The Influence of Anchor Characteristics on Consumers' Impulse Apparel Purchase Intention — Evidence from FsQCA

Live streaming has become the hottest marketing method for apparel brands nowadays. E-commerce anchors, as the core of live marketing, have an important role in influencing consumers' impulse purchase intention. However, the research on clothing brand anchor characteristics on consumers' impulse purchase intention based on live-streaming anchor characteristics has yet to be explored in related studies. The article uses consumer trust and attitude as mediating variables, and the cognitive-affective system as an analytical framework. All variables are tested by validated factor analysis, and FsQCA to analyze the grouping of anchor characteristics (attractiveness, popularity, professionalism, interactivity, and brand fit) in apparel live streaming marketing. The article discovers seven configurations that triggered consumers' high impulse purchase intention and five configurations of non-high impulse purchase intention. The findings indicate that (a) all dimensions of anchor characteristics have a significant effect on consumers' impulse purchase intention, in which the effect of popularity is most significant; (b) consumers' trust has a mediating effect between anchors' attractiveness, popularity, interactivity, brand fit, and impulse purchase intentions; (c) consumers' trust and consumers' attitude have a chain-mediated effect between popularity, brand fit, and impulse purchase intention. The results of the research will help clothing anchors to further understand consumers' information needs and perceived preferences, and provide references to improve the quality of live streaming.


INTRODUCTION
With the development of the Internet and the continuous upgrading of consumer demand, an increasing number of apparel companies to consumer demand as the entry point, have adopted the "online + offline" business marketing model.Web live streaming has gradually become an important online port for external marketing of apparel enterprises with its strong immediacy, good interactivity, and high sense of reality, which helps to enhance consumers' positive purchase experience and continuous purchase willingness, and plays an important role in stimulating consumption, expanding market share and promoting economic development.The e-commerce anchors as the core of the sales conversion of apparel live streaming, with the unique interactivity, visibility, and synchronization of live e-commerce, can effectively enhance the consumer's trust in the shopping process and ultimately https://doi.org/10.31881/TLR.2024.055induce the consumer to produce a strong willingness to purchase.In China, there are currently many anchors who utilize their abilities to sell products live, driving large numbers of fans to make purchases.
For example, Viya (a famous Chinese anchorwoman) sold more than 8 billion RMB on Taobao (China's largest online shopping platform) on Singles' Day 2021.Another example is Li Jiaqi (a famous Chinese male anchor), who sold 15,000 lipsticks in 5 minutes through live streaming [1].This can be seen based on the anchor live streaming under the network marketing model for product promotion and sales has a strong practical significance.However, the existing research on live streaming mainly analyzes its impact on consumers' purchase intention or purchase behaviour from the perspectives of product information, discount promotion, online interaction, and sense of social presence, etc. Research on how anchor characteristics influence consumers' behaviour is still in its infancy, and there is a dearth of research literature on the integration of anchor characteristics and consumers' perceptions from multiple perspectives, such as consumers' trust, consumers' attitudes, etc.There are fewer studies in the field of consumer perception, which leads to the failure to provide a more comprehensive and reasonable explanation of the decision-making process of forming impulse purchase intention when consumers watch live streaming.
As an important and indispensable role in the live-streaming marketing of apparel brands, the network anchor is an important factor that influences consumers' attitudes toward the holding of the brand and their willingness to purchase the product.Relevant studies have shown that the performance of different characteristics of anchors is an important factor influencing consumers' purchase intention [2].Therefore, distinct anchor characteristics help to improve live streaming activity and consumer perception, which in turn drives the continuous improvement of brand live streaming benefits.
Different from the traditional online celebrity endorsement method, live streaming is a new marketing model [3], for the anchor's professionalism and interactivity if the demand is higher, consumers will be for the anchor to show the qualities of judging whether to produce purchase behaviour [4].When brands are live streaming, anchors can stimulate consumers' emotions through their performances, which in turn enhances consumers' willingness to purchase the products recommended by the anchors.At the same time, the pleasure, excitement, appreciation, and emotional trust generated by consumers watching the anchor during live streaming will affect their purchase behaviour of cosmetics and apparel [5].From the perspective of marketing, the anchor strengthens the interaction between consumers, creating a warm shopping atmosphere, consumers watch the live streaming by the anchor to attract involuntarily order, which is due to the anchor affecting consumer psychology leading to impulse purchasing behaviour [6].Kurdi [7] found that social media internet celebrities' characteristics positively influence consumers' attitudes and purchase intentions by using the PLS-SEM method, which indicates that internet celebrities' characteristics influence consumers' intentions and attitudes toward products.In addition, anchors' characteristics were found to positively influence consumers' https://doi.org/10.31881/TLR.2024.055willingness to follow anchors' opinions and recommend them to others through prosocial interactions (PSI) and perceived anchors' trust in emotions [8], further confirming the impact of anchor characteristics on live streaming effectiveness and consumers' emotions.Through combing through the literature, it is found that the characterization shown by the anchor in live streaming marketing determines the anchor's ability to bring goods on the one hand, and on the other hand profoundly affects the consumers' perception of the brand and the product, which in turn affects the consumers' willingness to buy.Based on this, the author can't help but ask the question, when apparel brands live streaming, can the anchor characteristics prompt consumers to generate impulse purchase intention?
And how does the mechanism of action between the two proceed?Therefore, this article starts from the relationship between anchor characteristics and consumers' impulse purchase intention and digs deeper into the influence mechanism between both.
Based on the results of literature combing, this research also combines the unique attributes of apparel brands to construct the theoretical path of the influence of anchor characteristics on consumers' impulse purchase intention using cognitive-emotional theory as the analytical framework.The article divides the characteristics of clothing anchors into five dimensions, designs a research questionnaire to collect hypothesis-related data, conducts a reliability and validity test on the data, and then uses fuzzy set qualitative comparative analysis (FsQCA) to analyze the configuration of the variables [9], and ultimately puts forward the corresponding practical suggestions based on the empirical research findings.The article aims to reveal the mechanism of the influence of anchor characteristics on consumers' impulse purchase intention in apparel live streaming, to enrich the theoretical content in the field of apparel live streaming, and at the same time, to provide theoretical guidance for the development of marketing strategies for apparel brands' webcasting, and for the optimization of their characteristics by anchors.

Research Dimension Classification
According to consumer consumption behaviour, the image and traits shown by consumers in the process of watching the live streaming of clothing brand anchors on the web will affect their impulse Attractiveness, and Professional Ability [8].Wang et al. measured Netflix source characteristics in four dimensions: credibility, professionalism, interactivity, and attractiveness when investigating Netflix word-of-mouth in B2C live streaming [10].Huang explored the influence mechanism of virtual idol characteristics on consumers' purchase intention through questionnaire interviews and fuzzy-set qualitative comparative analysis, and classified virtual idol characteristics into seven dimensions: popularity, cuteness, anthropomorphism, attractiveness, professionalism, relevance, and homogeneity [11].Zakari explores the impact of celebrity characteristics on celebrity endorsement and company reputation in terms of four dimensions: attractiveness, likeability, expertise, and trustworthiness [12].Zhang in his study on celebrity characteristics on consumers' brand attitude measured celebrity endorsement characteristics in terms of four dimensions: expertise, attractiveness, homophily, and reverence [13].Calvo-Porral et al, in their study of internet celebrity endorsement mechanisms, categorized the characteristics of internet celebrities as expertise, trustworthiness, attractiveness, and credibility [14].This article combs through and summarizes the dimensions of anchor characteristics in the literature, taking into account the unique characteristics of clothing products and anchors.Clothing products especially rely on anchors to try on and explain, so the external attractiveness of the anchor's face and body and the professionalism of the product explanation can largely attract consumers to watch and promote impulse purchasing behaviour; consumers have different body shapes, and there are limitations such as skin colour, social roles, and personal preferences, which requires the anchor to actively interact with them in the process of personal demonstration to answer their questions and highlight the value of each piece of clothing; anchors with high visibility are more likely to get the attention and value of consumers.This requires anchors to actively interact with them during in-person demonstrations, answering their questions and highlighting the value of each piece of clothing; anchors with a high degree of visibility are more likely to gain consumers' attention and trust, which in turn leads to more sales in the live broadcast; and the higher the degree of match between the clothing brand and the clothing anchors, the easier it is for consumers to believe in what the anchors are describing during the live broadcast.Therefore, the article combines previous research and the characteristics of the clothing industry to classify anchor characteristics into five dimensions: attractiveness, popularity, professionalism, interactivity, and brand fit.

Anchor Characteristics and Impulse Purchase Intention
1) In the process of live-streaming clothing webcasts, the attractiveness of the anchor is an important factor that influences consumers to continue to watch the live-streaming.Specifically including the anchor's external image and internal qualities, when consumers are attracted by the anchor are more likely to generate impulse purchasing ideas.Luo et al. found through their research that the https://doi.org/10.31881/TLR.2024.055attractiveness of Netflix (social attraction, task attraction) affects consumers' purchase intention, and the higher the attractiveness of Netflix, the higher the consumers' purchase intention [15].
2) As an important incentive to stimulate consumers' impulse purchase intention, the popularity of the anchor is a crucial part of clothing online marketing.The popularity of the anchor represents its influence in live commerce, the higher the popularity of the anchor, the more resources and fans it has, and the easier it is for consumers to convince the anchor and generate impulse purchase intention [16].Chen et al. conducted a study on consumer impulse purchasing based on the characteristics of Internet celebrities and found that the popularity of Internet celebrities positively affects consumer trust and then influences consumer impulse buying willingness, and the higher the popularity of Internet celebrities, the higher the consumer impulse purchasing willingness [17].
3) Consumers need the information provided by the anchor when watching live streaming of clothing, and the professional product information provided by the anchor is the main factor influencing their purchase judgment.If consumers evaluate the quality of the product higher through the information provided by the anchor, they are more likely to have the impulse to purchase [18].LU et al. explored how interactivity affects consumers' impulse purchase intention and found that interactivity is mediated by consumers' immersion as a mediating variable, which in turn positively affects their impulse purchase intention [19].4) Live streaming marketing as an emerging online shopping mode, has the characteristics of high interactivity, a high interactivity anchor is more likely to guide and provoke the impulse purchasing behaviour of consumers.Li et al. used the interactive quality of live-streaming shopping as the independent variable and classified it into four dimensions: responsiveness, professionalism, informativeness, and personalization, and found that all four dimensions significantly and positively affect consumers' impulse purchase intention [20].5) Match between apparel brands and clothing anchors increases the ability to influence consumers during live streaming.Celebrity-brand fit or match is considered a key factor in advertising effectiveness and can have a positive impact [21].Pradhan et al. showed that brand fit with celebrity personality has a significant positive effect on both brand attitude and consumers' intention to purchase, which reduces consumers' perceived risk [16].
Hence, based on the results of the analysis of the above five points, the hypothesis between the five dimensions of apparel brand anchor characteristics and consumers' impulse purchase intention is proposed: H1a Anchor's attractiveness (AT) has a significant effect on consumers' impulse purchase intention (IPI) https://doi.org/10.31881/TLR.2024.055H1b Anchor's popularity (PO) has a significant effect on consumers' impulse purchase intention (IPI) H1c Anchor's professionalism (PR) has a significant effect on consumers' impulse purchase intention (IPI) H1d anchor's interactivity (IT) has a significant effect on consumers' impulse purchase intention (IPI) H1e Anchor's brand fit (BF) has a significant effect on consumers' impulse purchase intention (IPI)

Anchor Characteristics, Consumers' Trust and Impulse Purchase Intention
Giffin defines trust as the reliance of one party on another in a risky situation [22].Consumers' trust includes both consumers' trust in the product itself and their trust in the salesperson and the product brand.Consumers' trust is mainly based on the transaction behaviour between them and the brand and arises from the information acquired and accumulated by both parties in the transaction process.
Jha S found through their research that consumers' trust is the main influence on consumer behaviour in rationality-based situations [23].Hong et al. constructed a model of the relationship between perceived risk and consumers' purchase intention in online marketing and cited consumers' trust as a mediating variable, finding that trust plays an important positive transduction role in influencing consumers' purchase intention [24].Oliveira et al. classified consumers' trust into three dimensions: competence, honesty, and benevolence, which were empirically analyzed to show that consumers with high overall trust in e-commerce showed higher willingness to purchase online [25].Zhu et al. explored how anchors' characteristics affect consumers' purchase intention based on the perspective of pro-social interaction (PSI), and found that anchors' characteristics affect consumers' PSI, and then affect their affective trust, which ultimately affects consumers' intention to follow anchors' suggestions and recommend anchors, in which consumers' affective trust acts as a chain mediator [8].

Zhou et al. explored the mediating role of trust in consumers' purchase intention in live commerce
through a linear regression model and found that trust has a significant positive effect on consumers' purchase intention [26].
Hence, based on the above analysis of trust on consumers' impulse purchase intention, the hypothesis of the relationship between apparel brand anchor characteristics, consumers' trust, and impulse purchase intention is proposed:

Consumers' Trust and Consumers' Attitude
Consumer-perceived trust contributes to the generation of users' attitudes.The traits that apparel brand anchors display during live streaming and the competence they show when recommending products can affect consumers' trust, which in turn can lead to positive or negative consumer attitudes.
Research has found that consumers' trust in the brand is an essential factor influencing consumers' attitude toward perceived brands, and the higher the consumers' trust the higher their attitude will be [27].Waheed et al. investigated the impact of mobile social apps (MSAPs) on consumers' purchase attitude (COPY) based on IS theory and technology acceptance modelling and found that there is a significant positive effect of consumers' trust on their purchasing attitude [28].Lee, while exploring real estate intermediary loyalty using linear structural equation modelling, found that there was a significant positive effect of consumers' trust on attitude with the largest path coefficient [29].Bhalla in his study on motivation and constraints for collaborative consumption mediated by consumers' trust and attitude found that there is a statistically significant positive effect of consumers' trust on their attitude and attitude mediates between trust and consumers' intention to participate [30].Kim et al. found that consumers' trust in the apparel fashion cycle positively affects consumers' attitudes in their research on the online recycling of used clothing, which means that an increase in consumers' trust positively affects their attitudes [31].
Hence, based on the above analysis, the hypothesis of the relationship between trust and attitude is proposed: H3a Consumers' trust (CT) and consumers' attitude (CA) have a chain-mediated effect between attractiveness (AT) and impulse purchase intention (IPI) H3b Consumers' trust (CT) and consumers' attitude (CA) have a chain-mediated effect between popularity (PO) and impulse purchase intention (IPI) H3c Consumers' trust (CT) and consumers' attitude (CA) have a chain-mediated effect between professionalism (PR) and impulse purchase intention (IPI) https://doi.org/10.31881/TLR.2024.055H3d Consumers' trust (CT) and consumers' attitude (CA) have a chain-mediated effect between interactivity (IT) and impulse purchase intention (IPI) H3e Consumers' trust (CT) and consumers' attitude (CA) have a chain-mediated effect between brand fit (BF) and impulse purchase intention (IPI)

Theoretical Model
Based on the above literature analysis and research hypotheses, this article establishes a theoretical model of the influence mechanism of apparel brand anchor characteristics on consumers' impulse purchase intention, as shown in Figure 1.

Selection and Measurement of Indicators
The article adopts the form of questionnaire research, which mainly consists of two parts: ① the basic information of the respondents, and ② the related question items of five variables: anchor characteristics, consumers' trust, consumers' attitude and impulse purchase intention in live streaming.
Most of the measurement indicators originated from domestic and international literature, and a small part of the measurement indicators were designed independently.They were appropriately adjusted by combining anchor characteristics and live streaming marketing of apparel, which were designed to form a total of 33 question items including attractiveness, popularity, professionalism, interactivity, brand fit, consumers' trust, consumers' attitude, and impulse purchase intention in 8 variables.The items measuring attractiveness and popularity were based on the scales of He and Guo et al. [32,33].
The items measuring professionalism and interactivity were adapted from the scales of Yang and Ma et al. [34,35].Brand fit items were adapted from the scale in Park research [36].Consumers' trust and attitude were adapted from the scale in Wang and Al-Debei et al. [37,38].Measurement items for impulse purchase intention were borrowed from the measurement scales in the studies of Ye et al.
and Deng [39,40].After the formation of the initial questionnaire, the items were first reviewed by experts, teachers, and students in the field of apparel marketing, deleting or modifying unclear and ambiguous expressions, adjusting and enriching the content of the questionnaire according to the modifications, and then conducting a small pre-survey, and then modifying the questionnaire according to the results of the research.After forming the final questionnaire and distributing it, the measurement indicators were finally shown in Table 1.

Questionnaire Design and Collection of Sample Data
This study complied with the Declaration of Helsinki.All the ethical guidelines for data collection informed consent and pertinent disclaimers were reviewed and approved by the ethics committee.All subjects were fully informed of the content and purpose of the survey before participating in the survey.All subjects provided informed consent.All subjects were anonymous, and their data was protected.
First, the questionnaire sets screening questions to ensure that the respondents all have the experience of watching live apparel streaming and live purchasing; then the questionnaire sets the option to ask the respondents whether there is more impressive live streaming of clothing anchor, to stimulate the respondents' recollection of the characteristics of the apparel anchor; and then the questionnaire was issued to the subject, to investigate the consumer's trust, attitudes, and willingness to address the characteristics of the anchor after contact with the live streaming of the clothing.The questionnaire adopts a 5-point Likert scale, with integers from 1 to 5 indicating five different attitudes from strongly disagree to strongly agree.The formal questionnaire was designed on the questionnaire star platform and published online.We sent the accessible links of the questionnaire to WeChat, QQ, Weibo and other online communities, and invited relevant subjects to fill it out.The research object to apparel brand webcast consumer groups, a total of 543 questionnaires were issued, excluding the answer time is shorter, contradictory answers, and other invalid questionnaires 35, of which 508 valid questionnaires, the questionnaire validity rate of 93.55%.

Fuzzy-set Qualitative Comparative Analysis
Earlier, the qualitative comparative analysis (QCA) method was mainly used in sociology, political science, and other social disciplines.Ragin proposed the qualitative comparative analysis method, a research method that combines qualitative and quantitative methods [41].In recent years, the QCA method has attracted significant attention from management scholars and has become a key tool for resolving causal complexity in management, marketing, and management information systems https://doi.org/10.31881/TLR.2024.055[42,43].QCA method brings together the strengths of both qualitative and quantitative analysis, allowing complex causal analyses that are not limited to variable control and number of cases.In addition, the QCA methodology can analyze large samples of cases (advantage of quantitative analysis) as well as case-level configurations of conditions (advantage of qualitative analysis) and applies to small samples (< 10 or 15 cases), medium samples (10 or 15 -50 cases), and a large sample of case studies (> 100 cases) [44,45].Unlike causal inference methods based on correlations between variables (e.g., multiple regression, factor analysis, and SEM), the QCA method is based on the pooled relationship between the condition set and the outcome set of causal inference [46].QCA methodology fully integrates the strengths of qualitative and quantitative analysis, addressing the configuration problem rather than the traditional net effect problem.This research adopts Fuzzy-set Qualitative Comparative Analysis (FsQCA) to test and elucidate the correlation mechanism between anchor characteristics and impulse purchase intention, and to explore the theoretical paths and causal configurations of the formation of high impulse purchase intention and non-high impulse purchase intention.

Descriptive Statistical Analysis
Descriptive statistics were analyzed for the valid questionnaires, as shown in Table 2.As shown in Table 2, the proportion of male and female subjects is 49.02% and 50.98% respectively, which has a reasonable gender ratio; in terms of age distribution, 33.27% of the subjects were aged 31-40 and 32.09% were aged 26-30, which is much higher than the percentage of subjects under 18 and over 40, indicating that young people are more active in watching apparel live streaming; in terms of education level, the education level of undergraduates is predominantly undergraduate, and there are fewer master's degree and above, and the percentage of people who have received higher education is 84.84%, which indicates that most of the people who have received higher education are more conscious about live commerce;in terms of the subjects' occupations, private staff accounted for the largest number of them, followed by students, with a share of 31.5% and 23.43% respectively; in terms of disposable income, most of them were in the middle-and low-income groups, with 65.74% of them having an average monthly disposable income of less than 8,000 yuan; more than half of the subjects were in contact with live streaming for more than 3 years, and the number of people who had been in contact with live streaming for less than 1 year only accounted for 5.51% of the total sample; 77.56% of the subjects watched live streaming for an average of 0.5 to 2 hours per day, and the viewing time was in line with the subject characteristics of the testers, which indicated that the questionnaire collection was in line with the reality basis.https://doi.org/10.31881/TLR.2024.055

Normality Test
The normality test of the sample data was conducted using skewness and kurtosis.The absolute value of the skewness coefficient of the sample data in this article is less than 2 and the absolute value of the kurtosis coefficient is less than 7 [47], indicating that the data of each measurement item in this article satisfy the approximate normal distribution, which can be analyzed by using the Maximum Likelihood Estimation (MLE) method built into AMOS 24.0.https://doi.org/10.31881/TLR.2024.055

Reliability Test
The reliability of the data was tested using SPSS 24.0 software, and the overall Cronbach's α = 0.867 > 0.8 for the scale indicated that the overall reliability and stability of the scale were high.Meanwhile, the corrected item-total correlation of each dimension of anchor characteristics, consumers' trust, consumers' attitude, and impulse purchase intention are all greater than 0.3, and the Cronbach's alpha if the item deleted is > 0.8, which indicates that the scale has high reliability [48].

Validity Test
First, the validity of the questionnaire was tested.The results showed that the KMO value was 0.817 > 0.6 and the significance of Bartlett's spherical test was 0.000, indicating that the data was suitable for factor analysis [49].The questionnaire was then subjected to exploratory factor analysis using SPSS 24.0.The results show that the variable factor loadings are greater than the minimum requirement (0.500).Therefore, the structural validity of the data of this survey is good and can be continued for subsequent data processing.
Then the scale was subjected to a confirmatory factor analysis (CFA) using AMOS 24.0.The compliance of the factor structure of the study variables was judged based on the fit of the structural equation modelling.In identifying and ensuring the suitability of a factor for a particular construct, CFA must be carried out [50].CFA plotted the dimensions with the corresponding question items in a measurement model through AMOS and then judged the quality of the model fit by fitting it to the data.If the quality of model fit is good, it means that the relationship between the variables plotted by the measurement model and the question items is validated by the data.According to the results of the model fit test in Table 3, it can be seen that the data fit indicator of the basic model, χ2/df = 1.853, is in the range of 1-3, and RMSEA (root mean square error of approximation) = 0.041, is in the excellent range of <0.05.
The test results of the other GFI, AGFI, IFI, TLI, and CFI are 0.889, 0.880, 0.882, 0.865, and 0.880, respectively, which all reach a good level of more than 0.8 [51].Therefore, synthesizing the results of this analysis can show that the CFA model of clothing anchor characteristics has a good fit, and its AMOS path diagram is shown in Figure 2. Then the test was conducted for the convergent validity of the model, and the results of the analysis are shown in Table 4.The standardized regression weights, average variance extracted (AVE), and composite reliability (CR) of each variable meet the experimental criteria, which means that the standardized regression weights of all the observed variables are higher than 0.5 and have good significance at p < 0.001, and the AVE of the variables are greater than 0.5 and the CR is greater than 0.7, which indicates that the questionnaire has good convergent validity [52].https://doi.org/10.31881/TLR.2024.055

FUZZY-SET QUALITATIVE COMPARATIVE ANALYSIS Calibration
Before the FsQCA in this research, the sample data had to be calibrated.First, the average of the measurement terms of each variable is taken as the reflective value; second, using 95th percentile, 50th percentile, and 5th percentile as the data calibration values for full membership, crossover, and non-membership for the independent and dependent variables respectively, the variables were converted into fuzzy scores between 0-1 with the help of Calibrate (x, n1, n2, n3) function in FsQCA 3.0 software, where x is the variable and n1, n2, n3 are the full membership, crossover, and nonmembership from front to back, respectively.The specific calibrated anchor points for the variables are shown in Table 6.

Analysis of Necessary Conditions
In FsQCA, consistency was used to measure the necessity of the antecedent variable, which for the necessity condition should exceed a threshold criterion of 0.9 [53].Through the necessity analysis of FsQCA3.0 software, the specific results are shown in Table 7, the necessity of any independent condition in this study affecting high impulse purchase intention is a maximum of 0.7848 and a minimum of 0.5390, which are both less than 0.9, while the necessity of non-high impulse purchase intention is a maximum of 0.7667 and a minimum of 0.5430, which are also both less than 0.9.This suggests that no single independent condition is necessary for high or non-high impulse purchase intention, which means that these factors cannot individually explain high or non-high consumers' impulse purchase intention, therefore there is a need to explore the role of independent condition combinations in influencing the configuration of consumers' impulse purchase intention.

Constructing the Truth Table
By constructing a truth table it is possible to identify the logical combinations that led to the results.
The study contains eight independent variables, so there are a total of 2^8 = 256 configurations.There are two types of configurations: (1) configurations for which there are samples that match their conditional combinations； and (2) configurations for which there are no samples that can correspond https://doi.org/10.31881/TLR.2024.055 to their conditional combinations, known as logical residuals.In the truth table set, the configurations are filtered by a frequency threshold greater than 1, and the configurations for which the threshold is not satisfied, known as logical residuals, are deleted.According to the suggestion made by Ragin [41], the consistency threshold is generally set to 0.8 is better, if consistency is greater than 0.8, the result exists marked as 1; less than 0.8, the result does not exist marked as 0. Therefore, this study set the consistency threshold in the analysis of the constructs affecting willingness to impulse purchase clothing at 0.8.Identifying conditional configurations of impulse purchase intention using FsQCA 3.0 software.The partial truth table data are shown in Table 8, the first 10 rows are truth table data for the presence of impulse purchase intention results, and the last 1 row is truth table data for nonimpulse purchase intention, the condition variables are calibrated variables, the consistency indicates the degree to which the condition combination can interpret the results, and the number is the number of cases corresponding to the condition combination.From the truth table, it can be seen that the combinations of causes leading to consumers' impulse purchase intention are diverse, and it is initially verified that there is an interdependent and complex causal relationship between the antecedent conditions and the results of consumers' impulse purchase intention.Regarding previous studies, this study uses the intermediate solution as the main solution and the parsimonious solution as the supplement to derive the results of configuration analysis of high impulse purchase intention.As can be seen from Table 9, there are 7 conditional configurations, i.e. paths, affecting high impulse purchase intention, and the solution consistency of the 7 combinations is 0.9566, the consistency values of each combination are 0.9502, 0.9473, 0.9308, 0.9593, 0.9516, 0.9580, 0.9468, which are greater than the consistency criterion of 0.8, indicating that the 7 paths are https://doi.org/10.31881/TLR.2024.055all the high sufficient conditions for impulse purchase intention, while the solution coverage is 0.6318, generally, the solution coverage of large sample QCA study should be greater than 0.3, and this study meets the research criteria, indicating that the independent conditions explain the high impulse purchase intention to a large extent.
Table 9. Configuration results of high impulse purchase intention Six of the seven configurations identified PO and BF as one of the core conditions influencing consumers' impulse purchase intentions.In both Y2c and Y2d, AT and IT are among the core conditions that influence consumers' impulse purchase intention.In Y1, all five independent variables, AT, PO, PR, IT, and BF, are all core conditions that jointly influence consumers' impulse purchase intention.Among the five dimensions of anchor characteristics, PO has the highest number of occurrences as a core condition in the seven configurations and therefore has the most significant effect on consumers' highimpulse purchase intentions.The results of the FsQCA further support the hypotheses of the direct effects of H1a, H1b, H1c, H1d, and H1e.
All configurations except Y2a contain CT and PO, especially Y1 and Y2c, where M4 contains all the independent variables of anchor characteristics, and Y2c contains AT, PO, IT, BF, and CT as the core https://doi.org/10.31881/TLR.2024.055conditions, which suggests that AT, PO, IT, and BF affect consumers' clothing impulse purchase intention, which can be interpreted as increasing consumers' trust and thus enhancing their impulse purchase intention.Y1 and Y2c indirectly prove that hypotheses H2a, H2b, H2d, and H2e are true, suggesting that CT plays a mediating role between "AT→IPI, PO→IPI, IT→IPI, BF→IPI".

Configuration Analysis of Non-high Impulse Purchase Intention
Since the FsQCA follows the assumption of asymmetry of causality, i.e., the configuration of independent conditions that generate high impulse purchase intention is not symmetrically related to the configuration of independent conditions that generate non-high impulse purchase intention in the full sense of the word [54].Therefore, the dependent variable was changed to non-high impulse purchase intention to reveal the causal factors of the configuration that lead to non-high impulse purchase intention.Similarly, according to the results of the intermediate and parsimonious solutions, the results of the configuration analysis of non-high impulse purchase intention can be obtained, as shown in Table 10.There are 5 conditional configurations, i.e. paths, affecting non-high impulse purchase intention, and the solution consistency of the 5 combinations is 0.9604, the consistency values of each combination are 0.9581, 0.9487, 0.9493, 0.9679, 0.9727, which are greater than the consistency criterion of 0.8, suggesting that the 6 paths are all sufficient conditions for non-high impulse purchase intention, meanwhile, the solution coverage is 0.6232, indicating that the independent conditions explain the non-high impulse purchase intention to a large extent.In the five configurations of non-high impulse purchase intention, Y1a and Y1b are one category, and Y2a, Y2b, and Y2c are one category, and the difference between the two categories lies in whether or not the lack of CT is the core condition, when CT is absent, i.e., when CT is low, the lack of AT is the core condition that affects the consumers' intention to generate non-high impulse purchase intention, while when CT is non-absent, i.e., when CT is high, then PO and BF are the core conditions that influence their non-high impulse purchase intention.That is, low-trust consumers will be reluctant to impulse purchase clothing products because the anchors' attractiveness is too low, while high-trust consumers will be reluctant to impulse purchase clothing products because the PO and BF are too low.
It can be obtained that among the five dimensions of anchor characteristics, PO and BF have the highest number of times as non-existing condition variables in non-high impulse purchase intention among the five configurations, which can be interpreted as the most significant influence of these two anchor characteristics on consumers' impulse purchase intention.Hypotheses H2a, H2b, and H2e were further verified to be valid.
Among the five configurations, the lack of high IT is the core condition for all configurations, indicating that the lack of anchor interactivity is an important reason influencing consumers' non-high impulse purchase intention, which reflects the asymmetry in complex causal relationships.It can be said that high interactivity affects consumers' impulse purchase intention of hygiene factors rather than motivators.
The absence of PO, BF, and CA across the five configurations of non-high impulse purchase intention suggests that when consumers are watching clothing anchors with low PO and low BF, they develop a bad attitude towards the products of that clothing brand, which ultimately leads to non-high impulse purchase intention.This reflects that the lack of PO and low BF are the prerequisites for poor CA, while low attitudes lead to less impulse purchase intention for clothing products.The results of FsQCA further support the hypotheses of H4b and H4e, that is, the chain-mediated effects of CT and CA in the relationship between "PO→IPI" and "BF→IPI".

Robustness Check
After the configuration analysis, robustness analysis is also required to ensure the reliability of the results.The raw case data, calibration criteria, raw consistency thresholds, and case thresholds all affect the number of observed cases, which in turn has an impact on the results.In this study, the method of adjusting the consistency threshold is used to reanalyze the configurations of high-impulse purchase intention and non-high-impulse purchase intention by increasing the consistency threshold from 0.8 to 0.85 and keeping the frequency threshold unchanged.The analysis found that the change in the consistency threshold did not lead to changes in the results of the intermediate and simple solutions, and there were no significant changes in the solution consistency as well as the solution https://doi.org/10.31881/TLR.2024.055coverage, indicating that the paths of the influence of high impulse purchase intention and non-high impulse purchase intention have good robustness.In summary, the conclusions of this study are robust.

Research Conclusions
This article constructs a theoretical model of anchor characteristics and consumers' impulse purchase intention of apparel brands based on literature combing and theoretical analysis and analyzes the influence mechanism of consumers' trust and consumers' attitude between anchor characteristics and impulse purchase intention.
1) First, the FsQCA method shows that, on the one hand, overall high impulse purchase intention is directly driven by the interaction of two core factors (popularity and brand fit), while the attractiveness and interactivity of the anchor are also core factors influencing consumers' impulse purchase intention.
From the analysis of the configuration of high impulse purchase intention, it can be concluded that the most significant effect of the anchor characteristics of popularity is on impulse purchase intention.On the other hand, overall non-high impulse purchase intention is directly driven by the interaction of the three core factors (attractiveness, popularity, and brand fit), while the lack of interactivity is also a core factor contributing to consumers' non-high impulse purchase intention.Popularity, brand fit, and consumers' attitudes as the core factors of all configurations, and the lack of popularity and brand fit interacting with consumers' low attitudes directly drives consumers' non-high impulse purchase intention.From the analysis of the configuration of non-high impulse purchase intention, it can be concluded that popularity and brand fit among the anchor characteristics have the most significant effect on impulse purchase intention.Combining these two aspects, the influence of popularity among anchor characteristics on consumers' impulse purchase intention is the most significant.
2) Then, from the perspective of parallel mediation, there is a mediating effect of consumers' trust between anchors' attractiveness, popularity, interactivity, brand fit, and impulse purchase intention, and there is no mediating effect of consumers' trust between anchors' professionalism and impulse purchase intention.From the perspective of chain mediation, there is a chain mediation effect of consumers' trust and consumers' attitudes between anchors' popularity, brand fit, and impulse purchase intention, and consumers' trust and attitudes are similarly not chain mediated between anchors' professionalism and impulse purchase intention.https://doi.org/10.31881/TLR.2024.055

Theoretical Contributions
First, this research explains the mechanism of the influence of live-streaming apparel on consumers' purchasing behaviour from a new theoretical perspective.While a small number of researches have examined how anchor characteristics influence consumer behaviour [55], these studies have relatively simple explanatory mechanisms and lack research specific to the field of live streaming of apparel.This article based on the cognitive-affective framework emphasizes the impact of clothing anchor characteristics on consumers' impulse intention purchase, extends Sun and Zhu's studies [56,8], and clarifies the impact of anchors on consumers' emotions.
Second, this research advances the study of FsQCA in the clothing domain.The FsQCA method, based on the configuration analysis perspective, can analyze how the anchor characteristics dimensions lead to high or non-high impulse purchase intention in detail.Few previous researchers have applied the method to the field of apparel marketing while using FsQCA allows for testing the robustness of the hypothesized results after conducting the underlying empirical analysis [57].Therefore, based on the collected questionnaire data, this research adopts the FsQCA method to conduct a configuration analysis of anchor characteristics, consumers' trust, and consumers' attitudes, so that it can effectively find the influence paths of high and non-high impulse purchase intention.It advances the application of this method in the field of apparel live-streaming marketing and provides methodological references for subsequent research.
Finally, this research extends the trust and attitude perspective by finding that anchor characteristics have become important determinants of consumers' impulse purchases, as well as emphasizing consumers' trust and attitude as a potential bridge during live streaming of apparel.The results suggest that consumers' affective factors influence impulse purchase intention through the chain mediation of trust and attitude.Few previous researches have explored the mechanism from the two-level perspective of consumer emotions [58].The empirical results of this article extend Afzal's research that builds consumers' trust through brand reputation, brand competence, and brand predictability [59].

Managerial Implications
The results of this research will provide reference opinions for apparel brand anchors live streaming to enhance consumers' impulse purchase intention, and at the same time, it will provide valuable theoretical references for the apparel live streaming field to enhance the user scale and expand the benefits of live streaming, which specifically include the following two aspects: 1) First, this research provides a reference for apparel brands to choose their anchors.When apparel brands live streaming on the internet, they should focus on the popularity and brand fit of the anchor, https://doi.org/10.31881/TLR.2024.055which will affect consumers' trust and have a significant impact on the formation of impulse purchase intention [60].Therefore, when choosing an anchor, apparel brands can use the number of followers and the areas in which they have been live streaming for a long time as criteria to measure their effectiveness and make a decision on whether to cooperate with them.Since developing countries such as China have a high level of public self-consciousness and are highly susceptible to social pressures [61], it is even more important for apparel brands to consider the impact of an anchor's popularity on consumer blindness.Although the cost required for the anchor with high popularity is higher than that of the anchor with low popularity, the message conveyed by the former is more likely to have a psychological impact on consumers.In addition, apparel brands should choose anchors that fit with the brand's market positioning and marketing products, which are more likely to be recognized by consumers [62].
2) Furthermore, this research guides for apparel anchors to enhance their marketing effectiveness.
Anchors should strengthen the interaction with consumers during live streaming, and at the same time ensure that they can fully express their inner qualities during live streaming, to attract consumers to continue to watch live streaming, which will enhance their trust in the product and generate impulse buying behaviour.High-quality interaction can not only show the anchor's ability but also fully promote the products' good quality, so that consumers have a positive emotional attitude towards the anchor and the product, thus generating a willingness to purchase [63].The key to live interaction lies in the language skills of the anchor, responding positively and promptly to consumer questions, encouraging consumers to participate in the interaction, increasing consumer motivation, and creating a favourable live shopping environment for consumers [64].

LIMITATION AND FUTURE RESEARCH
This study still has several limitations which can be addressed in future research.First, the sample population for the empirical analysis mainly originated from consumers in China, so the applicability of the results of this study to consumers in different source countries may be low, limiting the generalisability of the findings.Second, consumers' memories of purchasing apparel whilst watching live streaming can be blurred and therefore can be inaccurate concerning the judgments made when completing the questionnaire.
Therefore, it is meaningful to use a wider range of data to validate the model based on different cultural contexts in future research.Not only can the sample size of the data in this study be enlarged, but it can also be further tested whether there are differences in consumers' sentiments towards anchor characteristics in apparel live-streaming marketing in different source countries.Meanwhile, future research would be better to use other methods to collect data or use experimental methods to purchase intention and continuous consumption behaviour.Consumers judge, analyze, and process the various features shown by the anchor when watching live streaming, and generate impulse purchase intention based on whether these features attract consumers to continue watching and whether they meet their purchase demands.Relevant researches show that the characteristics shown by anchors in clothing brand live streaming can influence consumers' purchase intention.Zhu et al. in their study classified anchors' characteristics into three dimensions: Physical Attractiveness, Social https://doi.org/10.31881/TLR.2024.055 H2a Consumers' trust (CT) mediates between attractiveness (AT) and impulse purchase intention (IPI) https://doi.org/10.31881/TLR.2024.055H2b Consumers' trust (CT) mediates between popularity (PO) and impulse purchase intention (IPI) H2c Consumers' trust (CT) mediating between professionalism (PR) and impulse purchase intention (IPI) H2d Consumers' trust (CT) mediates between interactivity (IT) and impulse purchase intention (IPI) H2e Consumers' trust (CT) mediates between brand fit (BF) and impulse purchase intention (IPI)

Figure 1 .
Figure 1.Research model has a good reputation PO3 The clothing anchor is more popular in the same industry PO4 The clothing anchor is accomplished in its industry Professionalism (PR) PR1 This clothing anchor has relevant knowledge in the product area Yang X et al.(2023) PR2 This clothing anchor has authority in the product area PR3 This clothing anchor can describe in detail the products he/she recommends PR4 This clothing anchor has extensive experience in product-related areas PR5 This clothing anchor can make professional assessments of recommended products Interactivity (IT) IT1 The clothing anchor was able to respond to my questions and comments promptly Ma X et al.(2023) IT2 The clothing anchor was responsive to my questions with pertinent information IT3 The clothing anchor was very positive in her communication with me IT4 The clothing anchor will keep a good interaction with the users during live-streaming Brand Fit (BF) BF1 The clothing anchor usually uses the brand's products as well Park H J et al.(2020) BF2 The image of the anchor of this brand of clothing has a high correlation with the product BF3 The brand's clothing anchor's temperament matches the endorsed product BF4 The brand's clothing anchors match the brand's image Consumers' Trust (CT) CT1 I believe the clothing anchor's description is pertinent and objective Wang X et al.(2022) CT2 This clothing anchor made me want to go out and buy the products it recommended CT3 I'm sure the clothing anchor will address all my questions during livestreaming CT4 Watching their live-streaming made me trust the clothing anchor even more Consumers' Attitude (CA) CA1 Watching that clothing anchor puts me in a good mood Lee D et al.(2012) CA2 I was pleased with the clothing anchor's ability to bring in the products CA3 It is wise to choose the products recommended by this clothing anchor Impulse Purchase Intention (IPI) IPI1 I would choose to purchase products I had no intention of buying after watching that clothing anchor live-streaming Ye Y et al.(2022)and Deng F et al.(2023) https://doi.org/10.31881/TLR.2024.clothinganchor, I found a lot of products that weren't in my cart but I wanted to purchase Ye Y et al.(2022)and Deng F et al.(2023) IPI3 I would have a strong purchase intention while watching this clothing anchor live-streaming IPI4 When I watch this clothing anchor I get an urge to buy the products that he/she recommends.

Table 1 .
Scale items and references AT5The personality of the clothing's anchor would arouse my interest in purchasing it https://doi.org/10.31881/TLR.2024.055

Table 2 .
Descriptive statistical analysis of the sample

Table 3 .
Model fit test

Table 4 .
Scale convergent validity test According to the results of the analysis in Table5, it can be seen that the standardized regression weights between each variable two and two in this test of discriminant validity are less than the square https://doi.org/10.31881/TLR.2024.055root of the AVE value corresponding to the variable, thus indicating that there is a good discriminant validity between each variable.

Table 5 .
Scale discriminant validity test

Table 7 .
Necessity analysis of impulse purchase intention

Table 8 .
Truth table

Table 10 .
Configuration results of non-high impulse purchase intention