Consumer trust in Thai street food vendors: implications for the post-pandemic era

Abstract

Purpose

This study aims to investigate the impact of service quality (SQ), perceived value (PV) and consumer satisfaction on Thai street food, with customer satisfaction (CS) as a mediator for customer loyalty and repurchase intention (RI). It also explores how consumer trust (CT) in Thai street food safety moderates these relationships.

Design/methodology/approach

Structural equation modelling (SEM) was utilised to analyse the complex interrelationships between various constructs. Multi-group analyses were conducted to investigate the moderating effects of CT on the structural model, considering two distinct groups based on trust levels: low and high.

Findings

The findings revealed that SQ and PV significantly influenced CS and behavioural intention, while the perceived quality of Thai street food had no significant impact on post-COVID-19 consumer satisfaction. The study highlighted the critical role of CT in moderating the relationships between SQ, PV and CS, with distinct effects observed in groups with varying trust levels.

Social implications

The research emphasises the importance of enhancing SQ and delivering value to customers in the context of Thai street food, which can contribute to increased CS, RI and positive word-of-mouth. Furthermore, the study underscores the critical role of building CT in fostering enduring customer relationships and promoting consumer satisfaction and loyalty.

Originality/value

This research offers valuable insights into consumer behaviour and decision-making processes, particularly within the realm of Thai street food. It underscores the significance of understanding and nurturing CT, especially in the post-COVID-19 landscape, emphasising the need for effective business strategies and consumer engagement.

Citation

Teangsompong, T., Yamapewan, P. and Sawangproh, W. (2024), "Consumer trust in Thai street food vendors: implications for the post-pandemic era", International Journal of Sociology and Social Policy, Vol. 44 No. 13/14, pp. 16-48. https://doi.org/10.1108/IJSSP-09-2023-0220

1. Introduction

As defined by the Food and Agriculture Organisation (FAO), street food is ready-to-eat food and beverage items sold and prepared by vendors or hawkers in streets as well as other public places (Lamuka, 2014). Approximately, 2.5 billion individuals consume street food daily, finding it an affordable, convenient and sometimes nutritious option (WHO, 2022). In Thailand, street food holds a position of immense significance (Suvittawat, 2023), serving as a vital element in the country's economic landscape and drawing in a considerable number of tourists (Torres Chavarria and Phakdee-auksorn, 2017). More than just a meal, Thai street food represents the nation's unique flavours and culinary traditions (Baldwin, 2017), integrates into daily life (von Bargen, 2016), fosters community ties and facilitates social interactions among locals and visitors (The Nation, 2018). Consequently, the thriving Thai street food commerce significantly contributes to the informal economy (Maglumtong and Fukushima, 2022), offering revenue to numerous vendors and impacting related sectors such as tourism (Boonpienpon, 2017), agriculture, food production and hospitality (Jeaheng et al., 2023).

The recent emergence of COVID-19 in December 2019 profoundly impacted the food industry globally, disrupting supply chains (Moosavi et al., 2022), changing consumer behaviour (Cai et al., 2023) and reshaping dining habits (Lamy et al., 2022). As the pandemic spread worldwide, governments enforced stringent measures, such as lockdowns and social distancing protocols, leading to the temporary closure of restaurants, cafes and eateries (Gursoy and Chi, 2020). This unprecedented situation resulted in a substantial decline in food service revenue, causing financial hardships for numerous businesses and the loss of numerous jobs (Maslach and Leiter, 2022). Concerns about illness spurred a shift towards home cooking (Goyal and Verma, 2021) and a surge in online food delivery services (Deepika and Joe Arun, 2021). Consequently, the industry rapidly adopted contactless delivery and reinforced hygiene standards, emphasising food safety and quality assurance to maintain customer trust and safety (Jiang et al., 2023; Gine-Garriga et al., 2021).

Consumer trust (CT) in the context of Thai street food vendors during the pandemic is a critical aspect influencing public health, consumer behaviour and the street food industry, as highlighted by Oloko and Ekpo (2021). Heightened concerns about viral transmission have amplified the importance of cleanliness and food handling practices, significantly impacting consumers' perceptions of food safety (Wu et al., 2021). A lack of trust in food safety might diminish street food demand, impeding vendors' recovery and expansion. Understanding consumers' trust in street food vendors is pivotal at this juncture, potentially influencing their choices and thereby impacting vendors' livelihoods and the sector's economic viability.

While previous studies have explored customer satisfaction (CS), perceived value (PV) and behavioural intention across service industries (McDougall and Levesque, 2000; Petrick, 2004; Yu et al., 2014), limited research has specifically delved into the role of CT concerning street food vendors, particularly in the post-pandemic era. This study employs the Stimulus–Organism–Response (S–O–R) theory to establish a framework linking service characteristics and customer behavioural intentions (BI) (Ryu et al., 2010). Perceived quality (PQ), service quality (SQ) and PV are considered stimuli (S), while CT and satisfaction represent emotional reactions (O), and BI serves as the response (R). The study aims to examine the causal relationship between these factors and explore the moderating role of CT in Thai street food vendors in the post-COVID-19 pandemic. These findings have the potential to enhance understanding of consumer perceptions, preferences and behavioural intentions, empowering street food vendors and policymakers to make informed decisions that strengthen food safety practices and sustain CT in the post-pandemic era.

2. Literature review

2.1 Essential role of street food in Thailand's economy and cultural identity

Street food holds an indispensable position in Thailand, serving as a vibrant and integral aspect of the country's economy and cultural identity (Boonpienpon, 2017; Chaiyasain, 2020; Jeaheng et al., 2023). The bustling street food scene has a profound impact on both the nation's financial prosperity and its rich heritage. From bustling night markets to humble street-side vendors, the ubiquity of street food reflects the heart and soul of Thailand's culinary landscape. This thriving street food culture plays a pivotal role in boosting the country's economy and preserving its cultural heritage (Kitlertsirivatana, 2022; Lawson, 2021; Tuohy, 2017). Economically, street food is a powerful driver of growth and employment in Thailand. It provides livelihoods for countless vendors, creating a web of entrepreneurial opportunities throughout the country (Chayada, 2017). In 2016, approximately 326 billion baht (ca 8,477 million euros) or roughly 20% of Thailand's total tourism revenue from international visitors, was spent on cuisine, according to the Tourism Authority of Thailand (TAT, 2017). Among many statistics, including CNN (2017), TAT (2019) and Prachachat (2018), Thailand is one of the culinary havens and street food capitals of the world. Many people throughout the globe have Thailand on their bucket lists for a variety of reasons. These include its awe-inspiring natural resources (McDowall and Choi, 2010), the friendliness of its people, its exotic cultures and both its remarkable local and international culinary scenes as well as remarkable experiences (Henkel et al., 2006). The restaurant industry in Thailand has been prosperous, and its contribution to the nation's gross domestic product (GDP) has increased steadily since 2009 and continued until the impact of the coronavirus (COVID-19) pandemic at the end of 2019. Despite this, the number of restaurants increased during that year (Statista Research Department, 2023). The relatively low-cost nature of street food attracts locals and tourists alike, bolstering tourism revenue and stimulating economic activity in urban centres and tourist destinations (Kitlertsirivatana, 2022). Culturally, street food serves as an edible diary of Thailand's traditions and history (Baldwin, 2017; Lee, 2009). The diverse range of flavours and cooking techniques found in street food reflects the country's regional variations and influences from neighbouring countries (Seubsman et al., 2009). The importance of street food in Thailand, both economically and culturally, cannot be overstated. By supporting the street food industry, Thailand not only bolsters its economy but also proudly showcases its rich cultural heritage to the world, solidifying its place as a culinary mecca (Su and Run, 2023).

2.2 S–O–R theory

The S–O–R theory outlines how various environmental factors influence an individual's internal cognitions and affections, subsequently impacting their behavioural responses. The environmental factors are considered stimuli (S), which have an impact on humans' perception, i.e. the organism (O), leading to the development of behavioural responses (R) (Mehrabian and Russell, 1974). This study applies the S–O–R theory to identify customer engagement behaviours for two main reasons. Firstly, previous studies have effectively utilised the S–O–R theory in food business research to elucidate consumer behaviours, including customer loyalty (CL) (Jang and Namkung, 2009; Lee and Yun, 2015; Yadav and Rahman, 2018) and repurchase intention (RI) (Gupta et al., 2018). Secondly, the S-O-R theory helps to predict the role of street food features and customer behaviour by offering a visual framework for exploring how environmental stimuli influence consumers' internal psychological reactions, thus shaping their behavioural responses (Aslam and de Luna, 2021). Consequently, we adopt the S–O–R theory as our research framework. More specifically, the current study treats PQ, SQ and PV as stimuli (S), CS and CT as reflections of emotional reactions or organisms (O) and consumer behaviour, i.e. CL and RI, as a response (R).

2.2.1 Environmental stimuli (S)

2.2.1.1 Perceived quality (PQ)

According to Parasuraman et al. (1985) and Zeithaml (1988), PQ is the subjective assessment a customer makes of a product's overall superiority to other products in their considered options. It represents customers' judgements and assessments based on their experiences, expectations and overall satisfaction with the product or service (Zeithaml, 1988). PQ holds substantial importance in influencing consumer behaviour, impacting various key factors such as CS, loyalty and repurchase decisions (Suttikun et al., 2021). Aaker (1991) emphasises the significance of PQ, highlighting its role in shaping customers' perceptions of value, which in turn influences their willingness to pay for a particular product or service. This aspect underscores the importance for businesses to focus on managing and enhancing PQ to meet and exceed customer expectations, thereby establishing a competitive advantage in the market (Matzler and Hinterhuber, 1998).

2.2.1.2 Service quality (SQ)

SQ represents the overall excellence of service delivery that meets or exceeds customers' expectations (Apte, 2004). It encompasses various aspects such as responsiveness, reliability, assurance, empathy and tangibles, which collectively contribute to customers' satisfaction and overall service experience. As highlighted by Cronin and Taylor (1992), SQ significantly influences customers' perceptions and plays a pivotal role in shaping their satisfaction levels and subsequent BI. Moreover, studies by Berry and Parasuraman (2004) emphasise the importance of SQ in fostering CL and positive word-of-mouth, ultimately enhancing the company's reputation and competitive edge. With the increasing emphasis on customer-centric approaches, businesses need to prioritise the continuous improvement of SQ to retain existing customers, attract new ones and establish a strong market position.

2.2.1.3 Perceived value (PV)

PV refers to the worth that a product or service has in the eyes of consumers based on their perception of the benefits and costs (Zeithaml, 1988). It is not merely determined by the intrinsic characteristics of the product but also by the psychological and emotional factors that influence a consumer's decision-making process (Gutman, 1997). As suggested by Lichtenthal et al. (1997), PV is a critical component in building a competitive advantage for businesses. By offering products or services that are perceived to be superior to those of competitors in terms of benefits relative to cost, businesses can attract and retain customers, foster brand loyalty and create a sustainable market position. Effective management of PV involves understanding customer needs and preferences, developing compelling marketing strategies and delivering exceptional customer experiences. The successful alignment of PV with customer expectations and satisfaction is crucial for achieving long-term profitability and business growth (Mawson, 2018).

2.2.2 Organism (O)

2.2.2.1 Customer satisfaction (CS)

Bitner and Hubbert (1994) defined CS as two outcomes: individual service transactions and total service interaction. Customers usually evaluate service performance according to expectations before buying or using it (Oliver, 1980). CS is influenced by instrumental (physical product performance) and expressive (psychological product performance) factors. Expressions above or equal to expectations are connected with satisfaction, while instrumental outputs below expectations are associated with dissatisfaction (Swan and Combs, 1976). Thus, CS and CL are significantly and positively related (Akbar and Parvez, 2009).

2.2.2.2 Consumer trust (CT)

CT refers to the confidence and reliance that customers have in the service provider's ability to deliver promised services consistently and reliably (Sirdeshmukh et al., 2018). It encompasses the belief that the service provider will act in the customer's best interest, maintain confidentiality and ensure the security of any information shared. Morgan and Hunt (1994) emphasised that trust is necessary for establishing and maintaining long-term relationships in service marketing, as it can only exist when one party has confidence in the reliability and integrity of the other. According to Lau and Lee (1999), if one party trusts another party, this will eventually result in favourable BI towards the second party.

2.2.3 Response (R)

BI refer to the individual's planned or intended behaviour in a specific situation (Dick and Basu, 1994). BI encompass two important concepts: CL and RI. CL is defined as the consumer's intention to revisit a particular business or brand in the future (Amin, 2016). It is a crucial factor for the long-term survival of organisations (Dick and Basu, 1994; Supriyanto and Wiyono, 2021).RI, on the other hand, refers to the customer's intention to make repeat purchases from a specific business or brand (Dick and Basu, 1994). Both CL and RI are influenced by various factors such as SQ, CS, trust and brand image (Saleem et al., 2017; Harisandi and Purwanto, 2022). Overall, BI play a crucial role in shaping customer behaviour and are influenced by a range of factors that businesses need to consider in order to foster CL and encourage repeat purchases.

2.3 Hypothesis development

2.3.1 Perceived quality (PQ) of street food

Food quality is a key element of perceived product quality in the restaurant industry (Namkung and Jang, 2007). The combined effects of quality factors such as food, service and physical environment contribute to CS and BI (Ryu and Han, 2009). It has also been found that the quality of street food service, including the food itself, the service from staff, the atmosphere, the price and how quickly you can get your food, has a positive effect on both utilitarian and hedonic values (Seo and Lee, 2021). Overall food quality significantly affects CS and BI, with satisfaction mediating the relationship between food quality and BI (Namkung and Jang, 2007). Cultural experiences, health and safety of the food, sensory appeal, staff service and physical environment also influence street food quality perception and RI among international tourists (Jeaheng et al., 2023). Factors such as cleanliness and organisation of the location have been found to positively influence the perception of product quality in Malaysian street food (Hanan et al., 2021). However, it is important to address food safety challenges and ensure satisfactory microbial quality in street-vended foods (Khairuzzaman et al., 2014). Therefore, perceived food quality is considered a determinant of CS in the restaurant industry (Ramanathan et al., 2016).

2.3.2 Service quality (SQ) of street food

SQ contributes to CS and BI in the restaurant industry (Ryu and Han, 2009). The dimensions of street food SQ, including food quality, employee service, physical environment, price and rapidity of service, have been found to positively impact customer values and RI (Seo and Lee, 2021). Additionally, Hanan et al. (2021) found that street food vendors' perceptions of their SQ had a positive impact on consumers' perceptions of their overall product quality. As a result, Hanan et al. (2021) argue that the level of service that vendors offer is crucial in influencing how consumers understand and perceive the quality of street food.

2.3.3 Perceived value (PV) of street food

PV plays a crucial role in shaping consumer behaviour and decision-making in the street food context (Khanna et al., 2021; Seo and Lee, 2021). The evaluation of PV in street food is based on both utilitarian and hedonic dimensions, considering factors such as taste, price, convenience and sensory appeal (Seo and Lee, 2021). The nutritional value of street food also contributes to its PV, with the ingredients used and cooking methods influencing consumers' perceptions of the food's healthiness and quality (Abrahale et al., 2018; Steyn et al., 2013). Additionally, the PV of street food is influenced by factors such as food neophobia, perceived risk and the overall SQ provided by street food vendors (Gupta et al., 2018; Khanna et al., 2021).

2.3.4 Customer satisfaction (CS) of street food

Previous research has highlighted the significance of CS as a predictor of BI (Ryu and Han, 2009). Street food vendors need to understand and meet customer expectations to ensure satisfaction and repeat patronage (Abrahale et al., 2018). Factors such as SQ, food quality and environmental attributes of street food establishments contribute to CS (Bannor et al., 2022; Sousa et al., 2022). Additionally, the perception of street food attributes, including taste, performance and nutritional value, influences CS (Hanan et al., 2021; Harthy et al., 2021). However, it is important to address potential risks and ensure food safety to enhance CS (Gupta et al., 2018; Sousa et al., 2021). Understanding customer perceptions, preferences and risk perceptions is crucial for street food vendors to deliver high-quality products and services that meet customer expectations and enhance satisfaction (Khanna et al., 2021; Seo and Lee, 2021). Overall, CS plays a vital role in the success and sustainability of street food businesses, emphasising the importance of meeting customer needs and preferences to foster positive experiences and CL.

2.3.5 Behavioural intentions (BI) in the context of street food

Suhartanto et al. (2018) found that e-SQ and food quality play a significant role in influencing CL towards online food delivery services. This suggests that customers are more likely to have positive BI, such as RI, when they perceive the food and service to be of high quality. Additionally, Abrahale et al. (2018) conducted a scoping review on street food research worldwide and found that the quality and safety of street food are important considerations for consumers. This implies that customers' BI, such as loyalty and RI, may be influenced by their perception of the safety and nutritional value of street food.

According to the aforementioned research, consumers' opinions of street food vendors depend on their perceptions of food quality, service and value. The attributes studied in this study reveal these preferences. Consumers' intentions to buy Thai street food again and suggest it depends on how they view its overall quality. Thus, the following assumptions were developed:

H1.

PQ will positively influence the CS.

H2.

SQ will positively influence the CS.

H3.

PV will positively influence the CS.

H4.

PQ will positively influence the BI.

H5.

SQ will positively influence the BI.

H6.

PV will positively influence the BI.

H7.

CS will positively influence the BI.

2.3.6 Moderating role of consumer trust (CT) in street food in the COVID-19 pandemic

The pandemic has heightened concerns about food safety and hygiene, leading to increased reliance on trusted sources of information and trusted vendors (Thomas and Feng, 2021). Trust in street food vendors is essential for consumers to feel confident in the safety and quality of the food they consume (Mucinhato et al., 2022). Trustworthy vendors who prioritise food safety measures and communicate transparently with customers can help alleviate concerns and enhance CT (Şahin and Gul, 2022). Additionally, trust in scientists and experts can influence consumer compliance with recommended behaviours, such as practicing safe food handling and following hygiene protocols (Matta et al., 2022). Building and maintaining CT is crucial for street food vendors to navigate the challenges posed by the pandemic and ensure CS and loyalty. Thus, the following hypotheses are proposed:

H8a.

CT moderates the effects of the PQ on the CS.

H8b.

CT moderates the effect of the SQ on the CS.

H8c.

CT moderates the effect of PV on CS.

H8d.

CT moderates the effect of PQ on BI.

H8e.

CT moderates the effect of SQ on BI.

H8f.

CT moderates the effect of PV on BI.

H8g.

CT moderates the effect of CS on BI.

3. Methodology

3.1 Theoretical framework

Figure 1 represents the conceptual model of Thai street food's PQ, SQ, PV, CS, BI and CT in its safety after the COVID-19 pandemic.

3.2 Participants and procedure

This research presents the analysis of data collected from Thai consumers residing in Bangkok and its surrounding areas who had recently purchased Thai street food within the last 6 months. For structural equation modelling (SEM), use a 5:1 to 10:1 subject-to-index ratio (Hair et al., 2010). The study comprised 31 items, so a 10:1 ratio was utilised to obtain a sample size of 310 with 5% precision and 95% confidence. This study selected 500 samples using simple random sampling, which is the proper size for this survey. The researchers reached out to members and followers of social media communities focused on Thai foods, inviting them to take part in an online survey. Interested participants were requested to provide their emails, from which 500 were randomly selected to receive a Google Form questionnaire. The data collection phase involved the distribution of 500 self-administered structured questionnaires, which took place from April to June 2023. The response rate was 84.8%, with 424 respondents returning 500 surveys. Later analysis used these 424 responses. The data collection approach includes emailing potential participants and requesting permission to take the Google Forms survey. The questionnaire's expected completion time was given, and respondents were asked if they wanted to take it in June 2023. Participants were also informed of confidentiality. Table 1 provides extensive demographic information about respondents for clarity.

The multiple-group analysis for each moderator variable, CT towards Thai street food vendor safety following the COVID-19 pandemic, used low and high CT subsamples. The summed CT scores of seven items (see Table A1 and Figure A1) were calculated to establish these subgroups, with a Cronbach's alpha of 0.955 indicating high internal consistency. Out of 424 respondents, 204 had low CT in Thai street food and 220 had high CT.

The study had 424 participants, 59.9% of whom were women. The largest age group was 18–30, with 179 (42.2%) responders. Regarding education, 169 (39.9%) respondents had education below a bachelor's degree, 215 (50.7%) held a bachelor's degree and 40 (9.4%) had an education above a bachelor's degree. Student occupation accounted for 265 (62.5%) of responses. 51.4% of participants earned less than 10,000 baht per month, followed by 25.5% of participants earning 10,001–20,000 baht per month. Most respondents (62.3%) ate street food once or twice a week.

3.3 Measures

A nine-section questionnaire was used to collect data for this investigation. Table 2 shows the constructs were measured using validated scales from prior research. The first segment included two questions to screen 18-year-olds who have bought street food in the past six months. The second portion collected demographic data on gender, age, education, occupation, monthly income and street food intake. Sections three to nine covered food quality, SQ, price, trust in food safety after the COVID-19 epidemic, satisfaction with street food vendors' adaptability, loyalty and RI. These sections had one open-ended question for respondents to express their ideas.

Sections three to nine variables were measured using a five-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). Three faculty members assessed the scale structure and content and 30 Thai street food consumers were tested for questionnaire consistency. The final instrument was refined using reliability and validity studies based on pilot sample feedback. Items with factor loadings below 0.5 or cross-loadings above 0.3 were deleted (Bagozzi and Yi, 1988). The variables showed high reliability, with Cronbach's α values ranging from 0.711 to 0.931 (Nunnally and Bernstein, 1994). The final 31-item questionnaire assessed the seven constructs.

3.4 Data analyses

The model was examined using SEM with AMOS 21. To compare the proposed model's measurement and structural invariance using data from the two groups, tests were conducted. Descriptive statistical analysis and reliability testing were carried out using SPSS 21. The SEM followed a two-step approach. Firstly, to validate the developed constructs, a confirmatory factor analysis (CFA) was conducted to ensure that all measurement items were appropriately loaded onto their respective constructs and that the constructs were correlated in the analysis. Subsequently, the SEM analysis was tested to assess the causal relationships between variables. Various fit indices were employed to assess the structural models, including the goodness-of-fit index (GFI), comparative fit index (CFI), root mean square error of approximation (RMSEA) (Anderson and Gerbing, 1988), normed fit index (NFI) (Byrne, 1994) and standardised root mean square residual (SRMR) (Byrne, 1994). Good model fit was indicated when GFI >0.9 (Pavlov et al., 2021), CFI >0.9, RMSEA <0.08 (Pavlov et al., 2021), NFI >0.9 (Anderson and Gerbing, 1988) and SRMR <0.05 (Curran et al., 1996). Moreover, a multi-group invariance analysis was performed on the structural model to examine potential group differences in the extent of paths (structural weight) between the two groups. Before estimating invariance models, a configural model was established as the baseline for all subsequent invariance tests.

The model was examined using SEM with AMOS 21. Tests compared the proposed model's measurement and structural invariance using data from the two groups. SPSS 21 was used for descriptive statistics and reliability testing. The SEM followed a two-step approach. Firstly, to validate the developed constructs, a CFA was conducted to ensure that all measurement items were appropriately loaded onto their respective constructs and that the constructs were correlated in the analysis. Subsequently, the SEM analysis was tested to assess the causal relationships between variables. Various fit indices were employed to assess the structural models, including the goodness-of-fit index (GFI), comparative fit index (CFI), root mean square error of approximation (RMSEA) (Anderson and Gerbing, 1988), normed fit index (NFI) (Byrne, 1994) and standardised root mean square residual (SRMR) (Byrne, 1994). Good model fit was indicated when GFI >0.9 (Pavlov et al., 2021), CFI >0.9, RMSEA <0.08 (Pavlov et al., 2021), NFI >0.9 (Anderson and Gerbing, 1988) and SRMR <0.05 (Curran et al., 1996). Moreover, a multi-group invariance analysis was performed on the structural model to examine potential group differences in the extent of paths (structural weight) between the two groups. Before estimating invariance models, a configural model was established as the baseline for all subsequent invariance tests.

4. Results

4.1 Confirmatory factor analysis (CFA) and reliability analysis

PQ, SQ, CS, CL and RI constructs and measurements were used in a CFA to assess scale construct validity. The CFA showed that the fit indices met the desired levels (χ² = 456.636 (p < 0.001); df = 232; χ²/df = 1.968; GFI = 0.918; CFI = 0.974; RMSEA = 0.048; NFI = 0.949; SRMR = 0.028). CFA and reliability analysis verified the measurement scales. All constructs have Cronbach's alpha values over 0.75 (Table 3), indicating acceptable internal consistency (Hair et al., 2021).

The CFA results were used as a reference point for constructing validity tests and gaining a deeper understanding of the measurement model outcomes. Based on the CFA results, the study further examined convergent validity, discriminant validity and the reliability of all multi-items. All indicators loaded significantly on their respective proposed constructs at p < 0.001.

The composite construct reliability (CR) estimations were 0.779–0.895, substantially over the proposed cut-off of 0.70 (Angwyn et al., 2022), indicating acceptable reliability. Furthermore, reliability coefficient (rhoA) estimates ranged from 0.854 to 0.934, satisfying the 0.70 cut-off.

Convergent validity refers to the average variance extracted (AVE) in a particular measure that converges to represent the underlying construct. The AVE is the mean of the squared loadings of each indicator associated with a construct. AVEs greater than 0.50 indicate convergent validity (Bagozzi et al., 1998). All construct indicators in this investigation had AVEs above 0.50, demonstrating convergent validity.

The square root of the AVE along the diagonal line (Table 4) should be greater than the correlation coefficients (r) between a construct and others in the model in the off-diagonal elements in the corresponding columns to assess discriminant validity. This study suggests discriminant validity because a component is more strongly correlated with its indicators than with the other model constructs (Teo and Noyes, 2014).

Table 5 shows discriminant validity results for conceptually similar (HTMT <0.90) and conceptually distinct (HTMT <0.85) constructs (Hair et al., 2021). The HTMT evaluates construct linking using the disattenuated construct score. This study showed no discriminant validity difficulties because all values fulfilled the predicted requirements, showing good CR and validity.

4.2 Structural equation modelling (SEM)

In this study, path analysis was used to investigate structural model hypotheses. SEM, which uses multiple regression and first-order CFA to estimate relationships between interconnected variables (causal factors like PQ, SQ and CS, which are antecedent variables of BI) (Hair et al., 2010), was used to test the hypotheses. Second-order CFA will be utilised to assess BI construct indicators CL and RI. All 424 samples were analysed. The initial SEM results did not match the acceptable fit indices for the suggested model (χ² = 566.100 (p < 0.001); df = 240; χ²/df = 2.359; GFI = 0.900; CFI = 0.962; RMSEA = 0.057; NFI = 0.936; SRMR = 0.031). The model was changed to increase fit indices, resulting in the desired alignment (χ² = 464.818 (p < 0.001); df = 235; χ²/df = 1.978; GFI = 0.917; CFI = 0.973; RMSEA = 0.048; NFI = 0.948; SRMR = 0.030). Table 6 shows the fit indexes and fit values of the SEM model, revealing the theoretical model's suitability through SEM analysis.

The estimated model's findings from pooled samples (N = 424) are provided in Table 7 and Figure 2 to show the standardised path coefficients' direction and magnitude. Five of seven hypotheses are accepted based on all standardised path coefficients, t-values and findings. PQ does not significantly impact CS (β = 0.080, p = 0.284). In other words, Thai street food quality does not improve CS. Thus, hypothesis 1 is not supported. Both SQ (β = 0.134, p = 0.030) and PV (β = 0.477, p = 0.000) significantly influenced CS, supporting hypotheses 2 and 3. PQ did not significantly affect BI (β = −0.016, p = 0.827), disproving hypothesis 4. The study found that SQ (β = 0.248, p = 0.000) and PV (β = 0.129, p = 0.019) significantly influenced BI, validating hypotheses 5 and 6. Finally, CS positively influenced BI (β = 0.558, p = 0.000), validating Hypothesis 7.

4.3 Multi-group analysis for moderating effects of consumer trust toward street food safety

The study used multi-group invariance analyses with AMOS 21.0 and the maximum likelihood (ML) estimation method to find out if the proposed structural model is seen differently by two groups of people when it comes to trusting the safety of Thai street food after the COVID-19 pandemic. A metric invariance test was done (see Table 8) to see if CT had an effect on the structural model that was proposed for hypotheses 8a–8g. Initially, the sample was divided into low CT (n = 204) and high CT (n = 220) groups. This study averaged the answers to seven CT scale questions and found the median score (mean = 6.569; SD = 3.796; median = 4.500). Therefore, this study identified two different groups: the low consumer trust (low CT) group (below 4.500) and the high consumer trust (high CT) group (above 4.500) (see Table A1 and Figure A1). However, the fit indices of the proposed model did not meet the desired levels in the multi-group analysis (χ² = 1587.704 (p < 0.001); df = 720; χ²/df = 2.205; GFI = 0.870; CFI = 0.949; RMSEA = 0.038; NFI = 0.911; SRMR = 0.038). To improve the model's strength, additional paths were added. The resulting modified model demonstrated greater statistical robustness than the initial proposal model (χ² = 1256.057 (p < 0.001); df = 690; χ²/df = 1.820; GFI = 0.895; CFI = 0.967; RMSEA = 0.031; NFI = 0.930; SRMR = 0.036). Table 8 provides the values of the fit indices and the fit values of the model to assess the adequacy of the structural invariance of the configural invariance (Model 1) through multi-group analysis. The results indicate that Model 1 is not invariant in form.

Then, the structural invariance of configural invariance (Model 1) and the measurement weight invariance (Model 2) were compared between the low CT and high CT groups. The results are shown in Table 8. Model 2 displayed good fit indicators: χ² = 1296.147 (p < 0.001); df = 726; GFI = 0.892; adjusted goodness-of-fit index (AGFI) = 0.866; NFI = 0.928; CFI = 0.967; RMSEA = 0.030. Δχ² = 40.090 and Δdf = 36 (p = 0.294) indicated no invariance in the measurement weight. Furthermore, the structural invariance of the configural invariance (Model 1) was compared to the structural weight invariance (Model 3) between the low CT and high CT groups. Model 3 demonstrated good fit indicators: χ² = 1343.065 (p < 0.001); df = 742; GFI = 0.889; adjusted goodness-of-fit index (AGFI) = 0.865; NFI = 0.925; CFI = 0.965; RMSEA = 0.031. Δχ² = 87.009 and Δdf = 52 (p = 0.002) indicated that at least one structural path exhibited different strengths between the groups. A path difference comparison between the two groups was conducted further (Table 9). After confirming configural invariance, metric invariance was tested with the help of the Stats Tools Package in Excel (Bryne, 2010) by using critical ratios for differences in a multi-group moderation test. A z-score greater than 1.96 indicates a difference (p-value <0.05) between the two groups (Aboobaker et al., 2019), suggesting differences in at least one structural path between the groups (Table 9).

According to Table 9, three of the seven paths in the model showed significant differences between the two groups regarding CT, implying that CT moderates the relationships between various factors and CS in Thai street food safety in the post-COVID-19 pandemic.

1. The high CT group had a higher effect on SQ (β = 0.363) than the low CT group (β = 0.001). This validates hypothesis H8b, showing that Thai street food CT greatly affects SQ and CS. Higher trust in Thai street food safety boosts SQ's impact on CS.

2. The PV effect was larger in the high CT group (β = 0.619) compared to the low CT group (β = 0.299). This supports hypothesis H8c, showing that CT considerably moderates the relationship between the PV of Thai street food and CS. Thus, PV affects CS more when buyers believe Thai street food is safe.

3. The high CT group had a higher influence on CS (β = 0.585) than the low CT group (β = 0.320). CT significantly moderates the association between CS with Thai street food and behaviour intention (supporting hypothesis H8g). In summary, CS affects BI more when buyers trust Thai street food safety.

This also shows that only high-CT consumers are affected by SQ (hypothesis H8b in Figure 3). This study examined SQ in terms of friendly, helpful, fast and comfortable service. Thus, street food consumers with high trust in Thai street food safety had high consumer satisfaction through SQ, but those with low trust did not. The partial moderating effect of CT in Thai street food safety in the post-COVID-19 pandemic is seen in Figure 3.

5. Discussion

This study confirms that SQ, PV and consumer satisfaction affect Thai street food satisfaction and BI. CS mediates RI and loyalty. CT in Thai street food safety moderates the relationship between SQ, PV and CS or BI. These findings corroborate the impact of satisfaction and intention, advancing street food consumer behaviour research.

By demonstrating that Thai street food quality does not significantly affect post-COVID-19 consumer satisfaction (hypothesis H1), this study contradicts earlier studies (Amir, 2021; Basir et al., 2022; Zibarzani et al., 2022). Quality, being a subjective concept influenced by various factors (Zeithaml, 1988), contrasts with the notion that CS is primarily based on meeting or exceeding expectations (Amir, 2021). Notably, despite difficulties with sanitation, street food consumption is resilient (Privitera and Nesci, 2015), frequently as a result of factors like reasonable prices (The Nation, 2018). Considering the study’s context, which primarily involves limited-income students, the findings suggest that PQ holds less significance compared to service and PV (Krommuang et al., 2017; Eresia-Eke et al., 2020).

The present research demonstrates that Thai street food SQ significantly influences CS (hypothesis H2). This supports prior studies (Gopi and Samat, 2020; Girdharwal et al., 2021; Basir et al., 2022). Meeting consumer expectations is highlighted as a crucial factor that contributes to the perception of high SQ (Mohamad et al., 2021). This aligns with findings showing that a clean and hygienic setting enhances tourist satisfaction with street food (Harthy et al., 2021). Correspondingly, direct relationships between perceived SQ and CS have also been documented in contexts such as Bangladesh (Ahmed et al., 2022) and Indonesia (Rinaldus and Christian, 2020). The provision of high-quality service remains a critical driving force behind enhancing CS (Rinaldus and Christian, 2020).

The current investigation affirms that PV has a significant influence on CS regarding Thai street food (hypothesis H3), indicating that meeting or surpassing customer expectations leads to satisfaction (Lee et al., 2007; Ryu et al., 2008). Various dimensions of PV contribute to satisfaction, including functional, image, emotional and trust value (Wuttipan et al., 2018). Similar effects of PV on satisfaction have been observed in Bangladesh (Uzir et al., 2021) and Indonesia (Gulam et al., 2023). Positive relationships between factors like food price, quality, hygiene, word-of-mouth and consumer satisfaction have been found, particularly in the post-COVID-19 period (Suraini et al., 2023), increasing the likelihood of future street food purchases.

The current study did not find a positive link between PQ and BI (hypothesis H4), contrasting with research in the United Arab Emirates by Ababneh et al. (2022). Their findings highlighted COVID-19 safety and food quality as significant predictors of CS and BI, even when considering SQ, PV and food quality. The importance of PQ has been emphasised due to safety concerns (Ryu and Han, 2009). Restaurants offering high-quality, reasonably priced food tend to have satisfied customers who return and recommend the restaurant to others. While this study did not directly connect PQ and BI, it is essential to consider multiple factors, including COVID-19 safety, food quality, SQ, PV and pricing, to understand customer behaviour and intention during challenging times.

The research underscores the substantial influence of SQ on BI to consume Thai street food, aligning with prior research across various contexts (hypothesis H5). For example, Kumar and Neha (2020) found that perceived SQ directly influences BI. Similarly, Mohamad et al. (2021) identified food hygiene, price, quality and SQ as accounting for 47.4% of the BI to revisit Penang street food spots. Eresia-Eke et al. (2020) established significant relationships between SQ dimensions and BI for street food in South Africa’s townships. Thatchinamoorthy and Meenambigai (2017) noted that CS with food quality, price and service, including a personal connection with vendors, leads to return visits. Research in the restaurant industry in China has also identified the importance of SQ, among other factors, in influencing BI and increasing CL (Liu, 2009). Villanueva et al. (2023) highlight the role of SQ in BI and CL, particularly when patrons perceive staff as competent, friendly and trustworthy. These consistent findings emphasise SQ's critical role in influencing BI and CL, extending beyond Thai street food to various settings and emphasising the importance of positive and trustworthy customer experiences.

The findings empirically support a direct link between the PV of Thai street food and BI (hypothesis H6), consistent with the findings of Nitchhote and Nuangjamnong (2022) during COVID-19's “new normal” situation, where PV, including risk and attitude, influenced purchase intention. Further support is found in the finding of Nitiwanakul (2014), showing PV and cost as primary factors influencing Thai consumers' intentions in fine dining. Service value, encompassing the PV of street food experience, quality, price, customer service, gastronomy and word of mouth, positively impacted revisit intention (Nitiwanakul, 2014). Similar findings were observed by Khanna et al. (2021), with PV through word of mouth positively affecting tourists' attitudes and street food consumption intention, though food neophobia and perceived risk had negative effects. This comprehensive understanding highlights the intricate dynamics of consumer behaviour, particularly in the context of Thai street food consumption during the COVID-19 pandemic.

According to this study, satisfied consumers are more likely to make repeat purchases and recommendations for Thai street food, reaffirming the direct relationship between customer happiness and BI (hypothesis H7). This aligns with Ababneh et al. (2022), who found a connection between satisfaction with COVID-19 safety measures and BI in the UAE's fast-food industry. Rehman et al. (2021) showed that tourist satisfaction with local food in Pakistan influenced BI during the pandemic. Hagen (2022) found satisfaction and loyalty were crucial for food delivery app use in Thailand during the pandemic. Namkung and Jang (2007) demonstrated that satisfaction, especially food quality, directly affects BI in mid-to-upscale restaurants. Qin and Prybutok (2017) found CS to be a key precursor to fast-food restaurant intentions among college students in the US. These findings underscore CS's enduring importance, especially in food consumption and during challenging circumstances like the COVID-19 pandemic.

The comparison of path differences reveals that CT is crucial in the context of Thai street food safety in the post-COVID-19 pandemic. Three paths (hypotheses 8b, 8c and 8g) are very different between the low CT and high CT groups. This shows how trust is very important in changing the link between SQ, PV and CS (Table 9). This investigation offers valuable insights into how trust impacts these variables, subsequently influencing BI. Ratasuk's findings (2023) suggest that the perception of food hygiene among Bangkok's street food customers during the COVID-19 pandemic influences their purchase intentions. Furthermore, the study highlights the positive influence of customer trust on purchase intentions while demonstrating a negative association with perceived risk. Trust is often considered a critical factor in nurturing positive consumer attitudes and beliefs by influencing their willingness to engage in transactions and make purchasing decisions, particularly under conditions of uncertainty and risk (Lee and Turban, 2001). According to Luhmann (1988), trust is a key factor in determining action when there is a perceived risk of a negative outcome. Previous meta-analyses have indicated that heightened levels of CT can lead to increased purchase intentions (Wang et al., 2022), as consumers are inclined to view products or services favourably and develop a strong sense of confidence in their purchase choices. Additionally, CT has been found to act as a key determinant of consumer loyalty, highlighting its important role in establishing enduring relationships between consumers and businesses (Singh and Sirdeshmukh, 2000; Gefen et al., 2003). Consequently, understanding and cultivating CT are vital for future research, food enterprises, tourism organisers, businesses and policymakers aiming to boost consumer buying intentions and foster enduring customer relationships.

6. Theoretical and practical implications

From a theoretical perspective, this study has significant implications for the fields of consumer behaviour and market research. Firstly, the extensive use of CFA and reliability analysis to assess scale construct validity and reliability reinforces the importance of robust measurement techniques in empirical studies. The study's demonstration of convergent validity and discriminant validity serves as a methodological benchmark for future research in similar contexts. Additionally, the application of SEM) to evaluate the complex relationships between various constructs sheds light on the dynamics of consumer behaviour and decision-making processes, contributing to the advancement of theoretical models in the field. The identification of CT as a critical moderating factor in the post-COVID-19 context further emphasises the need for a comprehensive understanding of the role of trust in shaping consumer attitudes and intentions.

From a practical standpoint, the implications of this research are substantial for the street food industry and related businesses. The validation of the influence of SQ and PV on CS highlights the importance of enhancing SQ and delivering value to customers. Strengthening these aspects can lead to increased CS and subsequent BI, promoting repeat purchases and positive word-of-mouth. Furthermore, the identification of CT as a key moderator underscores the significance of building trust among consumers, especially in the context of food safety post-COVID-19. Strengthening food preparation transparency, promotion of hygiene practices, contactless payment options, regular sanitisation and disinfection, collaboration with health authorities to undergo regular inspections and certifications, conducting educational campaigns on food safety and hygiene for both staff and customers and the use of fresh and high-quality ingredients sourced from reputable vendors can help establish and reinforce CT, leading to increased CS and loyalty. These practical insights can guide street food vendors, policymakers and marketers in devising effective strategies to enhance consumer experiences and drive business growth in the competitive market landscape.

7. Research limitations

The research provides significant insights into the factors influencing consumer behaviour in the context of Thai street food. While the study demonstrated a comprehensive analysis of the various factors affecting CT, satisfaction and BI, several limitations should be considered. (1) The research focused solely on Thai street food vendors, limiting its applicability to other geographical regions or cultural contexts. Thai street food culture may have unique characteristics that distinguish it from street food cultures in other countries, such as specific taste preferences, culinary traditions and hygiene standards. Therefore, the findings of the study may not be easily transferable to other regions, limiting the generalisability of the research outcomes. (2) Self-reported data can be prone to biases, including social desirability bias, where participants tend to respond in a manner that is socially acceptable rather than reflecting their true perceptions or experiences. Additionally, participants might not accurately recall their experiences with Thai street food vendors, potentially affecting the accuracy of the findings. Future research could consider incorporating observational data or qualitative methodologies to supplement the self-reported data and provide a more comprehensive understanding of consumer behaviour. (3) The study's focus on specific factors such as SQ, PV and CS might overlook other important variables that could potentially influence CT and behaviour. Factors such as digital technology, environmental sustainability practices, food safety regulations or vendor transparency could significantly impact CT and satisfaction but may not have been adequately addressed in the research. Future studies could consider a broader range of variables to provide a more comprehensive understanding of the complex dynamics influencing consumer behaviour in the context of street food vendors. (4) The study's cross-sectional design limits its ability to establish causal relationships between the identified factors. While the research provided valuable insights into the associations between CT, satisfaction and BI, it did not determine the direction of causality. Longitudinal studies or experimental designs could provide more robust evidence regarding the causal relationships among the identified variables, enhancing the validity of the research findings.

8. Conclusions

The research deeply explores the interplay between SQ, PV and consumer satisfaction in the domain of Thai street food, with a primary focus on the pivotal role of CS in mediating RI and loyalty. It elucidates how CT, particularly in the safety of Thai street food, moderates the intricate relationship between SQ, PV and CS. Contrary to previous studies, the findings challenge the belief that post-COVID-19 consumer satisfaction is significantly reliant on the quality of Thai street food. Instead, the study highlights the critical significance of SQ and PV in this context, emphasising their substantial influence despite the limited impact of food quality alone. Moreover, it underscores how SQ shapes CS and BI while shedding light on the direct impact of PV on CS. Additionally, the research identifies customer trust as a fundamental factor shaping BI, emphasising the crucial need to nurture and maintain this trust to build enduring and robust customer relationships.

Figures

Figure 1

The proposed model illustrates the resulting research framework based on an extensive review of the literature

Figure 2

Path results of the research model using pooled samples (n = 424)

Figure 3

The invariance model and path coefficients (standardised coefficients) are presented as the case of the low CT (n = 204) and high CT (n = 220) groups

Figure A1

Path results of the customer trust CFA model using pooled samples (n = 424)

Table 1

Demographic characteristics of the respondents categorised into two groups based on their consumer tendency (CT): low CT and high CT

	All		Low CT		High CT
	Frequency (n = 424)	Percentage (%)	Frequency (n = 204)	Percentage (%)	Frequency (n = 220)	Percentage (%)
Gender
Male	170	40.1	87	42.6	83	37.7
Female	254	59.9	117	57.4	137	62.3
Age
18–20	179	42.2	93	45.6	86	39.1
21–30	204	48.1	100	49	104	47.3
31–40	39	9.2	11	5.4	28	12.7
Above 40	2	0.5	–	–	2	0.9
Education level
Below bachelor’s degree	169	39.9	93	45.6	76	34.5
Bachelor’s degree	215	50.7	93	45.6	122	55.5
Above bachelor’s degree	40	9.4	18	8.8	22	10
Occupation
Students	265	62.5	134	65.7	131	59.5
Owner business	93	21.9	40	19.6	53	24.1
Private company employees	22	5.2	9	4.4	13	5.9
Government officer	18	4.2	11	5.4	7	3.2
State enterprise employees	24	5.7	8	3.9	16	7.3
Others	2	0.5	2	1	–	–
Monthly income (baht)
Below 10,000	218	51.4	111	54.4	107	48.6
10,001–20,000	108	25.5	54	26.5	54	24.5
20,001–30,000	58	13.7	28	13.7	30	13.6
30,001–40,000	19	4.5	4	2	15	6.8
40,001–50,000	6	1.4	2	1	4	1.8
Above 50,000	15	3.5	5	2.5	10	4.5
Frequency of street food consumption
1–2 times/week	264	62.3	204	100	60	27.3
3–4 times/week	83	19.6	–	–	83	37.7
5–6 times/week	58	13.7	–	–	58	26.4
More than 7 times/week	19	4.5	–	–	19	8.6

Source(s): Author's own creation/work

Table 2

Constructs and measurements

Constructs	Measurement	References
Antecedent
Perceived quality	4-item scale	Prasetya and Sianturi (2019)
Service quality	4-item scale	Prasetya and Sianturi (2019) Ishmael et al. (2018) Seo and Lee (2021)
Perceived value	4-item scale	Toni et al. (2017) Nikhashemi et al. (2016)
Consumer trust toward street food safety	7-item scale	Yeo et al. (2021) Rita et al. (2019)
Consumer satisfaction	4-item scale	Prasetya and Sianturi (2019) Fuentes-Blasco et al. (2014)
Behavioural intentions
Customer loyalty	3-item scale	Nikhashemi et al. (2016)
Repurchase intention	5-item scale	Seo and Lee (2021) Toni et al. (2017)

Source(s): Author's own creation/work

Table 3

The confirmatory factor analysis results

Construct	Mean (±SD)	Factor loadings (t-value)	CA (α)	CR (rhoc)	AVE	Reliability coefficient (rhoA)
Antecedent	4.220 (±0.661)		0.913	0.939	0.793	0.914
Perceived quality (PQ)
Food is visually appealing	4.245 (±0.709)	0.790 (fixed)
Food is being served at the right temperature	4.233 (±0.718)	0.849 (19.661***)
Food is always freshly cooked	4.208 (±0.765)	0.891 (20.887***)
Food is always clean	4.193 (±0.775)	0.875 (20.399***)
Service quality (SQ)	4.459 (±0.666)		0.925	0.947	0.817	0.926
The staff are friendly	4.507 (±0.688)	0.883 (fixed)
The staff is very helpful	4.465 (±0.746)	0.853 (23.788***)
Staff provide fast and timely service	4.434 (±0.750)	0.887 (25.705***)
I feel comfortable with the service of the staff	4.432 (±0.766)	0.855 (23.945***)
Perceived value (PV)	4.182 (±0.645)		0.908	0.935	0.783	0.910
I'll buy street food, although it's a bit more expensive	4.092 (±0.786)	0.851 (fixed)
I get to experience eating street food	4.208 (±0.701)	0.863 (22.844***)
Overall, the street food has better price options compared to other restaurants	4.205 (±0.720)	0.805 (20.302***)
The food prices of street food are reasonable	4.224 (±0.707)	0.856 (22.438***)
Customer satisfaction (CS)	4.146 (±0.646)		0.922	0.945	0.810	0.923
I am satisfied with my experience with street food	4.173 (±0.702)	0.870 (fixed)
I feel like I made the right decision to go to the street food	4.150 (±0.749)	0.874 (24.287***)
Compared to my expectation, I am satisfied with my street food	4.135 (±0.710)	0.877 (24.599***)
Compared to other food vendors, the chosen street food is the best	4.128 (±0.710)	0.838 (22.662***)
Behavioural intention (BI)
Customer loyalty (CL)	4.070 (±0.674)		0.848	0.908	0.767	0.854
I will recommend this street food restaurant to others	4.123 (±0.704)	0.834 (fixed)
I will come back to use the service at the street food shop	4.113 (±0.702)	0.852 (20.048***)
I'm not interested in going to other similar street food vendors	3.974 (±0.905)	0.730 (16.817***)
Repurchase intentions (RI)	4.119 (±0.648)		0.933	0.949	0.788	0.934
I will invite my friends and other close people to use the street food service	4.092 (±0.743)	0.816 (fixed)
Although other types of restaurants are very interesting, I still decided to use the same street food vendor	4.116 (±0.707)	0.891 (22.412***)
I have the intention of choosing street food as my first priority when eating out	4.078 (±0.766)	0.864 (21.868***)
Whenever I have to buy food I'll buy it from a street food vendor	4.123 (±0.730)	0.875 (22.774***)
I will be buying food from street food in the near future	4.189 (±0.703)	0.866 (21.793***)

Note(s): χ² = 456.636 (p < 0.001); df = 232; χ²/df = 1.968; GFI = 0.918; CFI = 0.974; RMSEA = 0.048; NFI = 0.949; SRMR = 0.028

Source(s): Author's own creation/work

Table 4

Comparison of AVE and squared correlations of paired constructs

	1(PQ)	2(SQ)	3(PV)	4(CS)	5(CL)	6(RI)
1. Perceived quality (PQ) of Thai street food	0.891
2. Service quality (SQ)	0.618	0.904
3. Perceived value (PV)	0.658	0.413	0.885
4. Customer satisfaction (CS)	0.489	0.384	0.556	0.900
5. Customer loyalty (CL)	0.473	0.454	0.522	0.656	0.876
6. Repurchase intention (RI)	0.453	0.469	0.451	0.577	0.746	0.888

Note(s): The AVE is written in italics on the diagonal and the squared correlations of paired constructs are on the off-diagonal

Source(s): Author's own creation/work

Table 5

The heterotrait–monotrait ratio (HTMT)

	1(PQ)	2(SQ)	3(PV)	4(CS)	5(CL)
1. Perceived quality (PQ) of Thai street food
2. Service quality (SQ)	0.673
3. Perceived value (PV)	0.722	0.449
4. Customer satisfaction (CS)	0.532	0.414	0.606
5. Customer loyalty (CL)	0.537	0.513	0.592	0.738
6. Repurchase intention (RI)	0.490	0.504	0.489	0.620	0.837

Source(s): Author's own creation/work

Table 6

Fit indices of models

		SEM model (pooled sample)			Multi-group model
Goodness-of-fit statistics	Acceptable compliance criteria	Proposal model	Modified model	Acceptable compliance criteria	Proposal model	Modified model	Acceptable compliance criteria
Chi-square (χ2)		566.100	464.818		1587.704	1256.057
Degree of freedom (df)		240	235		720	690
χ2/df	<2.000	2.359	1.978	Accepted	2.205	1.820	Accepted
Goodness-of-fit index (GFI)	≥0.900	0.900	0.917	Accepted	0.870	0.895	Accepted
Normed fit index (NFI)	≥0.900	0.936	0.948	Accepted	0.911	0.930	Accepted
Comparative fit index (CFI)	≥0.950	0.962	0.973	Accepted	0.949	0.967	Accepted
Root-mean-square-error of approximation (RMSEA)	<0.050	0.057	0.048	Accepted	0.038	0.031	Accepted
Standardised root mean square residual (SRMR)	<0.080	0.031	0.030	Accepted	0.038	0.036	Accepted
Variance explained (R2)
Consumer satisfaction (CS)		0.393	0.386		0.393	0.387
Behavioural intention (BI)		0.632	0.623		0.632	0.618

Source(s): Author's own creation/work

Table 7

Structural parameter estimates

Path coefficient	Unstandardised coefficients (β)	Standardised coefficients	S.E.	t-statistic	p	Hypothesis testing results
Pooled sample (n = 424)
H1: PQ → CS	0.096	0.088	0.090	1.070	0.284	Not supported
H2: SQ → CS	0.135	0.134	0.062	2.169	0.030*	Supported
H3: PV → CS	0.438	0.477	0.062	7.033	0.000**	Supported
H4: PQ → BI	−0.016	−0.016	0.074	−0.218	0.827	Not supported
H5: SQ → BI	0.235	0.248	0.053	4.474	0.000**	Supported
H6: PV → BI	0.129	0.149	0.055	2.338	0.019*	Supported
H7: CS → BI	0.527	0.558	0.052	10.173	0.000**	Supported

Note(s): *p < 0.05, **p < 0.01

Source(s): Author's own creation/work

Table 8

Fit indices for multi-group invariance tests

Model	χ2	df	p	GFI	AGFI	NFI	RMSEA	CFI	Δχ2	Δdf	p	Results
Pooled sample model	237.991	206	0.063	0.957	0.937	0.973	0.019	0.996
Configural invariance (Model 1)	1256.057	690	0.000	0.895	0.863	0.930	0.031	0.976	–	–	–	–
Measurement weight invariance (Model 2)	1296.147	726	0.000	0.892	0.866	0.928	0.030	0.967	40.090	36	0.294	No invariance
Structural weight invariance (Model 3)	1343.065	742	0.000	0.889	0.865	0.925	0.031	0.965	87.009	52	0.002	Invariance

Note(s): Configural invariance (factor patterns equal); measurement weight invariance (factor loadings equal); structural weight invariance (structural relation among latent variables equal)

Source(s): Author's own creation/work

Table 9

Comparison of parameter estimates of the structural model

Regression Path	High CT			Low CT			High – Low CT Path differences comparison
	Unstandardised Coefficients; β	p	Result	Unstandardised Coefficients; β	p	Result	Differences	Z-score	Result
H8a: PQ → CS	−0.142 (−0.125)	0.299	NS	0.196 (0.188)	0.113	NS	−0.338	1.835	NS
H8b: SQ → CS	0.363 (0.373)	0.000***	Supported	0.001 (0.001)	0.995	NS	0.362	−3.182***	Supported
H8c: PV → CS	0.619 (0.571)	0.000***	Supported	0.299 (0.316)	0.000***	Supported	0.320	−2.161**	Supported
H8d: PQ → BI	0.015 (0.013)	0.901	NS	−0.040 (−0.053)	0.662	NS	0.055	−0.362	NS
H8e: SQ → BI	0.324 (0.316)	0.000***	Supported	0.180 (0.307)	0.003***	Supported	0.144	−1.456	NS
H8f: PV → BI	0.097 (0.085)	0.412	NS	0.078 (0.113)	0.219	NS	0.019	−0.142	NS
H8g: CS → BI	0.585 (0.556)	0.000***	Supported	0.320 (0.442)	0.000***	Supported	0.265	−2.519**	Supported

Note(s): **p < 0.05; ***p < 0.01; NS: no support. Standardised coefficients are written in parentheses

Source(s): Author's own creation/work

Table A1

Confirmatory factor analysis of consumer trust (CT)

Construct	Mean (±SD)	Factor loadings (t-value)	CA (α)
Consumer Trust (CT)	4.041 (±0.686)		0.955
Street food vendor is reliable	4.054 (±0.781)	0.857 (fixed)
Street food vendor is responsible	4.002 (±0.812)	0.861 (26.669***)
Street food vendor meet my expectation	4.019 (±0.781)	0.855 (24.582***)
I expect good food recommendations from street food vendors	4.033 (±0.765)	0.896 (25.037***)
If problems occur, street food vendors will treat them fairly	4.047 (±0.786)	0.892 (25.212***)
I am satisfied with the standards of street food vendors that operate	4.078 ± (0.741)	0.895 (21.908***)
The street food vendor operates with prudence	5.054 (±0.743)	0.818 (23.761***)
Frequency of street food consumption	1.604 (±0.885)	(Median = 1.000)
CTWeighted by frequency of street food consumption*	6.569 (±3.796)	(Median = 4.500)

Note(s): χ² = 21.606 (p < 0.156); df = 16; χ²/df = 1.350; Goodness-of-fit index (GFI) = 0.987; Normed fit index (NFI) = 0.993; Comparative fit index (CFI) = 0.998; Root-mean-square-error of approximation (RMSEA) = 0.029; Standardised root mean square residual (SRMR) = 0.015

*To perform the multiple-group analysis for a moderator variable (consumer trust towards Thai street food safety), two subsamples were created: a low consumer trust (Low CT) group and a high consumer trust (High CT) group. In order to build the subgroups, the summed scores of the latent variable, consumer trust, were calculated. Seven items were added up (Cronbach’s alpha, or, CA = 0.955). Next, consumer trust (CT) was weighted by the frequency of street food consumption (mean = 6.569, SD = 3.796). The subsamples (high and low for the CT construct) were created based on the median (4.500) of CT_Weighted by frequency to split the groups. High CT (median > 4.500) = 220 samples and High CT (median ≤ 4.500) = 204 samples

Source(s): Author's own creation/work