Frontiers in Business, Economics and Management ISSN: 2766-824X | Vol. 19, No. 1, 2025 67 Research on the Impact of Online Celebrity Live- Streaming of Goods on Vietnamese Consumers’ Purchase Intention Thi Thuy Tien Vo School of Economic Management, Chongqing University of Posts and Telecommunication, Chongqing 400065, China Abstract: This study concentrates on the Vietnamese market, taking Vietnamese live-streaming e-commerce consumers as its research subjects, with the objective of deeply exploring the impact mechanism of influencer live-streaming e-commerce on their purchase intentions. Drawing upon the classic Stimulus-Response Theory (specifically, the Stimulus-Organism-Response or SOR model) and Matching Theory, the research constructs a comprehensive theoretical model incorporating direct effects, mediating effects, and moderating effects. Within this model, the key stimulus factors in the influencer live-streaming process are defined as influencer influence (encompassing social capital like popularity and fan base size), professionalism (reflecting cognitive capital such as product knowledge and recommendation skills), and interactivity (representing social capital like real- time communication and responsiveness). Consumers' perceived trust is conceptualized as the core organismic state variable, potentially playing a dual role: mediating the impact of influencer characteristics on purchase intention, and moderating the strength of this relationship. Concurrently, acknowledging the differential impact of product attributes on consumer decision - making paths, this study introduces product type (search goods vs. experience goods) as another significant moderating variable. The final behavioral response focuses on consumers' purchase intention. To validate the proposed theoretical model and research hypotheses, this study employed a quantitative research methodology. A structured questionnaire was designed based on a thorough literature review of relevant theories and existing research, adapting established scales from domestic and international studies to the Vietnamese context. The questionnaire covered multiple dimensions, including influencer influence, professionalism, interactivity, perceived trust, purchase intention, and demographic information. Data collection was conducted between September and November 2023, primarily in Ho Chi Minh City and Hanoi, Vietnam, using a combination of online (social media, email) and offline (public places) channels. A total of 500 valid responses were obtained. Data analysis was performed using statistical software such as SPSS 26.0 and AMOS (or similar tools like the PROCESS macro), involving descriptive statistics, reliability and validity tests (Cronbach's Alpha, Exploratory Factor Analysis - EFA), correlation analysis, multiple linear regression analysis, mediation effect testing using the Bootstrap method, and moderation effect testing via hierarchical regression analysis. The empirical results reveal the following key findings: (1) Influencer influence, professionalism, and interactivity all exert significant positive direct effects on Vietnamese consumers' purchase intentions, with influencer influence demonstrating the most prominent impact. (2) Perceived trust plays a significant partial mediating role in the relationships between influencer influence, professionalism, interactivity, and purchase intention; these three influencer characteristics effectively enhance consumer perceived trust, which, in turn, directly and positively drives purchase intention. (3) Perceived trust functions not only as a mediator but also as a significant positive moderator in the relationships between influencer characteristics and purchase intention, meaning higher levels of trust amplify the positive effects of influencer traits. (4) Product type significantly moderates the relationship between influencer characteristics and purchase intention, specifically: influencer influence and professionalism have a stronger positive impact on purchase intention for search goods, whereas interactivity exerts a stronger positive influence on purchase intention for experience goods. Keywords: Influencer Live-streaming E-commerce, Purchase Intention, Perceived Trust, Influencer Influence, Professionalism, Interactivity, Vietnamese Consumers, Stimulus-Organism-Response (SOR) Theory. 1. Introduction In recent years, Vietnamese e-commerce platforms have continuously innovated and created to bring consumers the best shopping experience. One of the solutions recently launched by e-commerce platforms is the live broadcast sales function. Especially after the Covid-19 pandemic in 2020, the level of live shopping has been pushed to a new level, and Internet celebrities have become increasingly important on the Internet. In the process of watching the sales live broadcast, the audience can go from understanding the product to confirming the purchase in a very short time, which helps to significantly shorten the customer's purchase time. Through the promotion of the network anchor, not only can the product sales be increased, but also the communication with fans can be strengthened. Since web hosting is an industry that has only become popular in recent years, domestic and foreign scholars have conducted a lot of discussions on this. Therefore, this paper starts from the sales environment of celebrity network anchors to explore which factors affect live broadcasts. The role of these factors is analyzed by streaming and discussion. 2. Literature Review and Research Model Based on the above theoretical framework and literature review, this study decomposes the influencing factors of live streaming sales by influencers into three core independent variables: influence, professionalism, and interactivity, and introduces perceived trust as a moderating variable and 68 purchase intention as a dependent variable. The following is a detailed definition of each variable: Hypothesis H1a: In the live streaming sales model of influencers, the influence of influencers has a positive impact on the purchase intention of Vietnamese consumers. Hypothesis H1b: In the live streaming sales model of influencers, the influence of influencers has a positive impact on the perceived trust of Vietnamese consumers. Hypothesis H2a: In the live streaming sales model of influencers, the professionalism of influencers has a positive impact on the purchase intention of Vietnamese consumers. Hypothesis H2b: In the live streaming sales model of influencers, the professionalism of influencers has a positive impact on the perceived trust of Vietnamese consumers. Hypothesis H3a: In the live streaming sales model of influencers, the interactivity of influencers has a positive impact on the purchase intention of Vietnamese consumers. Hypothesis H3b: In the live streaming sales model of influencers, the interactivity of influencers has a positive impact on the perceived trust of Vietnamese consumers. Hypothesis H4a: Perceived trust has a positive impact on Vietnamese consumers’ purchase intention Figure 1. Research model diagram 3. Research Methodology In order to ensure the scientificity and systematicness of the research, this paper adopts a combination of multiple research methods, mainly including the following three methods: (1) Literature research method The literature research method is the basic method of this study. Through the systematic review of relevant domestic and foreign literature, the theoretical framework and empirical results in the fields of live e-commerce, Internet celebrity economy and consumer purchase intention are obtained. This paper retrieved literature related to live streaming and consumer behavior from academic databases including CNKI, Web of Science, Scopus, etc., and focused on analyzing the research results on the development of e- commerce and consumer behavior characteristics in the Vietnamese market in recent years. Through the literature review, the shortcomings of existing research are clarified, and the basis for the construction of theoretical models and the formulation of hypotheses in this study is provided. In addition, the literature research method is also used to compare the differences in consumer behavior between the Vietnamese market and other countries and regions, providing a cross-cultural perspective for the formulation of research hypotheses. (2) Questionnaire survey method The questionnaire survey method is the core method for obtaining data in this study, which is used to collect behavioral and attitude data of Vietnamese consumers in the live streaming shopping scenario. This study uses a structured questionnaire and designs measurement items based on mature scales at home and abroad (such as Chen & Lin, 2018; Lin et al., 2017), covering variables such as influence of Internet celebrities, professionalism, interactivity, perceived trust and purchase intention. The questionnaire uses a 5-point Likert Scale, with scores ranging from "strongly disagree" to "strongly agree". To ensure the representativeness of the data, the questionnaire was distributed through a combination of online and offline methods, covering consumer groups in Ho Chi Minh City, Hanoi and other places in Vietnam, and finally collected no less than 500 valid samples. (3) Empirical analysis method Empirical analysis method is used to verify research hypotheses and models, mainly including descriptive statistical analysis, reliability and validity test, exploratory factor analysis (EFA), confirmatory factor analysis (CFA) and structural equation model (SEM) analysis. Specifically: Descriptive statistical analysis: Use SPSS software to describe the basic characteristics of sample data, including the distribution of demographic variables such as gender, age, and income. Reliability and validity test: Use Cronbach’s Alpha coefficient to test the reliability of the scale, and combine KMO test and Bartlett sphericity test to verify the structural validity of the scale to ensure the reliability of the measurement tool. Exploratory Factor Analysis (EFA): Screen out significant factor structures and verify whether the dimensional division of variables meets theoretical expectations. Structural Equation Model (SEM): Use AMOS software to build a path model to test the direct impact of influence, professionalism, and interactivity of influencers on purchase intention, as well as the moderating effect of perceived trust. Multiple regression analysis: Further verify the causal relationship between variables and analyze the differences in the impact of various factors under different product types (search-based and experience-based products). Through the comprehensive application of the above methods, this study aims to comprehensively explore the impact of influencers' live streaming on Vietnamese consumers' purchase intention from theory to empirical research. 4. Research Results The data collected from 500 survey participants, after being analysed by SPSS and AMOS software, were statistically described by the results of the sample and variable as shown in Table 1 and Table 2. 69 Table 1. Sample statistics feature category frequency % gender male 213 42.6% female 287 57.4% Age under 18 73 14.6% 18-25 210 42.0% 26-30 111 22.2% 31-40 69 13.8% over 40 37 7.4% Income (VND/month) 5.000.000 67 13.4% 5.000.000 - 10.000.000 233 46.6% 10.000.000 - 20.000.000 135 27.0% over 20.000.000 65 13.0% Learning High school 39 7.8% Intermediate 79 15.8% Bachelor 288 57.6% Master 57 11.4% Other 37 7.4% total 500 100.0% Source: Author SPSS data processing The reliability analysis results are shown in Table 2. The influence of Internet celebrities (DL) scale contains 9 items, with a Cronbach’s Alpha coefficient of 0.900 and a CITC value ranging from 0.545 to 0.772. The professionalism (KM) scale contains 5 items, with a Cronbach’s Alpha coefficient of 0.854 and a CITC value ranging from 0.529 to 0.769. The interactivity (TT) scale finally uses 7 items (the original TT_8 was excluded after preliminary analysis), with a Cronbach’s Alpha coefficient of 0.925 and a CITC value ranging from 0.674 to 0.883. The perceived trust (PT) scale contains 3 items, with a Cronbach’s Alpha coefficient of 0.862 and a CITC value ranging from 0.715 to 0.758. The purchase intention (YDM) scale contains 5 items, with a Cronbach’s Alpha coefficient of 0.804 and a CITC value ranging from 0.472 to 0.713. The Cronbach’s Alpha coefficients of all variables were significantly higher than 0.8, indicating that each scale had high internal consistency and reliability. At the same time, the CITC values of all items were greater than 0.4, and the Alpha coefficients after deleting any items were not significantly higher than the Alpha coefficients of the original scales, indicating that the currently retained items were necessary and effective for measuring their respective constructs. These results fully prove that the measurement tools used in this study have good reliability, laying a solid foundation for subsequent validity analysis and hypothesis testing. Table 2. Cronbach’s Alpha coefficient table of each variable Variable Scale means if item delete d Scale varianc e if item deleted Corrected item-total correlatio n Cronbach's Alpha if item deleted interactivity (TT, 7items) Cronbach’s Alpha = 0.925 TT_1 22.91 23.935 0.723 0.920 TT_2 23.33 21.407 0.789 0.914 TT_3 23.09 21.523 0.883 0.904 TT_4 23.14 21.734 0.781 0.914 TT_5 23.08 22.779 0.783 0.914 TT_6 23.15 22.299 0.770 0.915 TT_7 23.18 22.555 0.674 0.925 Influence of Internet Celebrities (DL, 9 items) Cronbach’s Alpha = 0.900 DL_1 31.02 24.256 0.718 0.884 DL_2 31.17 23.361 0.699 0.886 DL_3 31.22 23.664 0.772 0.880 DL_4 31.19 24.618 0.691 0.887 DL_5 31.28 24.206 0.707 0.885 DL_6 30.92 25.643 0.572 0.895 DL_7 30.97 24.604 0.666 0.888 DL_8 31.14 25.517 0.627 0.891 DL_9 31.00 26.218 0.545 0.897 Professionalis m (KM, 5 items) Cronbach’s Alpha = 0.854 KM_1 15.33 8.770 0.769 0.800 KM_2 15.49 8.571 0.692 0.818 KM_3 15.29 8.626 0.694 0.817 KM_4 15.49 9.721 0.529 0.858 KM_5 15.42 8.696 0.666 0.825 Perceived trust (PT, 3 items) Cronbach’s Alpha = 0.862 PT_1 6.85 2.150 0.715 0.830 PT_2 6.90 2.088 0.758 0.805 PT_3 6.88 2.120 0.730 0.821 Purchase intention (YDM, 5 items) Cronbach’s Alpha = 0.804 YDM_1 15.91 5.802 0.713 0.733 YDM_2 16.16 5.976 0.472 0.804 YDM_3 15.73 5.723 0.611 0.758 YDM_4 15.94 5.560 0.653 0.745 YDM_5 15.92 5.933 0.522 0.786 Source: The author based on SPSS 26.0 data processing results 70 Table 3. EFA analysis results of independent variables and moderating variables Variable Factor 1 (Interact ive TT) Factor 2 (Influence of Internet Celebritie s DL) Factor 3 (Profess ional KM) Factor 4 (Perceive d Trust PT) TT_1 0.815 TT_2 0.830 TT_3 0.912 TT_4 0.848 TT_5 0.855 TT_6 0.831 TT_7 0.762 DL_1 0.725 DL_2 0.678 DL_3 0.750 DL_4 0.773 DL_5 0.739 DL_6 0.695 DL_7 0.766 DL_8 0.690 DL_9 0.671 KM_1 0.860 KM_2 0.778 KM_3 0.805 KM_4 0.665 KM_5 0.769 PT_1 0.855 PT_2 0.880 PT_3 0.862 Eigenvalue 7.851 3.112 2.488 1.850 Explained variance (%) 32.713 12.967 10.367 7.705 Cumulative explained variance (%) 32.713 45.680 56.047 68.752 KMO 0.915 Sig. 0.000 Note: The rotation method is the maximum variance method. Only factor loadings greater than 0.4 are displayed. Source: Author based on SPSS 26.0 data processing results A total of 24 items, including interactivity (TT, 7 items), influence of Internet celebrities (DL, 9 items), professionalism (KM, 5 items) and perceived trust (PT, 3 items), were included in the EFA analysis. The KMO value was 0.915, far exceeding the standard of 0.7; the chi-square value of the Bartlett sphericity test was 8152.367 (df=276, p=0.000), indicating that the data is very suitable for factor analysis. The analysis extracted a total of 4 factors with characteristic roots greater than 1, and the cumulative variance explanation rate was 68.752%, exceeding the recommended standard of 60%, indicating that the extracted factors can well explain the total variance of the original variables. The rotated factor loading matrix is shown in Table 3. The results clearly show that all items converge to four independent factors as expected in theory: Factor 1 corresponds to interactivity (TT), Factor 2 corresponds to influence of Internet celebrities (DL), Factor 3 corresponds to professionalism (KM), and Factor 4 corresponds to perceived trust (PT). The loadings of all items on their respective factors are greater than 0.6, and the cross-loadings on other factors are all lower than 0.4, indicating that the scale has good convergent validity and discriminant validity (Fornell & Larcker, 1981). This verifies the structural validity of the independent variable and moderating variable measurement tools, and provides support for subsequent model testing. Table 4. EFA analysis results of dependent variables Variable factor loadings YDM_1 0.843 YDM_2 0.644 YDM_3 0.780 YDM_4 0.809 YDM_5 0.685 Eigenvalue 2.858 Extracted variance (%) 57.164 KMO 0.785 Sig. 0.000 Source: Author based on SPSS 26.0 data processing results EFA analysis was performed on the five items of the dependent variable purchase intention (YDM). The KMO value was 0.785 (>0.7), and the Bartlett sphericity test was significant (χ²=1143.2, df=10, p=0.000), indicating that the data were suitable for factor analysis. The analysis extracted a factor with a characteristic root greater than 1, with an eigenvalue of 2.858, which could explain 57.164% (>50%) of the total variance, which was consistent with the expectation of unidimensional construct. The loading of each item on this factor was greater than 0.6 (ranging from 0.644 to 0.843), as shown in Table 4. This shows that the purchase intention scale has good unidimensionality and structural validity. Pearson correlation analysis is used to preliminarily examine the linear correlation between the variables (including independent variables TT, DL, KM, mediating/moderating variables PT, and dependent variable YDM). The analysis results are shown in the following table: Table 5. Correlation analysis (Pearson) TT DL KM PT YDM TT 1 DL 0.342** 1 KM 0.216** 0.480** 1 PT 0.415 0.520 0.495 1 YDM 0.482** 0.595** 0.506** 0.610 1 The results of the correlation analysis show that perceived trust (PT) is significantly positively correlated with all independent variables (TT, DL, KM) and the dependent variable (YDM) (p<0.01). Specifically, the correlation coefficients between PT and TT, DL, and KM are 0.415, 0.520, and 0.495, respectively, indicating that the characteristics of Internet celebrities are indeed related to the improvement of consumer perceived trust, which preliminarily supports H1b, H2b, and H3b. The correlation coefficient between PT and YDM is 0.610, showing a strong positive correlation, which preliminarily supports H4a. At the same time, the correlation between the independent variables and the dependent variables is still significant. The correlation coefficients between all variables are lower than 0.7, indicating that the risk of multicollinearity is low and suitable for subsequent regression and path analysis. 71 In order to test the direct effect of independent variables (TT, DL, KM) on dependent variables (YDM) (H1a, H2a, H3a), a multiple linear regression analysis was first performed, ignoring the mediating variable PT. Table 6. Results of test model fitting and autocorrelation Model R R² Adjustm entR² estimated standard error Durbin- Watson 1 0.704 0.496 0.493 0.41838 1.521 The results show that the R value is 0.704, indicating that there is a strong linear relationship between the independent variable and the dependent variable; the R² value is 0.496, indicating that the independent variable explains 49.6% of the variation of the dependent variable; the adjusted R² value is 0.493, which still maintains a high explanatory power after considering the number of variables, indicating that the model fits well (Fornell & Larcker, 1981). The Durbin-Watson value is 1.521, close to 2, indicating that there is no autocorrelation problem in the residuals, which meets the basic assumptions of regression analysis. Table 7. Results of analysis of variance (ANOVA) Model sum of squares df mean square F Sig. Regression 85.551 3 28.517 162.914 0.000 Residuals 86.821 496 0.175 All 172.372 499 ANOVA analysis showed that the F value was 162.914 (p<0.001), indicating that the regression model was significant overall and the predictive ability of the independent variable on the dependent variable was statistically valid. Table 8. Regression model parameters Model Unstandardized coefficients (B) standard error Standardized coefficient (Beta) t Sig. Tolerance (Tolerance) VIF (Constant) 0.941 0.139 6.773 0.000 TT 0.225 0.026 0.300 8.818 0.000 0.880 1.137 DL 0.348 0.036 0.365 9.661 0.000 0.710 1.408 KM 0.213 0.029 0.266 7.299 0.000 0.766 1.305 The results of regression analysis (Table 4-8) show that after controlling for other independent variables, interactivity (TT, Beta=0.300, p<0.001), influence of influencers (DL, Beta=0.365, p<0.001) and professionalism (KM, Beta=0.266, p<0.001) all have significant positive direct effects on purchase intention (YDM). This supports the research hypotheses H1a, H2a and H3a. The model explains 49.3% of the variation in purchase intention (adjusted R²=0.493), and there is no multicollinearity problem (VIF<2). 5. Research Results 5.1. Research Conclusions Based on the stimulus-response theory (SOR) and matching theory, this study constructed and tested a comprehensive model that includes mediation and moderation effects, and systematically explored the impact mechanism of live streaming by influencers on the purchase intention of Vietnamese consumers. The main research conclusions are as follows: First, influence, professionalism, and interactivity of influencers all have a significant positive direct impact on Vietnamese consumers' purchase intention (supporting H1a, H2a, and H3a). Among them, influence (Beta=0.365) has the most prominent impact, followed by interactivity (Beta=0.300) and professionalism (Beta=0.266). This shows that in the Vietnamese market, the popularity and social status of influencers are the primary factors that attract consumers and prompt their purchase intention, while real-time interaction during live broadcasts and the professionalism displayed by the anchor also play an important role. Secondly, perceived trust plays a significant partial mediating role between influencer characteristics and purchase intention. The study found that influencer influence, professionalism, and interactivity can significantly enhance consumers' perceived trust (supporting H1b, H2b, and H3b), and perceived trust itself also has a significant positive direct impact on purchase intention (supporting H4a). The indirect effect through perceived trust is significant, indicating that influencer characteristics not only directly drive purchases, but also indirectly promote purchase decisions by building and enhancing consumers' trust. This highlights the core bridge role of trust in Vietnam's live e-commerce environment. Thirdly, perceived trust is not only a mediating variable, but also plays a significant positive moderating role in the relationship between influencer characteristics and purchase intention (supporting H4). The results show that when consumers have a high level of perceived trust in influencers, the positive impact of influencers’ influence, professionalism, and interactivity on purchase intention will be significantly enhanced. This shows that trust is not only the result of influencer characteristics, but also a key condition for amplifying their effects, especially in online shopping scenarios where information asymmetry and risk perception need to be overcome. Finally, product type plays a significant and differentiated moderating role in the impact of influencer characteristics on purchase intention. Specifically, interactivity has a greater impact on the purchase intention of experiential products (supporting H5c), while influence and professionalism of influencers have a more significant impact on the purchase intention of search-based products (supporting H5a, H5b). This reveals that when faced with different types of products, Vietnamese consumers have different degrees of reliance on different stimuli in influencer live streaming, reflecting the trade-offs and adaptability of their decision-making process. In summary, this study, through empirical analysis, reveals the multiple pathways through which online celebrity live streaming affects Vietnamese consumers’ purchasing intention, including direct effects, partial mediation effects through perceived trust, and moderating effects of perceived trust and product type. These findings provide a more comprehensive and in-depth perspective for understanding live streaming e-commerce consumer behavior. 72 5.2. Theoretical Significance The theoretical significance of this study is mainly reflected in the following aspects: First, this study verified and expanded the application of stimulus-response theory (SOR) in the live e-commerce scenario in emerging markets through empirical testing. The study not only confirmed the direct impact of external stimuli (influencer characteristics) on behavioral responses (purchase intention), but more importantly, it revealed the complex dual role played by internal psychological state - perceived trust: it not only transmits part of the impact as a mediating variable, but also changes the intensity of the main effect as a moderating variable. This deepens the theoretical depth of the SOR model in explaining complex consumer behavior, especially in the online environment where trust is at stake. Second, this study contributes to the study of consumer trust theory in the field of live e-commerce. By empirically analyzing the dual mediating and moderating effects of perceived trust, the study clearly demonstrates the core position and multi-dimensional functions of trust in the process of influencers’ live streaming on consumer decision- making. This not only echoes the general importance of trust in e-commerce (Gefen et al., 2003), but also specifically explains how trust is established (affected by the characteristics of influencers) and plays a role (directly affecting purchase intention and moderating the effects of other factors) in the highly social and instantly interactive live broadcast scenario. Third, this study combines the matching theory and examines the moderating role of product type to provide evidence for understanding the contextual dependence of live streaming sales. The study found that different characteristics of influencers have different effects on the willingness to purchase different types of products, emphasizing the importance of "stimulus-context-response" matching. This suggests that future consumer behavior research needs to pay more attention to how specific contextual factors (such as product type and cultural background) shape the applicability and predictive power of theoretical models. Fourth, this study focuses on Vietnam, an emerging market, and provides valuable insights into cross-cultural consumer behavior research. The results reflect the possible preference of Vietnamese consumers for social influence (influence of influencers) and social interaction (interactivity, trust) in live shopping, which may be related to the collectivist cultural background and reliance on social recommendations (Tuan & Pham, 2019). These findings help enrich our understanding of consumer behavior patterns in non-Western markets. 5.3. Practical Significance The practical significance of this study is mainly reflected in providing specific guidance and suggestions for live e- commerce platforms, anchors and merchants to optimize their marketing strategies and promote the healthy development of Vietnam's live e-commerce industry. First, this study provides a clear direction for anchors' marketing strategies. The results show that the influence, professionalism and interactivity of Internet celebrities have a significant impact on purchase intention, among which influence is the most critical. Therefore, anchors should focus on the construction of personal brands and the accumulation of fan bases to enhance social influence. For example, well- known anchors can increase their exposure by participating in charity activities or cross-platform cooperation, thereby attracting more consumer attention (Wang, 2020).At the same time, anchors need to improve their professionalism and interactive capabilities, for example, by systematically learning product knowledge, optimizing live broadcast explanation skills, and adding real-time question-and-answer sessions to enhance consumers' trust and sense of participation. In addition, anchors should adjust their live broadcast strategies according to different product types: for experience-based products (such as cosmetics), they should increase interactive sessions and personalized recommendations; for search-based products (such as electronic products), they should emphasize professional knowledge and authority to meet consumers' needs for objective information (Chen & Lin, 2018). Secondly, this study provides data support for e-commerce platforms to optimize their operations. The important moderating role of perceived trust in live streaming sales indicates that platforms should enhance consumers' trust in platforms and anchors by strengthening anchor qualification review and improving the stability of live streaming technology. For example, platforms such as Shopee and Lazada can introduce anchor certification mechanisms to ensure that anchors have a certain professional background and recommendation capabilities. At the same time, platforms can design differentiated live streaming functions based on product types: provide more interactive tools for experience- based products (such as real-time voting and pinned comments), and provide detailed product parameter display and comparison functions for search-based products. These measures can not only enhance consumer experience, but also increase purchase conversion rates and promote the long-term development of the platform. Finally, this study provides a reference for merchants to formulate product strategies. The study shows that different product types perform differently in live streaming, and merchants should choose appropriate anchors and live streaming methods based on product characteristics. For example, merchants of experience-based products should cooperate with anchors who are highly interactive and professional, and enhance consumer trust by using demonstrations and personalized recommendations; while merchants of search-based products should choose anchors with great influence to attract consumers through social recognition effects. In addition, merchants can analyze consumer preferences and needs based on live streaming data, optimize product portfolios and promotion strategies, and thus improve sales efficiency. In summary, the practical significance of this study is that it provides scientific guidance for anchors, platforms, and merchants to help them formulate more effective strategies in the Vietnamese live e-commerce market. These suggestions not only help to enhance consumers' willingness to buy, but also provide support for the standardized development and sustainable development of the industry. 5.4. Research Limitations and Prospects Although this study has achieved certain results in theory and practice, there are still some limitations, which provide directions for improvement and expansion of future research. First, the sample of this study has certain limitations. This study mainly focuses on consumers in large cities such as Ho Chi Minh City and Hanoi in Vietnam. The geographical distribution and cultural diversity of the sample may limit the generalizability of the research results. Although the sample 73 size reached 500, which met the requirements of statistical analysis, it failed to fully cover consumers in rural areas or small and medium-sized cities, where consumption behaviors may differ from those in large cities (Thang, 2016). Future research can improve the external validity and generalizability of the research conclusions by expanding the sample range, including consumers in more regions, and even conducting cross-national comparisons. Secondly, this study has room for improvement in variable selection and measurement. This study only selected influence, professionalism, and interactivity of influencers as independent variables, and failed to fully cover other factors that may affect purchase intention in live streaming, such as the personal charm of influencers, the entertainment of live streaming content, or the emotional connection of consumers (Chen & Lin, 2018). In addition, although the measurement of perceived trust is based on a mature scale (Gan et al., 2017), it does not further distinguish between trust in the anchor and trust in the product, which may obscure the complexity of the trust mechanism. Future research can introduce more variables, such as the influencer's sense of humor or the visual effects of the live broadcast, and refine the dimensions of perceived trust to more comprehensively reveal the factors that affect purchase intention. Third, the cross-sectional design of this research method limits the capture of dynamic changes in consumer behavior. This study uses a one-time questionnaire survey, and the data reflects consumer attitudes and behaviors at a certain point in time, which makes it difficult to reveal the long-term impact of live streaming by influencers on purchase intention. For example, consumers' trust and willingness to buy may change with multiple viewings of live streaming. Future research can use longitudinal research or experimental methods to explore the dynamic impact process and time effect of live streaming by influencers by tracking consumer behavior data or designing controlled experiments. Fourth, this study still does not consider cultural differences in depth. Although this study combines the cultural characteristics of Vietnamese consumers (such as the emphasis on social identity and group recommendations), it does not systematically analyze how cultural factors specifically affect the relationship between influencer characteristics and purchase intention. For example, differences in consumer culture between Vietnam and other Southeast Asian countries (such as Thailand or Indonesia) may lead to different purchasing behaviors (Trang & Vi, 2019). Future research can analyze the behavioral differences of consumers in different countries in live e-commerce through cross-cultural comparisons, and provide more targeted theoretical and practical suggestions for the global live e-commerce market. Finally, this study did not fully explore the potential impact of external environmental factors on the research results. For example, the policies and regulations of Vietnam's live e- commerce market, technological infrastructure (such as the popularization of 5G networks), and economic changes during the epidemic may have an indirect effect on consumers' purchasing intentions. Future research can introduce macro-environmental variables and use multi-level models to analyze the interaction effects between external factors and Internet celebrity characteristics, so as to have a more comprehensive understanding of the live e-commerce ecosystem. Through the above summary, this study provides a new perspective and empirical support for the research on consumer behavior and live e-commerce in theory, and provides scientific guidance for the healthy development of Vietnam's live e-commerce industry in practice. Future research should be further expanded in terms of sample coverage, variable design, research methods and cultural analysis to deepen the understanding of live e-commerce consumer behavior and provide more comprehensive theoretical and practical support for the development of the industry. Acknowledgements In the process of completing this master's thesis, I would like to express my gratitude to many people. First, I would like to express my sincere gratitude to my supervisor, Professor Wan Xiaoyu. Thank you for your guidance and support throughout the research process. Your expertise, patient listening and constructive suggestions played a key role in my research, allowing me to explore the research topic more deeply and achieve the results of this thesis. At the same time, I would also like to thank the other members of the research group led by Professor Wan Xiaoyu for giving me a lot of help in academic exchanges and resource support. Secondly, I would like to thank my family and friends. Thank you for their understanding, encouragement and support throughout my study career. Without their support and encouragement, I would not be able to complete this thesis successfully. I would also like to thank my classmates and colleagues who provided me with help and support during the research process. Thank you for sharing many valuable experiences and insights with me, and for giving me a lot of help and inspiration in academic and life. Their support provided me with good research conditions, allowing me to focus on my research work and make progress. Here, I would like to express my most sincere gratitude to all those who have helped me. Thank you for your support and help, which enabled me to complete this master's thesis and achieve certain research results. My most sincere thanks! References [1] Gan Chunmei, Zhong Qitong, & Luo Tingyu. (2017). Research on the influencing factors of consumer trust formation in socialized business environment. Information Science, 4, 68- 73 [2] He Junhong, Du Shangrong, & Li Zhongxiang. (2019). Research on the influence of online reviews on impulsive mobile shopping willingness. Contemporary Economic Management, 5. [3] Jia Xiaofeng. (2019). Exploring consumers’ purchase and engagement intention on E-Commerce live streaming platform (Master's thesis, Beijing University of Posts and Telecommunications) [4] Jiang Jiaqi. (2019). Analysis of Influencing Factors of Internet Celebrity Live Broadcast on Consumer Purchasing Decision. (Master's thesis, Beijing University of Posts and Telecommunications) [5] Lin Jiabao, Hu Qian, & Lu Yaobin. (2017). Effect of social commerce characteristics on Consumer Decision-making Behavior: A Guanxi Management perspective. Journal of Business Economics, 37(1), 52-63 74 [6] Lin Tong. (2017). Analysis of fashion live broadcast communication strategy and communication effect (Master's thesis, Wuhan University). [7] Lu Cheng, Zhu Yiyi, Zhao Zhuanye, & Wang Aiyun. (2018). The construction of online shopping index system and its influence on consumer purchase intention. Silk, 55(6), 45-51 [8] Linh, N. T. N, Hiền, D. D. (2019). Factors influencing the online fashion purchase intention of consumers in this city. Ho Chi Minh. Finance Magazine. [9] Lu Cheng, Zhu Yiyi, Zhao Zhuanye, & Wang Aiyun. (2018). The construction of online shopping index system and its influence on consumer purchase intention. Silk, 55(6), 45-51. [10] 2021Vietnam E-commerce White Paper https://trungtamwto.vn/an-pham/18190-sach-trang-thuong- mai-dien-tu-viet-nam-2021 [11] LeLinh-https://diendandoanhnghiep.vn/livestream-vu-khi-tan- cong-thi-truong-chau-au-cua-alibaba-194304.html [12] Nguyen The Anh, 2013. Factors affecting Hanoi residents’ willingness to purchase electronic products online. Master’s thesis of Ho Chi Minh City University of Economics [13] Lu Xueqing, Zhou Meihua. Problems and paths in the development of cross-border e-commerce platforms in my country [J]. Economic Perspectives, 2016(03):81-84. [13] Hoang Linh https://ictvietnam.vn/ban-hang-qua-livestream- bung-no-o-dong-nam-a-31127.htmlnewspaper, 2014, 02:22-23 [14] Deng Le. Research on the impact of online shopping experience on consumers' purchasing intention [J]. China Collective Economy, 2020(26): 50-51. [15] Wang Ziwei, Song Xiaoqing. Research on the impact of live streaming on consumers' irrational consumption behavior [J]. Shanghai Business, 2023, No.527(01):50-52. [16] Yan Zhen. Research on the marketing strategy of online live streaming [J]. Journal of Journalism Research, 2022, 13(23):253-256. [17] Yang Xi. Research on the influence of live streaming by internet celebrities on consumer purchasing behavior [D]. Sichuan Normal University, 2022. DOI:10.27347/d.cnki.gssdu.2022.000488. [18] Li Yi. Research on the business model of live streaming sales [J]. Financial management, 2022, No.457(09):11-13. [19] Thang, H.N. (2016). Factors influencing online shopping intention of Vietnamese consumers: A study extending the theory of planned behavior. VNU Journal of Science: Economics and Business, 32(4). [20] Trang, B. T. and Tien, H. X. (2020). Online commerce and purchasing behavior of consumer shopping. Journal of Industry and Trade. [21] Trang, N.T.M. and Vi, P.N.T. (2019). The intended uses of mobile Internet: Expanding technology acceptance patterns. Journal of Economic Development, 55-61 [22] Tuan, A. B. N. and Pham, M. Facebook The impact of live streaming content on online customers’ purchasing intentions. Sustainable business development software enters the connected age, 249 [23] Wang Chong, Li Yijun, & Ye Qiang. (2007). Research into Network Consumers Decision Behavior Based on Perceived Value in E-shopping Environment. Forecast, 26(3), 21-25. [24] Wang Ting, & Deng Yue. (2017). Analysis of the e-commerce model of online celebrity livestreaming in the background of the web livestreaming era. Modern Economic Information, 15. [25] Wang Tong (2020). Research on Consumers' Purchase Intention in the Ecommerce live streaming. (Master's thesis, Minzu University of China) [26] Wang, X. (2017). Live streaming from the perspective of communication. Visual, 2017(9). Retrieved from https://www.zzqklm.com/w/hxlw/21872.html [27] Xiao Tong (2010). A cross‐national investigation of an extended technology acceptance model in the online shopping context. International Journal of Retail & Distribution Management [28] Xu He, Qu Hongjian, & Cai Jianzhong (2020). Influencing Factors of Apparel Consumers' Impulse Purchase Intention in the Context of Network Broadcast. Journal of Donghua University. [29] Zhou Chao. Research on the theory, problems and countermeasures of the business model of live streaming by internet celebrities [J]. Economic Circle, 2021, No.151(01):43- 47. [30] Zhao Lulu, Liu Feng. The impact of user perceived information quality on purchase intention on mobile social shopping platforms: A case study of WeChat mini-programs [J]. Journal of the China Society for Scientific and Technological Information, 2020, 39(12): 1173-1185 [31] Wang Shanshan. Research on the influence of consumer perceived value on the purchase intention of new intelligent products: from the perspective of brand trust [D]. Tianjin University of Commerce, 2022. [32] Wang Lijun. Research on the influencing factors of “Internet celebrity live streaming” on the consumption behavior of higher vocational students [J]. Journal of Guangdong Communications Vocational and Technical College, 2023, 22(01):118-122. [33] Zhuang Qianqian. An empirical study on the factors affecting online consumers' purchasing intention [J]. Commercial Economic Research, 2019(7): 55-58. [34] Tao Huiyun. The influence of psychological distance on consumers' purchasing intention [D]. Suzhou University, 2018. [35] Cui Chao. Research on the influence of product combination and consumer purchasing intention [J]. Brand Research, 2020(27): 285-286, 296. [36] Huang Qingjuan, Li Min, Han Chunyan. Research on the impact of social media emotional information on consumers' online purchasing intention - an empirical analysis based on Weibo data [J]. Information Science, 2021, 39(9): 143-151 [37] Anderson, E. W., & Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Marketing science, 12(2), 125-143. [38] Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological bulletin, 103(3), 411. [39] Hanh, N. T. H., Hung, D. M., Trinh, V. D. P., Vy, C. T. T., & La Thi Nhu, Y. (2019). Factors affecting the intention to use online shopping apps in Ho Chi Minh City. Science & Technology Development Journal-Economics-Law and Management, 3(4), 390-401. [40] Butz Jr, H. E., & Goodstein, L. D. (1996). Measuring customer value: gaining the strategic advantage. Organizational dynamics, 24(3), 63-77. [41] Chen, C.-C., & Lin, Y.-C. (2018). What drives live-stream usage intention? The perspectives of flow, entertainment, social interaction, and endorsement. Telematics and Informatics, 35(1), 293–303. 75 [42] Chen, L., Gillenson, M. L., & Sherrell, D. L. Enticing online consumers: an extended technology acceptance perspective. Information & Management, Vol. 39(2002), p. 705-719. [43] Chen, Yu-Hui and Barnes, S. 2007. Initial trust and online buyer behavior, Industrial Management & Data Systems, 107(1), 21-36. [44] Chin, W. W., & Todd, P. A. (1995). On the use, usefulness, and ease of use of structural equation modeling in MIS research: A note of caution. MIS quarterly, 237-246. [45] Cục TMĐT và Kinh tế số, “Sách trắng TMĐT Việt Nam năm 2018”, 2018. [46] Dan Ming Xiao, & Wu Feng. (2018). An Empirical Study on the Influence of Web Live Marketing on Purchase Intention. Collection, 36. [47] Deng Feng. (2015). Research on the Composition of Customer Perceived Value in Online Shopping Mode. Business Economic Research, (2015 30), 66-67. [48] Dong Dahai, & Yang Yi. (2008). Theoretical Analysis of Consumer Perceived Value in the Network Environment. Chinese Journal of Management, 5(6), 856. [49] Eroglu, S. A., Machleit, K. A., & Davis, L. M. (2001). Atmospheric qualities of online retailing: A conceptual model and implications. Journal of Business research, 54(2), 177-184. [50] Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: an introduction to theory and research. [51] Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable variables and measurement error: Algebra and statistics. Journal of Marketing Research, 18(3), 382-388. [52] Gan Chunmei, Zhong Qitong, & Luo Tingyu. (2017). Research on the influencing factors of consumer trust formation in socialized business environment. Information Science, 4, 68- 73. [53] Gefen, D., Karahanna, E., & Straub, D. W. (2003). Trust and TAM in online shopping: An integrated model. MIS quarterly, 27(1), 51-90. [54] Geng Bo. (2012). Analysis of influencing factors of consumers' online shopping intention based on TAM model. Statistics and Decision, 23, 105-107. [55] Ghosh, A. (1990). Retail management. Chicago: Dryden press. [56] Grandón, E. E., Nasco, S. A., & Mykytyn Jr, P. P. (2011). Comparing theories to explain e-commerce adoption. Journal of Business Research, 64(3), 292-298