Introduction

E-commerce live streaming (ELS) is a form of online selling integrating real-time interaction through live streaming to create a more engaging and immersive experience for customers (Zheng et al., 2022). As one of the newest commercial modes, it was launched in 2016 by several e-commerce platforms in China. It has continued to rise in popularity, especially during online shopping festivals such as “Double Eleven”. Live-streaming shopping is now very popular among Chinese consumers, according to the Ministry of Commerce of China, with over 430 million ELS active users in 2021. The scale of China’s ELS market is expected to reach 3487.9 billion RMB in 2022. Austin Jiaqi Li, a famous live streamer in China with 64.4 million followers on Taobao Live and another 44.9 million followers on Douyin, reached 21.5 billion RMB pre-sales during the first live streaming of 2022’s “Double Eleven”, which is twice as much as last year. Besides, physicians are increasingly using live streaming to interact with patients in real-time and share their medical knowledge, and educate audiences on various platforms (Chen and Wu, 2024). New communication forms have emerged for live-stream shopping. For example, as an “immersive” platform, the metaverse can provide viewers with greater experiential value (Barta et al., 2023).

Live streamers, also known as live hosts, anchors, who perform live and interact with consumers through the internet, were developed from the news anchors in the American radio and television industry. ELS streamers are online influencers who impact consumers by displaying product information, creating promotional interactions, and offering personalized services (Zhang et al., 2022). According to Erving Goffman’s dramaturgical theory, our daily behaviors are performances displayed to others. Specifically, individuals can manage their image by controlling the behaviors in front of others. Live streaming behavior is with a clear sense of participation, like a performance in daily life (Sannicolas, 1997). Live streamers perform various roles to entertain, connect, and build relationships with viewers (You et al., 2023). They perform particular practices to attract audiences and (Chu et al., 2022) and manage the audiences’ sense-making process to build specific images in live streaming (Li et al., 2022). Driven by commercial objectives, ELS streamers strategically manipulate various characteristics, including their background, content, and interaction styles, to cultivate specific images that foster acceptance, trust, and popularity among viewers, ultimately benefiting their business goals (Hidayat, Hidayat (2020)). In light of the aforementioned research, ELS streamers’ image-building strategies may significantly influence consumer trust and purchase intention.

Streamers showcase various characteristics when introduce products and interact with consumers to sell products in the live broadcast room. However, characteristics in ELS room are not necessarily the complicated characteristics of the manifested social image publicly. Compared to self-image which refers to how people perceive themselves, social image is how they are perceived by others in society publicly. It can be influenced by various factors such as appearance, behavior, and language, which in turn influences consumers’ behaviors (Jiang et al., (2022)). The characteristics of ELS steamers have been studied in many previous studies, including attractiveness, expertise, popularity, affinity, interactivity, trustworthiness, entertainment, responsiveness, and more (Wang et al., 2022; Guo et al., 2022). Few studies have explored the effect characteristics of live streamers on buying behavior from the perspective of streamers’ social image which might be manifested on purpose. To reach the objective, the study attempts to address the following research questions:

RQ1. How do social image characteristics manifested by ELS streamers with different motivations impact the trust in ESL shopping?

RQ2. What is the effect of social image built by ELS streamers on purchase intention in ELS shopping?

RQ3. Does trust mediate the relationship between social image factors and purchase intention in ELS shopping?

As a contextual contribution, the present paper tries to bridge the gaps mentioned above by examining how different ELS streamers’ social image exert influences on consumers’ purchase intention in ELS in China. In which, a conceptual model with fourteen hypotheses based on the Stimulus-Organism-Response (SOR) framework and dramaturgical theory is crafted through the lens of the social image, dissecting it into three dimensions based on the intentional behavior of ELS streamers (Ming et al., 2021). The first dimension addresses the pursuit of an “I am attractive” image. ELS streamers meticulously sculpt their physical attractiveness, understanding that a visually appealing presence can captivate and retain an audience’s attention. The second dimension is rooted in the streamers’ endeavor to validate their role as “I am the right professional product endorser/spokesperson.” The ELS streamers invest considerable effort in ensuring there is a match-up between their persona and the products they endorse, in maintaining authenticity to foster trust, and in demonstrating expertise to affirm their authority and knowledge on the subject matter. The third dimension emerges from the aim to be perceived as “I am an interesting and effective communicator with customers.” To achieve this, ELS streamers focus on Entertainment to keep the audience engaged and on Responsiveness to interact and connect with the audience in real-time, ensuring an interactive and enjoyable experience for the viewers. The SOR framework is particularly well-suited for this study because it explains how specific stimuli (three dimensions of social image) affect consumer perceptions (consumer trust) and drive their actions (buying behavior) (Ming et al., 2021). The present study’s findings uniquely contribute to multiple academic fields. Meanwhile, the study indicates that e-retailers should pay attention to craft, sustain, and enhance ELS streamers’ social image to gain a competitive advantage over their rivals.

The rest of the paper is organized as follows: We first review the relevant literature and present our research model and hypotheses. Next, we describe the research methodology, data analysis, and results. Finally, we discuss the theoretical and managerial implications of the study’s findings, with limitations of the study and future research directions following thereafter.

Theoretical background and hypothesis development

Live streaming fosters live streamers’ identity construction and shapes the role of influencers (Hidayat, Hidayat (2020)). Typically, ELS streamers try to create, maintain, or improve their social images to make them more acceptable to ELS viewers. The social image of an ELS streamer can be summarized as a carefully constructed public image, crafted through a combination of external images, personality, unique traits, professionalism, and knowledge base (Wongsunopparat and Deng, 2021). The social image that ELS streamers try to build is similar to the streamer persona they want to show their viewers. Previous researches on streamer persona indicate that a streaming persona is a complex interplay of live performance, platform-specific features like streamer-specific emoticons and visual overlays, game choices, and individual playing style (Jackson, 2020).

Based on the characteristics of ELS streamers proposed by previous studies, considering the social image definition and persona research about live streamers, in this paper, the characteristics of streamers’ social images are categorized into three types according to the motivation of ELS streamers’ social image building behaviors. Firstly, characteristics that contribute to making a good first impression, such as facial attractiveness, language style, voice tone, and gestures (Lorenzo et al., 2010). Physical attractiveness aligns with the objective of presenting oneself as appealing and visually pleasing to viewers. In ELS, a streamer’s physical appearance can quickly capture attention and influence initial perceptions. Hence, physical attractiveness, the primary factor in forming first impressions, is the variable for social image objective 1 (“I am attractive”). Secondly, factors that enhance the professionalism of product introduction, such as authenticity, expertise, recommendation strength (influence), and broadcasting skills (Tian and Lee, 2022). The alignment between the streamer and the product is crucial for establishing credibility as a spokesperson. When viewers perceive a good fit, they are more likely to trust the endorsement. Genuine enthusiasm and belief in the product enhance the streamer’s persuasive power. Authenticity fosters trust and makes the endorsement more believable. Demonstrating knowledge and understanding of the product establishes the streamer as a reliable source of information, further strengthening their credibility. Therefore, in this paper, we have included variables “Matchup”, “Authenticity”, and “Expertise” for social image objective 2 (I am the right professional product endorser/spokesperson). Lastly, interaction factors that contribute to building a positive relationship with customers, such as entertainment, controllability, communication, and responsiveness (Huang 2012; Tian and Lee, 2022). Interacting with viewers in real-time builds rapport and a sense of connection. Responsiveness signals that the streamer values viewer participation and feedback, contributing to a positive relationship. Livestreaming is a form of entertainment. A streamer’s ability to engage and entertain viewers is key to maintaining their attention and fostering a positive experience. Hence, “Responsiveness” and “Entertainment” are the variables selected for social image objective 3 (“I am an interesting and effective communicator with customers”).

Physical attractiveness

First impressions are very important in everyday life. Making a good first impression is critical to keep viewers staying in the live broadcast room. Previous studies show that attractiveness has positive impacts on consumer attitudes toward advertising, brand, product, and purchase intention (Eisend and Langner, 2010). In ELS, although the viewers cannot meet the streamer in person, they watch and interact in real-time, forming impressions and judgments (Bargh et al., 2002). Prior studies on celebrity endorsements show that streamers’ attractiveness is powerful heuristic cues shaping viewers’ first impressions of the products due to their pivotal role of ELS (Park and Lin, 2020). In addition, attractive streamers are more likely to command attention build positive relationships with viewers, and generate positive perceptions of the products they recommended (Gao et al., 2021).

Dimensions of attractiveness varied by different scholars. Some studies focus on physical attractiveness, such as good looks, likability, and beautiful voice (Park and Lin, 2020), while others consider the broadcasting performance as well. In this study, we focus on physical attractiveness and define streamer physical attractiveness as the streamers’ appearance, including face and body attractiveness, makeup, dressing, voice, humor, and the streaming environment settings in a live streaming room (Karandashev et al., 2020; Yuan et al., (2016); Zhang et al., 2022). Physically attractive communicators are significantly more persuasive, as measured by both verbal and behavioral agreement (Chaiken, 1979). Experimental evidence from advertising and related disciplines indicates that physically attractive communicators and models are perceived more favorably than unattractive ones, leading to a positive impact on the products with which they are associated (Joseph, 1982). Moreover, physically attractive people are perceived more positively in first impressions and are likely to benefit from this enhanced positivity (Langlois et al., 2000; Lorenzo et al., 2010). Prior studies have shown that physical attractiveness determines the viewers’ first impression of the endorser (Choi and Lee, 2019).

As such, this study proposes the following hypotheses:

H1: The physical attractiveness of the social image built by ELS streamers has a significant impact on consumer trust.

H2: The physical attractiveness of the social image built by ELS streamers has a significant impact on consumer purchase intention.

Matchup, authenticity, and expertise

The main activities of an ELS streamer start with product selection, followed by showcasing and introducing the product and sharing the user experience details of a product. Additionally, they provide professional suggestions and answer the questions from viewers (Zhang et al., 2022). An ELS streamer with a high-level of professionalism in the presented product can improve the consumers’ shopping experience, which, in turn, can enhance customer satisfaction and encourage continuous ELS shopping (Chen et al., 2020a). There are numerous factors that influence the performance of ELS streamers’ product-presenting behaviors, including match-up congruency with the product, authenticity, expertise, popularity, broadcasting skills, and so on (Zhang et al., 2022). This study focuses on examining how the following three dimensions of ELS streamers’ social image affect consumers’ purchase intention.

Matchup congruency

Match-up congruency refers to the degree of consistency or similarity between the product and its endorser (Bergkvist and Zhou, 2016). It is based on the matchup hypothesis, which recognizes that different celebrity endorsers have varying effects on endorsed products. In the context of ELS, good matchup dimensions between a streamer and the product include the streamer’s lifestyle, physical fitness, broadcast style, and personality, which should align with the brand image and product category (Malodia et al., 2017). For instance, a streamer with a fashionable, contemporary, organized, youthful, and modest personality may match with a brand image representing sincerity, excitement, competence, sophistication, and ruggedness. Previous studies have explored how customers’ purchase intention for live streaming varies across matchup between ELS streamers and product types. For household products, a streamer promoting a healthy lifestyle is crucial, while for lifestyle products, streamers with physically fit and stylish lifestyles are more effective (Parmar et al., 2020). Research also indicates that consumers prefer experience-based styles that provide practical insights when purchasing experience products.

A good matchup between ELS streamer’s image and the product enhances believability, which in turn improves the purchase intention (Parmar et al., 2020). Studies on ELS hosted by celebrity further demonstrate that the matchup congruence between the celebrity and the brand/product positively influences celebrity trustworthiness and influence. The more aligned matchup between a product and its celebrity, the more likely consumers will create a strong mental link between the celebrity and the product in their memory (Till and Busler, 2000). This reinforcement of existing information leads to improved advertising effectiveness and brand attitude in terms of endorsement (McCormick, 2016; Parmar et al., 2020). In contrast, perceived mismatch can lead to a negative evaluation of the product and be less inclined to buy it (Ilicic and Webster, 2013). Previous studies also point out that anchor-product fit and live content-product fit can reduce perceived product uncertainty and promote consumer purchase behavior.

Hence, in view of the above references, it can be readily inferred that a close match between the ELS streamer image and the product image is necessary, and this study proposes the following hypotheses:

H3: Match-up congruence between the product and the social image built by ELS streamers has a significant impact on consumer trust.

H4: Match-up congruence between the product and the social image built by ELS streamers has a significant impact on consumer purchase intention.

Expertise

Expertise refers to the possession of professional knowledge in a particular area. (Shen et al., 2010). In ELS, streamers must provide detailed introductions of products, so expertise is defined as the knowledge, experience, and related skills that ELS streamers possess and convey to their audiences during live streaming (Ladhari et al., 2020). Expertise in ELS is demonstrated by the streamers’ ability to clearly introduce products, offer expert guidance with valuable advice, recommend personalized products to consumers, and address consumers’ doubts and so on (Zhang et al., 2022). Unlike product endorsers in traditional advertising, ELS streamers typically showcase a wide range of products in each livestream, rather than being limited to a particular product or catalog. Streamers with expertise in ELS often possess good business skills or specialized knowledge in a particular field, allowing them to combine their knowledge with products that resonate to the viewers and effectively meet customer needs (Zhang et al., 2022).

Prior studies on the key opinion leaders (KOLs) have shown that expertise significantly and positively correlates with consumers’ purchase intention. In addition, expertise fosters trust, solidifying the streamer’s credibility as a reliable source of information and an influential opinion leader (Casaló et al., 2020). In ELS, a streamer can help their viewers make comparisons between many choices based on the professional information, which in turn positively affects consumers’ trust on the product introduced by the streamer (Chen and Lin, 2018). The expertise of ELS streamer directly affects consumers’ perceptions of product quality, leading to higher levels of satisfaction, trust, and positively influencing their purchasing decisions (Chen et al., 2020b). Professionalism, as a synonym of expertise, has also been studied in the context of live streaming, also positively influences consumers’ purchasing decisions (Chen et al., 2022). Moreover, prior researches have tested the relationship between expertise and impulse purchase intention. In the ELS scenario, consumers are more likely to provide positive feedback on the recommended products if they perceive professionalism of the streamers, thus generating a strong impulse to make a purchase (Liu et al., 2022).

To sum up, it can be expected positive effects of the perception of expertise in the social image of ELS streamers on consumers’ trust and purchase intention. Therefore, this study proposes the following hypotheses:

H5: The expertise of the social image built by ELS streamers has a significant impact on consumer trust.

H6: The expertise of the social image built by ELS streamers has a significant impact on consumer purchase intention.

Authenticity

According to Trilling (1972), authenticity involves the display of hidden inner aspects, while its alternative meaning refers to passion and interiority. Other researchers argue that authenticity is being natural, i.e., the genuineness, reality, and originality of something (Fine, 2003). For internet user, authenticity describes how individuals assess the validity of information to analyze the quality of online (Beverland et al., 2008). In ELS context, where live broadcasts are unscripted and instantaneous, the content cannot be prerecorded or edited. Hence, authenticity refers to live streamers presenting the reality of things and finding a balance between commercial motivations and sincerity (Becker et al., 2019). This includes maintaining a consistent personal brand image, providing credible product information, and creating scenes that resemble real life (Morhart et al., (2015)). In this study, authenticity is associated with the ELS streamer, the content they deliver, and the manner in which they deliver the message.

Authenticity is important for brand trust and subsequently increasing purchase intention (Eggers et al., 2013). The authenticity of the streamer’s social image and live content helps viewers overcome skepticism about a product or a brand and build trust with the streamer by understanding the viewers’ needs and situations (Becker et al., 2019). Prior studies on social media celebrities emphasize the significance in maintaining the celebrities’ appeal and relationship with viewers (Marwick, 2013). Furthermore, the authenticity of live streamers plays a crucial role in enhancing viewers’ perceived intimacy, leading to increased online engagement and, consequently, higher purchase intention. (Becker et al., 2019). Several studies have focused on the relationship between authenticity and purchase intention. According to Tong (2017), the authenticity of live streaming enhances consumers’ purchasing intention by influencing their sense of immediacy and trust. Another study on fresh live broadcasting supports positive relationships between the authenticity of live streaming and consumers’ perceived trust, thus influencing their willingness to make a purchase (Song et al., 2022). ELS streamers strive to maintain the authenticity of their social image to enhance the viewers’ trust and purchase intention. For example, streamers establish their influence by maintaining authenticity through the disclosure of personal information or showcasing their daily activities in their streaming videos (Djafarova and Trofimenko, 2019).

Hence, this study proposes the following hypotheses:

H7: The authenticity of the social image maintained by ELS streamers has a significant impact on consumer trust.

H8: The authenticity of the social image maintained by ELS streamers has a significant impact on purchase intention.

Interactions

ELS streamers’ unique social images can be identified through the process of interactions (Li, 2018). The bullet screen feature in ELS allows viewers to ask questions or participate online games, and live streamers provide further details about products or special offers (Park and Lin, 2020). Unlike traditional online shopping, ELS enables real-time one-to-many high-intensity interactions, making the sale process more transparent, entertaining, and immersive. Interaction in ELS include entertainment activities, games, flash deals, and more. The sense of social presence and telepresence created by real-time interactions between ELS streamers and viewers in an online shopping environment can make viewers feel comfortable (Gao et al., 2021) and significantly enhancing consumer’s purchase intention (Wongkitrungrueng and Assarut, 2020).

Therefore, the responsive and enjoyable feature of interactions is important. Correspondingly, this study develops two interaction dimensions of ELS streamer’s social image, namely entertainment and responsiveness, to examine the effects on purchase intention.

Responsiveness

Responsiveness refers to the response speed to previous information to meet a user’s needs, such as reacting quickly, and responding with emotion. In ELS, viewers can ask questions and receive answers from the streamer almost in real-time, which can be considered a form of direct communication. Therefore, in ELS, responsiveness can be defined as the streamers’ timely reaction to incoming online messages (Guo and Sun, 2022). Additionally, responsiveness refers to the extent to which customer feedback is taken into consideration (Gummerus et al., 2004). It also reflects the willingness to help consumers and provide timely services (Zhang et al., 2022).

Many scholars have found that responsiveness is an important factor affecting consumers’ purchase intention (Rana et al., 2015). Previous studies have shown that responsiveness has a positive relationship with customer satisfaction, which, in turn has a positive relationship with purchase intention (Khatoon et al., 2020). Salespeople’s responsiveness can enhance customers’ perceptions of value. Moreover, responsiveness enhances commitment, trust, and purchase intention. It also promotes the willingness to recommend the provider to other potential customers (Guenzi et al., 2009). Personnel responsiveness plays a vital role in enhancing consumers’ perceived value, favorable behavioral intentions, positive word-of-mouth, and more. It has been confirmed that there is a positive relationship between personnel responsiveness and purchase intention (Abd Aziz et al., 2015). In ELS context, prior empirical study has confirmed that responsiveness is positively associated with perceived information usefulness and arousal, which, in turn positively affects purchase intention (Guo and Sun, 2022). Responsiveness is also one of the most effective ways to build trust and create better relationships. According to Lee (2005), perceived user responsiveness has a positive impact on a customer’s trust in online communities.

In view of the above references, we propose two following hypotheses:

H9: The responsiveness of interaction created by ELS streamers has a significant impact on consumer trust.

H10: The responsiveness of interaction created by ELS streamers has a significant impact on consumer purchase intention.

Entertainment

According to Ducoffe (1996), entertainment refers to the ability to fulfill an audience’s needs for escapism, diversion, esthetic enjoyment, or emotional enjoyment (Ducoffe, 1996). Its purpose is to make the audience feel happy or interested or to provide temporary relief from reality and help them put their worries behind them (Chen and Lin, 2018). Live streaming, at its core, is an audio and video medium that has possesses entertaining characteristics. During the covid-19 pandemic, ELS became a popular form of entertainment and a safer and effective way of shopping. The experience of entertainment (also known as hedonic value, which is subjective and can be derived from playfulness and fun) is defined as pleasure, thrill, relaxation, and diversion (Bosshart and Macconi, 1998). Watching ELS may be considered a form of entertainment. Enjoyable interactions with live-streaming viewers, such as games, help ELS streamers maintain engagement and alleviate boredom during product recommendations via live streaming (Chen and Lin, 2018).

According to past research, live-streaming broadcasts provides an entertaining media experience. The purchase decision is influenced by the experience of entertainment, as hedonic motivation directly impacts the intention to search and indirectly impacts the intention to purchase (To et al., 2007). The experience of entertainment positively affects attitude, thus influencing the willingness to recommend and the intention to use a specific social platform (Curras-Perez et al., 2014). Prior research on website flow shows that customer satisfaction and purchase intention are influenced by the perceived playfulness of a website (Hsu et al., 2012). Previous studies on e-commerce indicate that entertainment has positive effects on consumers’ affection, which ultimately affects their purchasing intention (Haile and Kang, 2020). Entertainment gratification leads to consumer loyalty toward live streaming channels as shown by Hsu et al. (2012). Additionally, consumers’ perception of entertainment (hedonic value) is positively associated with repeat purchase intention (Chiu et al., 2014; Gao et al., 2021). In esports live streaming, live streaming full of fun and entertainment can stimulate the viewers’ willingness to purchase virtual gifts through emotional attachment and flow experience (Chen and Wu, 2024).

In view of the above references, we propose two following hypotheses:

H11: The entertainment of interaction created by ELS streamers has a significant impact on consumer trust.

H12: The entertainment of interaction created by ELS streamers has a significant impact on consumer purchase intention.

Mediation effects: the role of trust

Trust has always been a key variable in the consumer buying behavior research models. It is essential in dealing with uncertainty and is one of the most important factors contributing to the success of e-business. Trust is also a significant driving force of social commerce intentions and can motivate social media users to make purchase decisions (Al-Tit et al., 2020). Lack of trust is frequently cited as a reason why consumers refrain from purchasing online. The main sources of trust are consumer characteristics, firm characteristics, website infrastructure, and interactions with consumer. Trust can be measures by the dimensions of competence, integrity, and benevolence (Oliveira et al., 2017). Previous studies on online shopping suggest that the antecedent of purchase intention affect trust, and in turn, trust increases purchase intention. Furthermore, consumers with high overall trust demonstrate a higher intention to purchase online (Oliveira et al., 2017).

Hence, this study proposes the following hypothesis:

H13: Consumers’ trust in streamers has a significant impact on their purchase intention.

In addition, the mediating effect of trust has been verified in numerous studies on online shopping. For example, previous studies show that trust acts as a mediator in the relationship between relative advantage and attitudes toward online shopping, as well as between eWOM and attitudes toward online shopping (Chetioui et al., 2021). Thus, this study also assesses the mediating effect of trust between the factors of ELS streamers’ social image and purchase intention:

H14a: Trust mediates the relationship between the physical attractiveness of ELS streamers’ social image and purchase intention.

H14b: Trust mediates the relationship between match-up congruence and purchase intention.

H14c: Trust mediates the relationship between the authenticity of ELS streamers’ social image and purchase intention.

H14d: Trust mediates the relationship between the expertise of ELS streamers’ social image and purchase intention.

H14e: Trust mediates the relationship between the responsiveness of ELS streamers’ social image and purchase intention.

H14f: Trust mediates the relationship between the entertainment of ELS streamers’ social image and purchase intention.

SOR Theory based research framework

The SOR theory, which proposes that an external stimulus (S) can affect individuals’ behavioral responses (R) through their internal cognitive state (O), was introduced by Mehrabian and Russell in 1974 based on environmental psychology. In the ELS context, the design features of the virtual environment with which consumers interact play a crucial role in shaping online interactions and purchase decisions. These environmental cues, as emphasized in social media studies utilizing the SOR theory, are crucial stimulus that manipulates the intervention processes and purchase decisions (Ming et al., 2021). The present study employs the SOR framework for several key reasons. First, this model has gained considerable popularity recently in the field of online consumer behavior research (Huang, 2023; Song et al., 2022; Tian and Li, 2022). Second, the SOR theory’s strength lies in its ability to integrate other theories, providing a comprehensive and interdisciplinary model (Zhu et al., 2023). Based on the SOR theory, ELS streamers’ social image, as an external stimulus, also influences customers’ cognitive reactions and subsequently impacts customer attitudes and purchasing behaviors. This study combines SOR theory with Erving Goffman’s dramaturgical theory to establish a structural equation model based on aforementioned hypothesis, as shown in Fig. 1, in which trust acts as a mediator between the stimulus (social image) and the response (purchase intention). Trust is a psychological state that reflects a consumer’s confidence and belief in streamers’ reliability and goodwill. It is not directly observable but is inferred from the consumers’ thoughts and feelings (Tuncer, 2021). It is shaped by the consumers’ perception of streamers’ social image and, in turn, influences their willingness to purchase. Purchase intention reflects the consumers’ decision-making process and their inclination to act based on their evaluation of the streamer and the product (Tian and Li, 2022)). In this paper, purchase intention is consumers’ self-reported likelihood of buying a product. It is the behavioral outcome or response resulting from the stimulus and the organism. Meanwhile, based on the literature review, we also examined the direct effect on purchase intention.

Fig. 1
figure 1

Research conceptual model.

Full size image

Research method

Measurement development

The measurement instruments employed in this study were mainly adapted from existing research. All the measures were carefully chosen and validated from prior studies to fulfill the objectives of the study. We measured physical attractiveness with four items adapted from an existing scale (Karandashev et al., 2020; Yuan et al., (2016)). Matchup congruence was measured by four items adapted from an existing congruence scale (Chen et. al, 2022, Parmar et al., 2020), including streamer and product fit, and streamer and viewer fit. Authenticity was measured with four items adapted from Becker et al. (2019) and Morhart et al., (2015). Expertise was measured with four items adapted from Ladhari et al., 2020 and Zhang et al. (2022). Responsiveness was measured with five items based on Guo and Sun (2022) and Gummerus et al. (2004). The entertainment of the interaction was measured with four items (Chen and Lin, 2018). Trust and purchase intention were measured with four items (Chetioui et al., 2021, Ho et al., 2022). All the items are measured on a five-point Likert scale ranging from one (“strongly disagree”) to five (“strongly agree”). Some of the items were reworded slightly to achieve a better fit with the context of the study.

To ensure the quality of the questionnaire, a focus group discussion and a pilot study were conducted with professors whose main research spot is e-commerce, two ELS streamers, and two ELS business managers. As the survey data is from China, we also invited two professors who are proficient in both Chinese and English helping us translate the questionnaire items clearly and precisely. Furthermore, the translation and original scale were further validated by a second bilingual expert to eliminate any potential inconsistencies or cultural misunderstandings. In addition, a pretest was conducted with 15 university students who are familiar with ELS. In presurvey, the respondents were asked to name ELS platforms they use and product categories introduced in ELS. Meanwhile, some of the statements in the presurvey were fine-tuned to form a final questionnaire.

Sampling and data collection

ELS has emerged as a dynamic force in the e-commerce landscape, transforming the way consumers discover and purchase products. Today, ELS has a global presence, with platforms like Taobao Live and Instagram Shopping taking root in various regions. China is not only the largest but also one of the fastest-growing live streaming markets globally (Chen and Wu, 2024). The sheer scale of the Chinese ELS market dwarfs that of most other countries. Besides, due to fierce competition among platforms and streamers, the ELS industry in China is constantly evolving with new features, engagement techniques, and sales strategies. Hence, China is a top choice for us to examine the impact of streamers’ social image on customers’ purchase intention.

Our data was collected from shoppers who had previously made purchases through ELS on popular live-streaming platforms (Taobao/Tmall, Douyin, Kuaishou, Jingdong, et al.) (Park and Lin, 2020), using both online and offline methods. Online survey data was collected through Wenjuanxing (https://www.wjx.cn), the largest online survey platform in China. Online questionnaires were disseminated through popular social media platforms in China (i.e. WeChat, Weibo, Zhihu) by providing a link from Wenjuanxing. WeChat has approximately 1.36 billion monthly active users, making it China’s most popular social media platform. Its WeChat’s Moments feature, similar to Instagram, contributes significantly to its widespread use (Bilal et al., 2024). Weibo, a Chinese microblogging website and social media platform, is similar to Twitter and Instagram with over 550 million monthly active users. Zhihu, a valuable platform in China for knowledge acquisition, professional networking, and engaging discussions surrounding various topics, has over 420 million registered users. For offline survey, paper-based questionnaires were distributed in person randomly on campuses and streets near universities in southern China, targeting both students and residents. Our data was collected from 365 respondents between March and April in 2023. Of these respondents 268 completed the survey online and 97 offline.

Participants were informed that their participation in the survey was entirely voluntary and anonymous before they began the survey. Considering that not all the respondents have the ELS shopping experience, we set up a screening question to facilitate the exclusion of respondents who had no ELS shopping experience. In the main study, firstly, we explained the research motivation and ensured the confidentiality of respondents’ information in the questionnaire. We emphasized that responses should be based on the streamer’s overall social image, rather than on the performance during a specific live-streaming session. Then, to gather background information, each respondent was asked about their age, income, education level, and time spent watching ELS. Lastly, each respondent was asked to recall a recent ELS viewing experience on one of the three leading platforms and name the products recommended by ELS streamers, and answer questions related to the study’s variables. To minimize the effects of response bias, we randomized the order in which Likert scale items are presented to respondents. Besides, we added one bogus item and one self-reported diligence item as attention checks to identify inattentive respondents. The survey questionnaire employed in this study included eight construct measures with 33 items (as given in Appendix A).

After cleaning the data and eliminating responses with substantial missing values and responses by inattentive respondents, there were 323 valid responses. Of these, 231 were collected online and 92 offline, resulting in an overall effective response rate of 88.49%. The background information of the respondents is presented in Table 1. Of the 323 respondents, 67.18% reported using ELS 1 to 3 times a week. Accordingly, 63.2% of the respondents were females, and 36.8% are males. The results are consistent with the prior studies in online retailing (e.g., Gao et al., 2021; Park and Lin, 2020). In addition, results also indicate that 76% of the respondents were between 20 and 34 years old. Respondent age groups ranged from 1.9% (10–17 years old), 32.51% (18–24 years old), 28.48% (25–30years old), 17.3% (31–35 years old), to 17.6% (above 35 years old). The most commonly mentioned platforms were Taobao, Douyin, and Kuaishou. ELS streamers recommended almost all kinds of products, with the most common categories being fashion, beauty, sports & outdoors, and electronics.

Table 1 Demographics of Respondents.
Full size table

Results

Model estimation

The main constructs (see Table 2) were first evaluated for reliability and convergence by using SPSS 26.0 software. To assess the reliability of the potential variables, we utilized Cronbach’s α coefficient and composite reliability (CR). Table 2 shows that all the Cronbach’s alpha and CR values were above 0.7 (Fornell and Larcker, 1981), and the Cronbach’s α of the entire sample data reached 0.925, indicating a satisfactory level of reliability. To evaluate convergent validity, we tested the factor loadings, composite reliability (CR), and average variance extracted (AVE). We used factor analysis in SPSS to determine the factor load of each item. As shown in Table 2, the loadings of each factor were higher than the accepted level of 0.7 (P < 0.001). Moreover, the AVE values were greater than the accepted level of 0.5 (Chin, 1998), indicating that the measurements had sufficient convergent validity. In addition, we tested the discriminant validity (see Table 3) before estimating the structural equation model (SEM) as well. To assess discriminant validity, we compared the square root of the average variance extracted (AVE) for each construct to its correlations with other constructs (Fornell and Larcker, 1981). As shown in Table 3, the square roots of the AVEs for all the constructs are greater than their correlations with the other constructs, indicating good discriminant validity.

Table 2 Analysis of reliability and validity.
Full size table
Table 3 Discriminant Validity Analysis.
Full size table

We used AMOS 24 to examine the path coefficients, associated p-value of the paths, and the proposed hypotheses. To evaluated the validity of the measured constructs, a confirmatory factor analysis (CFA) was conducted. The goodness-of-fit statistics verified that most criteria met the recommended values in the measurement model (χ2 = 625.3, df = 467, p < 0.001); CFI = 0.985; GFI = 0.915; RMR = 0.04; RMSEA = 0.037, AGFI = 0.889). Hence, the measurement model fit the data well (Marsh et al., 1998).

The path analysis results are presented in Table 4, while Fig. 2 illustrates standardized estimates of the structural equation model. The results show that all hypotheses were significant, with the exception of H6, H11, and H12. The attractiveness (\(\beta =0.15,\,p < 0.01\)), expertise (\(\beta =0.2,\,p < 0.001\)), authenticity (\(\beta =0.2,\,p < 0.01\)), and responsiveness (\(\beta =0.19,\,p < 0.01\)) of the social image built by ELS streamers had a significant impact on consumer trust. Moreover, the match-up congruence between the product and the social image positively impacted consumer trust as well (\(\beta =0.25,\,p < 0.001\)), supporting H1, H2, H3, H4, and H5. However, the effect of the entertainment of interaction created by ELS streamers on consumer trust was not statistically significant. Therefore, H6 was rejected.

Table 4 Summary of the path analysis results.
Full size table
Fig. 2
figure 2

Standardized Estimates of the Structural Equation Model.

Full size image

The match-up congruence between the product and the social image had an significant effect on consumer purchase intention (\(\beta =0.23,\,p < 0.001\)). The match-up congruence was the most influential variable, followed by expertise (\(\beta =0.16,\,p < 0.01\)), authenticity (\(\beta =0.15,\,p < 0.05\)), and attractiveness (\(\beta =0.1,\,p < 0.05\)) of the social image built by ELS streamers, supporting H7, H8, H9, and H10. However, the responsiveness and entertainment of interaction created by ELS streamers were not statistically significant, and thus, H11 and H12 were rejected. Additionally, consumers’ trust in streamers has a significant impact on their purchase intention (\(\beta =0.36,\,p < 0.001\)), supporting H13.

Mediation analysis

The study further explored the mediating role of consumer trust in the influence of the ELS streamers’ social image characteristics (physical attractiveness, matchup congruence, expertise, authenticity, responsiveness and entertainment) on consumer purchase intention. To examine the confidence intervals for the indirect effects, we conducted a bootstrapping procedure with 2,000 samples based on the process proposed by Nitzl et al. (2016). Regarding the relationship between ELS streamers’ social image characteristics and purchase intention, consumer trust has a significant indirect effect. As shown in Table 5, for physical attractiveness, matchup congruence, expertise, authenticity, and responsiveness, the bootstrap 95% confidence interval (CI) do not include zero with p-values less than 0.05, indicating support for hypotheses H14a, H14b, H14c, H14d, H14e. However, consumer trust did not significantly mediate the impact of entertainment interaction created by ELS streamers on purchase intention. Therefore, H14f was rejected. In summary, the mediating role of consumer trust between the ELS streamers’ social image characteristics and purchase intention is supported, except for the feature of entertainment.

Table 5 Intermediary test of trust and purchase intention.
Full size table

Discussions and implications

Using the S-O-R paradigm, this study investigated how ELS streamers’ social image characteristics affect consumers’ purchase intention toward ELS in China. The study also examined the mediation effects of consumer trust. Altogether, ten of the thirteen hypotheses were supported by empirical data.

Generally, the majority of the conclusions are consistent with expectations. Specifically, the study found that the characteristics of social image created by ELS streamers have different effects on consumer trust and purchase intention. First, consistent with Eisend and Langner (2010), a physically attractive social image has a significant positive effect on consumer trust(H1), which in turn positively affect purchase intention(H2). In addition, in ELS, consumer trust mediates the relationship between the physical attractiveness of ELS streamers’ social image and purchase intention(H13a). The results are in line with expectations, and there may be two possible reasons for this. First, it’s hard to select a live streaming room among the vast number of diverse ELS programs, physical attractiveness helps build a good first impression which is critical to attracting viewers and keeping them engaged in the live broadcast room (Park and Lin, 2020). An ELS broadcast usually lasts for a long time, the viewers are more attracted by the good looks, humor, beautiful voice, and so on. In addition, these findings might differ due to other cultural contexts. For example, compared to Westerners, Easterners are more likely to have greater motivation to conform to the set of criteria considered beautiful in their society (Madan et al., 2018). Hence, Westerners with less emphasis on physical beauty might show a weaker influence of this factor.

Additionally, consistent with the previous studies (Parmar et al., 2020; Tong, 2017; Zhang et al., 2022), the study found that the social image of the right professional product endorser with matchup congruence, expertise, and authenticity have a significant effect on both consumer trust and purchase intention (H3, H4, H5, H6, H7, H8). Additionally, the mediating role of consumer trust between ELS streamers matchup congruence (expertise, authenticity) and purchase intention is supported (H14b, H14c, H14d). However, the three characteristics do not affect consumer trust and purchase intention equally. The effect of matchup congruence on consumer trust is stronger than those of expertise or authenticity. Moreover, matchup congruence exerts the strongest influence on purchase intention in ELS shopping. The results can be explained by social comparison theory, comparing oneself with others in order to evaluate or to enhance some aspects of the self (Festinger, 1954). Consumers may engage in social comparison with the ELS streamer, evaluating their own status, appearance, or lifestyle in relation to the streamers’ portrayed image. Negative social comparison to streamers in ELS positively correlates with impulsive buying behavior (Mundel et al., 2023). In this study, we took into account both the matchup between the ELS streamer and the products introduced, as well as the matchup between the ELS streamer and the viewers (McCormick, 2016; Malodia et al., 2017). The results also implicate that consistency between an ELS streamer and the viewer can evoke an emotional connection, which can develop a stronger emotional bond, leading to increased purchase intention (Chiu and Ho, 2023).

Furthermore, as expected, this study results confirmed the crucial role of expertise and authenticity in ELS. Shopping is the main motivation for watching live broadcasts in ELS, so the product itself is the most crucial aspect. ELS streamers with expertise can introduce products in detail, provide valuable advice, and recommend appropriate products to their viewers, which is of great importance in online shopping. Therefore, ELS streamers with expertise can help resolve consumer doubts and meet customer needs effectively (Ladhari et al., 2020). The results also be explained by social comparison theory. According to Markus and Nurius (1986), viewers compare their current selves with ELS streamers, idealized versions of what they might become, particularly those with perceived expertise, influencing viewers to mimic their consumption choices (Van Tran et al., (2023)). In our study, authenticity focuses on the originality of the product content delivered and the consistency of streamer personality in the ELS room. Thus, the authenticity of the streamer’s social image and live content can help viewers overcome their skepticism about a product and build trust with the streamer. Hence, our research results are in agreement with the prior studies (e.g., Becker et al., 2019). The authenticity of ELS streamer also leads to relational trust, which contributes to positive social proof. When consumers have positive experiences and perceive it as authentic, they are more likely to build trust and increase their purchase intention (Kim and Kim, 2021).

Finally, the study also examined the characteristics of interaction. The responsiveness feature of the ELS streamers’ social image positively affects consumer trust, consistent with Lee (2005). However, contrary to our expectation (Abd Aziz et al., 2015; Guo and Sun, 2022), our study found no statistically significant effect of responsiveness and entertainment feature on purchase intention, and no statistically significant effect of entertainment on consumer trust either. Our findings are not without precedent in existing research. Studies on web design shows that responsive web design positively affects users’ aesthetic evaluations and usability perceptions, its direct impact on purchase intentions was not significant when isolated from other factors, such as content quality and seamless checkout process (Thielsch et al., 2014). Furthermore, Zhu et al. (2023) found that perceived usefulness driven by information-task fit exerts a stronger effect than sociability, which encompasses interactive features like responsiveness. From a psychological perspective, if viewers in the ELS broadcast room are aware of the commercial intent behind entertainment programs, the persuasion of live-streaming shopping might trigger their persuasion knowledge which makes them take responses strategically and become less receptive to the promotions, reducing the significant impact on purchase intention (Chen et al., 2020). The results suggest that while responsiveness and entertainment might contribute to a positive viewing experience, they may not be the primary drivers of purchase decisions in ELS. Based on our analysis, we propose four potential explanations. First, due to the one-to-many nature of ELS, streamers may not be able to respond to every viewer’s comment or question and cannot interact deeply with each viewer. As such, viewers might prioritize other aspects over immediate interaction, such as like physical attractiveness, matchup, expertise and authenticity. Second, compared to traditional brick-and-mortar stores, in ELS, the primary focus is often on the product demonstration and the streamer’s presentation. Viewers can gather information and make decisions independently, relying less on real-time interaction with the streamer. Hence, responsiveness is not as important in ELS as it is in offline stores. Third, ELS live streaming may be perceived by viewers as a platform for leisure and enjoyment rather than a platform for purchasing products. Therefore, the effect of entertainment has no significant effect on purchase intention. Last but not least, from a regulatory perspective, transparency requirements on live streaming platforms regarding the disclosure of commercial activities, such as sponsorships, endorsements, and advertisements. These requirements enhance transparency and reduce the likelihood of misleading or deceptive practices, thus the persuasion effect of entertainment on purchase intention is limited.

Theoretical contribution

This study attempted to make several theoretical contributions to existing literature. First, previous studies on social image have explored how ELS streamers can create, maintain, or improve their social image to make them more acceptable to viewers (Wongsunopparat and Deng, 2021). Specifically, ELS streamers can build particular images by adjusting their behaviors. Nevertheless, few studies have investigated the effect of ELS streamers’ social image characteristics on purchase intention. This study offers novel insights into social image preferences by exploring the impact of social image characteristics on consumer trust and purchase intention in ELS shopping.

Second, unlike most prior research, which focused on the characteristics of live streamers or celebrities (e.g. Park and Lin, 2020; Zhang et al., 2022), our study distinguished between the external characteristics and the real characteristics and focus on the external characteristics of streamers’ social image. In which, a conceptual model with fourteen hypotheses base on the Stimulus-Organism-Response (SOR) framework is crafted through the lens of social image, dissecting it into three dimensions based on the intentional behavior of ELS streamers. Our results show that different ELS streamers’ image-building preferences have different impacts on consumer trust and purchase intention. Contrary to our expectations, purchase intention and consumer trust are not significantly affected by shaping the social image into an entertaining ELS streamer according to the results. Therefore, this paper consolidates the literature of live streamers characteristics in ELS, providing a comprehensive understanding of the influence of social image on consumers’ purchase intention.

Third, this study contributes to dramaturgical theory by offering empirical evidence and validating the claim that the social image cultivated by ELS streamers significantly influences consumer trust and purchase intention (Hidayat, Hidayat (2020)). Moreover, the study analyzed the detailed social image features. Based on the definition by Wongsunopparat and Deng (2021), three aspects of social image are described and several dimensions of each aspect are analyzed successively. Therefore, our results enhanced understanding of which aspect of social image factors contributes to consumer trust and influence buyers’ purchase intention.

Finally, this research added new evidence about the effects of factors on consumer trust and purchase intention in live streaming shopping, including physical attractiveness, matchup congruence, expertise, and authenticity. The study also provided new evidence about the mediating role of consumer trust, which is consistent with prior research (e.g. Chetioui et al., 2021). However, the study found different results from previous studies on the effects of responsiveness and entertainment on purchase intention and consumer trust, which is contrary to the research results by Rana et al. (2015) and Haile and Kang (2020).

Managerial implications

First, our paper contributes to improving the management of e-commerce companies that apply live streaming technology to their businesses. The study helps the managers realize the importance of the social image of a live streamer and the significance of finding a streamer with good matchup congruence between their personality and the products introduced in live streaming broadcasts. Care must be taken in designing and presenting social image characteristics of e-commerce live streamers. The study also highlights the importance of physical attractiveness which can help e-commerce live stream viewers build a positive first impression on streamers and induce greater persuasion leading to a positive attitude towards the related product. From an ethical standpoint, emphasizing physical attractiveness can lead to discrimination based on appearance in ELS streamers selection. There is an ethical responsibility for companies to hire streamers with a range of body types, ages, races, and other characteristics. The study also suggests the significance of strengthening training and guidance on product expertise to build the professionalism of ELS streamers’ social image.

Second, the results of this study can help e-commerce live streamers enhance their understanding of social image and adjust their behaviors to create, maintain, or improve their social image to make it more acceptable for ELS viewers. Matchup congruence is of great importance, so it is crucial to consider whether the nature of the selected products endorsed in ELS broadcasts can match the ELS streamers’ lifestyles or personality. The expertise and authenticity conveyed help build consumer trust. Thus, they can improve their social image by making themselves more professional, such as using high-quality equipment to present products, sharing their honest experience with viewers based on their actual use of products, and providing valuable product information or recommendation through relevant training. They also need to be aware that physical attractiveness is not only about their appearance but also their beautiful voice, humorous language style, appropriate dressing, and so on (Xu et al., 2020). While pursuing commercial interests is important, maintaining a positive social image as a live streamer is crucial. However, from an ethical standpoint, honesty is of utmost importance, and one should not resort to unscrupulous means to achieve their goals.

Finally, although watching ELS can be a form of entertainment that enables ELS streamers to reduce their boredom when recommending products, it is essential to be aware that the primary function of ELS is selling and shopping rather than entertainment. While appropriate entertainment can keep viewers engaged and bring pleasure, thrill, relaxation, however, it is not the most critical factor that affects consumer trust and purchase intention due to the limited effect. Instead, it is important to focus on creating high-quality content, building relationships with viewers, being transparent in your business practices.

Limitations and future research

A few limitations in this paper need to be clarified. First, as the present study focuses on Chinese consumers and their attitudes towards ELS shopping, the empirical data used were collected online and offline from customers of ELS platforms in China, which may limit our research findings. Due to the impact of policy and cultural differences, caution should be exercised when applying these findings to other contexts. Future research that investigates the impact of policy and cultural background on ELS in different countries is needed to provide a more comprehensive understanding of this phenomenon.

Second, the current study does not account for the full spectrum of ELS formats and situations. This study does not explore the possible moderating effects of product category. As a general study, this paper shows the important features of ELS streamers’ social image that positively affect consumers’ trust, which in turn affects their purchase intention. However, the effects of characteristics on purchase intention may vary under different product categories. Therefore, future research should explore the impact of these variations on the measurement and generalizability of findings. Moreover, the investigation is limited by the inherent difficulty of comprehensively tracking post-purchase behavior. Because of the difficulty in conducting a detailed purchase investigation, repeated purchase and return behaviors are not included in this study. Future studies can adopt new techniques and methods to research repeated purchase behaviors and return behaviors in ELS. In addition, since ELS is still in its early stage for practitioners in the industry, consumers’ response to may change as it develops, which could affect the research findings too.

Third, regulatory policies are considered as a background factor in the current study which focuses on understanding the direct relationships between social image characteristics, trust, and purchase intention. Within the context of a single country, the regulatory environment for live streaming tends to be homogeneous. Regulatory effects are discussed to interpret the research findings rather than being explicitly included in the conceptual model. However, it is important to acknowledge that regulatory policies can vary significantly across different countries and ELS development stages. This highlights a valuable direction for future research, which could explore the moderating or mediating effects of specific regulations on the relationships investigated in this study.

Fourth, e-commerce livestreaming platform characteristics are excluded in this study. Research suggests that factors like information-task fit, visual effects, and sociability significantly influence consumers’ online shopping intentions (Zhu et al., 2023; Chen et al., 2024). Moreover, platform-related factors, such as aesthetic and recommendation relevance, can differentially affect interaction and engagement (Chen et al., 2024). By not accounting for these platform features, our findings may not fully capture the complex interplay between platform characteristics, consumer trust, and purchase intention. Future research should consider incorporating cross-platform studies to investigate how the proposed model operates across different ELS platforms. Additionally, comparative studies examining the model’s results under varying platform characteristics (e.g., high vs. low recommendation relevance) could provide valuable insights into the moderating role of platforms in shaping consumer behavior.

Finally, price and discount factors were not included in this study. As a relatively new business model for practitioners to attract new customers, it is common to offer discounts or hot promotions during ELS. Hence, follow-up research may explore the possible moderating effects of product discount level. We hope that this study provides a foundation for understanding the effect of ELS streams’ social image features on consumer attitudes toward ELS shopping.