This study aims to explore the antecedent factors that directly and indirectly influence electronic word of mouth (eWoM) for social commerce (s-commerce) in two developing countries (e.g. Kuwait and the United Arab Emirates [UAE]) by extending social cognitive theory.
This study uses a previous robust model (Rouibah et al., 2021) as theoretical background to investigate and compares the antecedents (trust in Instagram, perceived risks) on eWoM for s-commerce through the mediation of three mediators (perceived enjoyment, perceived value and customer satisfaction) among two Arab countries. Data was collected from Kuwait (n = 1,132) and the UAE (n = 190). Different statistical analyses and structured equation modeling-based analysis of moment structure are used to test the robustness of the research model.
This study found customer satisfaction to be most important factor that mediates the effect independent factors on eWoM for s-commerce in both countries. Surprisingly, perceived enjoyment has no effect, and trust in Instagram and perceived risks are the most important factors that are considered imperative for customer satisfaction and positive feedback.
One limitation of this study is that the author does not focus on the difference between the effects of textual and graphical information on customers’ decisions and trust in buying merchandise. Another limitation is that this study focuses on Kuwait and the UAE. Other Gulf Cooperation Council countries are also growing exponentially, and mobile and internet penetration rates are booming; they could be a trigger for more studies on whether differences occur among all of them.
The first implication is that it is the first in its field to extend the effects of eWoM. To the best of the author’s knowledge, compared to the online research this study is unique because the authors examine six factors for eWoM in s-commerce using the Instagram platform as opposed to other platforms.
The third implication of this study is that the previous ones have applied eWoM to different subjects of e-commerce such as tourism and marketing but have concentrated less on s-commerce, where in-depth research is needed much more to explore factors and theories that explain human behavior.
Furthermore, most of these studies have focused on the intention to use (Dincer and Dincer, 2023; X. Hu, Chen, Davison, and Liu, 2022; Zhou et al., 2023). However, the attention in this research is on the actual use.
Introduction
Since the introduction of Web 2.0, more and more people use eWoM to communicate and exchange opinions and product information (Cheung and Thadani, 2012; Rouibah et al., 2021). Through eWoM, customers can evaluate products they purchased online by providing positive or negative feedback about their experiences.
Studying the effect of eWoM on potential customers is an important research area. Indeed, prior studies have posited that when customers make decisions regarding purchasing products from e-commerce websites, they tend to trust the online reviews of people they do not know more than the reports and ads in the traditional media (Cheung and Thadani, 2012). EWoM can be a vital information source for customers in deciding whether the product or service meets their levels of satisfaction. As eWoM takes place on the internet where consumers and posters seldom meet, consumers are usually unsure about the truthfulness and reliability of posters’ feedback. Therefore, companies are very concerned about how eWoM affects their customers’ attitudes. Furthermore, eWoM has significant effects on consumers’ decisions to purchase. For example, 93% of users claim that online reviews have a significant effect on their purchase behavior, 81% of consumers use Google to evaluate local companies, and 50% of all internet users post online reviews. Moreover, only a few percentage (2%) have never read an online review (Howarth, 2023). In general, large companies in the market invest millions of dollars to ensure they have a strong presence on social media. For example, General Motors announced it would invest 30 million annually to generate content on Facebook (Nadeem et al., 2021). According to MacRae (2024), 4.9 billion social media users across the world and 90% of marketers use social media as a tool for their marketing campaigns. In 2023 brands and big companies had spent more than $207bn globally while they spent $930bn between 2017 and 2023 (MacRae, 2024). Several scholars have argued that investing in the effect of eWoM based on social media results in many benefits for companies such as developing significant numbers of followers and facilitating customer relationships (Ashley and Tuten, 2015; López et al., 2017).
The literature about eWoM has received extensive research because of huge investments and technological developments. However, prior studies in this field are very fragmented and have many inconsistencies (Cheung and Thadani, 2012; Line et al., 2024; Liu et al., 2021). As a result, J. Lee and Lee (2009) have focused on two major levels of analysis: market and individual.
Unlike prior studies that focused on the effect of eWoM for individual social usage context, this paper investigates the effect of psychological factors on eWoM for s-commerce at the individual level for business context (i.e. for purchase reasons) across two different Arab countries (Al-Okaily, 2023; Rouibah et al., 2021). This focus differs from those of prior studies that investigated different eWoM issues (Rouibah et al., 2021) including the following: Why do consumers share eWoM about some products but not others (Y. Hu and Kim, 2018; Line et al., 2024)? Why do people accept or reject technological products even in crisis periods such as the COVID-19 pandemic (Al-Okaily, 2024a, 2024b)? Why do some customers use eWoM while others do not, and can the effect of positive and negative eWoM be distinguished (Bronner and De Hoog, 2011; Y. Hu and Kim, 2018)? Therefore, this study focuses on studying the effect of psychological factors (fear and trust in social media) on eWoM for s-commerce purposes based on Instagram. Prior studies have shown that enjoyment, risk and trust play large roles in the mood to interact with unknown individuals in social media (Rouibah et al., 2021) and other forms of information and communication technologies (Al-Okaily, 2024a; Al-Okaily et al., 2023).
This study compares the effect of antecedents of eWoM for s-commerce across two Arab countries (Kuwait and the United Arab Emirates [UAE])) for three reasons: First, they are among the largest economies in the Gulf Cooperation Council (GCC) (after Saudi Arabia) and have similar size e-markets (see Figure 1).
However, the UAE tends to be much faster at adopting new technologies and integrating them into business due to its culturally open environment and liberal political system. Second, we focus on these two countries for convenience objective, because we have easier access to local data and therefore to follow-up on research issues. Third, very few studies compared the adoption of different technologies among countries of the Gulf Cooperation Council including mobile telephone (AlMutairi and Yen, 2017), m-banking (Thrassou and Philip, 2008) and m-learning (Alsswey et al., 2020). However, neither of them focused on s-commerce (social media) nor used social theory. Following this analysis, we use the social theory as a lens for our comparative study, and we used the model developed and validated previously in Kuwait that includes three antecedents (propensity to trust – PT, perceived risks – PR, trust in Instagram – TRIN) and two mediating factors (perceived enjoyment – PE, perceived customer satisfaction – CS, perceived value – PV) which influence eWoM for s-commerce (Rouibah et al., 2021).
Our research question is whether the independent factors (PT, PR, TRIN) have association on eWoM for s-commerce through the three mediation factors (PV, CS and PE).
The study consists of several sections. In the next one, we present the related literature followed by the research model and research methodology. After that, we present the data analysis for the comparison of the two sample countries. In the last section, we conclude by offering a summary of our research contributions, implications for both research and management and research limitations.
Literature review and model development
Importance of social networks for s-commerce
This study investigates antecedents of eWoM for s-commerce based on Instagram in Kuwait and UAE. We chose Kuwait because it ranks first in the GCC for the depth of penetration and speed of the internet, and the penetration of its mobile phone network. A majority of people use social media such as Instagram and Snapchat. The UAE has a similar situation, but a majority of its users are younger (Blogger, 2023; Kemp, 2023).
We followed Rouibah et al. (2021) and defined s-commerce as:
The exchange-related activities that can be influenced by an individual’s online social network (as Facebook and Instagram) or computer-mediated social environments (e.g. WhatsApp, etc.) – where the activities correspond to need recognition, pre-purchase, purchase and post-purchase stages.
eWoM-based s-commerce, according to Dincer and Dincer (2023), plays an important role in creating the intention to purchase and increasing the loyalty of customers. Through social media and computer-mediated social environments, people exchange opinions with each other and build a collective decision to purchase. Moreover, eWoM-based s-commerce enhances interactions that consequently lead to the wisdom of the crowd after discussing, adjusting and enhancing customers’ understanding (Xu et al., 2023).
Two theories can explain why people are influenced by eWoM: the social information processing theory (Salancik and Pfeffer, 1978) and the information richness theory (Draft and Lengel, 1986). According to these theories, the people who have more tools to communicate are socially more powerful in influencing others compared to people who lack such tools. In addition, they justify the use of Instagram as a social tool for s-commerce because it is rich in information and provides plenty of functions for people such as taking pictures, publishing multimedia data and providing different feedback (like, share, etc.) (Rouibah et al., 2021).
Social cognitive theory
Social cognitive theory (SCT) is a cognitive formulation of social learning theory that explains human behavior as a dynamic interaction between personal factors. It supports the influential role of factors in the adaptation and change of humans. Bussey and Bandura (1999) believe that social, psychological, behavioral and technological factors combine to shape humans’ decisions. STC is built around four distinctive human capabilities:
symbolization helps in the understanding of the environment by triggering people’s ability to create and regulate environmental conditions;
learning through observation that expands people’s knowledge and skills;
self-regulation that helps people exercise self-directedness; and
self-reflection in which people evaluate the adequacy of their thinking and actions and “judge one’s agentic efficacy to produce effects” (Bussey and Bandura, 1999).
SCT (Bandura, 1991, 2001) is relevant to our study because it has the power to explain the motivation behind eWoM as a “function of the reciprocal interaction involving behavior, cognition, and social environmental influences” (Lee et al., 2012). In our context, we consider satisfaction as a cognitive factor (Del Bosque and San Martín, 2008; Homburg et al., 2006), trust as a social influence (Cho et al., 2015) and risk as a behavior (Dowling and Staelin, 1994). This theory is the basis of the model developed by Rouibah et al. (2021) that will be used as a comparative research model between Kuwait and UAE.
Electronic word of mouth literature review
Originally, eWoM was defined as a “network phenomenon: People create ties to other people with the exchange of units of discourse (that is, messages) that link to create an information network while the people create a social network” (Dwyer, 2007, p. 64). Furthermore, Hennig-Thurau et al. (2004) extend eWoM to:
Any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions for commercial and non-commercial activity.
EWoM has increasingly attracted attention from academics, researchers and practitioners. Traditionally, studies have shown how WoM has played a critical role in a customer’s decision to purchase and in e-commerce (Richins and Root-Shaffer, 1988). The intention to purchase, in turn, is important and has many applications in the field of business (Onita et al., 2022). Since the beginning of the Internet, researchers have explored the importance of eWoM with the help of information technology. Hennig-Thurau et al. (2004) have explained that:
The advent of the Internet has extended consumers’ options for gathering unbiased product information from other consumers and provides the opportunity for consumers to offer their own consumption-related advice by engaging in eWoM.
Researchers advocate that eWoM is important in many dimensions. One is the spread of rumors (Mirbabaie et al., 2021). Another is the marketing field because 62% of customers trust peer recommendations (Nilashi et al., 2022).
(Rouibah et al., 2021; Rouibah et al., 2015) observed that prior s-commerce studies suffer from the following limitations:
they focused more on customers’ intentions rather than their actual behavior;
they do not follow the sequence of belief-attitude-behavior found in well-established theories such as TAM; and
they do not integrate hedonic factor (perceived enjoyment) and utilitarian factor (perceived value) in their theoretical research models that are important to the customers’ actual decisions to purchase.
Therefore, a better understanding would help companies to direct their efforts toward better control that would lead to incremental improvements in actual purchases that would then hopefully have a positive effect on customer loyalty. Another important shortage in the previous studies is the lack of understanding the impact of enjoyment on eWoM. According to Cheung and Lee (2012) “research on why consumers engage in eWoM in online consumer-opinion platforms remains relatively limited”. Prior enjoyment studies focused on other dimensions such as enjoyment helping other customers (Tong et al., 2007). Mainly, they treated enjoyment as an altruistic motivation (Hennig-Thurau et al., 2004; Kankanhalli et al., 2005; Tong et al., 2007). Except Rouibah et al. (2021), none explored in-depth the impact of enjoyment as a hedonic factor on eWoM. We use eWoM to represent the expected social outcome. Table 1 gives a brief review of prior studies on eWoM, dependent factor, used theories and context.
Brief review of eWoM literature
| Study | Dependent factors | Theory | Context |
|---|---|---|---|
| Rouibah et al. (2021) | Behavioral intention | Value–Satisfaction–Loyalty | s-commerce |
| Ismagilova et al. (2020) | Intention to buy | Meta-analysis | e-commerce |
| Qahri-Saremi and Montazemi (2019) | eWoM adoption | Meta-analysis | Messages |
| Liang et al. (2021) | eWoM adoption | Meta-analysis | Communication |
| Hong et al. (2017) | eWoM adoption | Meta-analysis | e-commerce |
| Alnoor et al. (2024) | eWoM and intention to use | Social exchange theory | Intention to use of s-commerce |
| Li et al. (2024) | Negative-WoM and Positive-WoM | Meta-analysis | Consumer decision-making process |
Research model
Customer satisfaction.
CS is critical for the existence of organizations and their continued development. CS is not only measured by the product’s value to customers or their feelings about the purchase but also to the atmosphere surrounding the purchasing process (Biesok and Wyród-Wróbel, 2011). Following Rouibah et al. (2015), we defined customer satisfaction as the degree to which a customer is satisfied with his purchasing from s-commerce.
CS is related to specific experiences compared to previous similar experiences. For a current transaction, CS increases if that the felt experience is better than the previous one. Woisetschläger et al. (2008) have identified that satisfaction with the social media community positively affects participation in it. According to the SCT, CS is the result of personal cognitive processes that are developed from the effects of external factors. Following Rouibah et al. (2021), we propose that CS has a direct effect on the use of eWoM in s-commerce. Thus, we hypothesize the following:
CS positively affects the eWoM for s-commerce.
Perceived enjoyment.
PE refers to the value a customer receives in terms of subjective experiences of fun and playfulness (Rouibah et al., 2021). According to many scholars (Hwang and Kim, 2007), the influence of affect (enjoyment, hate, etc.) is neglected in the s-commerce research that tests an individual’s decision-making and their reactions to using technologies. This is why after an in-depth literature review, Sun and Zhang (2015) called for paying more attention to the effects of affect factors. According to Venkatesh et al. (2012), hedonic is “fun or pleasure from using technology” and “the search for happiness, enjoyment, fantasy, awakening, and sensuality” (Akram et al., 2021).
Given that Instagram is a hedonic tool equipped with many features that increase the user’s enjoyment – such as their profile, pictures, videos, sharing and interacting with others to ask for better recommendations and opinions (Rouibah et al., 2021) – we expect that CS will increase. This increased satisfaction will be transformed into more eWoM for s-commerce purposes-based Instagram. Thus, we replicate Rouibah et al. (2021) two hypotheses:
PE positively affects CS during s-commerce.
PE positively affects eWoM for s-commerce.
Perceived value.
PV is defined as a consumer’s overall evaluation regarding the utility of products and services based on his or her perceptions of what is received and given (Rouibah et al., 2015). In a systematic review, Cui et al. (2018) consider PV to be among the most important perceptions that influence satisfaction, engagement, behavioral intention and the loyalty of customers. The customer’s PV comes from their estimation of the importance of the product or service. The literature has shown that PV is a crucial motivational factor that affects CS and eWoM (Hajli et al., 2015; Rouibah et al., 2021; Zeithaml, 1988). Following, Rouibah et al. (2021), we hypothesize the following:
PV positively affects the CS during s-commerce.
PV positively affects eWoM for s-commerce.
Perceived risk.
PR relates to the expectations of uncertainty and perceived consequences from the purchase of goods and services, fraud and product quality (Forsythe et al., 2006; Rouibah et al., 2021; Rouibah et al., 2016). S-commerce differs from e-commerce in that the vendors on s-commerce platform are small companies with no official or legal business license. In such cases, PR is high, who may feel different risks (such as quality of product, lack of good feel of product, after-sale service, etc.). These feelings are logical, as these small companies lack the product expertise and return policies that official and large vendors have.
Given the high uncertainty avoidance characteristic of the Arab culture, Arab consumers who are happy with more face-to-face types of interactions will negatively perceive the impact of PR on PV and CS in s-commerce over Instagram. Accordingly, we follow Rouibah et al. (2021) and replicate the following two hypotheses:
PR negatively affects the PV during s-commerce.
PR negatively affects the CS during s-commerce.
Trust factors (propensity to trust and trust in Instagram).
Trust is necessary to cope with PR inherent to s-commerce platforms and conduct business in an online environment. We included in our research model two types of trust: PT and TRIN. PT, also known as trust toward other people, refers to the extent of character that a person demonstrates in a consistent tendency to depend on others across a broad spectrum of situations and individuals (McKnight et al., 1998; Rouibah et al., 2021). While TRIN refers to the degree of people’s trust in social media tools (here trust in Instagram) used to interact with other parties.
Some scholars (Bugshan and Attar, 2020) believe that the effects of negative factors (e.g. PR) surpass the effects of positive factors (e.g. trust factors). Unlike other scholars and studies who claim PT directly influences intention, we follow Rouibah et al. (2021) and posit that people with high PT will feel less risk-threatening and thus increase customer PV and CS. Also, normal people who trust Instagram technology will have high PV and CS too. Following Rouibah et al. (2021), we replicate the following four hypotheses:
PT positively affects PE during s-commerce.
PT positively affects CS during s-commerce.
TRIN positively affects PV during s-commerce
TRIN positively affects the CS during s-commerce.
Based on the previous sections, we reused the research model of Rouibah et al. (2021) (see Figure 2).
Research methodology
In answering the research question, this study used a methodology based on a quantitative method that consisted of a survey questionnaire.
Sample and procedure of data collection
We reused well pretested, reliable and accurate items from Rouibah et al. (2021). The research instrument was available for both Arabic and English since these two languages are dominant in both Kuwait and the UAE. In addition, the research instrument was pretested in a pilot group of 10 Kuwaiti Instagram users and 5 Emirates. The research instrument was also reviewed by two information system professors to double-check it and to identify any deviations in meaning between the two languages. The population consists of all adults in Kuwait and the UAE. Data was collected through the online survey (Quatrics.com) based on random sampling, the snowball technique. The survey link was promoted by invitation e-mails to thousands of people who were asked to promote it in their networks. Participants contacted other friends, family members who were encouraged to answer the questionnaire and to distribute the uniform resource locator to others.
The original sample size was 1,716 (total from both countries). Before going further in the analysis, we went through a data pretest to double-check whether they reflected participants’ seriousness in engagement and their full attention to the questionnaire. Thus, we first removed responses with missing values (respondents did not give their opinion at all and skipped many questions). Also, we tested the data to measure if the standard deviation was more than 0.4. Otherwise, the responses were removed. In total, we deleted 394 participations from the collected data leaving a total of 1,322 for further analysis.
The sample consisted of 1,132 Kuwaiti participants and 190 UAE participants. The reason for the large size difference is because of different legal systems of the two countries.
Construct measurements
Items were reused from past validated scales to ensure content validity. Table 2 shows the literature sources for the factors/items in the study.
Study items
| Latent item | Abr. | Items | Source |
|---|---|---|---|
| PT | PT1 | A high degree of trust exists in my family | Cheung and Thadani (2012); Rouibah et al. (2021) |
| PT2 | My friends are generally trustworthy | Cheung and Thadani (2012); Rouibah et al. (2021) | |
| PT3 | People of my community trust each other | Cheung and Thadani (2012); Rouibah et al. (2021) | |
| PT4 | Overall, I am living in a high trust society | Cheung and Thadani (2012); Rouibah et al. (2021) | |
| PR | PR1 | I am worried that the price on Instagram may be higher than in the mall | Lim (2003); Rouibah et al. (2021) |
| PR2 | I am worried that purchasing from Instagram may take me too much time, including choosing products from a wide selection and a delay in shipment | Lim (2003); Rouibah et al. (2021) | |
| PR3 | I am worried that products sold through Instagram are fake, copied, or imitated | Lim (2003); Rouibah et al. (2021) | |
| PR4 | I am worried that the products I buy from Instagram do not meet my expectations due to being unable to touch them or try them in person | Lim (2003); Rouibah et al. (2021) | |
| PR5 | I am worried about after-sale services based on purchasing on Instagram | Lim (2003); Rouibah et al. (2021) | |
| PR6 | I am worried about the value of the product I purchase through Instagram does not meet its price | Lim (2003); Rouibah et al. (2021) | |
| TRIN | TRIN1 | Instagram has safeguards to make me feel comfortable using it for e-shopping | Rouibah et al. (2021); Schaupp et al. (2010) |
| TRIN2 | I feel assured that legal and technological structures adequately protect me from e-shopping problems | Rouibah et al. (2021); Schaupp et al. (2010) | |
| TRIN3 | In general, the internet is a robust and safe environment in which to e-shop | Rouibah et al. (2021); Schaupp et al. (2010) | |
| PE | PE1 | I have fun interacting with Instagram | Agarwal and Karahanna (2000); Rouibah et al. (2021) |
| PE2 | Using Instagram provides me with a lot of enjoyment | Agarwal and Karahanna (2000); Rouibah et al. (2021) | |
| PE3 | I enjoy using Instagram | Agarwal and Karahanna (2000); Rouibah et al. (2021) | |
| CS | CS1 | I am willing to use eWoM to recommend things to my relatives and friends | Lee (2014); Rouibah et al. (2021) |
| CS2 | I am willing to try new products introduced by eWoM | Lee (2014); Rouibah et al. (2021) | |
| CS3 | Based on my prior experience, I am satisfied with my online shopping through Instagram | Murray and Howat (2002); Rouibah et al. (2021) | |
| CS4 | Based on my prior experience, I think purchasing through Instagram is a wise choice | Murray and Howat (2002); Rouibah et al. (2021) | |
| CS5 | Based on my prior experience, I am planning to repurchase through Instagram in the future | Murray and Howat (2002); Rouibah et al. (2021) | |
| CS6 | Based on my past purchasing experiences, I will reuse Instagram to repurchase in the future | Murray and Howat (2002); Rouibah et al. (2021) | |
| PV | PV1 | Compared to the fee I need to pay, the use of Instagram offers value for money | Forsythe et al. (2006); Rouibah et al. (2021) |
| PV2 | Compared to the effort I need to put in, the use of Instagram is beneficial to me | Forsythe et al. (2006); Rouibah et al. (2021) | |
| PV3 | Compared to the time I need to spend, the use of Instagram is worthwhile to me | Forsythe et al. (2006); Rouibah et al. (2021) | |
| eWoM | eWoM1 | When I am not sure about the quality of the product, I rely on eWoM recommendations to obtain information | Lee (2014); Rouibah et al. (2021) |
| eWoM2 | Whenever I required product information, I first seek eWoM recommendations on Instagram accounts | Lee (2014); Rouibah et al. (2021) | |
| eWoM3 | I consider online customer reviews when I make a purchasing decision through Instagram | Lee (2014); Rouibah et al. (2021) |
Data analysis
Demographic data of the two samples (Kuwait and United Arab Emirates)
Table 3 shows that the Kuwaiti sample comprises 194 males (17.2%) and 933 females (82.8%). In this sample, there were 727 students (64.7%), 305 people looking for a new job (27.2%) and 39 retirees (3.5%). The work status for the UAE sample was 118 students (62.8%), 41 looking for a new job (21.8%) and 24 retirees (12.8%). The distribution of the academic levels was 334 attended high school or lower (29.5%), 139 had high school diplomas, i.e. two years degree (12.3%), 609 had bachelor’s degrees (53.8%), 31 had master’s degrees (2.7%) and 18 had PhD degrees (1.6%). The distribution of academic levels for the UAE sample was 56 attended high school or lower (29.5%), 10 had high school diplomas (5.3%), 103 had bachelor’s degrees (56.3%), 14 had master’s degrees (7.4%) and 3 had PhD degrees (1.6%).
Demographics of sample study
| Frequency | % | ||||
|---|---|---|---|---|---|
| Item | Categories | Kuwait | UAE | Kuwait | UAE |
| Gender | Male | 194 | 33 | 17.2 | 17.4 |
| Female | 933 | 157 | 82.8 | 82.6 | |
| Work status | Student | 727 | 118 | 64.7 | 62.8 |
| Job seeker | 305 | 41 | 27.2 | 21.8 | |
| Retiree | 39 | 24 | 3.5 | 12.8 | |
| Not specified | 52 | 5 | 4.6 | 2.7 | |
| Academic level | High school or less | 334 | 56 | 29.5 | 29.5 |
| Diploma | 139 | 10 | 12.3 | 5.3 | |
| Bachelor | 609 | 107 | 53.8 | 56.3 | |
| MS | 31 | 14 | 2.7 | 7.4 | |
| PhD | 18 | 3 | 1.6 | 1.6 | |
Assessing construct validity and reliability
We first assessed the convergent validity (measured by Cronbach’s α) and discriminant validity of the structural model, which were achieved through factor analysis in conjunction with analysis of moment structure (AMOS). As shown in Table 4, factor analysis resulted in seven factors that accounted for 73% of the total variance explained. We note that two items (PT1 and TRIN2) were deleted, as they do not meet the requirements. Also, each of the factors met the base criteria for retention as follows (Hair, 2011):
Factor analysis test
| Factors/items | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| PT2 | 0.720 | ||||||
| PT3 | 0.866 | ||||||
| PT4 | 0.801 | ||||||
| PR3 | 0.841 | ||||||
| PR4 | 0.844 | ||||||
| PR6 | 0.828 | ||||||
| TRIN1 | 0.737 | ||||||
| TRIN2 | 0.785 | ||||||
| TRIN3 | 0.727 | ||||||
| PE1 | 0.838 | ||||||
| PE2 | 0.860 | ||||||
| PE3 | 0.845 | ||||||
| PV1 | 0.753 | ||||||
| PV2 | 0.840 | ||||||
| PV3 | 0.848 | ||||||
| eWOM1 | 0.839 | ||||||
| eWOM2 | 0.856 | ||||||
| eWOM3 | 0.678 | ||||||
| CS1 | 0.758 | ||||||
| CS2 | 0.772 | ||||||
| CS3 | 0.799 | ||||||
| CS4 | 0.803 | ||||||
| CS5 | 0.670 | ||||||
| CS6 | 0.702 |
items defining the various factors all had communalities greater than 0.50;
extracted factors accounted for greater than 50% of the variance in their sets of items;
all factors have eigenvalues greater than 1.0; (iv) all item loadings were greater than 0.50; and
all factors are clearly interpretable. Tables 4 and 5 show that these conditions are met.
Explained variance and Cronbach’s reliability of latent factors
| Factors | Variance explained (%) | Cronbach reliability (%) |
|---|---|---|
| PT | 66 | 74 |
| PR | 73 | 81 |
| TRIN | 69 | 77 |
| PE | 81 | 88 |
| PV | 76 | 84 |
| eWoM | 72 | 81 |
| CS | 69 | 91 |
Table 5 shows the results of explained variance and reliabilities of study latent factors. The lowest value of coefficient Cronbach’s α is 74%, which is more than the 0.73% required in behavioral studies (Hair et al., 1998). These high values witness that items reflect the factors they intend to measure. Therefore, data collected using the research instrument is reliable and valid.
Table 6 presents the model’s fitness measures. The results show that the research model is a good fit.
Model fit and model validation of the Kuwaiti sample
This study tested the research model and hypotheses using the AMOS software. The fit indices of the measurement model for the Kuwaiti sample show converged and reasonably fit indices (see Table 6).
Table 7 shows the tested hypotheses, independent factors, dependent factors, path coefficients (betas, βs) and whether the hypotheses are significant or not. Of the 11 hypotheses, two associations were insignificant. These are H3b and H5b. The remaining nine hypotheses are statistically significant (0.5 or more). Table 8 shows the β values for items per each latent factor (factor in the research model). The associations between the factors in the Kuwaiti sample are shown in Table 7.
Significance and effect power of hypotheses for Kuwait
| Hypothesis | Independent factor | Dependent factor | β-value | p-value | Decision |
|---|---|---|---|---|---|
| H1 | CS | eWoM | 0.42 | 0.02 | Yes |
| H2a | PE | CS | 0.248 | 0.03 | Yes |
| H2b | PE | eWoM | 0.090 | 0.05 | Yes (minor impact) |
| H3a | PV | CS | 0.261 | 0.03 | Yes |
| H3b | PV | eWoM | – | No | |
| H4a | PR | PV | 0.090 | 0.05 | Yes (minor impact) |
| H4b | PR | CS | −0.071 | 0.05 | Yes (minor impact) |
| H5a | PT | PE | 0.304 | 0.03 | Yes |
| H5b | PT | CS | – | No | |
| H6a | TRIN | PV | 0.437 | 0.02 | Yes |
| H6b | TRIN | CS | 0.494 | 0.01 | Yes |
β values of research model’s associations for the Kuwait sample
| Latent factor | Items | β-value |
|---|---|---|
| PT | PT2 | 0.58 |
| PT3 | 0.79 | |
| PT4 | 0.69 | |
| PR | PR3 | 0.61 |
| PR4 | 0.80 | |
| PR6 | 0.48 | |
| TRIN | TRIN1 | 0.66 |
| TRIN2 | 0.71 | |
| TRIN3 | 0.82 | |
| PE | PE1 | 0.82 |
| PE2 | 0.89 | |
| PE3 | 0.85 | |
| CS | CS1 | 0.67 |
| CS2 | 0.68 | |
| CS3 | 0.80 | |
| CS4 | 0.78 | |
| CS5 | 0.85 | |
| CS6 | 0.85 | |
| PV | PV1 | 0.71 |
| PV2 | 0.83 | |
| PV3 | 0.87 | |
| eWoM | eWoM1 | 0.83 |
| eWoM2 | 0.89 | |
| eWoM3 | 0.55 |
Figure 3 shows the effects of the associations for Kuwaiti users. It reveals that PT has an indirect effect on eWoM in s-commerce through the mediation of PT. PR has an indirect and negative effect on eWoM through the mediation of CS and PV. TRIN has an indirect and positive effect on eWoM in s-commerce via the mediation of CS and PV.
Model fit and model validation of the Emirati sample
The fit indices of the measurement model for the Emirati sample show converged and reasonably fit indices (see Table 9).
Model’s fitness measures for UAE
| Fitness measurement | UAE | Acceptance threshold |
|---|---|---|
| Chi-square (CMIN) | 414.472 | |
| Normed Chi-Square (PCMIN/DF) | 1.919 | 4.0 or less |
| CFI | 0.917 | 0.9 or more |
| TLI | 0.894 | 0.9 or more |
| IFI | 0.919 | 0.9 or more |
| AIC | 580.472 | Smaller, the better |
| BCC | 604.618 | Smaller, the better |
Table 10 shows the tested hypotheses, independent factor, dependent factor, path coefficients (betas, βs) and whether the hypotheses are significant or not for the Emirati sample. Of the 11 hypotheses, five associations were insignificant. These are H2b, H3b, H4a, H4b and H5b. The remaining six hypotheses are statistically significant (0.5 or more). Table 11 shows the β values for items per each latent factor (factor in the research model).
Significance and effect power of hypotheses for UAE
| Hypothesis | Independent factor | Dependent factor | Impact | p-value | Significance |
|---|---|---|---|---|---|
| H1 | CS | eWoM | 0.54 | 0.02 | Yes |
| H2a | PE | CS | 0.292 | 0.03 | Yes |
| H2b | PE | eWoM | – | No | |
| H3a | PV | CS | 0.287 | 0.03 | Yes |
| H3b | PV | eWoM | – | No | |
| H4a | PR | PV | – | No | |
| H4b | PR | CS | – | No | |
| H5a | PT | PE | 0.208 | 0.05 | Yes |
| H5b | PT | CS | – | No | |
| H6a | TRIN | PV | 0.606 | 0.01 | Yes |
| H6b | TRIN | CS | 0.494 | 0.02 | Yes |
β values for research model’s associations for the UAE sample
| Latent factor | Item | β-value |
|---|---|---|
| PT | PT3 | 0.904 |
| PT4 | 0.800 | |
| PR | PR3 | 0.791 |
| PR4 | 0.842 | |
| PR5 | 0.768 | |
| PR6 | 0.779 | |
| TRIN | TRIN1 | 0.727 |
| TRIN2 | 0.672 | |
| TRIN3 | 0.769 | |
| PE | PE1 | 0.867 |
| PE2 | 0.794 | |
| PE3 | 0.851 | |
| CS | CS1 | 0.733 |
| CS2 | 0.852 | |
| CS3 | 0.847 | |
| CS4 | 0.869 | |
| CS5 | 0.851 | |
| PV | PV1 | 0.769 |
| PV2 | 0.829 | |
| PV3 | 0.879 | |
| eWoM | eWoM1 | 0.870 |
| eWoM2 | 0.877 | |
| eWoM3 | 0.777 |
Figure 4 shows the effects of the associations for UAE Instagram users. It reveals that PT has an indirect effect on eWoM in s-commerce through one path (PT → PE → CS → eWoM). PR has no effect (direct or indirect) on eWoM. TRIN has indirect and positive effect on eWoM during s-commerce through two paths (TRIN → CS → eWoM and TRIN → PV → CS → eWoM).
Results discussion
The results show some similarities, and some differences occur between the behavior of the sampled participants in the two countries (Kuwait and UAE, see Table 12). We can infer four main observations on the associations between factors among the two countries.
Comparative table between the significance of the two samples (Kuwait vs UAE)
| Hypothesis | Independent factor | Dependent factor | β-value | Significance | Comparison (difference in β-value) |
|---|---|---|---|---|---|
| H1a [Kuwait] | CS | eWoM | 0.42 | Yes | 0.12 [UAE> Kuwait] |
| H1b [UAE] | 0.54 | Yes | |||
| H2a [Kuwait] | PE | CS | 0.248 | Yes | 0.044 [UAE> Kuwait] |
| H2a [UAE] | 0.292 | Yes | |||
| H2b [Kuwait] | PE | eWoM | 0.09 | Yes (minor impact) | 0.09 [Kuwait> UAE] |
| H2b [UAE] | – | No | |||
| H3a [Kuwait] | PV | CS | 0.261 | Yes | 0.026 [UAE> Kuwait] |
| H3a [UAE] | 0.287 | Yes | |||
| H3b [Kuwait] | PV | eWoM | – | No | Same [ns] |
| H3b [UAE] | – | No | |||
| H4a [Kuwait] | PR | PV | 0.09 | Yes (minor impact) | 0.09 [Kuwait> UAE] |
| H4a [UAE] | – | No | |||
| H4b [Kuwait] | PR | CS | −0.071 | Yes (minor impact) | −0.071 [Kuwait> UAE] |
| H4b [UAE] | – | No | |||
| H5a [Kuwait] | PT | PE | 0.304 | Yes | 0.096 [Kuwait> UAE] |
| H5a [UAE] | 0.208 | Yes | |||
| H5b [Kuwait] | PT | CS | – | No | Same [ns] |
| H5b [UAE] | – | No | |||
| H6a [Kuwait] | TRIN | PV | 0.437 | Yes | 0.169 [UAE> Kuwait] |
| H6a [UAE] | 0.606 | Yes | |||
| H6b [Kuwait] | TRIN | CS | 0.494 | Yes | Same |
| H6b [UAE] | 0.494 | Yes |
First, four associations in the research model are stronger in the UAE sample than those in the Kuwaiti sample. These are as follows: CS → eWoM, PE → CS, PV → CS, TRIN → PV. In addition, four associations in the research model are stronger in the Kuwaiti sample than those in the UAE sample. These are as follows: PE → eWoM, PR → PV, PR → CS; PT → PE. In addition, two associations were insignificant for both samples. These are as follows: PV → eWoM, and PT → CS, and in one association (TRIN → CS) the two effects are similar.
Second, the two samples exhibit dissimilarities in the significances of the weakest effects weight (β-value) in two associations. These are represented by H2a and H5a. They are the weakest among the other hypotheses. For the UAE sample, H5a (PT → PE) is the weakest while the weakest for the Kuwaiti sample is H2a (PE → CS). However, they are still roughly in the same range. For example, H2a is almost the same between Kuwait (β = 0.25) and the UAE (β = 0.292). This finding is similar to Rouibah et al. (2021), who found the association between PE and CS was also 0.25 (weakest association).
Third, the two samples exhibit similarities in the significance of the importance effects weight (β-value) in two associations. These are represented by the association between TRIN and CS (the strongest effect in the research model) followed by the association between CS and eWoM. These close similarities are perhaps related to the cultural effect since the two countries are categorized as collectivist (compared to individualistic) (Hofstede, 2009). In other words, the cultural and religious backgrounds of both societies are similar and many Kuwaitis and Emiratis are relatives (Hopkyns and Zoghbor, 2022). A total of 5 hypotheses (H2b, H3b, H4a, H4b and H5b) out of the 11 are insignificant (or of lower strength), except for H2b (PE → eWoM), for both countries. Our results confirm the findings of other studies. These results lead to the conclusion that the two countries are very similar in terms of their eWoM adoption in s-commerce purposes based on Instagram. These results extend the studies of some scholars who believe that the similarities between GCC countries are greater than their differences (Hopkyns and Zoghbor, 2022; Karolak and Allam, 2020; Younis et al., 2022).
Fourth, unlike previous studies that found PE to be an important mediator between independent factors and intention to use new technologies in Arab countries (Rouibah, 2008; Rouibah et al., 2021; Rouibah et al., 2016) our study is the first to found that it does not play any role in s-commerce. Accordingly, it is possible that a high level of satisfaction reduced the effect of enjoyment perceived by customers, which requires more research investigation.
Conclusion, practical and theoretical implications
This study investigated the impact of three independent factors (PT, TRIN and PR) on eWoM for s-commerce purpose through the mediation of three factors (PE, PV and CS) based on the robust model of Rouibah et al. (2021).
Research contributions
Our study offers four main theoretical and practical implications.
The first theoretical implication is that it is among the few studies in the s-commerce field to compare the effect of antecedents and mediating factors on eWoM use in two Arab countries. In line with Rouibah et al. (2021), to the best of our knowledge, our study is among the few ones that focused on Instagram platform for s-commerce as opposed to other platforms (Rouibah et al., 2021; Silaban et al., 2023). Our study results show many factors that play a critical role in eWoM for s-commerce purposes.
The second implication is that the scarcity of literature on the subject of eWoM for s-commerce is an important motivation for scholars to correct that deficiency (Silaban et al., 2023). In our study, we explore the use of eWoM in the context of the GCC that has received less attention (Bugshan and Attar, 2020).
The third implication is that many studies investigated eWoM in e-commerce such as tourism and marketing but have concentrated less on s-commerce. Furthermore, most of these studies have focused on the intention to use (Dincer and Dincer, 2023; X. Hu et al., 2022; Zhou et al., 2023) as opposed to the actual use (Rouibah et al., 2021).
The last theoretical implication, and in line with Rouibah et al. (2021), our study contributed to enriching the literature on comparative studies between Arab countries related to eWoM in s-commerce and contributed to reduce the scarcity of research that uses models and theories to understand s-commerce adoption in two Arab countries.
From a managerial perspective, this study shows which factors (drivers and obstacles) drive/inhibit the use of eWoM for s-commerce purposes. As such companies may use these results to increase the use of s-commerce, which in turn may create more jobs, and therefore this study has a positive impact on the two countries (Kuwait and UAE). In addition, satisfaction plays the most important effect on eWoM for s-commerce in both countries and probably can be applied to other GCC countries such as Saudi Arabia that share similar cultural characteristics. It is even more important to know that Instagram is a cross-border application that the majority of Arab and multinational companies can use to promote their products. Many scholars have called for more efforts to understand customer behavior and how online companies disseminate knowledge about themselves (Rouibah et al., 2021).
Limitations and future studies
This study suffers from the following three limitations:
First, we did not focus on the difference between the effects of textual and graphical information on customers’ decisions and trust in buying merchandise. Zinko et al. (2020) find that text may affect the decision to buy on s-commerce platforms. Furthermore, their results show that when a company provides an improper amount of text, images play a critical role in moderating the negative effect of text on customers’ trust and intention to purchase. We, therefore, encourage future studies to investigate the role of eWoM in s-commerce based either on text and graphical information.
Second, our study focuses on Kuwait and the UAE. We ignored other GCC countries even though the mobile and internet penetration rates and s-commerce are booming. Accordingly, a future study could be initiated to study the potential differences that may occur among the six GCC countries. Although many studies claim that the similarities are obvious in Arab countries as we claimed previously, some differences also occur among them that still need to be explored and test the stability of the research model across the six countries of GCC.
Third, we compared the effect of size factors on eWoM in s-commerce across two Arab countries. However, we omitted to include cultural factors. Therefore, we encourage future studies to undertake comparative studies among Arab countries (including GCC and Northern African Arab countries) using cultural factors from a well-known model (Hofstede) including power distance, individualism/collectivism, masculinity vs. femininity, uncertainty avoidance, long-term vs short-term orientation and indulgence vs restraint.
