1. Introduction
Over the past few years, e-commerce has become an essential component of universal retail. The introduction of the Internet has engendered dramatic changes in the retail landscape. In addition, the digitization of modern life has enabled customers around the globe to benefit from online transactions [
1]. The number of online shoppers is growing yearly as the availability and usage of the Internet grow steadily. In 2020, around two billion customers purchased products/services online, with worldwide e-commerce sales surpassing USD 4.2 trillion [
2]. Across the planet, B2C (business-to-consumer) e-commerce is increasing and progressively becoming an important component in the retail landscape [
3]. It is attracting ever more customers due to its price advantages and convenience [
4,
5]. Furthermore, B2C e-commerce has been recognized as a highly significant alternative for businesses as e-retailers benefit from the lower costs of operations [
6]. E-commerce has been already adopted by a considerable number of businesses as an essential trading tool in their daily processes. However, many small and medium enterprises (SMEs) face pressures in adopting e-commerce due to the intense rivalry from large firms [
7]. Hence, numerous countries have now recognized the prominence of e-commerce, particularly B2C e-commerce, in their economies and take account of it in their economic development strategies [
8].
Nevertheless, the global diffusion of e-commerce remains extremely uneven across nations [
9]. Such imbalanced development and the readiness of e-commerce are attributed to various factors at a national level, such as market factors, GDP, culture, and educational level [
10,
11]. Ayob [
12] points out that the diffusion of e-commerce is not only limited by the quality of formal institutions (e.g., laws, infrastructure), but also by cultural dimensions such as uncertainty avoidance (UA) and risk tolerance. Research has indicated that nations with high UA are content with existing conditions and are resistant to change, consequently, they act conservatively [
13]. People in such societies are unlikely to accept risks when trying new technologies and thus are slower in adopting them [
14]. Compared to physical stores, e-commerce as a new model of trade is recognized as more uncertain, and only customers with a solid trust value are keen to adopt it. In general, e-commerce is less developed in the Arab world than in other regions [
15]. According to a report by [
16], e-commerce contributed to 16% and 14% of all retail sales in the UK and the USA, respectively, compared to less than 2% in the Arab world. Although there has been a rapid growth in usage of the Internet, e-commerce is growing at a slow rate in the Arab region [
17]. Jordan is an Arabic and developing country located in the Middle East and is considered a high-UA country [
18]. Internet penetration in Jordan stood at 66.8% in 2021 (6.84 million Internet users), with social media penetration reaching 61.5% of the population (6.3 million) [
19]. E-commerce in Jordan is considered more advanced than in its counterparts in the region. In 2021, 8% of the population in Jordan were reported to have made online purchases and/or paid bills online [
19]. According to one report, e-commerce market revenue is likely to report a yearly growth rate of 17.24% in 2022, resulting in an expected market volume of USD 4646 m by 2025 [
20]. Despite the development of e-commerce in Jordan, a lack of legislation supporting the e-commerce sphere and protecting consumers has been recognized as the main obstacle impeding e-commerce growth [
21].
The primary objective of the present study is to explore how e-retailer-based signals mitigate transaction uncertainty in e-commerce, increasing customers’ trust and subsequently their purchase intentions in emerging markets. This study examines e-retailer-based signals from the perspective of e-commerce customers in the emerging market of a developing country (Jordan). A considerable number of micro and small businesses in emerging and developing markets have benefited from e-commerce by reaching a larger number of customers and conducting transactions online. This study offers insights into how, in developing e-commerce environments, customers reduce the uncertainty related to online transactions by relying on specific signals to build trust, which in turn develops purchase intention. Given that online customers cannot physically assess products before purchasing them online, the current study uses signaling theory (ST) and relational signaling theory (RST) to increase knowledge of the retailer-based signals e-commerce customers use to alleviate uncertainty by increasing their trust, thereby positively affecting their online purchase intentions. In particular, ST and RST are used to determine whether e-retailer-based signals such as return policy leniency (RPL), cash on delivery (COD), and social commerce constructs (SCCs) can effectively decrease transaction uncertainty in online transactions. Hence, RPL, COD, and SCCs can be employed as information signals to mitigate uncertainty in e-commerce transactions.
The paper is organized as follows: given that the introduction is the
Section 1, the
Section 2 provides a review of literature related to the current research, presents the theoretical foundation, and discusses the constructs of the research model and the hypotheses formulated. The research methodology is introduced in
Section 3. The statistical analysis is presented in
Section 4 and
Section 5 discusses the study findings.
Section 6 then considers the implications of the research The
Section 7 summarizes the research objective, main findings, and limitations.
3. Methodology
A survey method was employed for this study. The items used to measure the research model’s constructs were adopted from previous research (see
Appendix A). All measurement items were assessed on a 5-point Likert scale. Prior to administration, the think-aloud technique was applied to discuss the questionnaire with three academicians and four experienced e-commerce users for content and face validity. Small changes to the phrasing and layout of the measurement items were made based on these conversations, and the final online questionnaire was designed using Google Forms. The first part was dedicated to collecting demographic information about the participants (see
Table 1), whereas the second section was composed of 26 items that measure the constructs of the research model (see
Appendix A).
Data were collected in Jordan from 15 December 2021 to 18 February 2022. Due to the absence of a sample frame for e-commerce users, the survey link was distributed to potential participants via various WhatsApp groups and social media pages (after obtaining permission from the administrators). This ensured that all recruited participants were conversant with social media and increased the likelihood of finding users who purchased products through e-commerce. Most WhatsApp groups and social media members tend to belong to numerous other groups and pages. Therefore, a snowball sampling method was adopted. The administrators of the WhatsApp groups and social media pages were instructed to distribute the survey link to other pages and groups and to encourage their members to do the same. Accordingly, the online survey resulted in 573 returned questionnaires, 13 of which were discarded due to a high level of incompleteness. Consequently, a total of 560 questionnaires were valid and subjected to analysis. The respondents’ demographic characteristics are displayed in
Table 1.
In survey research, Common Method Variance (CMV) is a potential issue [
140]. This issue arises for various reasons, including item ambiguity, participants attempting to remain consistent in their responses, scale length, common scale, anchors/formats, and gathering data about independent and dependent variables from the same participant and measuring them in the same location. According to Sharma [
141], the validity of the relationships among variables is threatened by CMV, which inflates observed correlations and provides erroneous support for the hypotheses. Furthermore, Kock [
142] states that CMV deflates the size of correlations among variables, hence making the outcomes insignificant. As suggested by Podsakoff [
140], CMV was procedurally controlled during the questionnaire design by utilizing clear and simple language, fragmenting the measurement items for independent and dependent variables, eliminating “double-barreled” questions, and separating. These processes were effective in controlling CMV, as the result of a “Harman Single Factor” analysis indicated that the total variance extracted by one factor was 49.01% (<50%), demonstrating no bias in the dataset [
140].
4. Data Analysis
The convergent validity of the constructs was assessed by examining the internal consistency of the indicators using the Dijkstra–Henseler’s rho (rho_A) and the “average variance extracted” (AVE). As displayed in
Table 2, all rho_A, composite reliability (CR), Cronbach’s alpha (α) (>0.7), and AVE (>0.5) values satisfied the recommended cut-off values [
143].
Hair et al. [
143] state that to confirm the presence of discriminant validity, each construct’s
value should be higher than the correlations involving the constructs. The diagonal numbers in
Table 3 represent the
. These are larger than the off-diagonal numbers (correlation values) in the corresponding columns and rows, indicating discriminant validity. Additionally, the factor loadings and cross-loadings for each indicator were calculated and are displayed in
Table 4. The indicators (items) of each construct yielded a factor loading higher than 0.707, except for RPL5 and SCCs, which had a factor loading less than 0.707 and were consequently deleted. Furthermore, each indicator loads higher on its intended theoretical construct than on any other construct, indicating the presence of adequate discriminant and convergent validities [
143]. Finally, the “heterotrait–monotrait ratio of correlations” (HTMT) was employed to examine discriminant validity.
Table 5 shows that the values of the HTMT were all <0.85 [
143], reconfirming the existence of discriminant validity.
SEM-PLS modeling with Smart PLS was utilized to examine the suggested hypotheses. The Kolmogorov–Smirnov test indicates that the goodness of fitness for all the measurement items was <0.05, demonstrating that the data in this study were non-normally distributed [
144]. SmartPLS is widely used for SEM and can effectively manage non-normal data and small samples [
145]. The results of the hypotheses testing are presented in
Table 4. All hypotheses were supported. The RPL (β = 0.371,
p < 0.001), COD (β = 0.411,
p < 0.001), and SCCs (β = 0.15,
p < 0.001) exhibited a significant positive influence on TR. Furthermore, TR (β = −0.677,
p < 0.001) exhibited a significant positive influence on purchase intention. The total variances explained for PI and TR were 45.9% (R
2 = 0.459) and 64% (R
2 = 0.64), respectively (see
Table 6), indicating that the research model has moderate explanatory power [
146]. The indices of model fitness were all within the recommended range [
143], including the NFI “Normed Fit Index” = 0.91 (>0.9), SRMR “Standardized Root Mean Square Residual” = 0.063 (<0.9), and RMS Theta = 0.11 (<0.12).
As demonstrated in
Table 7, all the VIF “variance inflation factor” values for the independent variables (COD, RPL, SCCs) were <3, indicating that there were no collinearity issues [
143]. Although the effect sizes (
f2) of COD (0.246), RPL (0.191), and SCCs (0.048) on TR were all medium (see
Table 7), TR exerted a large effect size of 0.847 on PI [
147].
5. Discussion
This study examined how RPL, COD, and SCCs affect customers’ PI through customer trust. The analysis was based on a structured survey dataset from e-commerce users in Jordan. SEM-PLS was used to validate the research model. The results demonstrate that RPL has a significant positive impact on TR, indicating that H1 is supported. This is consistent with the findings of previous research [
50]. This suggests that RPL acts as an effective mechanism to enhance customer trust. The more lenient the return policy, the more customer trust will be increased. Customers evaluate return policies according to their degree of leniency before committing to a purchase. Lenient return policies are considered by customers as a signal that e-retailers are eager to share with customers the transaction-related risks. This builds goodwill and trust, which leads to customers’ purchase intentions. Enabling customers to easily return a wide range of products within a reasonable timeframe and without imposing fees means they are more likely to develop trust.
H2 was also supported. This indicates that COD exerts the strongest significant positive impact on TR. This suggests offering a COD payment option increases customer trust. Although prior research found COD positively affects purchase intentions [
67], the effect of COD on customer trust has been less widely explored. Prior payment is likely to be a problem for many customers as they are uncertain as to whether their order will be dispatched or if they will receive the right products. COD solves such uncertainties and reduces customer stress by allowing them to make the payment only after checking the shipment. If customers receive inaccurate or low-quality products, they can instantly return them. Furthermore, COD is a secure payment option that does not require customers to share their financial information online [
66]. The simple structure of COD makes it a simple method for making an online transaction, which in turn enables customers with average computer competency to engage in online shopping. Hence, the provision of a COD payment option by e-retailers increases customer trust. Importantly, the effect of COD is higher than RPL on customer trust. A plausible explanation for this is that COD conveys a shopping experience that simulates an offline shopping experience as customers can inspect products before paying and can instantly return products if they are inadequate or faulty.
The results confirm the significant positive effect of SCCs on TR supporting H3. This finding is confirmed by previous research [
104,
112,
115]. It implies that providing customers with social commerce tools (e.g., ratings, social media, recommendation systems, reviews) to access the opinions and feedback of former customers regarding their purchase experiences increases the trust of potential/actual customers in transacting with e-retailers. Through SCCs, customers are able to use different channels to share their purchasing experiences with respected e-retailers and products/services without any interruption from e-retailers. Electronic Word-of-Mouth (EWOM) provided through SCCs is recognized as a trustworthy information source for customers. SCCs are methods used to communicate and exchange information online about e-retailers and products/services between senders (former customers) and receivers (actual or potential customers). For the receivers, the information provided by the senders (former customers) has no commercial intent, and as such is viewed as more credible than other information sources such as advertisements or e-retailers’ websites.
The results suggest that TR is a key enabler of PI as it has a significant positive effect on this variable; hence, H4 is supported. This finding aligns with previous research [
50,
136,
138,
139]. This implies that the more customers’ trust increases, the more they intend to purchase. Trust is an effective mechanism that reduces the inherent uncertainty and risk related to e-commerce. If customers perceive the integrity, benevolence, and ability of e-retailers to be sufficient they will develop an inclination to be vulnerable to e-retailers. If the extent of a customer’s trust in an e-retailer surpasses their perceived risk, then the customer will become involved in a risky relationship with the e-retailer. This means that trust is the main antecedent of purchase intention in online shopping settings where there is a perceived risk of a negative consequence [
123].
6. Managerial Implications
The main findings of this study show that to enhance customer trust as a key determinant of purchase intention, e-retailers should provide customers with COD, SCCs, and RPL. The strategic use of return policies by retailers can generate a significant increase in customers’ lifetime value [
148]. The leniency of return policies was found in this study to be a key predictor of customer trust. Thus, e-retail managers should actively realize the importance of customers’ trust in converting return policies into purchase behavior. Although customer trust is a necessary aspect to be considered in product purchasing, e-retailers need to be aware that building higher trust will allow them to introduce new products and renew their offers as customers will trust them in the event of a service recovery [
50]. Thus, managers should employ return policies as a method to boost the competitive position of their businesses by gaining customer trust to increase future sales. This requires offering lenient return policies in terms of momentary costs, longer return windows, convenience, a wider range of products that can be returned and exchanged, and full refunds. These aspects should be considered when developing return policies. Furthermore, it is important to determine the main reasons for the returns as the factors that influence the return experience will help to clarify why returns occur, facilitating the process of identifying effective solutions [
149]. Importantly, because of technological advancements, unethical/opportunistic returns may not have a substantial impact on e-retailers as they can detect unusually frequent returns by an individual customer.
The effect of SSCs on customer trust in this study was significant. This suggests that Web 2.0 technologies should be considered a key element when designing e-commerce websites. Increasing customer trust requires integrating Web 2.0 technologies into e-commerce websites and connecting these websites to various social network sites (e.g., Facebook). In so doing, customers will be allowed to access more social and trustworthy information based on previous shopping experiences and feedback related to products and e-retailers. By providing customers with credible sources of information other than e-commerce websites, the perceived uncertainty of customers will be reduced. Moreover, the implementation of SCCs will enhance the trustworthiness of e-retailers as it is an indicator of transparency that discourages the act of information concealing from customers. SCCs can also aid e-retailers in monitoring consumer interactions, allowing them to predict and prevent negative WOM that might imperil their reputation and, therefore, reduce customers’ willingness to buy their products [
110]. Furthermore, these tools can be a valuable source of information for two-way communication as well as assist e-retailers in successfully and promptly resolving consumer problems. Importantly, e-retailers should identify strategies that will encourage customers to use SCCs to generate content and enhance profits as a result of attracting new customers [
150]. Positive WOM is an effective marketing approach employed to endorse products/services, attract more customers, and deepen relationships with existing customers. However, e-retailers should also be aware that negative WOM can significantly overshadow positive WOM and thus increase customer uncertainty [
31]. High-quality customer service, lenient return policies, and high logistics service quality can increase customer satisfaction, motivating customers to convey positive WOM through SCCs [
50,
63,
151].
The findings indicate that customer trust is increased by the availability of a COD payment option. Thus, e-retailers, particularly new e-retailers planning to enter the e-market, should consider providing COD as a payment option, among others. It has also been claimed that COD could be employed as a strategic approach for e-retailers to increase sales as it is deemed to appeal to a broader demographic [
66]. Furthermore, a study by Kidane and Sharma [
152] found that nearly 67% of customers dismiss e-commerce transactions when e-retailers request authentication of their banking information. Hence, COD can be used by e-retailers to decrease customers’ anxieties about online fraud. Because COD increases the risk of returns [
64], e-retailers and their logistics service providers (LSPs) should ensure that customers’ orders are checked with respect to quality and accuracy before shipment [
63]. Furthermore, LSPs should bear in mind that delivering the right orders to customers on time with the expected condition(s) requires adequate logistics infrastructures. In addition, it is important for e-retailers to use reliable logistics partners to ensure the accuracy and condition of shipments.