Frontiers in Business, Economics and Management ISSN: 2766-824X | Vol. 9, No. 3, 2023 224 Factors Effecting of the Consumer Purchase Intention by Local Life Services Live Streaming Qi Yang1, Anaspree Chaiwan2, Chaiwat Nimanussornkul3 1Master’s Degree Program in Economics, Faculty of Economics, Chiang Mai University, Thailand 2Associate Professor, Faculty of Economics, Chiang Mai University, Thailand 3Assistant Professor, Faculty of Economics, Chiang Mai University, Thailand Abstract: As new e-commerce and local life service model, local service live streaming has great potential. This study intends to investigate the connections between live streamers, viewers, the product, the platform (i.e., ability, interaction, comment, promotion, trust and popularity) and the consumer's purchase intention. It also investigates the moderating role of consumer perceived value (i.e. perceived risk, perceived usefulness and perceived social value). Collected data from questionnaire and examined the hypotheses with structural equation model. The results suggest that interaction and ability have direct positive effects on purchase intention, and consumer perceived risk, perceived usefulness and perceived social value has significant moderating effect between factors and intention. This study gives some management implications for local service live streaming while extending the field of theoretical research on consumers purchase intention. Keywords: Live streaming, Local life services, Consumer perceived value, Consumer purchase intention. 1. Introduction 1.1. Background Today, with the gradual development of 5G internet and financial payment technology, the local life service market has emerged with wider and new ways. Especially, in China local life live streaming is developing rapidly. Live streaming is a two-way circular release of information networks, which enables information to be released simultaneously with occurrence 0. And it is very common that live streaming is the preferred choice for both private and professional events now. Digital media like Douyin has played a major role in recent years. China's live-streaming market has developed into a crucial platform for the country's economic recovery throughout the epidemic. The "live streaming with local services" consumption model has grown in popularity, and the Chinese market has demonstrated significant potential and vitality. According to report from CNNIC, the total amount of China's digital economy was about 50 trillion yuan in 2022, accounting for over 40% of GDP. The local life services sector has continued to grow steadily with live streaming thanks to the fast development of the digital economy. The local life market in China is closely related to the O2O (Online-to-offline) market, a business model that encourages potential online customers to make purchases in physical businesses using online channels [2]. Today, It is focused on connecting online transactions to offline service-oriented goods and local life services including dining, movies, travel, and other leisure activities. Douyin (the Chinese version of TikTok) local life refers to content that focuses on local living. It is divided into three categories: catering, tourism and comprehensive industry. Based on the live-stream model, it is expected to involve real- time interaction [3]. Today online local service market size reaches 345.5 billion yuan, and it will exceed 2.5 trillion yuan by 2025. As a new entrant, Douyin has built up a large number of traffic and user gracefulness in this fields. 2. Literature A number of authors have considered the effect of live streaming on different fields, such as tourism, e-game, education. During pandemic, live streaming enables interactions in real-time between many viewers in tourism [4], which gives them a better understanding and trust of the products. It is now well established from a variety of studies that factors effecting consumers purchase behavior in live streaming. From the perspective of platform, Platform loyalty have a significant impact on consumers’ purchase decisions [5]. From the consumer's perspective, sense of community, entertainment, information seeking, and a lack of external support in real life can increase engagement [6]. Additionally, it was proposed that the viewer's behavioral intentions for live streaming will be significantly influenced by gender and comedy appeals, social status presentation, and interaction [7]. Due to the relatively short history of of Douyin local life, the academic research may be limited at this stage. But industry reports, case studies, and user surveys could provide valuable insights into the research about local life services live streaming on Douyin. 3. Hypothesis Development 3.1. Effect of Local Life Live Streaming 3.1.1. Effect of Ability As scholars have suggested, the professionalism of the messenger can go some way to reducing consumer concerns about product quality [8]. Live streamer’s ability allows viewers to obtain useful information and reduce the cost of information acquisition. Through introduction and real-time communication, live streamer can stimulate fans' positive perception of the quality and value of the product and reduce consumers’ purchase doubts [9]. It also creates a positive social environment and enhance the perceived social value of the live streaming. Hypothesis 1a (H1a). Ability of live streamers negatively affects consumer perceived risk. 225 Hypothesis 1b (H1b). Ability of live streamers positively affects consumer perceived usefulness. Hypothesis 1c (H1c). Ability of live streamers positively affects consumer social perceived value. 3.1.2. Effect of Interaction Interaction in real time between consumers and live streamers helps enhance consumers participation, immersion, product involvement and perceived utility value [10]. Live streamers who actively respond to concerns, and provide detailed can reduce consumers perceived risks. Hypothesis 2a (H2a). Interaction between live streamers and consumers negatively affects consumer perceived risk. Hypothesis 2b (H2b). Interaction between live streamers and consumers positively affects consumer perceived usefulness. Hypothesis 2c (H2c). Interaction between live streamers and consumers positively affects consumer social perceived value. 3.1.3. Effect of Popularity Live streaming room’s popularity refers to the number of people in the room and the number of products sold. With a large number of viewers, they may experience a sense of social presence and there is a higher likelihood of generating conformity behavior. Conformity behavior can reduce consumers' risk perception in uncertain situations. One study showed that the number of viewers and products sold in a livestream can create a popular live atmosphere, which can make consumers make a sense of trust and belonging [11]. Meanwhile, Katz and Shapiro defined when users perceive the number of users on that platform to be large in size, it can increase the utility of the users and their willingness to continue using the platform [12]. Hypothesis 3a (H3a). The popularity of live streaming room negatively affects consumer perceived risk. Hypothesis 3b (H3b). The popularity of live streaming room positively affects consumer perceived usefulness. Hypothesis 3c (H3c). The popularity of live streaming room positively affects consumer perceived social value. 3.1.4. Effect of Trust Flavián and Guinalíu revealed that consumer trust is mostly influenced by consumer perceptions of private data security [13]. A higher platform reputation will reduce the consumer's purchase risk and leads to better expected behaviour and positive shopping attitudes [14]. When consumers perceive a higher level of trust in the platform, they are more inclined to have a positive attitude towards the content and activities on the platform. The trust in the platform can also enhance the perception of social presence among consumers. Hypothesis 4a (H4a). The trust on live streaming platform negatively affects consumer perceived risk. Hypothesis 4b (H4b). The trust on live streaming platform positively affects consumer perceived usefulness. Hypothesis 4c (H4c). The trust on live streaming platform positively affects consumer social perceived value. 3.1.5. Effect of Comment When viewers see other viewers actively engaging in comment box, they may feel more confident and secure. And richness of interaction has a positive effect on consumer purchase intention [15]. It provides to help viewers enhance perception of the usefulness. Additionally, other findings suggested that a person's participation to a task positively affects how customers perceive their social interactions in a group [16]. Hypothesis 5a (H5a). Comment negatively affects consumer perceived risk. Hypothesis 5b (H5b). Comment positively affects consumer perceived usefulness. Hypothesis 5c (H5c). Comment positively affects consumer social perceived value. 3.1.6. Effect of Promotion The live-stream promotion refers to free shipping, coupons, and discounts. These discounts can increase perception of usefulness as they can acquire the desired products or services at a lower cost. It also causes consumers to exaggerate the perceptions and ignore the perceived risks in the promotion [17]. In promotional activities, consumers often interact with other consumers, it provides a sense of belonging. Hypothesis 6a (H6a).Promotion negatively affects consumer perceived risk. Hypothesis 6b (H6b). Promotion positively affects consumer perceived usefulness. Hypothesis 6c (H6c). Promotion positively affects consumer perceived social value. 3.2. Mediating role of Customer Perceived Value Consumers' perceived value of products or services has a positive impact on purchase intention [18]. Sweeney's research provided the foundation for further refining the consumer's perceived value model to four dimensions: quality value, social value, emotional value, and pricing value [19]. Hypothesis 7 (H7). Consumer perceived usefulness mediates the effect of local life service live streaming on consumer purchase intention. Hypothesis 8 (H8). Consumer perceived risk mediates the effect of local life service live streaming on consumer purchase intention. Hypothesis 9 (H9). Consumer perceived social value mediates the effect of local life service live streaming on consumer purchase intention. 3.3. Directly Impact on Purchase Intention Consumer purchase intention means how likely consumers are to be willing to take a particular purchase action. Effect of Douyin local life services live streaming (ability and interaction, popularity, trust, comment, promotion) may not pass through the intermediary variables, but directly affect the actual consumption behavior of consumers. Hypothesis 10a (H10a). Ability of live streamers positively affects consumer purchase intention. Hypothesis 10b (H10b). Interaction between live streamers and consumers positively affects consumer purchase intention. Hypothesis 10c (H10c). The popularity of live-stream room positively affects consumer purchase intention. Hypothesis 10d (H10d). The trust on livestream platforms positively affects consumer purchase intention. Hypothesis 10e (H10e). Comment positively affects consumer purchase intention. Hypothesis 10f (H10f). Promotion positively affects consumer purchase intention. 4. Method This study uses questionnaire to collect data, taking into account the purpose of the study and the effect of local life services live streaming on Douyin. The original questionnaire was constructed in English, but the respondents on Douyin 226 were presented with a questionnaire that was translated to Chinese. These scales have been previously developed and tested by researchers in the relevant field and all the items in questionnaire were measured by a 5-point Likert scale ranging from “1 = strongly disagree” to “5 = strongly agree”. Through the survey platform (https://www.wjx.cn), a total of 600 questionnaires were distributed, of which 522 were complete and valid responses. 4.1. Structural Equation Modeling Structural Equation Modeling is a very general statistical modeling technique, which is widely used in the behavioral sciences. It can be viewed as a combination of factor analysis and regression or path analysis. The relationships between the theoretical constructs are represented by regression or path coefficients between the factors. The structural equation model implies a structure for the covariances between the observed variables, which provides the alternative name covariance structure modeling [20]. A key contribution of the structural equation model is the incorporation of customer perceptions of equity and value and customer brand preference into an integrated repurchase intention analysis. Hellier developed a general service sector model of repurchase intention from consumer theory literature [21]. The structural equation model can realize the factor analysis, path and causal effect analysis and other aspects of the analysis between the indicators. The theoretical framework of this study is presented in Figure 1. Figure 1. Structural Equation Model 5. Result 5.1. Descriptive Statistical Analysis There are 522 complete and valid responses, of which 318 are from females and 204 are from males. Most of the respondents were young people. The descriptive statistical analysis of this result of questionnaire is presented in table 1. Table 1. Descriptive statistical analysis Frequency % Gender Female 318 60.92% male 204 39.08% Age =<20 5 0.96% 21-30 120 22.99% 31-40 314 60.15% 41-50 62 11.88% >=51 21 4.02% Time Evening 320 61.3 Moon 204 39.1 Midnight 202 38.7 Morning 136 26.1 Afternoon 134 25.7 Category Food 299 57.3 Clothing 292 55.9 Cosmetic 239 45.8 Household appliances 184 35.2 Daily necessities 179 34.3 Travel 113 21.6 Others 75 14.4 227 5.2. Reliability In order to ensure the reliability of the survey results, the reliability of the questions will be tested. Cronbach 's Alpha reliability coefficient will be used to check the consistency of the variables on each measurement item. The higher the coefficient value of the result, the higher the reliability (0-1). It is generally considered that the reliability coefficient is higher than 0.6. For all the latent variables, the Cronbach’s alpha in table 2 is higher than 0.9, indicating acceptable internal consistency. Table 2. Cronbach's Alpha Cronbach's Alpha Items Number Interaction 0.962 3 ability 0.95 3 popularity 0.968 3 comment 0.948 3 trust 0.965 3 promotion 0.963 3 Perceived Usefulness 0.95 3 Perceived Risk 0.946 3 Perceived social value 0.948 3 purchase intention 0.965 3 5.3. Validity 5.3.1. Confirmatory factor analysis According to table 3, CMIN/DF (X2) is 1.636 in the range of 1-3, RMSEA is 0.035 in the excellent range of less than 0.05, and IFI, TLI, CFI, GFI and AGFI greater than 0.9. Therefore, based on results, it can be seen that this scale model has a good fit. Table 3. Model Fit Summary Index Reference Results of fit CMIN/DF 1-3 excellent，3-5 good 1.636 RMSEA <0.05 excellent，<0.08 good 0.035 IFI >0.9 excellent，>0.8 good 0.988 TLI 0.986 CFI 0.988 GFI 0.929 AGFI 0.91 On the basis of the good fit of the CFA model, The convergence validity of each dimension of the scale will be further tested. The standardized factor loading of each measurement item in the corresponding dimension is calculated by the established CFA model, and then calculates the convergence validity value and composite reliability value of each dimension through the calculation formulas of AVE and CR. Based on the suggested average variance extracted (AVE) value higher than 0.5, composite reliability (CR) requires a minimum of 0.7 to show adequate validity. AVE= (∑λ2)/n CR= (∑λ)2/（ (∑λ)2+∑δ） The AVE values of each dimension reach more than 0.5, and the AR values reach more than 0.7 in table 4, indicating that each dimension has good convergence validity and combined reliability. Table 4. AVE and CR Parameter significance estimation Factor Loadings U.std S.E. std AVE CR Interaction Q1 1.127 .957 .897 .963 Q2 1.050 .024 .943 Q3 1.000 .024 .942 Ability Q4 1.000 .973 .865 .950 Q5 .878 .022 .905 Q6 .870 .022 .910 Popularity Q7 1.000 .966 .911 .968 Q8 1.015 .020 .953 Q9 .957 .019 .944 Comment Q10 1.000 .946 .880 .956 Q11 .937 .023 .933 Q12 .920 .025 .935 Trust Q13 1.000 .962 .903 .966 Q14 .976 .020 .952 Q15 .917 .020 .937 Promotion Q16 1.000 .960 .898 .963 Q17 .922 .020 .936 Q18 1.009 .021 .946 PU Q19 1.000 .940 .865 .950 Q20 .961 .024 .944 Q21 .954 .026 .905 PR Q22 1.000 .979 .857 .947 Q23 .854 .023 .890 Q24 .924 .023 .906 PSV Q25 1.000 .965 .870 .952 Q26 .734 .020 .886 Q27 .967 .021 .945 PI Q28 1.000 .959 .904 .966 Q29 1.050 .022 .944 Q30 1.022 .021 .950 In Table 5, it can be seen that the standardized correlation coefficients between the two of each dimension are all smaller 228 than the square of the ave value corresponding to the dimension, so it is explained that each dimension has a good discriminant validity. Table 5. Discriminant validity. Interaction Ability Popularity Comment Trust Promotion Interaction .898 .490 .499 .512 .526 .497 Ability .865 .485 .527 .499 .461 Popularity .910 .424 .471 .466 Comment .860 .521 .435 Trust .903 .447 Promotion .898 Square root of AVE .947 .930 .954 .928 .950 .947 5.4. Normal distribution Statistical analysis and normal distribution test results are presented in table 6, the mean scores of each variable are between 3-4, and the scale scoring method is 1-5 positive scoring. Therefore, it can be seen that the cognition and behavior of the target group in the local life services live streaming of this study are all above average level. The normality test of each measurement item is tested by SPSS skewness and kurtosis. According to the standard proposed by Kline, if the absolute value of the skewness coefficient is less than 3 and the absolute value of the kurtosis coefficient is less than 8, it can be considered that the data meets the approximate normal distribution requirements [22]. The absolute values of the skewness and kurtosis coefficients of each measurement item in table 6 are within the standard range. Therefore, it can be stated that the data of each measurement item matches normal distribution. Table 6. Normal distribution test results. M SD Skewness Kurtosis Total M Total SD Interaction 3.780 1.317 -.793 -.580 3.589 1.206 3.440 1.244 -.479 -.810 3.550 1.186 -.491 -.662 Ability 3.360 1.197 -.383 -.653 3.268 1.094 3.110 1.130 -.104 -.545 3.330 1.114 -.418 -.691 Popularity 3.640 1.267 -.627 -.691 3.575 1.232 3.650 1.303 -.576 -.852 3.430 1.240 -.484 -.760 Comment 3.840 1.214 -.786 -.404 3.499 1.115 3.520 1.151 -.444 -.638 3.150 1.148 -.211 -.539 Trust 3.740 1.302 -.747 -.634 3.572 1.229 3.570 1.285 -.523 -.839 3.410 1.225 -.396 -.820 Promotion 3.420 1.242 -.455 -.695 3.386 1.187 3.340 1.174 -.403 -.814 3.390 1.272 -.457 -.876 Perceived Usefulness 3.820 1.206 -.719 -.493 3.497 1.130 3.480 1.154 -.434 -.605 3.190 1.194 -.084 -.747 Perceived Risk 3.730 1.231 -.674 -.558 3.434 1.146 3.130 1.156 -.084 -.729 3.440 1.229 -.358 -.855 Perception social value 3.650 1.279 -.520 -.963 3.350 1.137 2.840 1.023 .222 -.506 3.560 1.264 -.544 -.793 Purchase intention 3.400 1.248 -.533 -.611 3.548 1.247 3.660 1.332 -.578 -.927 3.580 1.288 -.579 -.789 5.5. Correlation analysis Pearson correlation coefficient is used to explore the correlation between multiple variables [23]. According to the analysis results, the correlation coefficients r between each variable are all greater than 0, so there is a significant positive correlation between each variable. 5.6. Structural model The results of the path analysis for the model are in table 7. It includes the standardized path coefficients (Std), standard errors (S.E.), critical ratio (C.R.) of the test statistic, and the significance level (P-value) for testing the hypothesis paths between variables. If C.R. >1.96 and P<0.05, hypothesis is supported. 229 Table 7. Measurement items of hypothesis STD.E S.E. T-value P Result Ability > PR .10 .05 2.06 .04 H1a:supported Interaction > PR .15 .05 3.09 .00 H2a:supported Popularity > PR .10 .04 2.14 .03 H3a:supported Trust > PR .15 .05 3.07 .00 H4a:supported Comment > PR .20 .05 4.16 *** H5a:supported Promotion > PR .15 .04 3.27 .00 H6a:supported Ability > PU .20 .06 4.02 *** H1b:supported Interaction > PU .10 .05 1.98 .05 H2b:supported Popularity > PU .18 .05 3.84 *** H3b:supported Trust > PU .14 .05 2.79 .01 H4b:supported Comment > PU .14 .05 2.92 .00 H5b:supported Promotion > PU .05 .04 .98 .33 H6b: no Ability > PSV .10 .06 2.05 .04 H1c:supported Interaction > PSV .08 .05 1.66 .10 H2c: no Popularity > PSV .19 .05 4.23 *** H3c:supported Trust > PSV .15 .05 3.32 *** H4c:supported Comment > PSV .20 .06 4.21 *** H5c:supported Promotion > PSV .16 .05 3.66 *** H6c:supported PU > PI .09 .05 2.00 .05 H7:supported PR > PI .13 .05 2.76 .01 H8:supported PSV > PI .12 .05 2.63 .01 H9:supported Ability > PI .17 .06 3.52 *** H10a:supported Interaction > PI .15 .05 3.19 .00 H10b:supported Popularity > PI .07 .05 1.55 .12 H10c: no Trust > PI .06 .05 1.35 .18 H10d: no Comment > PI .05 .06 1.10 .27 H10e: no Promotion > PI .05 .05 1.00 .32 H10f: no Based on the above results, remove H6b,H2c for mediation analysis. And as for H10c H10d H10e H10f, it is possible that there is a mediating effect in the case where the coefficient c is not significant [24]. 5.7. Mediation Analysis Bootstrap method is a non-parametric statistical technique used to estimate the distribution of a statistic and construct confidence intervals by repeatedly sampling with replacement from the original data [25]. In this study, I aim to employ the Bootstrap method to assess the significance of the mediating effect and calculate a 95% confidence interval. Table 8 displays the bootstrapping results of the mediating effect with PR as the mediating variable. The Z values for the Interaction, Ability, and Popularity effects are all more than 1.96 when Perceived Risk is used as the mediating variable, and the confidence intervals exclude zero. Furthermore, confidence ranges for direct effects do not include zero, indicating partial mediation. Direct effects for Comment and Trust are all smaller than 1.96, with confidence ranges that include zero, indicating complete mediation. Promotion's total and direct impacts are both smaller than 1.96, demonstrating the absence of a mediating effect. Table 8. Bootstrapping Results with PR Bootstrapping Effect Path P.C B.C 95%CI P. 95%CI SE Z Lower Upper Lower Upper Total Effects Interaction>PI .058 3.534 .086 .316 .090 .320 Direct Effects Interaction>PI .060 2.917 .053 .284 .057 .288 Indirect Effects Interaction>PR>PI .014 2.143 .009 .064 .008 .062 Total Effects Ability>PI .061 4.033 .122 .360 .128 .364 Direct Effects Ability>PI .060 3.750 .102 .339 .106 .343 Indirect Effects Ability>PR>PI .013 1.615 .001 .057 -.002 .051 Total Effects Popularity>PI .052 2.423 .025 .229 .025 .228 Direct Effects Popularity>PI .052 2.077 .005 .212 .005 .211 Indirect Effects Popularity>PR>PI .011 1.636 .002 .045 .000 .043 Total Effects Comment>PI .059 2.237 .017 .249 .017 .248 Direct Effects Comment>PI .060 1.517 -.030 .211 -.030 .211 Indirect Effects Comment>PR>PI .016 2.625 .016 .080 .014 .077 Total Effects Trust>PI .055 2.145 .010 .229 .010 .229 Direct Effects Trust>PI .054 1.667 -.018 .199 -.017 .200 Indirect Effects Trust>PR>PI .013 2.154 .007 .059 .007 .058 Total Effects Promotion>PI .048 1.813 -.005 .184 -.006 .183 Direct Effects Promotion>PI .047 1.277 -.031 .157 -.035 .153 Indirect Effects Promotion>PR>PI .013 2.077 .008 .060 .007 .057 230 Table 9 displays the bootstrapping results of the mediating effect with Perceived social value as the mediating variable. The findings indicate that the mediating influence on ability is only partial. However, the values for Popularity, Comment, and Trust are complete mediators. But as for Promotion, here is no mediating effect. Table 9. Bootstrapping Results with PSV Bootstrapping Effect Path Product of Coefficients B.C 95%CI P. 95%CI SE Z Lower Upper Lower Upper Total Effects Ability->PI .061 4.066 .123 .362 .129 .368 Direct Effects Ability->PI .060 3.800 .096 .339 .107 .346 Indirect Effects Ability->PSV->PI .012 1.667 .001 .055 -.001 .048 Total Effects Popularity->PI .052 2.462 .029 .231 .027 .230 Direct Effects Popularity->PI .052 1.808 -.006 .201 -.008 .198 Indirect Effects Popularity->PSV->PI .014 2.429 .012 .069 .010 .066 Total Effects Comment->PI .059 2.288 .022 .253 .020 .253 Direct Effects Comment->PI .059 1.627 -.020 .215 -.021 .212 Indirect Effects Comment->PSV->PI .016 2.438 .014 .078 .012 .073 Total Effects Trust->PI .055 2.200 .012 .231 .012 .231 Direct Effects Trust->PI .057 1.614 -.021 .205 -.020 .207 Indirect Effects Trust->PSV->PI .014 2.071 .010 .065 .007 .061 Total Effects Promotion->PI .048 1.854 -.002 .187 -.005 .185 Direct Effects Promotion->PI .048 1.271 -.031 .159 -.034 .155 Indirect Effects Promotion->PSV->PI .012 2.333 .009 .058 .008 .056 Table 10 displays the bootstrapping results of the mediating effect with Perceived usefulness as the mediating variable. It implies that there are partial mediating factors for both the Interaction and Ability effects. But for popularity, comment, and trust, there are significant complete mediating influence. Table 10. Bootstrapping Results with PU Bootstrapping Effect Path P.C B.C 95%CI P. 95%CI SE Z Lower Upper Lower Upper Total Effects Interaction->PI .058 3.552 .087 .318 .090 .322 Direct Effects Interaction->PI .058 3.241 .071 .298 .076 .302 Indirect Effects Interaction->PU->PI .011 1.727 .002 .047 .000 .045 Total Effects Ability->PI .061 4.066 .123 .361 .129 .366 Direct Effects Ability->PI .061 3.426 .084 .331 .085 .331 Indirect Effects Ability->PU-> PI .016 2.438 .016 .078 .014 .074 Total Effects Popularity->PI .052 2.442 .026 .228 .026 .228 Direct Effects Popularity->PI .052 1.846 -.009 .197 -.009 .196 Indirect Effects Popularity->PU-> PI .013 2.385 .011 .064 .010 .060 Total Effects Comment->PI .059 2.254 .019 .250 .017 .248 Direct Effects Comment->PI .060 1.783 -.012 .227 -.011 .227 Indirect Effects Comment->PU-> PI .012 2.167 .007 .058 .005 .052 Total Effects Trust->PI .055 2.164 .011 .230 .011 .230 Direct Effects Trust->PI .056 1.696 -.015 .207 -.014 .207 Indirect Effects Trust->PU->PI .012 2.000 .006 .054 .004 .052 6. Result The main focus of this study is to examine the consumer purchasing psychological mechanisms in Douyin local service live streaming. Distinguishing from traditional e- commerce, this study analyzes the impact of interaction with live streamer and consumer, live streamer’s ability, livestream popularity, comments, trust in the platform and promotion on consumer purchase intention. It also examines the mediating effects of perceived usefulness (PU), perceived social value (PSV), and perceived risk (PR). The results indicate that interaction, ability, popularity, comment, and trust have different psychological mechanisms influencing purchase intention. First, interaction and ability have direct positive effects on purchase intention, and PU and PR partially mediate the relationship between interaction and purchase intention. PU, PSV, and PR partially mediate the relationship between ability and purchase intention. In other words, interaction and ability have direct positive influence on purchase intention, and consumer interaction can reduce perceived risk and partially increase purchase intention. The host's ability can also increase purchase intention partially by reducing perceived risk, increasing perceived usefulness, and enhancing social value perception. Second, popularity, comment, and trust do not have a direct effect on purchase intention, but they indirectly influence purchase intention through the mediating variables PU, PSV, and PR. Particularly, PU, PSV, and PR completely mediate the relationship between comment, trust, and purchase intention. PR partially mediates the relationship between popularity and purchase intention, and PU and PSV fully 231 mediate the relationship between popularity and purchase intention. Third, Promotion does not have an impact on purchase intention. Figure 2 represents the revised structural analysis of the study based on the results. Figure 2. Results of structural model In summary, the ability and interaction of the live streamer have direct positive influence on consumers purchase intention. Through interactions and introduction with viewers, live streamer can establish emotional connections and trust, increasing viewer engagement and purchase intention. Online stores can better select suitable live streamers to improve sales conversion rates and customer satisfaction. Trust on platform and consumers’ comment can also indirectly effect on consumers purchase intention by consumers perceived value. So platforms can optimize platform design and functionality, providing a better interactive experience and increasing user engagement and trust. And researching on the factors of local life services live streaming on purchase intention contributes to the formulation of industry standards and regulations, and enhances the overall integrity and credibility of the local life services live-stream industry. Findings of this study have provided a deeper understanding and development to the SEM model in the context of digital consumer behavior in local life services live streaming. References [1] Ni, Z.h. (2022). Innovation and development of live streaming industry in the era of 5G. National Circulation Economy (11), 25-27. doi:10.16834/j. cnki.issn1009-5292.2022.11.034. [2] Fitzgerald, M. (2012). O2O: O2 for local business. Retrieved from the website: http://www. Online economy. org/tag/online- to-offline. [3] Wongkitrungrueng, A., & Assarut, N. (2020). The role of live streaming in building consumer trust and engagement with social commerce sellers. Journal of Business Research, 117, 543-556. doi: 10.1016/j.jbusres.2018.08.032. [4] Lin, K., Fong, L. H. N., & Law, R. (2022). Live streaming in tourism and hospitality: a literature review. Asia Pacific Journal of Tourism Research, 27(3), 290-304. doi: 10.1080/10941665.2022.2061365. [5] Wongsunopparat, S., & Deng, B. (2021). Factors Influencing Purchase Decision of Chinese Consumer under Live Streaming E-Commerce Model. Journal of Small Business and Entrepreneurship, 9(2), 1-15. doi:10.15640/jsbed.v9n2a1. [6] Hilvert-Bruce, Z., Neill, J. T., Sjöblom, M., & Hamari, J. (2018). Social motivations of live-streaming viewer engagement on Twitch. Computers in Human Behavior, 84, 58-67. doi:10.1016/j.chb.2018.02.013. [7] Hou, F., Guan, Z., Li, B., & Chong, A. Y. L. (2019). Factors influencing people’s continuous watching intention and consumption intention in live streaming: Evidence from China. Internet Research. https://doi.org/10.1108/INTR-04- 2018-0177. [8] Gilly, M. C., Graham, J. L., Wolfinbarger, M. F., & Yale, L. J. (1998). A dyadic study of interpersonal information search. Journal of the academy of marketing science, 26(2), 83- 100. doi:10.1177/0092070398262001. [9] Meng, L., Liu, F.J., Chen, S.Y., & Duan, S.H.. (2020). Can I evoke you - A study on the mechanism of different types of live ceblebrity information source characteristics on consumers' purchase intention. Nankai Management Review (01), 131-143. [10] Liu, Z.Y., Zhao, X.H., & Long, W. .(2020). The formation mechanism of consumer's purchase intention under the live streaming of celebrity e-commerce - an analysis based on rooting theory. China Circulation Economy (08), 48-57. doi:10.14089/j.cnki.cn11-3664/f.2020.08.005. [11] Wang, Q. Z., Yao, Q., & Ye, Y.. (2014). A study on the mechanism of price discount and number of buyers on consumers' impulse purchase intention in online group purchase scenario. Journal of Management Engineering (04), 37-47. doi:10.13587/j.cnki.jieem.2014.04.034. [12] Katz, M. L., & Shapiro, C. (1985). Network externalities, competition, and compatibility. The American economic review, 75(3), 424-440. https://www.jstor.org/stable/1814809. [13] Flavián, C., & Guinalíu, M. (2006). Consumer trust, perceived security and privacy policy: three basic elements of loyalty to a web site. Industrial management & data Systems. doi:10.1108/02635570610666403. [14] De Ruyter, K., Wetzels, M., & Kleijnen, M. (2001). Customer adoption of e‐service: an experimental study. International journal of service industry management, 12(2), 184-207. doi:10.1108/09564230110387542. [15] Jahng, J., Jain, H., & Ramamurthy, K. (2007). Effects of interaction richness on consumer attitudes and behavioral intentions in e-commerce: some experimental results. European Journal of Information Systems, 16(3), 254-269. doi:10.1057/palgrave.ejis.3000665. [16] Finsterwalder, J., & Kuppelwieser, V. G. (2011). Co-creation by engaging beyond oneself: the influence of task contribution on perceived customer-to-customer social interaction during a group service encounter. Journal of Strategic Marketing, 19(7), 607-618. doi:10.1080/0965254X.2011.599494. [17] Lu, C.B., Qin, Q.X., & Lin, Y.Y. .(2013). Cognitive mechanisms of consumer purchase decisions in false promotions: An empirical study based on time pressure and overconfidence. Nankai Management Review (02), 92-103. [18] Zeithaml, V. A. (1988). Consumer perceptions of price, quality, and value: a means-end model and synthesis of evidence. Journal of marketing, 52(3), 2-22. [19] Sweeney, J. C., & Soutar, G. N. (2001). Consumer perceived value: The development of a multiple item scale. Journal of retailing, 77(2), 203-220.doi:10.1016/S0022-4359(01)00041-0. 232 [20] Hox, J. J., & Bechger, T. M. (1998). An introduction to structural equation modeling. Doi:10.1007/978-3-030-80519- 7_1. [21] Hellier, P. K., Geursen, G. M., Carr, R. A., & Rickard, J. A. (2003). Customer repurchase intention: A general structural equation model. European journal of marketing. doi:10.1108/03090560310495456. [22] Kline, R. B. (1998). Software review: Software programs for structural equation modeling: Amos, EQS, and LISREL. Journal of psychoeducational assessment, 16(4), 343-364. [23] Cohen, I., Huang, Y., Chen, J., Benesty, J., Benesty, J., Chen, J., ... & Cohen, I. (2009). Pearson correlation coefficient. Noise reduction in speech processing, 1-4. [24] Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: new procedures and recommendations. Psychological methods, 7(4), 422. [25] Johnson, R. W. (2001). An introduction to the bootstrap. Teaching statistics, 23(2), 49-54.