The Effect of Attractiveness, Expertise, Visual Attractiveness and Interactivity on Online Impulsive Buying Behavior

An innovative approach to product marketing through live streaming on TikTok can potentially encourage impulsive buying behavior among Indonesian customers, offering opportunities for e-commerce business owners to increase sales. This research focuses on identifying factors influencing impulsive purchasing decisions for fashion products through live streaming on TikTok, namely attractiveness, expertise, visual appeal, and interactivity. Using the Stimulus-Organism-Response (SOR) theoretical framework, this study explores consumer reactions and behaviors caused by stimulus factors which are then analyzed using Structural Equation Modeling (SEM) PLS techniques with the SmartPLS application based on 156 valid respondents on TikTok. The results are expected to guide fashion business actors in social commerce to design effective marketing strategies using TikTok live streaming by emphasizing aspects of perceived enjoyment to encourage impulse buying.


INTRODUCTION
The need for social media has changed the way people live, especially in communication, thus opening up new opportunities in business activities.Social media is used for various purposes such as communicating with family, spending free time, reading news or stories, searching for information, and even buying goods (Kepios, 2023).Live shopping, as a new feature on social media, has developed into a popular marketing tool worldwide.As reported by McKinsey Digital 2021, live shopping has become an

1338
innovative sales channel in China, with the market value increasing by more than 280% from 2017 to 2020, reaching $171 billion in 2020 (Tuncer, 2021).With 109.90 million users, Indonesia has become the second largest user of TikTok globally, driving sales trends in Indonesia and making live shopping increasingly popular and widely performed via the TikTok platform (Sadya, 2023).TikTok live streaming is now an interesting phenomenon due to its popularity in facilitating online transactions, especially for clothing products (Xue & Liu, 2023).The rapid advancement of digital marketing encourages many apparel sellers to turn to ecommerce platforms (Guercini et al., 2018).In the third quarter of 2019, the growth of the textile and clothing industry reached its peak with an increase of 15.08 percent, surpassing the overall economic growth of only 5.02 percent.This industry becomes a priority in the Making Indonesia 4.0 roadmap, emphasizing the commitment to the industrial era 4.0 (Ministry of Industry of the Republic of Indonesia, 2019).People's desire to look fashionable and follow social trends encourages impulse buying (Trimartati, 2014;Widuri et al., 20201).An Ipsos survey of 3,000 respondents in Southeast Asia, including Indonesia, shows that fashion products are the top category in online purchases via live streaming, reaching 72% (Ipsos, 2022).Live streaming allows real-time interaction between streamers and consumer which influences impulse buying behavior.The ease of obtaining information through TikTok live streaming tends to encourage consumptive behavior, allowing consumers to make purchases without prior planning (Lee & Chen, 2021; Susanto & Savira, 2021).
Impulse buying refers to the act of buying suddenly without planning, often triggered by emotional impulses and instant gratification (Wu et al., 2016).This phenomenon is common in the context of online purchases, where visual appeals play an important role in attracting customer attention and increasing impulsive behavior (Zheng et al., 2019;Chopdar & Balakrishnan, 2020).This research uses the S-O-R (stimulus-organismresponse) framework to study how situational factors influence the perception and impulse purchase behavior of fashion products through TikTok live streaming in Indonesia.The focus is to explore impulse buying in the context of live streaming, which is an innovative form of electronic commerce.This research is relevant because live streaming is increasingly popular as a sales channel and understanding the factors that influence impulse buying can assist businesses in designing more effective marketing strategies.By using appropriate analytical techniques, this research aims to provide indepth insight into how stimuli in TikTok live streaming influence consumer purchasing decisions and contribute to a better understanding of consumer behavior in the digital age.

LITERATURE REVIEW
According to Pramesti & Dwiridotjahjono (2022), consumer behavior is a factor that influences consumers in the purchasing decision-making process.Consumer behavior is closely related to the buying process which includes decisions about what to buy, where to buy, and how to pay.The consumer buying process can be categorized into three types, namely fully planned purchases, partially planned purchases, and unplanned purchases.As explained by Kotler (Chu & Liang, 2018).The SOR model explains how a certain stimulus triggers an individual to choose an option and give a certain response.There are three main components in this model: Stimulus (S) which triggers consumer behavior and response; Organism (O) which includes internal factors that affect consumers; and Response (R) which is the actual reaction of consumers to the stimulus.
Social commerce combines buying and selling with social media to enhance the online shopping experience for consumers.Wongkitrungrueng & Assarut (2020) identified two streams of social commerce: first, using cultural dimensions such as social capital.social

Attractiveness and Interactivity on Online
Impulse Buying 1339 exchange, social influence, and social support theory that focus on the socio-cultural component; second, using the Stimulus-Organism-Response (SOR) model to explore consumer motives, benefits, values, and adoption of social commerce through word of mouth, involving value or motivation theory, the stimulus-organism-response (SOR) model, technology acceptance model, and reasoned action theory.Live Streaming Shopping is a practice where sellers or companies go live via e-commerce platforms or social media to showcase and sell goods in real time to audiences.Consumers can view items, interact with sellers or hosts, and make purchases directly during the broadcast.This live interaction enhances consumer interest and encourages purchases (Li et al., 2021).
Impulsive buying is a condition where individuals make purchases with a feeling of urgency, difficult to resist, or postponed (Saricam & Okur, 2019).Although it occurs without planning, this spontaneous buying phenomenon generally gives consumers confidence that the action is appropriate and natural.Zheng defines impulse buying as purchasing behavior that is unplanned, attractive, and reflects the hedonic side of consumers (Zheng & Liu, 2023).The urge to make impulse purchases is a manifestation of unplanned emotions where individuals do not deeply consider the need for the product to be purchased.Perceived enjoyment reflects the level of customer satisfaction when purchasing goods or services via the internet (Kian et al., 2017).Akroush et al. (2011) emphasizes that perceived enjoyment in online shopping is comparable to enjoyment in physical shopping, significantly impacting customer intentions and behaviors.Research shows that perceived enjoyment has a positive influence on impulsive buying in online transactions.Consumers who enjoy online shopping are more likely to make impulse purchases (Lu et al., 2023) because perceived enjoyment increases mood and positive emotions, which can reduce self-control and increase the likelihood of impulse purchases (Li, 2016).
Attractiveness includes physical attractiveness as well as other characteristics such as personality and athletic ability (Erdogan, 1999).Influencers with high attractiveness are more likely to shape their followers' purchase intentions (Guo et al., 2018;Van der Waldt et al., 2009).Additionally, attractiveness is considered a determining factor in promoting valuable messages (Wang & Scheinbaum, 2018).Expertise is defined as the degree to which a support provider is perceived to have sufficient knowledge, experience, or skills to promote a product (Van der Waldt et al., 2009).Some researchers categorize ecommerce streamer attributes that positively impact consumers' online purchase decisions into four categories: charm, recommendation, appearance, and interaction (Han & Xu, 2020).Prior to purchase.visual appeal factors emerge as a critical factor in attracting consumer attention where visual appeal is a key factor influencing consumer purchasing decisions (Fang et al., 2017).The online visual experience parallels the in-store environment, where the presentation of information through visual media can subtly influence consumers to add items to their purchase, a phenomenon closely linked to impulse buying behavior (Bhalla & Anuraag, 2010).
Interactivity refers to the capacity of participants to control the information flow and play an active role in reciprocal conversations (Hoffman et al., 1999).As a critical component of social stimuli within online advertising, interactivity can evoke physiological responses in consumers, shape their attitudes, and impact purchase intentions (Sreejesh et al., 2020).Perceived interactivity is a key factor in marketing and communication on both online and mobile platforms (Park & Yoo, 2020).Park and Yoo identified four primary indicators of perceived interactivity: speed of information processing, rapid information delivery, minimal information delivery delay, and instant information delivery.

METHOD
This study uses a quantitative approach involving 156 respondents who actively use TikTok live-streaming e-commerce for approximately the past six months.Data was collected through an online questionnaire and distributed through various social media platforms and analyzed using Partial Least Square Path Modeling (PLS-SEM).This research uses simple random sampling technique by taking respondents randomly from the population without paying without considering any specific status.Partial Least Squares (PLS) uses the Ten Times Rule method to determine sample size.This method specifies a minimum sample size must be 10 times the number of arrows pointing to a variable (Margarian, 2022).These arrows can be formative indicators to variables, structural paths to constructs, or endogenous variables (Johansson, 2018).Therefore, this study determined a minimum of 110 samples obtained from 11 construct paths multiplied by 10 that divided into several variables including independent variables (attractiveness to perceived enjoyment (X1), expertise to perceived enjoyment (X2), visual appeal to perceived enjoyment (X3), interactivity to perceived enjoyment (X4), the dependent variable is the urge to buy impulsively (Y), and the mediating variable is perceived enjoyment (M).

RESULT
This study employs data analysis with the PLS-SEM method using SmartPLS version 3 software.Data  The discriminant test in this study uses the heterotrait-monotrait ratio (HTMT).The aim is to measure the extent of the uniqueness of a latent variable against other latent variables and meet several criteria that must be met to state that the discriminant validity value of a variable is good according to the heterotrait-monotrait ratio (HTMT) value that has criteria of ≤ 0.90.(Hair et al. 2021).Based on the results of the discriminant validity test as shown in Table 2, it is known that the HTMT value has met the criteria.Analysis of Goodness of Fit utilize Coefficient of Determination (R^2), Predictive Relevance (Q2), and Standardized Root Mean Square Residual (SRMR).The results of the calculation of the coefficient of determination (R square) as presented in Table 3 are used as an indicator to measure how much variance in endogenous constructs can be explained by the predictor constructs obtained R2 = 0.532.These results indicate that 53.2% of the variation in the Perceived Enjoyment variable is influenced by the variables of attractiveness, expertise, interactivity, and visual appeal.Meanwhile, for the urge to buy impulsively variable, the coefficient of determination is 0.182, means that 18.2% of the variation in the urge to buy impulsively variable is influenced by perceived enjoyment.Table 4. presented the results of the calculation of the predictive relevance of Q square to show how well the prediction of the observation value produced in this study.Predictive relevance (Q2) is an indicator to assess the predictive power of a model using blindfolding.Based on Table 3. the Q square value for all variables meets these criteria, which is more

Attractiveness and Interactivity on Online
Impulse Buying 1342 than 0 and more than 0.36 (Al-Fraihat et al., 2020).This shows that all variables have predictive relevance values.This means that the independent variables used to predict the dependent variable are correct.This model fit is measured using the standardized root mean square residual (SRMR) with a criterion value indicating the model fit (fit) (Margarian, 2022).The results of data processing seen in Table 4, shows that the standardized root mean square residual (SRMR) value is 0.062 where this value meets the criteria, namely SRMR <0.08.Therefore, it can be concluded that the model built is considered a good fit.The inner model analysis of the direct effect aims to test hypotheses related to the direct effect of influencing variables (exogenous) on the affected variables (endogenous) and the indirect effect refers to the influence caused by a construct or independent variable on the dependent variable indirectly through intermediary or intervening variables.The tstatistics and p-values are used to determine the statistical significance of the relationship between variables.A relationship is concluded to be statistically significant if the t-statistic value is greater than >1.65 (Henseler et al., 2009).In addition to t-statistics, p-values also need to be considered.The research hypothesis is considered accepted if the p-value is smaller than or equal to ≤ 0.05 (Margarian, 2022).The results of hypothesis testing seen in Table 6 show that all variables that have a direct or indirect effect on other variables are accepted.

1343
The results of the first hypothesis (H1) show that the relationship between the Attractiveness (AT) variable and Perceived Enjoyment (PET) has a path coefficient of 0.346, t-statistics of 4.363 and p-values of 0.000.The t-statistics value in the first hypothesis is higher than the t-statistics table (1.65).In addition, the p-values obtained from the relationship between these two variables are lower than the cut-off value (0.05).Therefore, from this hypothesis, it can be seen that there is a positive and significant relationship.This finding is in line with previous research by Lee and Chen (who showed a positive influence of attractiveness on perceived enjoyment (Lee & Chen, 2021).Attractiveness, in this case streamer attractiveness, can influence consumers emotional state and trigger positive associations between the brand and the product being promoted (Dang- Van et al., 2023).Respondents in this study also felt attractiveness in the sale of fashion products through live streaming on TikTok, which in turn increased their enjoyment in watching and shopping for these products.
The results of the second hypothesis (H2) show that the relationship between the Expertise (EP) variable and Perceived Enjoyment (PET) has a path coefficient of 0.143, t-statistics of 2.121 and p-values of 0.034.The t-statistics value in the second hypothesis is higher than the t-statistics table (1.65).In addition, the p-values obtained from the relationship between these two variables are lower than the cut-off value (0.05).In this hypothesis, it can be seen that there is a positive and significant relationship.The findings of this study support the results of Lee and Chen's research which shows that streamer expertise has a positive impact on users' perceived enjoyment in TikTok live streaming (Lee & Chen, 2021).This is in line with Lou and Yuan's research which found that consumers are more likely to engage with fellow users who have expertise or can present brands and/or products (Hughes et al., 2019).These studies emphasize that streamer expertise can influence consumers emotional states.Other research states, where the expertise of the seller or streamer in presenting and explaining the details of fashion products in live streaming on TikTok is considered by respondents as a factor that can increase their enjoyment in watching live streaming of fashion product sales on TikTok.
The results of the third hypothesis (H3) show that the relationship between the Visual Appeal (VA) variable and Perceived Enjoyment (PET) has a path coefficient of 0.250, tstatistics of 2.499 and p-values of 0.015.The t-statistics value in the third hypothesis is higher than the t-statistics table (1.65).In addition, the p-values obtained from the relationship between these two variables are lower than the cut-off value (0.05).So, in this hypothesis, it can be seen that there is a positive and significant relationship.The findings of this study support the results of Wen et al. (2024) which show that visual appeal has a positive impact on users' perceived enjoyment in TikTok live streaming.In line with the research of Haridasan & Fernando (2018) regarding online stores, the more attractive the visualization presented by online stores in displaying the products they offer, the more it will give consumers a sense of pleasure.With the appearance of an attractive website created by an online store, it will determine the feeling of consumer pleasure.Variations in background color, table color, and font color need to be chosen carefully to make it easier for visitors to read the content.This convenience will create pleasure for visitors when viewing and selecting the desired product (Vassiliadis et al., 2021).This approach is in line with the experience of visual appeal in the context of shopping in physical stores, where the presentation of information through visual media can suggestively influence consumers to add items to purchases that are closely related to impulse buying behavior (Bhalla & Anuraag, 2010).
Results of the fourth hypothesis (H4) show that the relationship between the Interactivity (IN) variable and Perceived Enjoyment (PET) has a path coefficient of 0.203, t-statistics of 2.144 and p-values of 0.002.The t-statistics value in the fourth hypothesis is higher than the t-statistics table (1.65).In addition, the p-values obtained from the relationship between these two variables are lower than the cut-off value (0.05).Thus, this hypothesis shows that there is a positive and significant relationship.The findings of this study support the results of research (Dwitya & Hartono, 2023) which shows that interactivity has a positive impact on users perceived enjoyment in TikTok live streaming.

1344
Closeness and audience involvement can be created in live streaming if the streamer is able to play the audience emotions through interesting interactions and make them feel comfortable watching until the end.The TikTok Live feature provides a more exciting and entertaining online shopping experience.It opens up new channels of communication between sellers and consumers.Live Streaming Commerce provides an interactive, engaging, and user-focused environment.This makes live streaming more enjoyable for viewers as they feel engaged and valued by streamers who are willing to answer questions in real-time.This research supports the findings of Sheng (2019) which show that the perception of audience interactivity through the comment column has an influence on consumer buying interest.
The results of the fifth hypothesis (H5) show that the relationship between the Perceived Enjoyment (PET) variable and Urge to Buy Impulsively (UBI) has a path coefficient of 0.426, t-statistics of 6.665 and p-values of 0.000.The t-statistics value in the fifth hypothesis is higher than the t-statistics table (1.65).In addition, the p-values obtained from the relationship between these two variables are lower than the cut-off value (0.05).From this hypothesis, it can be seen that there is a positive and significant relationship.This study found that perceived enjoyment from selling fashion products through TikTok live streaming mediates the influence of attractiveness, expertise, visual appeal, and interactivity on the urge to buy impulsively.This finding is in line with Lee and Chen's research, which shows the role of perceived enjoyment as a mediator in the relationship between these variables (Lee & Chen, 2021).Research conducted by Xiang also found similar results, where perceived enjoyment of social commerce platforms positively influences users' tendency to make impulse purchases (Zafar et al., 2021).This suggests that perceived enjoyment is an important factor driving impulse purchases, both in the context of social commerce platforms on TikTok live streaming.

CONCLUSION
Using the Stimulus-Organism-Response (SOR) model, which includes perceived enjoyment and perceived usefulness as cognitive and emotional states for consumers, and urge to buy impulsively as a behavioral response, this study identifies that perceived enjoyment has a positive and significant effect on urge to buy impulsively.The perceived enjoyment variable is influenced by attractiveness, namely the attractiveness of TikTok live streaming streamers or hosts which is proven to increase consumers enjoyment of seeing real-time promotions.This variable is also influenced by the streamer expertise in marketing products which increases trust in the brand associated with the product being marketed.In addition, the visual appeal of the product and store increases consumers pleasure to add more their shopping items.Interactivity from the audience through the chat column has an influence on consumers buying interest because they can get the information quickly, which also affects consumers buying interest.The research data was collected through questionnaires to 156 respondents and analyzed using PLS-SEM.The findings show that attractiveness, expertise, visual appeal, and interactivity have a positive and significant influence on perceived enjoyment in purchasing fashion products through TikTok live streaming.Overall, perceived enjoyment plays a crucial role in creating a pleasant shopping experience and encouraging impulse purchase decisions.

Table 1 .
analysis using this method begins with analyzing the research model, namely the outer model, model goodness test, inner model.The Outer Model Analysis (Measurement Model) at this stage is analyzed to test internal reliability, convergent validity, and discriminant validity of constructs used to measure research variables.The resulting calculations include Cronbach's Alpha, Composite Reliability, Factor Loading.AVE and HTMT.The convergent validity test results as shown in Table 1, it is known that all research construct indicators have an average variance extracted ≥0.5, factor loading ≥0.7, Composite Reliability ≥0.7 (Hair et al., 2019).The smallest factor loading value is 0.729 (VA1 and VA5) and the largest is 0.908 (EP2), so that all factor loading is greater than 0.7.The results of Table1 show that all indicators used in the study are valid.While the reliability test has Cronbach's Alpha> 0.7 (Hair et al. 2013).The smallest Cronbach's Alpha value is 0.811 (Expertise) and the largest is 0.895 (Interactivity).The lowest CR value is 0.876 (Expertise) and the largest is 0.925 (Interactivity).While the lowest AVE value is 0.638 (Expertise) and the largest is 0.757 (Visual Appeal).These results indicate that all indicators used in this study are reliable.Research Validity and Reliability Testing

Table 5 .
Hypothesis Test Results