Analysis of Factors Contributing to Impulse Buying Behavior of E-Commerce Users

Ease of accessing the internet and information technology triggers impulsive buying behavior through increased accessibility to products and services that facilitate the buying process. Through internet facilities by browsing, consumers can feel like window shopping in a mall. This study focuses on the Java-Bali area which is a strategic location in access to products. Data were collected from 273 respondents of E-commerce users using a survey method. Structural Equation Modeling (SEM) modeling technique with SmartPLS 3.0 software. The results show that ecommerce browsing and the big five models have a significant effect on urge to purchase and impulse purchase. E-commerce usage intensity and urge to purchase have no effect on impulse purchases and the results of the mediation role test of urge to purchase have no effect on e-commerce browsing, e-commerce usage intensity, and the big five model on impulse purchases. e-commerce purchases. quality of also the of each to provide to


Analysis of Factors Contributing to Impulse Buying
Behavior of E-Commerce Users access a product. The existence of online channels and information technology has triggered impulsive buying behaviors among consumers through increased accessibility to products and services because it facilitates the purchase process [3].
Many experts agree that the lack of planning is a contributing factor to the purchase to be categorized as impulsive [4]. Through internet facilities by browsing consumers can feel like window shopping in a mall. The time consumers spend browsing e-commerce can trigger interest in purchasing the desired goods. So that through this browsing can provide information to consumers regarding the product to be purchased. The more time allocated will provide stimulation in impulse buying because you feel the need for the product. Bweb rowser is used for product purchase activities in a timely and efficient manner to achieve the goal comfortably and efficiently at the same time a little effort. Web browsing is the initial phase of online purchases that involves shoppers searching for information and making choices through the website. Shoppers place great emphasis on information gathering and browsing during online shopping. Theintensity of Facebook usage can be used to measure Facebook usage and this scale consists of the number of friends, the average time spent per day, and six additional items about a user's connection and engagement with Facebook. Impulse purchase behavior is to make an unplanned or sudden purchase and unthinkable in advance to buy that particular product. Impulsive buyers usually have the characteristics of not doing long-term thinking. Emotionally they feel attracted to the product of the object and the presence of a desire to fulfill satisfaction [5]. According to Ref. [6] Impulse purchase behavior is experienced by consumers at the age of around 18-39 years. In other studies it was shown that women have a greater tendency in impulse purchase than men [7]. Impulse purchase behavior can be influenced by internal or intrinsic factors in the form of self-control. Personality in consumers is an important personal factor that determines and reflects a person's response to the environment in making a purchase. The psychological features of a person are reflected in his consumption behavior. This behavior of this individual person is depicted on a dimension known as the Big Five Personality [8]. Currently, the Big Five Personality which includes extraversion, emotional, conscientiousness, agreeableness and openness is considered the benchmark of personality trait theory. An individual may have all five of the Big Five Personality traits but may score high on one or several traits and lower on other traits. This research focuses on the Java-Bali area which is a strategic location in access to obtain products. In addition, according to BPS data, there are 10 metropolitan cities covering P-ISSN 2962-5955 • E- ISSN 2962-5467 Bali. In addition, the population of Java, West Java, East Java and Central Java is the province with the largest number. Looking at the data, the area can be a consideration for retailers in offering products to meet consumer needs. Based on several existing studies, researchers identified several differences in research settings. Previous research conducted by [8] shopping malls in Malang, Indonesia. [9] research focus on hypermarkets and supermarkets in India which are influenced by consumer traits and situational factors. Finally, the research [10] conducted research related to impulse purchases in Surabaya malls, Indonesia related to fashion products.
This study focuses on what factors influence impulse purchase behavior in Ecommerce users. The Big Five Model theory is the theoretical background used in this study. Ease and convenience in using Ecommerce is one of the factors consumers choose online shopping. The selection of Big Five Model (BFM) variables to be studied is related to impulse purchase behavior in Ecommerce users. It is used to fill in the gaps in previous studies that have been carried out [2]. The Big Five Model (BFM) is an accurate theory to be used as a benchmark theory of personality traits [2]. BFM has five dimensions, namely agreeableness, neuroticism, extraversion, openness, and conscientiousness. Everyone has those 5 dimensions on himself, but with different values. Previous research, stated that BFM affects online consumer behavior.
In previous studies, impulse purchases were only associated with F-commerce browsing and F-commerce usage Intensity only. F-commerce browsing was found to have no significant effect on impulse purchases but had a positive effect on the impulse to make purchases, while F-commerce usage intensity was significant in influencing impulse purchases.
The contribution of this study is in accordance with the advice of the study [2] to test the role of the Big Five Model Theory (BFM) in solving what factors affect impulse purchase behavior in Ecommerce users. The personality of the individual has a tremendous impact on impulse purchases [11]. Different personalities, ages and genders will also give a tendency to engage in different impulse purchase behaviors. The use of age and gender moderation is based on the findings that impulse buying behaviors are experienced by consumers at the age of about 18-39 [6] and are more carried out by women [7]. BFM has also been shown to have a relationship with urge to purchase in studies conducted by [2]. In the same study, it was shown that personality traits in the BFM theory have an important role in relation to urge to purchase and will eventually affect impulse purchase. Urge to purchase indicates the condition under which desire arises when facing an object, which means that this happens before the appearance of impulse buying [12]. This study aims to explain the relationship between Ecommerce browsing, urge to purchase, usage intensity, Big Five Model, and impulse purchase.

A. E-commerce browsing dan urge to purchase
Ref. [12] explains that websites used by retailers have a significant impact on unplanned purchases or impulse buying. Through browsing sites, one of them with the platform becomes a situational factor that can encourage impulsive purchases by consumers. This situation is caused because the internet has browsing facilities for all circles of society. So that it encourages people to shop anytime and anywhere. Ref. [13] browsing in the research conducted is divided into hedonic browsing and utilitarian browsing which encourages to behave impulse buying. In addition, these browsing variables motivate consumer searches.
Browsing is the first step for consumers in finding information and making decisions. The reason for this is that some consumers spend time browsing rather than purchasing. There is an influence of browsing on the urge to buy. In the study, it was explained that users by browsing while shopping, will give a higher probability of unplanned purchases than nonbrowsing. The situation is caused by browsing for a long time will get stimulation to make purchases unplanned.
H1. E-commerce browsing has a positive effect on the urge to purchase

B. E-commerce usage intensity dan urge to purchase
Ref. [2] say that e-commerce consumers can shop at any time and from any location. They can now browse the internet easily due to these ubiquitous ecommerce features. As a result, e-commerce allows online customers to visit their favorite e-commerce pages regularly. The likelihood of consumers being interested in a particular item increases as the frequency of browsing e-commerce sites increases. The pleasant mood of the consumer has to do with the desire to make impulse purchases. Social networks can increase a person's self-esteem. An increase in self-esteem as a result of the use of online social networks can lead to a loss of selfcontrol and more impulsive behaviors. They went on to say that the frequency with which people use e-commerce can cause them to make irrational decisions, such as increasing their spending. Based on the already existing literature, we propose hypotheses: H2. E-commerce usage intensity has a positive effect on the urge to purchase

C. Big Five Model dan urge to purchase
Urge to purchase is "a state of desire experienced when facing an object in the environment.
This clearly precedes the actual and spontaneous and sudden action of impulses". Urge to purchase is associated with a poor ability to intentionally suppress stronger or automatic responses. Although there is no established theory linking the big five model to the urge to purchase, there is a correlation between consumer responses to product design and openness to experience does exist. "The degree at which an individual feels the urge to acquire things P-ISSN 2962-5955 • E-ISSN 2962-5467 that have an attractive design (e.g., 'If the product design really "talks" to me, I feel compelled to buy it.')" is how the "Response" scale is measured. Based on these findings, we believe that the big five models will have a huge impact on the urge to purchase in e-commerce, leading to the following hypothesis: H3. Big Five Model has a positive effect on the urge to purchase

D. Urge to purchase dan E-commerce Impulse Purchase
Urge to purchase is a spontaneous behavior and precedes impulse purchase behavior.
Existing literature shows a significant relationship between urge to purchase and impulse purchase [2]. The impulse to buy is impulsively high then they tend to make impulsive purchases. in line with other studies. findings from Huang (2016) stated that urge to purchase was found to be significant in predicting impulsive buying behavior. Consumers will be aroused in impulsive behavior when looking around the store. However, in this study, impulses are explained offline by searching the store. Besides they can't prevent not to impulse buying despite trying their best. Urges have a positive and significant relationship with impulse buying.
But peer communication has a bigger relationship for people to do impulse buying. Urge to buy has no effect because it only happens when before consumers do shopping impulsively. Based on the already existing literature, we propose hypotheses: H4. Urge to purchase has a positive effect on E-commerce Impulse Purchase

E. E-commerce browsing dan E-commerce Impulse Purchase
Exploring the initial stages of online purchases made by buyers to find information about related products/services through ecommerce or websites. Previous literature found a positive relationship between internet browsing and online impulse purchases. This is due to the tendency of people who have the pleasure of browsing and ignoring outcomes, tend to give rise to hedonists who have an impact on buying impulses. In contrast to previous findings, Ref. [12] found an insignificant relationship between e-commerce browsing and impulse purchases.
Browsing is a search for information carried out by consumers, for the later stage of decision making. This browsing provides sensory stimulation. However, the results showed that browsing was insignificant. This is due to the possibility that the time consumers spend looking through products does not fully encourage impulsive buying behavior. In addition, because the context of this search is only the scope of Facebook. Both directly and mediation browsing had a significant influence on impulse buying. Browsing provides hedonistic motivation in making impulsive purchases. Consumers who browse an e-commerce site will get their own pleasure in finding information about a product. They are more free and widely informed although it is not certain to actually buy or not. Therefore, we propose a hypothesis: H5. E-commerce browsing negatively affects Ecommerce Impulse Purchase

F. Ecommerce usage intensity dan E-commerce Impulse Purchase
Ecommerce makes it easier for customers to reach all stores not limited to distant times and locations so that it encourages impulse purchases more [7]. found that the increase in the use of f-commerce drove an increase in purchases. Therefore, we propose a hypothesis : H6. E-commerce usage intensity has a positive effect on E-commerce Impulse Purchase

G. Big Five Model dan Ecommerce Impulse Purchase
Personality is one of the factors that causes a person to make purchases unplanned. There are personality traits that explain individual differences. They will get low or high marks on certain dimensions. There is an influence of personality on impulse buying. Especially openness to change has a positive effect. In addition, some dimensions have a positive and negative relationship. The dimensions that have a positive influence on impulse buying are openness but not significant, extraversion, while negative are conscientiousness, agreeableness, neuroticism. Therefore, we propose a hypothesis : H7. Big Five Models positively affect Ecommerce Impulse Purchase

H. The role of mediation urge to purchase
Both directly and browsing mediation had a significant influence on impulse buying. In this study, the focus of the product is fashion, impulse buying has a relationship with browsing which influences impulsive buying. Context of impulse buying in digital celebrities or influencers in encouraging motivation to do impulse buying. Searching for information through those celebrities in a virtual environment can convince others. Their role becomes an agent to shape the behavior of their audience. Urge to purchase has a significant positive relationship with browsing and impulse purchase. It is urgent as mediation has an insignificant influence.
The context of this study is offline impulse in a mall. While the independent variable is the store environment, this is influenced by the lack of attractiveness of the stores in the Mall. If a site or browsing feature is not interesting, it does not cause an urge to buy. This urgent relationship in mediating is positive. Research shows that urge has a positive relationship in mediating but is not significant because of environmental differences. Examples of the intended environment are visual appeal, quality of service and characteristics of customers especially women.

H8. Urge to Purchase mediates the relationship between E-Commerce Browsing and E-Commerce Impulse Purchase
The psychological effect of personality traits provides a construction of understanding of consumers who have different reactions in various conditions. Urge to buy has a role can play as a mediation of emotions such as the anticipated regret that retailers can successfully engage deep feelings in the minds of consumers on impulse purchases. According to research by Ref. [12] urge to buy can mediate consumers in making spontaneous purchases. Urge to P-ISSN 2962-5955 • E-ISSN 2962-5467 buy impulsively is a picture of the state of consumers who want an object in a certain environment. Previous studies on impulsive purchases have shown that it is difficult to control urge during proximity to products, while attractive displays also create more urge to buy. The main dependent variable is impulse purchase and that includes the purchase of real goods or the satisfaction of impulses. Therefore, the higher the urge to buy consumers, the tendency to make impulsive purchases also increases. Therefore, we propose a hypothesis:

A. Research Design
This research is a quantitative research with the aim of testing the influence between variables and conducting hypothesis tests. The type of data used is primary data. Primary data is obtained from the results of distributing questionnaires online and is cross-sectional. The population in this study is e-commerce consumers in Indonesia in the age range of 16-64 years.
This population was selected based on the age range, percentage of users and e-commerce purchase activity in Indonesia according to GWI data survey results of users aged 16 to 64 years have high purchasing power on e-commerce purchases. The percentage of using online shopping applications on HP/Tablet is 78.2%, buying products online is 87.1%, visiting retail websites and online stores is 87.3%. The category of e-commerce purchases based on the age 55 of 16-24 years is 84.8%, the age of 25-34 years is 88.5%, the age of 35-44 years is 89%, and the age of 45-54 years is 89.4%.
Sampling in this study was carried out in a non-probability manner. The sample was selected using purposive sampling, with the criteria that active e-commerce consumers in Indonesia at least make one transaction in the age range of 16-64 years. The selection of this sample was carried out based on data from the GWI survey about e-commerce activities in Indonesia. This study used partial least square-structural equation modeling (PLS-SEM).
Therefore, it is necessary to take into account the sample amount required for analysis using the method (PLS-SEM). To determine the sample size using the Cohen approach based on statistical power and effect size when determining the minimum sample size, for statistical power 80%, significance level 5 %, minimum R2 10%, with the maximum number of arrows leading to a construct amounting to 7, then the sample size required is 137. However, a large sample size can improve the accuracy and consistency of PLS-SEM estimation results.
Therefore, the sample in this study will be greater than ten times the number of structural paths in the model.

B. Data Analysis Techniques
This study uses quantitative analysis to test hypotheses or measure the influence between variables with structural equation models (Structural Equation Modeling / SEM). There are two types of SEM that are widely known, namely covariance-based SEM and partial least square SEM. This study used a partial least square structural equation model. Partial least square (PLS-SEM) is a type of SEM that aims to test predictive relationships between constructs by looking at whether there is a relationship or influence between the constructs. The consequence of the use of PLS-SEM is that the test can be carried out without a solid theoretical basis, ignoring some assumptions and parameters of the accuracy of the prediction model judging from the value of the coefficient of determination (R-Square) [14], for which reason the use of PLS-SEM is very appropriate in this study aimed at developing a theory. The estimation of parameters obtained with PLS can be categorized into three [15].The first category, is the reflective measurement model, which is the weighting estimate used to create the latent variable score, then the second category of formative measurement models reflects the estimated path connecting the latent variable and between the latent variable and its indicator block (charge). Further, the third category is structural models relating to the average variance and location of parameters (regression constant values) for indicators and latent variables. A summary of the data analysis index criteria can be seen in Table 1.  Test criteria: H0 is not supported if the critical ratio value is significant, greater than or equal to 1.96.

A. Data Description
Data collection was carried out by distributing questionnaires to respondents through online dissemination using google forms and through whatsapp grub to reach a wider scope of  Table 2.  Table 3. The result of the AVE value contained in Table 5 for all the constructs in this study, some have met the criteria, which is greater than 0.5. This means that the convergent validity test based on the factor charge and the AVE value has already been met, because on average the variance value described by each indicator present in each construct tested is greater than the error value on the construct, so that all existing indicators can explain its construct compared to other factors that are not measured in this measurement [19] . After conducting convergent validity testing, the next stage in the construct validity test is a discriminant validity test that aims to measure the extent to which a construct is completely different from one another. The high validity of discriminants provides evidence that a construct is unique and captures some phenomena that cannot be measured by any other 59 construct. The results of the calculation of convergent validity for each of the constructs contained in this study can be seen in Table 4. Table 4 shows the result of the calculation of the validity of the discriminant by looking at the value of the Fornell-Larcker Criterion having a value greater than the value of the quadratic correlation between variables. A construct is said to have discriminant validity if the indicator has the highest loading value (AVE root) in its own construct group [15]. The value indicates that the validity of the discriminant has been fulfilled, so it can be concluded that each variable is able to explain something unique and different from one another.

C. Reliability Test Results
After testing the validity of the construct, the construct reliability test is then carried out.
Reliability tests are carried out to find out the extent to which the measuring instruments (instruments) used in research are consistent in measuring [17]. Composite reliability is considered to have reliability in presenting a measure of reliability in research using structural equation models. A construct can be said to have a good reliability value if it has a value greater than 0.7 [15]. The Reliability Test in this study was carried out using the help of the SmartPLS 3.0 analysis tool. Detailed data on the calculation of composite reliability shoewed that all constructs in this study have a composite reliability value of > 0.7 which indicates that all proposed constructs have good reliability.

D. Formative Measurement Model Evaluation
The second step in evaluating the results of PLS-SEM involves the examination of the formative measurement model [15]. For constructs measured formatively, convergent validity is assessed by the correlation of the construct with alternative measures of the same concept and is already described in the evaluation of reflective measurement models.
At the testing stage of the formative measurement model, there is an evaluation based on the cholinearity of indicators using the variance inflation factor (VIF). According to the criteria of the VALUE OF VIF 5 or more indicates a problem of critical cholinearity among formatively measured constructive indicators. However, cholinearity problems can also occur at VIF values lower than 3. Ideally, the VIF value should be close to 3 and lower [15]. to provide an explanation of the construct being measured. The results of the estimation of the structural model are presented in the form of a model chart on Fig. 2.

E. Structural Model Evaluation
Hypothesis testing in research using the PLS-SEM method is important to understand that adjusting the model to the sample data in order to get the best parameter estimates is by maximizing the variance described from endogenous latent variables [18]. There are stages that need to be carried out in the evaluation of structural models, namely the significance test  Table 5 for the results. can also be stated that H10 is not supported by a p value > 0.05 or 0.373.

Discussion
The results of this study prove that e-commerce browsing has a significant effect on the urge to purchase. The results of the study are in line with those carried out by Ref. [12], show that there is a significant influence of browsing on the urge to purchase. Through browsing, one of them with the platform becomes a situational factor that can encourage impulsive purchases by consumers. This situation is caused because the internet has facilities in browsing for all circles of society [13]. So that it encourages people to shop anytime and anywhere. The more often consumers search for information (browsing) on online media, it affects the level of purchases impulsively in online stores. This is because consumers sometimes search for information on online media to add shopping references so that the possibility of a purchase impulse when the consumer is browsing relatif is high.
The results of this study prove that e-commerce usage intensity has an insignificant effect on the urge to purchase. The results of the study are different from previous studies conducted by Ref. [2] shows there is a significant positive influence between [20] usage intensity and urge to purchase. This result explains that if the higher the intensity of ecommerce use, it is likely that there will not be a strong impulse from within to make purchases on e-commerce shopping sites. This can be caused because with the low intensity of using ecommerce, users will look around the user's home page more often. With this, it can be predicted that the post or store in the e-commerce that is shared does not have a big influence on users in providing encouragement to make purchases.
The results of this study prove that the big five models have a significant effect on the urge to purchase. There is correlation between consumer response to product design and openness to experience does exist. The degree at which an individual feels an urge to acquire things that have an attractive design is how the "Response" scale is measured. The results of the study in line with those conducted by Ref. [12] show that there is a significant influence of the big five model factor on the urge to purchase. Through the picture of consumer personality as measured through the dimensions of the big five models, consumers get the impetus to make purchases of an object in the e-commerce environment.
The results of this study prove that urge to purchase has an insignificant effect on ecommerce impulse purchases. The results of the study have an inequality of the research carried out by Ref. [2] showed that urge to purchase has a significant positive effect on impulse purchases. This explains that consumers with a low drive to buy have no tendency to make a purchase compared to consumers whose level of drive to buy it is higher. The impulse to buy becomes a variable that precedes the variability of impulse buying behavior. Based on the results of this research, it shows that e-commerce activities still do not have a big influence in providing encouragement to make purchases. This can be caused because the quality of information and also images of the content of each product is still unable to provide interest to users.
The results of this study prove that e-commerce browsing has a significant effect on ecommerce impulse purchases. The results of the study in line with those conducted by Ref. [12] showed that there was a significant influence of browsing on impulse purchases. This shows that when consumers on e-commerce platforms browse (search for information), there will be impulsive purchases. That is, the higher their level of information search, the higher the impulsive purchase rate will be. This happens because when someone is happy with looking through the content of the website, comparing products, looking for information, then without realizing it, the urge of the heart to buy arises by itself so that the possibility of an impulse purchase will also be higher. This happens because some Indonesian e-commerce platforms provide very large discounts.
The results of this study prove that e-commerce usage intensity has an insignificant effect on e-commerce impulse purchases. The results of the study have an inequality of the research carried out by Ref. [2] showed that there was a significant positive influence between usage intensity and impulse purchase. Where the development of technology makes consumers faster to shop online, because they no longer spend time on the way to the store, but simply via cellphone, and online purchases can be made anywhere and anytime. But online purchases can not always encourage someone to make purchases impulsively, because in online shopping, it could be that buyers think they cannot see and touch the goods directly to assess the quality of the goods and refuse to make purchases. This can be caused because with the low intensity of using an e-commerce platform, users will more often see products that are not in demand on the user's home page.
The results of this study prove that the big five models have a significant effect on ecommerce impulse purchases. These results are in line with the research of Ref. [12], showing that there is a significant influence of the big five model factor on impulse purchases.
Personality is one of the factors that causes a person to make purchases unplanned. There are personality traits that explain individual differences. They will get low or high marks on certain dimensions. There is a personality influence on impulse purchases. Especially openness to change has a positive effect.
The results of this study prove that e-commerce browsing, e-commerce usage intensity, and the big five models have an insignificant effect on e-commerce impulse purchases through urge to purchase. The results of this study are different from previous studies [2] who explained that there is a relationship between browsing, usage intensity and the big five personality model factor to impulse purchase. Urge to purchase is a strong desire to make a purchase that arises when facing an object and this desire is a condition that occurs before an impulsive purchase is made, but for e-commerce users who see a product continuously do not experience impulsive impulses. Highinformation searches affect impulsive purchase rates in certain situations. This is because consumers sometimes search for information on online media to add shopping references. One of the differences in these results is influenced by demographic factors. Age has a negative influence on impulse buying behavior, meaning that the older a person is, the lower the tendency to impulsive purchases. Another demographic factor that decreases purchasing behavior is income. Consumers who have higher incomes are shown to have more tendency to have impulsive purchasing behavior compared to those with low incomes. Low-income individuals tend to use Ecommerce not for purchases, but just to look at it or just to add shopping references. Facts have proved that a low-income individual is less likely to manage to buy all the goods he wants.

Conclusion
The purpose of this study is to determine the factors that influence impulse purchase behavior in e-commerce users. Some of the new findings have been validated in the context of e-commerce. The results of the study found that e-commerce browsing and the big five models had a significant effect on the urge to purchase and impulse purchase. E-commerce usage intensity and urge to purchase have no effect on impulse purchase. In addition, there is no significant influence of the role of urgent to purchase mediation in e-commerce browsing, ecommerce usage intensity, and the big five model on impulse purchases. In summary, the findings of this study will allow practitioners to gain more insight and understanding of ecommerce while providing useful advice to e-commerce players in the drive to buy and impulse purchases among e-commerce consumers. For future research can add external factors such as situational factors. In addition, future research needs to add other variables such as lifestyle, seeing the existence of increasingly developing technology that encourages individuals to compete to show their social class.

Conflict of Interest
Authors declare that there is no conflict of interest.