse sample. However, future research could further enhance these insights by incorporating mixed-method approaches, such as interviews or case studies, to provide deeper qualitative perspectives. Additionally, integrating objective behavioral data, such as digital service usage logs or administrative records, could offer more comprehensive and actionable insights into data governance effectiveness.

Second, by focusing on Thailand’s public sector, this study contributes to a deeper understanding of digital governance challenges in a developing country context. Future research could expand on these findings by conducting comparative studies across multiple countries or regions. Such studies could provide cross-cultural perspectives on the adoption of digital accessibility frameworks, policy effectiveness, and technological infrastructure development, offering insights into globally adaptable best practices.

Lastly, this study applies MG-SEM to examine group-based differences in audit data governance, providing a robust analytical foundation. Future research could further enhance methodological rigor by employing longitudinal modeling to track evolving trends over time or experimental designs to assess the real-world impact of digital governance interventions. These approaches would help policymakers and practitioners refine strategies for ensuring inclusive, transparent, and efficient digital governance frameworks.

By addressing these areas, future research can play a crucial role in advancing digital public administration, optimizing data governance strategies, and ensuring equitable service delivery for all citizens.

Based on the multi-group SEM results ( Tables 3 and 4), the findings reveal important distinctions between government officials and people with disabilities regarding the perceived importance of factors influencing audit data governance. Government officials emphasized data quality and organizational culture as essential drivers for effective data governance, reflecting their role in ensuring accurate and reliable data management systems. Conversely, people with disabilities highlighted data accessibility as a primary concern, underscoring the need for improved inclusivity in digital public service delivery.

These results are consistent with international evidence that highlights the necessity of inclusive data practices within public governance frameworks. The OECD (2024) emphasizes that data-driven practices in public sectors must prioritize both quality and accessibility to foster public trust and ensure equitable service delivery. Similarly, the European Commission’s eGovernment Benchmark ( European Commission, 2023) illustrates successful practices across European governments, such as Estonia and Denmark, where comprehensive data governance and inclusive digital platforms have effectively narrowed the digital accessibility gap for vulnerable populations, including people with disabilities.

This study aligns with global recommendations from the United Nations (2022), suggesting that inclusive and accessible public digital services significantly enhance trust and participation among vulnerable communities. The observed differences between government officials and people with disabilities highlight gaps that policymakers must address. Specifically, officials prioritize data-driven organizational culture and service quality, whereas people with disabilities emphasize accessibility and ease of data use.

Practically, these insights advocate for tailored policy frameworks to bridge gaps in perceptions and practices. For instance, governments can leverage strategies such as those implemented in European Union countries, which successfully employ comprehensive accessibility guidelines in digital service frameworks ( European Commission, 2023). Thailand could benefit from adopting similar inclusive digital governance practices, potentially enhancing the transparency, accessibility, and quality of public audit services.

Moreover, developing comprehensive guidelines that explicitly integrate accessibility considerations into digital audits can contribute significantly to closing the accessibility gap for people with disabilities. These guidelines should encompass clear mandates on data collection processes, analytical transparency, and effective dissemination practices aligned with international best practices ( OECD, 2019b).

Table 5 summarizes practical implications and policy recommendations, clearly highlighting the target audience, international alignment, and expected outcomes.

Data are available under the terms of the Creative Commons Attribution 4.0 International (CC-BY 4.0) license ( https://creativecommons.org/licenses/by/4.0/), permitting unrestricted use, distribution, and reproduction, provided the original authorship is properly cited.