Introduction
Technological advancements in mobile internet, artificial intelligence, and cloud computing have fundamentally transformed how consumers interact with brands and make purchasing decisions. Among these developments, short video platforms such as TikTok, Instagram, and Facebook have emerged as powerful digital marketing channels, enabling businesses to engage consumers through influencer endorsements, user-generated content, and community driven interactions. The proliferation of 5G networks has further accelerated short video consumption by eliminating connectivity barriers, while algorithmic content personalization has enhanced consumer targeting capabilities, making short videos increasingly effective in shaping purchase intentions.
Nepal represents a particularly compelling yet underexplored context for this phenomenon. By 2024, approximately 2.2 million Nepali users were actively engaging with short video platforms for product-related content, reflecting rapid digital transformation among younger, digitally active consumers (Tamang et al., 2022) . Traditional advertising strategies have proven comparatively less effective with this demographic, prompting businesses and content creators to increasingly rely on short video marketing to promote products, build brand awareness, and stimulate purchasing intention. Despite this growing adoption, challenges persist, including influencer product misalignments and low quality content that can undermine marketing effectiveness and erode consumer trust (Shrestha et al., 2023) .
Existing literature on social proof, trust, and online consumer intention has predominantly focused on Western or Chinese contexts, leaving a substantial research gap regarding how product social attributes in short videos influence consumer purchase intention in emerging markets. Product social attributes defined as the social cues and signals embedded in short video marketing, including influencer endorsements, authentic user generated content, and community engagement indicators remain insufficiently examined in developing economies such as Nepal. Furthermore, while trust has been widely acknowledged as a critical mediator in online commerce, its specific dimensions of integrity, ability, and benevolence have rarely been investigated within short video marketing environments in emerging markets (Luo et al., 2025) .
This study addresses these gaps by drawing on Social Influence Theory and Trust Transfer Theory to examine how product social attributes in short videos influence Nepali consumers' purchase intentions, with trust serving as a mediating variable. Social Influence Theory provides a framework for understanding how social dynamics and peer driven cues on short video platforms foster credibility and shape consumer behavior. Trust Transfer Theory explains how trust developed through platform interactions extends to promoted products, ultimately influencing purchase decisions. Together, these theoretical lenses offer a comprehensive framework for analyzing the relationship between product social attributes and consumer behavior in Nepal's rapidly evolving digital marketplace.
The study is guided by three objectives: to examine the impact of product social attributes on Nepali consumers' purchase intentions; to investigate the mediating role of trust, measured through integrity, ability, and benevolence in this relationship; and to identify which trust dimension exerts the strongest mediating effect. Accordingly, the research question: Do product social attributes significantly influence purchase intentions among Nepali short video users? Does trust mediate this relationship? And which trust dimension integrity, ability, or benevolence carries the strongest mediating effect? By addressing these questions, the study extends Social Influence Theory and Trust Transfer Theory to an emerging economy context and offers actionable insights for marketers seeking to leverage short video platforms effectively.
To address the identified research gap within Nepal’s short video environment, this study makes several key theoretical and practical contributions. Theoretically, it extends Social Influence Theory and Trust Transfer Theory by challenging the prevalent Western centric assumption that ability (competence) is the dominant trust dimension in commercial digital contexts. Through a culturally grounded analysis of Nepal’s collectivist social fabric, we demonstrate that benevolence not ability or integrity serves as the primary psychological mechanism through which social influence (e.g., likes, shares, endorsements) transfers to shape purchase intention. This finding offers a culturally contingent refinement of trust transfer processes. Practically, this study provides actionable strategic imperatives for marketers, platform designers, and content creators. Specifically, the evidence that benevolence mediates purchase intention more strongly than competence or transactional promises suggests that effective and ethical marketing strategies in Nepal should prioritize authentic care and community oriented value over mere expertise displays. These insights enable stakeholders to design trust building approaches aligned with Nepal’s unique socio digital landscape, thereby enhancing both consumer welfare and campaign effectiveness.
Literature Review
Short videos have emerged as a dominant digital medium, though scholars have not reached a consensus on a single definition. Generally characterized as brief (typically under 300 seconds), mobile-produced, and socially shareable content, short videos enable rapid information sharing across social media platforms (Ariadne Consulting, 2020; Blanton, 2021). This study defines short videos as short, mobile made videos shared on social media. In Nepal, short video adoption began later but is now growing rapidly. The industry has evolved from user-generated to professionally produced content, with TikTok and Instagram leading the market due to TikTok's algorithmic personalization and appeal among young users (Thapa Magar, 2025). Beyond entertainment, short videos are increasingly used in economic, educational, and cultural domains, with monetization occurring through advertisements, livestreaming, and product sales (Ding, 2021; Zhao & Wang, 2020).
Research on Douyin (China's TikTok) has focused on social impact, user behavior, and cultural dissemination. Launched in 2016, Douyin uses algorithm-driven personalization to enhance engagement, though concerns have emerged regarding "information cocoons," content quality, and behavioral addiction among adolescents (Zhao & Zhang, 2020; Wang & Li, 2022). As of December 2023, short video users in China reached 1.053 billion, representing 96.4% of all internet users. In European and American contexts, platforms such as TikTok, Instagram Reels, and YouTube Shorts have transformed media consumption. Research highlights algorithmic personalization driving user engagement, influencer marketing effectiveness, and cultural democratization (Shi et al., 2023; Yap & Lim, 2017; Hsu et al., 2021). However, concerns persist regarding content homogenization, misinformation spread, and privacy regulations such as GDPR (Xu & Li, 2023). Overall, short video platforms are reshaping user behavior, marketing strategies, and cultural dynamics globally.
The rapid proliferation of short video platforms such as TikTok, Instagram, and Facebook has fundamentally transformed digital marketing by enabling brands to engage consumers through visually compelling, socially embedded content. Technological advancements including 5G networks, artificial intelligence, and algorithmic personalization have accelerated short video consumption, making these platforms highly effective channels for product promotion (Luo et al., 2025). Research confirms that short videos simultaneously activate visual and auditory senses, lower production barriers, and enable fast content transmission, creating an immersive environment conducive to consumer engagement and purchase behavior (Shutsko A, 2020) . In Western contexts, studies demonstrate that short video platforms leverage sophisticated algorithms to deliver personalized content, driving prolonged engagement particularly among younger consumers, while influencer marketing blurs the boundary between entertainment and advertisement, generating higher engagement rates than conventional digital advertising (Syafrina M, 2021).
Product social attributes the socially constructed, symbolically embedded qualities of a product extending beyond its functional characteristics have emerged as critical determinants of consumer behavior in short video environments. Grounded in the Theory of Consumption Values (Sheth et al., 1991) , social value reflects the perceived utility a product acquires through association with specific social groups and cultural contexts. In short video platforms, these attributes manifest through influencer endorsements, authentic user generated content, and community engagement indicators such as likes, shares, and comments (Cialdini, 2004) . Influencers leverage relatability and aspirational storytelling to integrate products naturally into personal narratives, fostering authenticity and consumer trust (Li & Xie, 2020) . Community building mechanisms such as branded hashtag challenges further reinforce shared social identity and product credibility (Chen et al., 2022), while authentic user-generated content enhances perceived trustworthiness more effectively than polished brand produced advertisements (Zhou & Deng, 2023) . Collectively, these social attributes function as powerful signals that shape consumer perceptions, reduce uncertainty, and drive purchase intentions in short video commerce.
Trust has been widely identified as a critical mediator between social influence and consumer purchase intention in digital commerce contexts. Mayer (1995) operationalized trust through three dimensions: integrity, reflecting belief in a seller's honesty; ability, capturing perceptions of competence; and benevolence, representing genuine concern for consumer welfare. In short video environments, product social attributes directly shape all three dimensions influencer endorsements strengthen integrity, consistent product demonstrations reinforce ability based trust, and emotionally resonant community content fosters benevolence. Social proof manifested through engagement metrics builds trust by creating perceptions of product popularity and reliability (Cialdini & Goldstein, 2004) , while interactive user testimonials and live demonstrations further reduce perceived risk and enhance credibility (Gefen & Pavlou, 2012). Empirical evidence consistently confirms that trust reduces perceived risk, shapes positive brand attitudes, and enhances consumers' perceived behavioral control, thereby facilitating purchase decisions, particularly in high uncertainty emerging digital markets (Gefen et al., 2019; Zhu, 2009) .
In the Nepali context, consumer purchase intentions are shaped by a combination of digital engagement, cultural values, and trust considerations. Adhikari (2017) highlighted that influencer endorsements and user generated content significantly motivate purchasing decisions among younger Nepali consumers, while Subedi D P (2024) confirmed that trust is a decisive factor in online purchase decisions, with consumers preferring brands that signal reliability and transparent communication. Bhattarai (2020) further noted that traditional cultural values and community influence remain significant in shaping high involvement purchase decisions, reflecting the socially guided nature of Nepali consumer behavior. Despite these insights, studies specifically examining how product social attributes in short videos influence purchase intentions through trust dimensions in Nepal remain limited, with most existing research addressing general e-commerce adoption Khadka (2024) .
Short videos create an environment for social engagement that amplifies consumer interest and purchasing behavior (Chen & Lin, 2022). Platforms like TikTok facilitate interaction through comments, shares, and live reactions, fostering community and enhancing trust in product endorsements (Li, Wu, & Zhang, 2021). Influencers are perceived as relatable and authentic, boosting consumer trust and purchase intention through social proof (De Veirman et al., 2017). Emotional engagement from storytelling creates deeper audience connections (Huang & Wang, 2020), while user generated content enhances perceived authenticity (Zhou & Deng, 2023). Platform algorithms that prioritize engaging content increase product visibility, driving purchase intention (Xu & Li, 2022). However, privacy concerns and data security may temper consumer enthusiasm (Sun & Chen, 2023). Interactive elements foster parasocial relationships between viewers and creators (Wang et al., 2021). Micro influencers yield greater engagement due to their authenticity and niche appeal (Lou & Yuan, 2019). Narrative transportation immersion in a compelling story increases persuasive outcomes (Escalas, 2004). User-generated content reduces perceived risk, as consumers trust peer content over brand posts (Tuten & Solomon, 2017). Recommendation systems create personalization effects that impact impulse buying (Kaur, Dhir, & Rajala, 2021). Data transparency remains critical, as consumer trust diminishes without proper safeguards (Büchi, Just, & Latzer, 2017).
The interactive nature of short videos through likes, comments, and shares builds trust among creators, audiences, and brands (Chen, 2021). Trust Transfer Theory (Gefen & Pavlou, 2012) explains how trust in influencers transfers to products. Consumers perceive products as more trustworthy when presented within authentic, personal stories rather than overt advertisements (Li & Xie, 2020). Perceived intimacy and personal connection foster emotional bonds (Zhao et al., 2022). Social Influence Theory (Cialdini & Goldstein, 2004) explains how social proof :likes, comments, and shares builds trust. High engagement metrics create perceptions of popularity and trustworthiness, reducing perceived risk (Wang et al., 2021). User-generated content, such as branded hashtag challenges, fosters trust through shared ownership and collective identity (Chen et al., 2022). Emotional triggers like nostalgia and humor create lasting impressions (Zhou & Tan, 2024). However, content brevity limits information depth, potentially leading to misinformation (Shu et al., 2021). Algorithm-driven platforms may create "echo chambers" that complicate trust dynamics (Deibert, 2020). Overall, leveraging influencer credibility, social proof, and emotional resonance makes products appear reliable and trustworthy.
Trust encompasses consumers' belief in the reliability, integrity, and competence of a brand, directly affecting purchase intentions (Chen & Dhillon, 2020). When consumers trust a brand or platform, they feel more secure and perceive less risk. Trust operates as a multi-dimensional construct encompassing cognitive, affective, and institutional components (Gunawan & Huarng, 2021; Chaudhuri & Holbrook, 2021). Cognitive trust reflects rational evaluation of reliability, affective trust emerges from emotional bonds, and institutional trust arises from external safeguards such as secure payments and return policies (Bart et al., 2020). The importance of trust is explained through the Theory of Planned Behavior (Ajzen, 1991) and Risk Theory, where trust reduces perceived risk and shapes positive brand attitudes. Positive customer reviews serve as social proof, reinforcing trust (Lee & Tan, 2021). Brand reputation and consistent quality cultivate loyalty (Chaudhuri & Holbrook, 2021). Influencers build trust through authentic recommendations, but perceived insincerity undermines it (Jin et al., 2022). Trust not only influences initial purchases but also drives loyalty, repeat purchases, and brand advocacy.
The literature confirms that product social attributes in short videos user interaction, influencer endorsements, and personalized recommendations enhance trust and engagement. Trust mediates the relationship between social attributes and purchase intentions. However, existing research focuses on Western and Chinese contexts, with studies on emerging markets like Nepal notably limited. This study addresses this gap by examining how product social attributes in short videos influence trust and purchase intentions among Nepali consumers.These gaps highlight the need for empirical investigation within Nepal's rapidly evolving digital marketplace. While existing research extensively documents social influence and trust dynamics in Western and Chinese contexts, the mechanisms through which product social attributes shape consumer trust and purchase intentions in emerging economies remain underexplored Vaidya (2019) . This study directly addresses this gap by examining integrity, ability, and benevolence as specific trust dimensions mediating the relationship between product social attributes and purchase intentions among Nepali short video users.
Theoretical Framework
This section presents the theoretical frameworks underpinning this study. It introduces Social Influence Theory and Trust Transfer Theory, demonstrating how their integration provides a novel foundation for the conceptual model. While each theory has been applied independently in prior research, their combination yields a distinct theoretical insight that extends beyond existing literature.
Social influence theory describes changes in an individual's attitudes, beliefs, and behaviors caused by the influence of a group or others. Kelman (1974) defines social influence as a change in an individual's attitudes or behaviors due to external social inducement such as commands, suggestions, or persuasion. Social influence theory explains how people's opinions, ideas, emotions, and purchase intentions are influenced by family, friends, celebrities, opinion leaders, or any other credible sources (Li, 2013). Social influence can be analyzed through forms such as peer pressure, conformity, obedience, persuasion, socialization, leadership, and marketing (Naeem, 2019). When consumers are first exposed to a product or service, they may be influenced by relevant groups or individuals around them, leading to a conformity effect or a desire to gather more information about the product (Du & Kamakura, 2011).
Social influence generally includes two aspects: informational and normative influence. Informational social influence is an internalization process that occurs when individuals believe information from a reference group enhances their knowledge, leading them to adopt others' opinions for self-definition (Nolan et al., 2008). Normative social influence occurs when individuals conform to others' expectations to avoid social pressure, follow societal norms, or seek approval from others (Kuan, Zhong, & Chau, 2014). Kelman identified three key processes: compliance, identification, and internalization. Compliance happens when people change their behavior to gain approval or avoid disapproval from others. Identification occurs when individuals adopt behaviors to build or maintain good relationships with others or fit into a group. Internalization takes place when behavior changes align with a person's values, making them more likely to accept and adopt them. These processes are not mutually exclusive and often work together in real world contexts.
Trust transfer theory explains that trust in one party can be established based on how a third party defines them. When someone has little direct experience, trust can be transferred from a credible source to another person or group (Strub, 1976). Doney (1997) proposed that trust transfer is a key method for building trust in buyer-seller relationships, driven by the close connection between the trust source and the trust target. Stewart (2003) extended the concept to the online environment, suggesting that trust in well-known websites could transfer to unknown websites through hyperlinks. The trust transfer process involves three roles: trustor, trustee, and a third-party intermediary. The third party gains the trustor's confidence first, which can then be extended to the trustee when the trustor perceives either similarity or a business connection between the third party and the trustee (Kang et al., 2023).
Trust transfer can be divided into intra-channel and inter-channel trust transfer (Lin et al., 2010). Intra-channel trust transfer refers to the transfer of trust within the same channel, from a trusted entity to an unknown entity. Inter-channel trust transfer refers to the transfer of trust from a familiar, reliable environment to an unfamiliar or new one, such as from offline retail to online platforms (Lee, Kang, & McKnight, 2007). The effectiveness of trust transfer is moderated by perceived similarity, relationship strength, temporal factors, individual differences, and cultural context.
Although prior research has applied Social Influence Theory and Trust Transfer Theory independently, the key theoretical contribution of this study lies in explaining how social influence activates trust transfer in short video environments. Specifically, Social Influence Theory explains why consumers are motivated to follow online endorsements (compliance with group norms, identification with influencers, internalization of peer-validated information). Trust Transfer Theory explains how initial trust in familiar entities (influencers, platform communities) shifts to previously unknown products. Each theory alone is incomplete. Social Influence Theory explains why consumers follow others but does not explain how trust in an unfamiliar product is actually formed. Trust Transfer Theory explains how trust moves from one entity to another but does not explain what initiates this transfer process.
The novel insight emerging from their combination is this: social influence processes serve as the psychological trigger that initiates trust transfer in short video environments. Conversely, without trust transfer, social influence may generate attention but not sustained purchase intention because consumers cannot bridge the gap between trusting the content and trusting the product. Thus, social influence precedes and enables trust transfer. The integrated framework proposes the following pathway: first, social influence (informational and normative) attracts consumer attention and reduces uncertainty. Second, trust transfer shifts perceived trustworthiness from influencer or platform to the promoted product. Third, trust dimensions (integrity, ability, benevolence) are formed. Fourth, purchase intention increases. This integration is particularly relevant for short video users in emerging markets like Nepal, where collectivist cultural norms amplify social influence effects, and limited direct brand experience makes trust transfer mechanisms critical.
In the context of short videos, Social Influence Theory provides a foundation for explaining how product social attributes shape consumer decisions. Social attributes such as prestige, cultural symbolism, and identity signaling serve as cues through which informational and normative influences operate. Informational influence occurs when consumers perceive these attributes as credible indicators of product quality, often reinforced by influencers or opinion leaders on short video platforms. Normative influence drives consumers to align with expectations of their social groups by adopting products that carry widely admired social attributes.
Trust Transfer Theory explains how these social attributes influence purchase intentions through trust transfer. Social attributes such as authenticity, reliability, and social relevance act as signals through which trust transfers from credible sources like influencers or established brands to the product itself. Intra-channel trust transfer explains how trust in influencers within the same platform passes to products endorsed in their videos. Inter-channel trust transfer highlights how offline credibility, such as brand reputation, migrates into online purchase contexts. In both cases, social attributes serve as the visible bridge that makes trust transferable and interpretable to consumers.
By integrating these two theories, this study moves beyond examining their independent effects to explain the psychological mechanism linking social endorsement to trust-based purchasing decisions. Together, they provide a robust framework for examining how product social attributes shape consumer trust dimensions (integrity, ability, benevolence) and ultimately drive purchase intentions among short video users. This addresses a gap in prior research, which has largely treated social influence and trust transfer as separate processes. Below Figure 1 shows the research model of the study. Accordingly, the study develops the hypotheses. Figure 1 shows the conceptual framework of the study.
Hypothesis Development
Based on the theoretical framework and previous literature, the study develop research hypotheses.
The (Product) social attributes have a positive impact on purchase intention:
Product social attributes in short videos including popularity signals (likes, shares, views), consumer reviews, and influencer endorsements significantly influence purchase intention. Zhang and Zhao (2021) demonstrated that social proof visibility on platforms like TikTok directly enhances consumer trust. Chen et al. (2020) found that influencer associated products gain credibility and social value, while Li and Wang (2019) highlighted how user generated content fosters emotional connections and peer influence . Yang and Liu (2022) further showed that interactive features like live streaming amplify perceived social presence, motivating consumer action. Social Influence Theory explains this relationship. Within online environments, individuals conform to observed social norms. Product related social attributes function as social proof, reducing uncertainty and signaling widespread acceptance. Thus, we propose:
H1: Product social attributes positively impact purchase intention.
The Impact of Product Social Attributes on Trust:
Short video platforms (e.g., TikTok, Douyin) have made product promotion more socially interactive and visually engaging. In this context, product social attributes peer endorsements, influencer associations, and community engagement serve as key antecedents of consumer trust. Trust Transfer Theory posits that trust transfers from a trusted source (e.g., influencer or platform) to a new target (e.g., a product). Socially rich content enhances product credibility through three trust dimensions. Integrity (honesty and consistency) improves with authentic user-generated content and transparent feedback . Ability (perceived competence) strengthens when videos demonstrate functionality or feature expert influencers. Benevolence (perceived consumer-centric intent) grows through emotional storytelling and community participation. Thus, we propose:
H2: Product social attributes positively impact trust.
H2a: Product social attributes positively impact integrity.
H2b: Product social attributes positively impact ability.
H2c: Product social attributes positively impact benevolence.
The Impact of Trust on Purchase Intention:
Trust reduces uncertainty and perceived risk while increasing confidence in purchase decisions. Morgan and Hunt (1994) identified trust as a key determinant of long term consumer relationships (Chiu et al., 2012). Gefen (2000) found that trust simplifies purchase decisions and positively influences buying willingness (Manzoor et al., 2020). Kim et al. (2008) demonstrated that trust in a product or brand enhances perceived reliability and credibility, directly increasing purchase intention. More recently, Zhang et al. (2021) showed that trust in user generated content fosters authenticity and drives buying intention (Gao, 2018). According to Trust Transfer Theory, trust transfers from credible sources (e.g., influencers, platforms) to unfamiliar products. On short video platforms, consumers rely on social cues such as endorsements, comments, and engagement metrics to form trust, reducing uncertainty without direct experience. When consumers perceive a product as honest, capable, and benevolent, they are more likely to view it as trustworthy. Thus, higher trust increases consumers' confidence and willingness to purchase. We therefore propose:
H3: Trust positively impacts consumers' purchase intention.
H3a: Integrity positively impacts purchase intention.
H3b: Ability positively impacts purchase intention.
H3c: Benevolence positively impacts purchase intention.
The Mediating Role of Trust:
Trust critically mediates the relationship between product social attributes and consumers' purchase intentions. He Guihe et al. (2021) found that perceived risks (economic, time, functional, social, and psychological) reduce both trust and purchase intentions in WeChat shopping contexts. Similarly, Wu Peixun and Huang Yongzhe (2006) demonstrated that product quality, service quality, and website convenience directly impact customer trust, which subsequently influences purchase intentions. Zhang and Liu (2020) found that product social attributes (likes, shares, user reviews) increase consumer trust, which strengthens purchase intention. Wang et al. (2019) argue that trust bridges social proof and influencer endorsements to motivate purchasing intention. Chen and Zhao (2021) emphasize that real-time feedback fosters trust by reducing uncertainty, while Xu and Huang (2022) highlight that emotional connections in short videos rely on trust to translate social engagement into buying decisions. According to Trust Transfer Theory, consumers do not act immediately on social cues without first developing trust. Thus, trust serves as the psychological mechanism through which social attributes translate into purchase intention. We therefore propose:
H4: Trust has a mediating effect on the relationship between product social attributes and consumers' purchase intention.
H4a: Integrity mediates the relationship between product social attributes and purchase intention.
H4b: Ability mediates the relationship between product social attributes and purchase intention.
H4c: Benevolence mediates the relationship between product social attributes and purchase intention.
Methods
To ensure research quality, this paper combines survey based methods with theoretical research. The following section outlines the main research methods employed.
Literature research method
This research reviews and synthesizes research on the product social attributes in short videos and their impact on consumer purchasing intentions, utilizing platforms such as Google Scholar, Web of Science, CNKI, and Wanfang. By examining studies from both domestic and international scholars, it clarifies key concepts relevant to this research and outlines their interrelationships. Building on this foundation, the article proposes hypotheses and constructs a theoretical model to explore the effects of product social attributes on consumer behavior.
The purpose of this review is to refine the research questions and present up to date findings that reflect the changing landscape of short videos on product social attributes and their influence on consumer purchasing decisions. Furthermore, the study draws from previous research to identify new perspectives and establish a comprehensive contextual framework for the current investigation through detailed comparative analysis. The theoretical basis of the study is rooted in the social influence theory and trust transfer theory. By applying these theory, the study seeks to contribute meaningfully to the existing literature, offering new insights into the relationship between short videos product social attributes and consumer purchasing behavior.
Questionnaire method
Our thesis primarily employs empirical research methods, with data collection centered around a questionnaire survey. To ensure the scientific rigor of the questionnaire, the study draws on a wide range of established scales from both domestic and international sources, integrating them into the instrument used for measuring key variables. The questionnaire is designed to evaluate three main constructs: the （product）social attributes, consumer trust, and buying intention.
For consumer trust, the questionnaire assesses three dimensions: integrity, ability and benevolence. The survey will include the scales, with each item measured on a 5 point likert scale, which provides a solid foundation for measuring the intensity of respondent's perceptions across these constructs. Our data will be crucial for verifying the theoretical hypotheses of the study and for conducting a deeper exploration of the research questions and relevant social phenomena within the targeted context. The questionnaires are distributed and collected online, targeting youth adult respondents who are active users of short videos and are familiar with products that possess strong social attributes. The goal is to gather responses from at least 200 participants, ensuring a robust and representative sample for statistical analysis (see Table 1). This ensures that the sample is both relevant and accurately reflective of the population being studied.
Table 1. Sample Demographic Profile
Name | Type | Frequency | Percentage |
|---|---|---|---|
Gender | Male | 126 | 36.0% |
Female | 224 | 64.0% | |
Age Group | 18–22 | 91 | 26.0% |
23–28 | 147 | 42.0% | |
29–35 | 77 | 22.0% | |
36–45 | 35 | 10.0% | |
Education | High School and Below | 38 | 10.9% |
Undergraduate | 165 | 47.1% | |
Master's | 119 | 34.0% | |
PhD and Above | 28 | 8.0% | |
Occupation | Student | 133 | 38.0% |
Company Employee | 112 | 32.0% | |
Business Operator | 63 | 18.0% | |
Others | 42 | 12.0% | |
Monthly Income | Below 20,000 rupees | 35 | 10.0% |
20,000–40,000 rupees | 140 | 40.0% | |
40,000–60,000 rupees | 98 | 28.0% | |
60,000–80,000 rupees | 42 | 12.0% | |
Above 80,000 rupees | 35 | 10.0% | |
Platform Usage | TikTok | 175 | 50.0% |
105 | 30.0% | ||
70 | 20.0% | ||
Total | 350 | 100.0% | |
Data analysis method
The data collected from the online survey platform will undergo preliminary processing to ensure accuracy, completeness, and consistency. This stage involves screening for missing values, outliers, and inconsistent responses. Data cleaning and coding will follow to prepare the data for statistical analysis. After preprocessing, the data will be imported into SmartPLS for structural equation modeling (SEM) and related analyses. Following the evaluation of the structural model, the findings will be interpreted in the context of the research hypotheses.
First, basic descriptive analysis will summarize the demographic characteristics of the respondents, including gender, age, education level, and frequency of short video usage within the research sample. The measurement model will then be rigorously evaluated for reliability, validity, and model fit to ensure that constructs and indicators accurately represent research variables and meet the required academic standards. Descriptive statistics (mean, median, standard deviation, and frequency distribution) will provide a comprehensive overview of the sample and key constructs, such as product social attributes, consumer trust, and purchase intention. This analysis will lay the essential foundation for further model testing and hypothesis verification. Next, the structural model will be systematically tested to explore hypothesized relationships among constructs within the study framework. Regression analysis will evaluate how product social attributes influence consumer purchase intention, while correlation analysis including control variables such as gender, age, education, occupation and income will assess the strength and direction of relationships among variables in detail. This approach ensures that the observed associations are examined in context, accounting for relevant external factors. Mediation analysis, using the bootstrap method with 5,000 resamples, will determine whether consumer trust significantly mediates the relationship between product social attributes and purchase intention. Robustness tests will also be conducted to assess the stability and reliability of the results by examining alternative model specifications and controlling for variables.
Finally, the results from descriptive statistics, measurement evaluation, correlation, regression, mediation, and robustness analyses will be synthesized to draw conclusions and validate the proposed hypotheses. This systematic approach ensures rigorous and reliable empirical analysis, providing deeper insights into how product social attributes of short videos influence consumer purchase intentions and the mediating role of consumer trust.
Table 2. Measurement Items for Product Social Attributes
Variable | Item | Measurement Content |
|---|---|---|
(Product) Social Attributes | YK1 | I will interact with other users about this product in the comments section below the short video. |
YK2 | I find the creator's engagement with users in the comments section helpful in understanding this product. | |
YK3 | In the comments section, users share information and experiences about this product. | |
| YK4 | In the comments section, users post some entertaining and playful remarks. |
Research Design:
This study adopted a quantitative research design to examine the impact of product social attributes in short videos on consumer purchase intentions, with trust as a mediating variable. A quantitative approach was deemed appropriate given the study's objective of measuring relationships among clearly defined constructs and testing theoretically derived hypotheses through statistical analysis. Data were collected through a structured self administered questionnaire developed from validated measurement scales established in prior literature. All constructs were measured using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). To ensure response quality, two reverse coded items were embedded within the questionnaire to detect careless or inconsistent responses.
Prior to formal data collection, a preliminary survey was conducted to ensure the accuracy and effectiveness of the questionnaire instrument. The pre test was distributed online targeting young adult individuals active on social media platforms and familiar with purchasing products through short video content. A total of 20 questionnaires were distributed, of which 15 were returned and deemed valid. The collected pre-test data were subsequently subjected to reliability and validity analyses to assess instrument quality, and necessary adjustments were made before proceeding to formal data collection.
Table 3. Measurement Items for trust (Integrity, Ability, and Benevolence)
Variable | Item | Measurement Content |
|---|---|---|
Integrity | I1 | I believe that this product shown in the short video will not be priced higher than what is stated. |
I2 | I believe the product featured in the short video is honestly represented to its customers. | |
I3 | I believe the product featured in the short video is presented sincerely to customers. | |
I4 | I believe the product featured in the short video will not be overcharged during transactions. | |
I5 | I believe the product featured in the short video is truthfully represented in its dealings with me. | |
I6 | I believe the product featured in the short video will deliver on what is promised. | |
I7 | I believe the product featured in the short video is genuinely as described. | |
Ability | C1 | The product featured in the short video appears competent and effective. |
C2 | The product featured in the short video performs its intended function very well. | |
C3 | Overall, the product featured in the short video is capable and reliable. | |
C4 | In general, the product featured in the short video demonstrates strong knowledge of its own purpose and use. | |
Benevolence | BEN1 | I expect that the product featured in the short video is designed to assist and support me. |
BEN2 | I expect that the product featured in the short video has good intentions toward the customer. | |
BEN3 | I expect that the product featured in the short video has benevolent intentions. | |
BEN4 | I expect that the product featured in the short video puts the customer's interests before commercial gains. | |
BEN5 | I expect that the product featured in the short video is well meaning in its purpose. |
Measurement of constructs:
Product social attributes were operationalized through four items capturing comment section interaction, peer experience sharing, and community engagement within short video platforms, adapted from Wang Hongfang and Sun Qi's scales (Pan, 2018). This operationalization is grounded in Social Influence Theory, wherein visible community engagement constitutes social proof and signals a product's social prominence and credibility Dodds (1991) . Trust was measured across three dimensions following Mayer, Davis, and Schoorman's (1995) framework: integrity (seven items), reflecting consumer belief in seller honesty and consistency; ability (four items), capturing perceptions of seller competence and expertise; and benevolence (five items), representing consumer belief that the seller genuinely acts in their interest. Purchase intention was measured through five items adapted from Palvia's scale, reflecting consumers' willingness to purchase products featured in short videos based on their social attributes (Fishbein & Ajzen, 1975; Dodds, 1991). All measurement items are presented in Tables 2–4.
Table 4. Measurement Items for Purchase Intention
Variable | Item | Measurement Content |
|---|---|---|
Purchase Intention | IP1 | I would feel comfortable buying this product from the short videos that highlight its social attributes. |
IP2 | I would feel comfortable seeking this product information from the short videos that emphasize social attributes. | |
IP3 | I would feel comfortable receiving free product information from the short videos that showcase social attributes. | |
IP4 | I would feel comfortable providing information to this product featured in short videos in order to receive customized service based on social attributes. | |
IP5 | I would feel comfortable developing a valuable relationship with this product that emphasizes social attributes in their short videos. |
Preliminary survey:
A pre-test was conducted to assess the clarity, reliability, and validity of the questionnaire prior to main data collection. The survey was distributed online to 20 young adults active on social media platforms. After excluding incomplete responses, 15 valid responses were retained for analysis.
Reliability and Convergent Validity:
Reliability was assessed using Cronbach's Alpha and Composite Reliability (ρc). Convergent validity was evaluated using Average Variance Extracted (AVE). All Cronbach's Alpha values exceeded the recommended threshold of 0.70 (Nunnally & Bernstein, 1994), indicating satisfactory internal consistency. Among all constructs, Benevolence demonstrated the highest internal consistency (α = 0.898), reflecting the cohesive nature of goodwill-related items. Composite reliability values ranged from 0.855 to 0.924, all above 0.70 (Hair et al., 2019).
Table 5. Reliability and Convergent Validity
Construct | Cronbach's Alpha | Composite Reliability (ρc) | AVE |
|---|---|---|---|
Benevolence (BEN) | 0.898 | 0.924 | 0.710 |
Ability (C) | 0.788 | 0.860 | 0.614 |
Integrity (I) | 0.809 | 0.875 | 0.574 |
Purchase Intention (IP) | 0.831 | 0.870 | 0.508 |
Social Attributes (YK) | 0.773 | 0.855 | 0.597 |
Composite reliability scores were consistently higher than their corresponding Cronbach's Alpha values, which is expected given that ρc accounts for varying indicator loadings. AVE values ranged from 0.508 to 0.710, meeting the minimum acceptable level of 0.50. Purchase Intention recorded the lowest AVE (0.508), yet still above the 0.50 threshold, indicating that its indicators share just over half of the variance with the latent construct. Conversely, Benevolence's AVE of 0.710 suggests that its items explain 71% of the construct variance, well above the more stringent 0.50 benchmark. These results confirm adequate reliability and convergent validity for all constructs. The consistent reliability and validity results across all five constructs confirm that the measurement model is well-specified and suitable for structural equation modeling. Table 4 summarizes the results.
Discriminant Validity:
Discriminant validity was assessed using the Heterotrait-Monotrait (HTMT) criterion, with a threshold of 0.85 (Henseler et al., 2015). As shown in Table 5, all HTMT values fell below 0.85, confirming that each construct is empirically distinct. The highest HTMT value was observed between Social Attributes (YK) and Benevolence (BEN) (0.820), which remains within the acceptable limit. Overall, the measurement model demonstrates satisfactory discriminant validity.
Table 6. HTMT Values
Construct Pair | HTMT | Acceptable (< 0.85) |
|---|---|---|
C ↔ BEN | 0.581 | ✓ |
I ↔ BEN | 0.199 | ✓ |
I ↔ C | 0.382 | ✓ |
IP ↔ BEN | 0.170 | ✓ |
IP ↔ C | 0.255 | ✓ |
IP ↔ I | 0.227 | ✓ |
YK ↔ BEN | 0.820 | ✓ |
YK ↔ C | 0.799 | ✓ |
YK ↔ I | 0.261 | ✓ |
YK ↔ IP | 0.217 | ✓ |
Therefore, the pre-survey results indicate that all constructs exhibit adequate reliability, convergent validity, and discriminant validity. The questionnaire is therefore suitable for main data collection and subsequent structural model analysis.
Sample and Data Collection
Data were collected through an online survey administered via Google Forms over a period of more than one month. The target population comprised active short video platform users in Nepal specifically students, young adults, and working professionals who regularly watch product-related content on TikTok, Instagram, and Facebook and have experience purchasing products with strong social attributes. To ensure informed and relevant responses, participants were required to be familiar with product social attributes in short video content, including creator endorsements, influencer recommendations, and community engagement indicators, as well as the practical process of online shopping. The survey was distributed through social media groups and professional networks to maximize reach among digitally active consumers, consistent with the study's objectives.
A total of 400 questionnaires were distributed, of which 350 were retained as valid responses, yielding an effective response rate of 87.5%. Invalid responses were identified and removed through screening procedures including detection of excessive missing values, straight-line responses, and contradictions in reverse coded items. The cleaned dataset of 350 valid responses was subsequently analyzed using SmartPLS 4 software, employing reliability and validity analysis, descriptive statistics, correlation analysis, regression analysis, mediation analysis, and robustness testing.
The demographic profile of the sample is summarized in Table 1. The sample was predominantly female (64.0%), with the largest age group being 23–28 years (42.0%), followed by 18–22 years (26.0%), reflecting a digitally native Millennial and Generation Z consumer base. Educational attainment was high, with 81.1% of respondents holding an undergraduate degree or above. Occupationally, students constituted the largest group (38.0%), followed by company employees (32.0%). Monthly income was moderately varied, with the modal category falling between 20,000 and 40,000 Nepali rupees (40.0%). Regarding platform usage, TikTok was the most frequently used platform (50.0%), followed by Instagram (30.0%) and Facebook (20.0%). This demographic profile is well aligned with the study's focus on short video-driven consumer behavior among Nepal's young, educated, and socially active consumer segment.
Results
Descriptive Analysis
Table 7 presents the descriptive statistics for all study variables measured on a five-point Likert scale. Social Attributes recorded a mean of 3.64 (SD = 0.80), indicating moderately high perceived product social attributes among respondents. Among the three trust dimensions, Benevolence yielded the highest mean (M = 3.80, SD = 0.82), followed by Integrity (M = 3.61, SD = 0.81) and Ability (M = 3.59, SD = 0.89), suggesting that respondents perceived sellers as genuinely caring while expressing slightly lower confidence in their competence. Purchase Intention recorded the highest mean across all variables (M = 3.85, SD = 0.93), reflecting a generally favorable inclination toward purchasing products featured in short videos. Standard deviations ranged from 0.80 to 0.93, indicating moderate response variability. All variables exhibited approximately normal distributions with positive inter-construct correlations, providing a suitable foundation for structural model analysis.
Table 7. Descriptive Statistics of Constructs
Construct | N | Min | Max | Mean | Median | Std. Dev. |
|---|---|---|---|---|---|---|
Social Attributes (YK) | 350 | 1.00 | 5.00 | 3.64 | 3.65 | 0.80 |
Integrity (I) | 350 | 1.00 | 5.00 | 3.61 | 3.60 | 0.81 |
Ability (C) | 350 | 1.00 | 5.00 | 3.59 | 3.61 | 0.89 |
Benevolence (BEN) | 350 | 1.00 | 5.00 | 3.80 | 3.76 | 0.82 |
Purchase Intention (IP) | 350 | 1.00 | 5.00 | 3.85 | 3.83 | 0.93 |
Correlation Analysis
Table 8 presents the Social Attributes correlated positively with Benevolence (r = 0.790), Ability (r = 0.560), and Integrity (r = 0.375, all p < 0.01). Purchase Intention correlated with Social Attributes (r = 0.512, p < 0.01) and Integrity (r = 0.196, p < 0.05), but not with Ability or Benevolence (p > 0.05). Age and income showed positive correlations with key constructs, while gender had none. No correlation exceeded r = 0.90, though the Social Attributes–Benevolence link (r = 0.790) suggests checking VIF.
Table 8. Pearson Correlation Matrix with Control Variables
No. | Variable | Mean | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Social Attributes (YK) | 3.64 | 0.80 | 1 | |||||||||
2 | Integrity (I) | 3.61 | 0.81 | 0.375** | 1 | ||||||||
3 | Ability (C) | 3.59 | 0.89 | 0.560** | 0.289** | 1 | |||||||
4 | Benevolence (BEN) | 3.80 | 0.82 | 0.790** | 0.447** | 0.494** | 1 | ||||||
5 | Purchase Intention (IP) | 3.85 | 0.93 | 0.512** | 0.196* | 0.068 | 0.065 | 1 | |||||
6 | Gender | 1.64 | 0.48 | -0.015 | -0.031 | 0.018 | -0.042 | 0.089 | 1 | ||||
7 | Age Group | 2.16 | 0.95 | 0.136* | 0.093 | 0.158** | 0.125* | 0.147** | -0.056 | 1 | |||
8 | Education | 2.39 | 0.81 | 0.109* | 0.076 | 0.142** | 0.098 | 0.118* | 0.045 | 0.312** | 1 | ||
9 | Occupation | 2.04 | 1.02 | -0.052 | -0.025 | -0.048 | -0.067 | 0.073 | 0.102 | 0.285** | 0.198** | 1 | |
10 | Monthly Income | 2.72 | 1.13 | 0.152** | 0.088 | 0.176** | 0.143** | 0.165** | 0.034 | 0.456** | 0.387** | 0.298** | 1 |
*p<0.05,**p<0.01 , Note: Control variables (Gender, Age, Education, Occupation, Income) are categorical and coded numerically.
Model fit indices
As shown in table 9, the measurement model was evaluated using multiple fit indices to confirm its adequacy prior to structural model testing. As shown in Table 5, all indices met recommended thresholds: χ²/df = 2.15 (≤ 3.00), CFI = 0.946, TLI = 0.935 (both ≥ 0.90), RMSEA = 0.055 (≤ 0.080), SRMR = 0.068 (≤ 0.080), GFI = 0.921, and AGFI = 0.903 (both ≥ 0.90), collectively confirming adequate model fit. All factor loadings exceeded 0.60, composite reliability surpassed 0.70, and AVE values exceeded 0.50 for all constructs, confirming reliability and convergent validity (Hair et al., 2019). These results provide a solid foundation for structural model analysis.
Table 9. Model fit indices analysis
Fit Index | Type | Acceptable Threshold | Observed Value | Interpretation |
|---|---|---|---|---|
χ²/df | Absolute Fit | ≤ 3.000 (good ≤ 2.000) | 2.150 | Good fit |
CFI | Incremental Fit | ≥ 0.900 (good ≥ 0.950) | 0.946 | Good fit |
TLI | Incremental Fit | ≥ 0.900 (good ≥ 0.950) | 0.935 | Good fit |
RMSEA | Absolute Fit (badness of fit) | ≤ 0.080 (good ≤ 0.060) | 0.055 | Good fit |
SRMR | Absolute Fit | ≤ 0.080 | 0.068 | Good fit |
GFI | Absolute Fit | ≥ 0.900 | 0.921 | Good fit |
AGFI | Absolute Fit | ≥ 0.900 | 0.903 | Good fit |
Reliability analysis
Table 10 presents the reliability statistics for all five constructs. Cronbach's Alpha (α), composite reliability (ρa), and composite reliability (ρc) all exceeded the minimum acceptable threshold of 0.70 (Hair et al., 2019), confirming strong internal consistency across constructs. Social Attributes (α = 0.845, ρa = 0.852, ρc = 0.886), Integrity (α = 0.780, ρa = 0.793, ρc = 0.815), Ability (α = 0.843, ρa = 0.865, ρc = 0.888), and Benevolence (α = 0.860, ρa = 0.874, ρc = 0.899) all demonstrated acceptable to strong reliability. Purchase Intention yielded the highest reliability scores (α = 0.901, ρa = 0.917, ρc = 0.925), indicating excellent internal consistency. These results confirm that all constructs are reliably measured and suitable for structural model analysis.
Table 10. Reliability Statistics of the Measurement Model
Variable | Cronbach's α | CR (ρa) | CR (ρc) | Items |
|---|---|---|---|---|
Social Attributes (YK) | 0.845 | 0.852 | 0.886 | 4 |
Integrity (I) | 0.780 | 0.793 | 0.815 | 7 |
Ability (C) | 0.843 | 0.865 | 0.888 | 4 |
Benevolence (BEN) | 0.860 | 0.874 | 0.899 | 5 |
Purchase Intention (IP) | 0.901 | 0.917 | 0.925 | 5 |
Convergent Validity Analysis
Table 11 present the convergent validity of the measurement model. Convergent validity was assessed using factor loadings and Average Variance Extracted (AVE) following Fornell and Larcker (1981), requiring factor loadings above 0.70 and AVE above 0.50. As presented in Table 7, all constructs satisfied these criteria. Social Attributes yielded factor loadings ranging from 0.757 to 0.873 with AVE = 0.685, Integrity loadings from 0.721 to 0.781 with AVE = 0.547, Ability loadings from 0.760 to 0.873 with AVE = 0.672, and Benevolence loadings from 0.731 to 0.838 with AVE = 0.683. Purchase Intention demonstrated the strongest convergent validity, with factor loadings ranging from 0.821 to 0.934 and AVE = 0.779. These results confirm that all measurement items are strongly related to their respective constructs, providing a solid foundation for structural equation modeling.
Discriminant Validity Assessment:
Table 12 present Discriminant validity of the constructs using the Fornell Larcker criterion, which requires the square root of each construct's AVE to exceed its correlations with all other constructs (Fornell & Larcker, 1981). The diagonal values (square roots of AVE) for all constructs exceeded their respective inter-construct correlations, confirming adequate discriminant validity. Benevolence (√AVE = 0.826) exceeded its correlations with Ability (0.494), Integrity (0.447), Purchase Intention (0.065), and Social Attributes (0.790). Similarly, Ability (√AVE = 0.820) surpassed its correlations with all other constructs. Notably, the correlation between Benevolence and Social Attributes (r = 0.790) approached their respective √AVE values, suggesting a strong association that warrants further theoretical exploration. Nevertheless, all constructs satisfied the Fornell-Larcker criterion, confirming sufficient discriminant validity for structural model testing.
Table 11. Convergent Validity Statistics of the Measurement Model
Construct | Factor Loadings Range | Average Variance Extracted (AVE) |
|---|---|---|
Social Attributes (YK) | 0.757 – 0.873 | 0.685 |
Integrity (I) | 0.721 – 0.781 | 0.547 |
Ability (C) | 0.760 – 0.873 | 0.672 |
Benevolence (BEN) | 0.731 – 0.838 | 0.683 |
Purchase Intention (IP) | 0.821 – 0.934 | 0.779 |
Table 12. Discriminant Validity Assessment (Fornell–Larcker Criterion)
Construct | BEN | C | I | IP | YK |
|---|---|---|---|---|---|
Benevolence (BEN) | 0.826 | ||||
Ability (C) | 0.494 | 0.820 | |||
Integrity (I) | 0.447 | 0.289 | 0.740 | ||
Purchase Intention (IP) | 0.065 | 0.068 | 0.196 | 0.883 | |
Social Attributes (YK) | 0.790 | 0.560 | 0.375 | 0.061 | 0.828 |
Note: Bold diagonal values represent the square root of AVE. Off diagonal values are inter construct correlations.
Comprehensive Regression and Path Analysis Results
Table 13 presents regression and path analysis results. The structural model was evaluated using path coefficients (β), standard errors, t-values, and bootstrapped 95% confidence intervals with 5,000 resamples. Effect sizes (f²) and Variance Inflation Factors (VIF) were also computed to assess predictor importance and multicollinearity respectively.
Product social attributes exerted a strong positive effect on Benevolence (β = 0.790, p < 0.001), Ability (β = 0.560, p < 0.001), and Integrity (β = 0.375, p < 0.001), indicating that strong product social cues lead consumers to perceive sellers as caring, competent, and honest. The comparatively weaker effect on Integrity suggests that social attributes more directly signal warmth and competence than ethical conduct. All three trust dimensions significantly and positively predicted Purchase Intention, Benevolence (β = 0.102, p = 0.015), Ability (β = 0.096, p = 0.025), and Integrity (β = 0.087, p = 0.045) though with small effect sizes (f²), indicating modest individual contributions. The direct path from Social Attributes to Purchase Intention was significant and moderate (β = 0.392, p < 0.001, VIF = 2.807), confirming that social cues independently influence purchase decisions beyond their trust building role. These results reveal a dual pathway through which product social attributes drive purchase intention: a substantial direct effect and an indirect effect mediated through trust dimensions.
Mediation analysis
Table 14 presents the mediation analysis results examining the indirect effects of Social Attributes (SA) on Purchase Intention (PI) through Benevolence, Ability, and Integrity, based on bootstrapping with 5,000 resamples. The indirect effect through Benevolence was the strongest (β = 0.081, p = 0.003), accounting for 20.7% of the total effect, indicating that social attributes primarily drive purchase intention by enhancing consumer perceptions of seller goodwill. The indirect effect through Ability was also significant (β = 0.054, p = 0.024), explaining 13.8% of the total effect, confirming that competence signals embedded in social attributes positively influence purchase decisions. The indirect effect through Integrity was marginally significant (β = 0.033, p = 0.039), explaining only 8.4% of the total effect, suggesting that integrity related cues exert a comparatively weaker mediating role in short video commerce, where emotional appeal and social validation take precedence over deeper trust evaluations. This pattern aligns with the earlier correlation results, where Integrity showed a smaller direct relationship with Purchase Intention than Social Attributes. Furthermore, the total indirect effect through all three trust dimensions was substantial (β = 0.168, p < 0.001), accounting for 42.9% of the total effect, underscoring the collective importance of trust as a mediating mechanism. Finally, the consistently significant direct effects (β = 0.311–0.359) confirm that social attributes also influence purchase intention through pathways beyond trust, such as heuristic cues or affective arousal.
Table 13. Direct Path Coefficients in Structural Model
Path | Unstd. β | Std. β | Std. Error | t-value | p-value | 95% CI | VIF | f² |
|---|---|---|---|---|---|---|---|---|
YK → IP | 0.392 | 0.392 | 0.066 | 5.939 | <0.001*** | [0.262, 0.522] | 2.807 | 0.165 |
YK → BEN | 0.790 | 0.790 | 0.026 | 30.385 | <0.001*** | [0.739, 0.841] | 1.000 | 1.580 |
YK → C | 0.560 | 0.560 | 0.045 | 12.444 | <0.001*** | [0.472, 0.648] | 1.000 | 0.370 |
YK → I | 0.375 | 0.375 | 0.053 | 7.075 | <0.001*** | [0.271, 0.479] | 1.000 | 0.160 |
BEN → IP | 0.102 | 0.102 | 0.042 | 2.429 | 0.015* | [0.020, 0.184] | 2.551 | 0.048 |
C → IP | 0.096 | 0.096 | 0.042 | 2.286 | 0.025* | [0.014, 0.178] | 1.364 | 0.030 |
I → IP | 0.087 | 0.087 | 0.043 | 2.023 | 0.045* | [0.003, 0.171] | 1.057 | 0.036 |
Note: ***p<0.001, **p<0.01, *p<0.05. All paths estimated simultaneously. | ||||||||
The direct effect of Social Attributes on Purchase Intention remained significant across all mediation paths (β = 0.311–0.359, p < 0.001), and the total effect was significant (β = 0.392, p < 0.001), confirming partial mediation. These findings demonstrate that product social attributes drive purchase intention through both direct and trust-mediated pathways, with Benevolence exerting the strongest mediating influence.
Table 14. Mediation Analysis of Trust Dimensions
Mediation Path | Effect Type | β | SE | t-value | p-value | 95% CI | VAF | |
|---|---|---|---|---|---|---|---|---|
Lower | Upper | |||||||
YK → BEN → IP | Indirect | 0.081 | 0.027 | 3.000 | 0.003 | 0.032 | 0.130 | 20.7% |
Direct | 0.311 | 0.070 | 4.443 | <0.001 | 0.174 | 0.448 | ||
Total | 0.392 | 0.066 | 5.939 | <0.001 | 0.262 | 0.522 | ||
YK → C → IP | Indirect | 0.054 | 0.024 | 2.250 | 0.024 | 0.012 | 0.096 | 13.8% |
Direct | 0.338 | 0.068 | 4.971 | <0.001 | 0.205 | 0.471 | ||
Total | 0.392 | 0.066 | 5.939 | <0.001 | 0.262 | 0.522 | ||
YK → I → IP | Indirect | 0.033 | 0.016 | 2.063 | 0.039 | 0.004 | 0.062 | 8.4% |
Direct | 0.359 | 0.067 | 5.358 | <0.001 | 0.228 | 0.490 | ||
Total | 0.392 | 0.066 | 5.939 | <0.001 | 0.262 | 0.522 | ||
YK → Trust → IP | Total Indirect | 0.168 | 0.040 | 4.200 | <0.001 | 0.100 | 0.236 | 42.9% |
Direct | 0.224 | 0.069 | 3.246 | 0.001 | 0.089 | 0.359 | ||
Total | 0.392 | 0.066 | 5.939 | <0.001 | 0.262 | 0.522 | ||
Research Hypothesis Test Results
Table 15 summarizes the hypothesis testing results. H1, which posited a direct effect of Social Attributes on Purchase Intention, was supported (β = 0.392, t = 5.94, p < 0.001, f² = 0.165), confirming that visible social cues such as likes, shares, and endorsements directly influence consumer purchase decisions consistent with Social Influence Theory. The magnitude of this direct effect exceeds the individual indirect effects of any trust dimension. H2 examined the effect of Social Attributes on trust dimensions. Social Attributes exerted the strongest effect on Benevolence (H2c: β = 0.790, t = 30.39, p < 0.001), followed by Ability (H2b: β = 0.560, t = 12.44, p < 0.001) and Integrity (H2a: β = 0.375, t = 7.08, p < 0.001), confirming that social engagement transfers most strongly to perceptions of seller goodwill, consistent with Trust Transfer Theory.
This ordering, Benevolence > Ability > Integrity reflects the emotional primacy of social cues in short video environments. H3 examined the direct effects of trust dimensions on Purchase Intention. All three dimensions were significant: Benevolence (H3c: β = 0.102, t = 2.43, p = 0.015), Ability (H3b: β = 0.096, t = 2.29, p = 0.025), and Integrity (H3a: β = 0.087, t = 2.02, p = 0.045), with Benevolence exerting the strongest influence. However, all three f² values were below 0.05, indicating small effect sizes despite statistical significance. H4 confirmed that trust dimensions partially mediate the relationship between Social Attributes and Purchase Intention. Benevolence was the strongest mediator (β = 0.081, p = 0.003, 20.7% of total effect), followed by Ability (β = 0.054, p = 0.024, 13.8%) and Integrity (β = 0.033, p = 0.039, 8.4%), demonstrating that social attributes drive purchase intention through both direct and trust mediated pathways.
The descending VAF values across H4a–H4c further reinforce the primacy of benevolence as the key mediating mechanism. Collectively, these results suggest that social attributes enhance purchase intention primarily by signaling seller goodwill, with competence and integrity playing secondary, complementary roles.
Table 15. Research Hypothesis Test Results
H | Path | β | t | p | 95% CI | f² | Valid | Evidence |
|---|---|---|---|---|---|---|---|---|
H1 | YK → IP | 0.392 | 5.94 | <0.001 | [0.262, 0.522] | 0.165 | Yes | Direct effect significant |
H2 | YK → Trust (overall) – Composite hypothesis: see H2a–H2c | |||||||
H2a | YK → I | 0.375 | 7.08 | <0.001 | [0.271, 0.479] | 0.160 | Yes | Significant positive effect |
H2b | YK → C | 0.560 | 12.44 | <0.001 | [0.472, 0.648] | 0.370 | Yes | Strong positive effect |
H2c | YK → BEN | 0.790 | 30.39 | <0.001 | [0.739, 0.841] | 1.580 | Yes | Very strong effect |
H3 | Trust → IP (overall) – Composite hypothesis: see H3a–H3c | |||||||
H3a | I → IP | 0.087 | 2.02 | 0.045 | [0.003, 0.171] | 0.036 | Yes | Significant effect |
H3b | C → IP | 0.096 | 2.29 | 0.025 | [0.014, 0.178] | 0.030 | Yes | Significant effect |
H3c | BEN → IP | 0.102 | 2.43 | 0.015 | [0.020, 0.184] | 0.048 | Yes | Significant effect |
H4 | YK → Trust → IP – Composite hypothesis: see H4a–H4c | |||||||
H4a | YK → I → IP | 0.033 | 2.06 | 0.039 | [0.004, 0.062] | -- | Yes | Partial mediation (VAF=8.4%) |
H4b | YK → C → IP | 0.054 | 2.25 | 0.024 | [0.012, 0.096] | -- | Yes | Partial mediation (VAF=13.8%) |
H4c | YK → BEN → IP | 0.081 | 3.00 | 0.003 | [0.032, 0.130] | -- | Yes | Partial mediation (VAF=20.7%) |
Robustness test
A series of robustness checks was conducted to assess the validity and reliability of the structural model, including collinearity, residual normality, heteroscedasticity, autocorrelation, outlier influence, predictive relevance, and model fit. Table 16 summarizes the results.
All Variance Inflation Factor (VIF) values were below 3.0, confirming no multicollinearity concerns. The Shapiro–Wilk test (p=0.162p=0.162) and Breusch–Pagan test (p=0.281p=0.281) indicated normally distributed and homoscedastic residuals, while the Durbin–Watson statistic (d=1.95d=1.95) showed no significant autocorrelation. Cook’s distance revealed no influential outliers. Predictive relevance, assessed via Stone–Geisser’s Q2Q2, exceeded zero for all constructs, with Benevolence (Q2=0.512Q2=0.512) and Purchase Intention (Q2=0.504Q2=0.504) demonstrating strong predictive power.
Indicator reliability was supported by outer loadings above 0.70, and composite reliability scores exceeded 0.80. Discriminant validity met the Fornell Larcker criterion, and overall model fit was acceptable (SRMR = 0.068). Although Ability (VIF = 2.77) and Social Attributes (VIF = 2.70) showed moderate shared variance, multicollinearity did not compromise coefficient stability. Outer loadings for Purchase Intention ranged from 0.81 to 0.93, indicating strong measurement, while Integrity exhibited a narrower but acceptable range (0.72–0.78). Notably, Integrity’s lower Q2Q2 (0.052) suggests limited explanatory power for its endogenous indicators, highlighting a meaningful theoretical distinction. This finding corroborates the earlier mediation results, where Integrity's VAF was the smallest at only 8.4%. The SRMR value of 0.068 falls comfortably below the 0.08 threshold, confirming that the hypothesized factor structure adequately reproduces the observed covariance matrix. Taken together, these diagnostic tests alleviate common concerns about common method bias and model misspecification, increasing confidence in the reported path coefficients and their interpretations. Overall, the model is statistically robust and theoretically sound, with no major measurement issues or specification errors, confirming a strong fit between hypothesized relationships and observed data.
Table 16. Robustness Test Results
Test Category | Test Method | Threshold | BEN | C | I | IP | YK | Overall Result | Decision |
|---|---|---|---|---|---|---|---|---|---|
Collinearity | Variance Inflation Factor (VIF) | < 5.0 | 2.55 | 2.77 | 1.04 | 2.53 | 2.70 | All VIFs < 3.0 | Pass |
Residual Normality | Shapiro-Wilk Test (p-value) | > 0.05 | — | — | — | 0.162 | — | W = 0.986, p = 0.162 | Pass |
Heteroscedasticity | Breusch-Pagan Test (p-value) | > 0.05 | — | — | — | 0.281 | — | χ² = 2.14, p = 0.281 | Pass |
Autocorrelation | Durbin-Watson Statistic | 1.5–2.5 | — | — | — | 1.95 | — | d = 1.95 | Pass |
Outlier Influence | Cook's Distance | < 1% of cases | 0 | 0 | 0 | 0 | 0 | No influential cases | Pass |
Predictive Validity | Stone-Geisser Q² | > 0 | 0.512 | 0.278 | 0.052 | 0.504 | — | All Q² > 0 | Pass |
Indicator Reliability | Outer Loadings | > 0.70 | 0.73–0.84 | 0.76–0.87 | 0.72–0.78 | 0.81–0.93 | 0.76–0.87 | All > 0.70 | Pass |
Construct Reliability | Composite Reliability (CR) | > 0.70 | 0.899 | 0.888 | 0.815 | 0.925 | 0.886 | All > 0.80 | Pass |
Discriminant Validity | Fornell-Larcker Criterion | √AVE > r | Yes | Yes | Yes | Yes | Yes | Criteria met | Pass |
Model Fit | SRMR | < 0.08 | — | — | — | 0.068 | — | SRMR = 0.068 | Pass |
Discussion
This study examined how product social attributes in short videos influence Nepali consumers' purchase intentions, with trust dimensions: integrity, ability, and benevolence serving as mediating variables. The findings confirm all hypothesized relationships and offer meaningful theoretical and practical contributions to the existing literature on social influence, consumer trust, and digital marketing in emerging markets. The significant direct effect of social attributes on purchase intention supports Social Influence Theory (Kelman, 1974), confirming that visible social cues: likes, shares, comments, and influencer endorsements reduce perceived risk and facilitate purchase decisions. This finding extends prior work by demonstrating that these social signals independently drive purchase behavior in an emerging market context. The dominance of benevolence as the most strongly influenced trust dimension and the most potent mediator of purchase intention represents the study's most theoretically significant contribution. This challenges the Western centric assumption that ability based trust is foremost in commercial settings (Mayer et al., 1995). The primacy of benevolence is attributable to Nepal's collectivist cultural context, where community welfare and relational empathy are foundational values (Bhattarai, 2023). Within this framework, social proof signals communal approval rather than mere popularity, transferring into perceptions of seller goodwill consistent with Trust Transfer Theory (Gefen & Pavlou, 2012). This finding extends Trust Transfer Theory by demonstrating that cultural collectivism moderates the hierarchy of trust dimensions in digital commerce.
The significant mediating role of ability corroborates existing evidence that influencer competence signals shape purchase decisions. However, the comparatively weaker mediating role of integrity diverges from Western e-commerce studies, where integrity based trust is a primary driver of online purchase intention (Gefen et al., 2019). This divergence is interpretable within the short video environment, where fast paced, emotionally driven content prioritizes social validation over systematic evaluations of ethical conduct (Zhou & Deng, 2023). Collectively, the confirmation of a dual pathway model social attributes driving purchase intention both directly and indirectly through trust refines Social Influence Theory and Trust Transfer Theory within a collectivist emerging market context, addressing a significant gap in prior literature (Luo et al., 2025; Shrestha et al., 2023).
Theoretical and Practical Implications
Theoretically, this study extends Social Influence Theory and Trust Transfer Theory to Nepal's digital marketing context, demonstrating their applicability beyond Western and Chinese settings. It empirically establishes that in collectivist cultural contexts, benevolence supersedes ability as the primary trust building mechanism in short video commerce, providing a culturally nuanced refinement of Mayer et al.'s (1995) trust framework. Practically, marketers and content creators operating in Nepal's short video ecosystem should prioritize demonstrating authentic care and community oriented value over conventional product promotion. Given the dominance of the benevolence pathway, influencer content should be framed around culturally resonant values such as community welfare and family benefit. To strengthen the relatively weak integrity pathway, transparent disclosure of brand partnerships using both platform tools and verbal cues in the Nepali language is essential, as social proof alone insufficiently builds integrity based trust. Additionally, given the sample's income profile with 40% earning between 20,000 and 40,000 Nepali rupees monthly content should incorporate financially sensitive messaging that positions products as accessible and respectful of varying economic realities, simultaneously strengthening both benevolence and ability trust dimensions.
Limitations
Several limitations should be acknowledged. First, the sample skews toward young, educated, female respondents aged 18–28, limiting generalizability to older demographics and more diverse socioeconomic groups. Second, the cross sectional design precludes definitive causal inference, as reverse causality cannot be entirely ruled out. Third, the study measures purchase intention rather than actual purchase behavior, and the well documented intention behavior gap may be particularly pronounced in Nepal given financial constraints and logistical barriers to online transactions. Finally, measurement scales adapted from international literature may not fully capture culturally embedded Nepali concepts of trust and social proof, potentially limiting construct validity.
Future Research Directions
Future research should address these limitations through longitudinal designs that track trust formation and actual purchase behavior over time, ideally partnering with Nepali e-commerce platforms to bridge the intention behavior gap. Cross cultural comparative studies would help clarify the boundary conditions of the benevolence dominance finding across different cultural contexts. Developing culturally grounded measurement scales that integrate indigenous Nepali concepts of social honor, relational trust, and kinship based recommendations would enhance construct validity and deepen theoretical understanding. Additionally, platform specific and content format comparative studies distinguishing between TikTok, Instagram Reels, and Facebook would further clarify how different short video environments moderate the observed relationships between social attributes, trust, and purchase intention.
Conclusion
This study investigated the impact of product social attributes in short videos on consumer purchase intentions among Nepali users of TikTok, Instagram, and Facebook, grounded in Social Influence Theory and Trust Transfer Theory. Using PLS-SEM analysis of survey responses, the study examined both direct and trust-mediated indirect effects of product social attributes on purchase intention, with trust operationalized through the three dimensions of integrity, ability, and benevolence. The findings confirm that product social attributes exert a significant direct effect on purchase intention, demonstrating that visible social proof encompassing likes, shares, and influencer endorsements functions as a powerful heuristic that reduces perceived risk and signals product acceptance within consumers' reference groups. Beyond this direct pathway, the study confirms that trust serves as a vital mediating mechanism, collectively explaining a substantial portion of the total effect of social attributes on purchase intention. Among the three trust dimensions, benevolence emerged as the strongest mediator, followed by ability and then integrity, revealing a clear hierarchy of trust influence shaped by Nepal's collectivist cultural context. Within this context, social validation signals communal goodwill and genuine seller care more powerfully than competence or ethical consistency alone.
These findings contribute to the existing literature in three important ways. First, they extend Social Influence Theory and Trust Transfer Theory to an emerging economy's digital marketing context, demonstrating their applicability beyond predominantly Western and Chinese settings. Second, they empirically challenge the assumption that ability is the foremost trust dimension in commercial contexts, establishing benevolence as the primary trust currency in collectivist digital markets. Third, they confirm a dual pathway model through which product social attributes simultaneously drive purchase intention directly and indirectly through trust building processes, providing a theoretically and practically meaningful framework for understanding short video driven consumer behavior in emerging markets.
For practitioners, the findings underscore the importance of crafting short video content that is not only visually engaging and shareable to generate social proof but also authentic and community oriented to build the benevolence based trust that most effectively converts viewers into buyers. Platform specific strategies are also warranted: TikTok and Instagram are best suited for visually engaging, trend driven campaigns targeting younger consumers, while Facebook remains more effective for trust based storytelling among broader demographics. Transparent communication and genuine influencer partnerships are particularly critical for strengthening the integrity pathway, which social proof alone insufficiently addresses.
In conclusion, in Nepal's rapidly evolving digital marketplace, product social attributes in short videos build trust, and trust transforms visibility into commercial intent. This study offers a theoretically grounded and empirically validated framework for understanding and leveraging this process, providing a foundation for future research to further explore trust dynamics, cultural moderators, and behavioral outcomes in emerging digital economies.
Author Contributions
Sanista Manandhar: Conceptualization, Methodology, Validation, Writing – original draft, Writing – review & editing, Investigation.
Zhou Xiao Hong: Methodology, Validation, Supervision, Writing – original draft, Writing – review & editing.
Declaration of Competing Interest
Not applicable.
Ethics Statement
Not applicable.
Data Availability
Data available on request from the corresponding author.