The Effect of Brand Ambassador, Website Quality, and E-WOM on Purchase Decision in Shopee E-commerce

. This study aims to examine the effect of Brand Ambassador, Website Quality, and Electronic Word of Mouth (E-WOM) partially and simultaneously on Purchase Decisions in Shopee e-commerce. This study uses a quantitative approach with primary data sources collected through questionnaires. The data sample used was 125 respondents from the Management Study Program FEB-UMM (Faculty of Economics and Business, University of Muhammadiyah Malang) as Shopee consumers. The sampling technique is non-probability sampling using the purposive sampling method which gives certain criteria to respondents. In analyzing the data of this study using multiple linear regression analysis tools. The results of this study are Brand Ambassador, Website Quality, and E-WOM have a positive and significant effect on Purchase Decision partially, and simultaneously have a significant effect on Purchase Decision in Shopee e-commerce.


BACKGROUND
The purchase decision is one of the main components in consumer behavior that leads to the purchase of a product or service (Ramli, 2020). Purchase decision is a basis that determines consumer action in buying a product (Pratiwi, Juliati, & Sa'diyah, 2022). A purchase decision is one of the important factors that need to be identified by a marketer, because with this social factors will emerge that need to be developed so that consumer enthusiasm in making purchasing decisions increases (Hanum & Hidayat, 2017). Purchasing decisions are one of the important aspects that companies need to pay attention to, where a company must be able to attract consumers to decide to buy. In conditions of increasingly fierce competition, every company needs to monitor competition in its business world, and efforts are needed to win the market (Pratama & Santoso, 2018). In winning the market, companies must come up with interesting ideas and analyze them so that consumers continue to use the products or services that have been offered by the company (Prasetya, Yulianto, & Sunarti et al., 2018).
Brand Ambassadors can be termed as supporters of a brand by using celebrities and are chosen interchangeably. Celebrities are known as one of the psychological supporting factors that can influence consumer attitudes and beliefs about a product (Ramadhan & Saputri, 2021). Brand Ambassadors also act as advertising promotion models (Faradasya & Trianasari, 2021). Brand Ambassadors can be judged as advertising attractiveness and able to influence consumer preferences, because Brand Ambassadors can be a trendsetter for the products sold by the company. Brand Ambassador is one of the strategies used by companies in the product marketing process with the aim of increasing sales to the wider community (Nuryanti & Istiyanto, 2020). The right use of Brand Ambassadors can improve consumer purchasing decisions (Brestilliani & Suhermin, 2020).
Website quality is a seller's effort in displaying products and information that aims to attract buyers in making purchasing decisions. In this case, a good quality website is needed because a consumer will be more interested in a website that has design features and is used effectively (Chen, Huang, & Davison, 2017). Website quality is something that all online sellers need to pay attention to. Website quality has an important influence on consumers in the buying process. Therefore, a good website will have a good impact on the company, i.e. a consumer to make a purchase decision. Because of buyers will be interested in the quality of the website and can make purchasing decisions if they can use it easily and efficiently (Lelani, Welsa, & Cahyani, 2019). In e-commerce, the role of the website is very important in attracting, and maintaining consumer enthusiasm in making purchasing decisions. Because consumers expect a high-quality website when they shop online. In this case, improving the website by the company can increase the effect on its users so that it affects purchasing decisions (Sa'diyah, Nur, & Kusuma, 2017). The website is no longer just an information system but also a marketing channel (Ardhi & Wijaksana, 2020). This means that the better the quality of the website, the higher the enthusiasm of consumers in making purchasing decisions (Harsanto & Hidayat, 2017).
The emergence of the internet provides a new pattern in Word of Mouth communication, and this is the beginning of the emergence of the term Electronic Word of Mouth (E-WOM) (Kartika & Ganarsih, 2019). E-WOM is an activity to disseminate information about products, brands, services, or companies that are considered important as sources of information in influencing human behavior through the internet (Ismagilova, Slade, Rana, & Dwivedi, 2020). E-WOM is carried out through social media networks so that consumers can experience more effective and practical communication that can positively influence consumers in making purchasing decisions.
The shift in behavior that occurs at this time, makes people prefer shopping with an online system. Based on the survey results of the Indonesian Internet Service Providers Association (APJII) regarding internet users for 2019-2020 (Q2), it shows that the number of internet users in Indonesia has increased by 8.9%, equivalent to 196.7 million users. This proves that most people in Indonesia have access to the internet. Based on this phenomenon, it has opened up new opportunities for online business people. Shopee is one of the many parties who take advantage of this opportunity (Nurjanah, Kurniati, & Zunaida, 2019).
Shopee is an e-commerce company that uses a customer to customer (C2C) system, which is an online transaction activity in a Marketplace space. Users can present their products on the site (C2C) so that other users can see and be interested in buying (Siregar et al., 2020). Shopee offers various features that make it easy for sellers and buyers to interact with each other quickly and easily (Kartika & Ganarsih, 2019). Shopee also provides attractive offers in conducting transactions for its users (Sastika, 2018). Therefore, Shopee has the most number one downloaders in Indonesia and is the Top 1 most website visitor in Southeast Asia (Japarianto & Adelia, 2020). Based on the explanation of the background data, the purpose of this study is to examine the effect of Brand Ambassador, Website Quality, and E-WOM on Purchase Decisions at Shopee's e-commerce.

THEORETICAL REVIEW
A brand ambassador is someone who describes the best image of a product in an advertisement. This person is usually from among celebrities or other popular people (Rozalia, Rinuastuti, & Rusminah Hs., 2020). Brand ambassador is generally synonymous with a celebrity who is already well-known in the public's mind. Celebrities are known as one of the psychological supporting factors that can influence consumer attitudes and beliefs about a product (Ramadhan & Saputri, 2021). Brand ambassador is one of the strategies used by companies in the product marketing process with the aim of increasing sales to the wider community (Nuryanti & Istiyanto, 2020). Brand ambassador variable indicators are visibility (popularity), credibility (credibility), attraction (attractiveness), and power (strength) (Nurhasanah et al., 2021). Various studies have shown that brand ambassadors have an influence on purchase decisions, i.e. in research of Nurhasanah et al. (2021); and Dewi, Edyanto, & Siagian (2020). Based on the explanation above, this research formulate the first hypothesis (H1):

H1: Allegedly brand ambassador affects purchase decisions on Shopee e-commerce.
Website quality is an effort by sellers to display products and information that aims to attract buyers in making purchasing decisions. Website quality is one of the techniques or methods in measuring quality on a website based on end user perceptions (Safuan & Khuzaini, 2018). Website quality can be reviewed as an attribute of a website that contributes to its usefulness to consumers. There are three dimensions to website quality, namely in the usability dimension with indicators that the Shopee website is easy to use or operate, the Shopee website has a good and attractive appearance; on the information quality dimension there are indicators that the Shopee website provides accurate information, the Shopee website provides up to date information, the Shopee website provides easy-to-understand information, the Shopee website provides detailed information; On the service interaction dimension, there is an indicator of the Shopee website ensuring security in transactions, the Shopee website is known as a trusted site (Barnes & Vidgen, 2002). Various studies have shown that website quality has an influence on purchase decisions, i.e. in research of Barnes and Vidgen (2002), and Safuan and Khuzaini (2018). Based on the explanation above, this research formulate the second hypothesis (H2):

H2: Allegedly website quality affects purchase decisions on Shopee e-commerce.
Electronic word of mouth (E-WOM) is a positive or negative statement made by actual, potential, or previous consumers of a product or company where this information can be reached by the wider community through internet media (Sari, 2019). E-WOM is carried out through social media networks so that consumers can experience more effective and practical communication that can positively influence consumers in making purchasing decisions (Pristiwasa & Widodo, 2019). There are three dimensions to E-WOM, namely in the intensity dimension there are indicators of the frequency of accessing information, the frequency of interacting with users, the number of reviews written by social media users; in the content dimension there are indicators of product choice information, product quality information, price information, security information in transactions and the internet. network site. which has been provided; on the valence of opinion dimension there are indicators of positive statements from social media users, as well as recommendations from social media users (Goyette, Ricard, Bergeron, & Marticotte, 2010). Various studies have shown that E-WOM has an influence on purchase decisions, i.e. in research of Pristiwasa and Widodo (2019); Rupayana, Suartina, & Mashyuni (2021); Sari (2019); Santosa (2019); and Purnamasari and Yulianto (2018). Based on the explanation above, this research formulate the third hypothesis (H3):

H3: Allegedly E-WOM affects purchase decisions on Shopee e-commerce.
The purchase decision is a consumer's final act in deciding to buy a product or service (Nuryanti & Istiyanto, 2020). Purchasing decisions are the result of a thought process on choices made based on several alternative choices used in the buying process (Nurjanah et al., 2019). In this case, a consumer will go through a process of assessing to choosing from these various alternatives to determine an option that is considered the most profitable according to his interests (Keren & Sulistiono, 2019). Purchase decision indicators include stability in making purchases at Shopee, habits in making purchases at Shopee, providing recommendations to others in making purchasing decisions at Shopee e-commerce, making repeat purchase decisions at Shopee e-commerce (Kotler & Keller, 2009). Nurpermana (2019) examines the influence of website quality, brand ambassador, and electronic word of mouth (E-WOM) on decisions purchases on ecommerce in the Yogyakarta Province community and have results findings that the quality of the website, brand ambassador, and electronic word of mouth (E-WOM) has an influence together (simultaneously) to purchase decisions. Based on the explanation above, this research formulate the fourth hypothesis (H4):

H4: Allegedly brand ambassador, website quality, and E-WOM affect purchase decisions on Shopee e-commerce simultaneously.
Based on the theoretical review and previous research, this research model is illustrated in Figure 1.

RESEARCH METHODS
This type of research is explanatory research with a quantitative approach. The location of this research was conducted in Malang City. This research uses primary data type, where primary data is obtained directly by the researcher. Primary data was obtained by distributing questionnaires to respondents. The scale used to measure the answer score in this study was using a Likert scale. The Likert scale is used to measure agree or disagree with certain subjects, objects or events (Sayidah, 2020). This scale consists of five alternative answers, i.e. Strongly Agree (5); Agree (4); Enough (3); Disagree (2); and Strongly Disagree (1). The population in this study were all students of the FEB-UMM Management Study Program as Shopee consumers. In taking the number of samples using the indicator formula multiplied by 5 (Ferdinand, 2014 In this study there were 25 indicators, so the total sample was 125 respondents. The sampling technique in this research is non-probability sampling using purposive sampling method where this sampling aims to identify the types of people that can provide information as desired based on criteria that have been determined by researchers (Sayidah, 2020). The sample used in this study consisted of the following criteria: Respondents who have accessed the Shopee application or website; Respondents who have read previous customer reviews that have make a purchase at Shopee via the internet, before making a purchase on Shopee e-commerce; and Respondents who have made purchasing decisions on Shopee e-commerce. In analyzing the data, this study used multiple linear regression analysis tools. In addition, this research also uses instrument test (validity, reliability, and normality test), classical assumption test (multicollinearity, autocorrelation, and heteroscedasticity test), coefficient of determination, hypothesis test (F-test and t-test). The definition of operational variables in this study is as follows: 1. Brand Ambassador: A celebrity who was chosen by Shopee e-commerce to promote Shopee e-commerce so that consumers are interested in making purchasing decisions on e-commerce Shopee. As for the indicators according to Nurhasanah et al. (2021) are: on the popularity that is how far is the popularity celebrity; on the credibility that is related to product knowledge that is known celebrity; on the attraction that is focus more on attractiveness celebrity; and on the power that is celebrity abilities in attract consumers to buy. 2. Website Quality: It is a component and feature in the Shopee e-commerce website that is designed as good and attractive as possible to provide convenience and convenience for Shopee e-commerce users. According to Barnes and Vidgen (2002), there are three dimensions to website quality, i.e. usability, information quality, and service interaction. The indicators are as follows: in usability there is an indicator Shopee website is easy to use or operate, the Shopee website has a nice and attractive appearance; in information quality there is an indicator the Shopee website provides accurate information, the Shopee website provides up to date information, the Shopee website provides easy-to-understand information, the Shopee website provides information that details; in service interaction there is an indicator the Shopee website ensures security in transactions, the Shopee website is known as a trusted site. 3. E-WOM: Is a review containing a statement positive as well as recommendations from regular and ex-customers customers to e-commerce Shopee through internet media. According to Goyette et al. (2010) there are three dimensions to E-WOM, namely intensity, content, and valence of opinion. The indicators are follows: in intensity there is an indicator the frequency of accessing information, the frequency of interacting with users, the number of reviews written by social media users; in content there is an indicator information on product choices, information on product quality, information on prices, information on security in conducting transactions and internet networking sites; and in valence of opinion there is an indicator that have been provided, positive statements from social media users, as well as recommendations from social media users. 4. Purchase Decision: It is a final decision that is owned by a consumer in doing shopping on e-commerce Shopee. According to Kotler and Keller (2009) purchase decision indicators include stability in making purchases at Shopee, habits in making purchases at Shopee, providing recommendations to others in making purchasing decisions at Shopee e-commerce, and making repeat purchase decisions at Shopee ecommerce.

Validity test
The validity test is an index test that illustrates that the instrument in the study is correct in measuring what is to be measured (Amanda, Yanuar, & Devianto, 2019

Reliability test
Reliability test is an index test that aims to determine the extent to which the instrument can be trusted or relied on (Amanda et al., 2019). Criteria in making decisions on reliability tests can be measured using the Cronbach's alpha formula with the condition that if the Cronbach's alpha coefficient>0.6, then it can be said to be reliable. The results of the reliability test results are shown in Table 2.

Normality test
The normality test is a test that has the aim of knowing whether the sample from the population has a normal distribution (Darma, Firdaus, & Haryadi, 2016). Testing for normality using the Kolmogorov-Smirnov test, the data is said to be normal if the resulting significance value is more than 0.05 (5%). The results of the Kolmogorov-Smirnov test can be seen in Table 3.

Multiple Linear Regression Analysis
Linear regression analysis is a statistical technique that is used to create models and investigate the effect of one or several independent variables on the response to dependent variables (Nihayah, 2019). In this study, multiple linear regression analysis aims to determine the effect of Brand Ambassador, Website Quality, Electronic Word of Mouth (E-WOM) on Purchase Decisions. The results of processing regression data using SPSS is shown at equations [1].

Y = ----------------------------[1]
The constant value (a) is 2.894, it means that if there is no brand ambassador, website quality, and electronic word of mouth or equal to zero, the purchase decision has a fixed value of 2.894. The coefficient value (b1) of brand ambassador has a positive effect based on the regression coefficient value of 0.112 or 11.2%. It can also be interpreted that the brand ambassador has a positive (unidirectional) impact on the purchase decision of 11.2%. Coefficient value (b2) of website quality has a positive effect based on the regression coefficient value of 0.163 or 16.3%. It can also be interpreted that website quality has a positive (unidirectional) impact on purchase decisions by 16.3%. The coefficient value (b3) of E-WOM has a positive effect based on the regression coefficient value of 0.179 or 17.9%. It can also be interpreted that E-WOM has a positive (unidirectional) effect on purchase decisions by 17.9%.

Multicollinearity test
The multicollinearity test was carried out with the aim of determining whether a model has multicollinearity symptoms. In this study, the multicollinearity test was carried out by means of the VIF test, this test was carried out by looking at the Tolerance value if it was greater than 0.10 then multicollinearity did not occur and if the VIF value was less than 10 then multicollinearity did not occur (Perdana K., 2016). The results of the multicoloniarity test can be seen in Table 4. Based on the Table 4, it can be concluded that the three independent variables have a tolerance value above 0.10 and a VIF value below 10.0. It can be stated that brand ambassador, website quality, and electronic word of mouth (E-WOM) is free from multicollinearity symptoms.

Autocorrelation Test
The autocorrelation test is a test that aims to test whether there is a correlation between the confounding error in period t and the confounding error in period t-1 in the linear regression model. To detect autocorrelation symptoms, it can be tested using the Durbin-Watson test (d) by looking at the Durbin-Watson table. If the value of dU<d<4-dU, then there is no autocorrelation symptom in the regression model (Perdana K., 2016). The autocorrelation test results can be seen in Table 5. Based on the Table 5, it can be concluded that there is no relationship between confounding error in period t and confounding error in period t-1 in the linear regression model. This is obtained from the value of Durbin-Watson test, the test results are between dU<d<4-dU with a value of (1.757<1.965<2.242).

Heteroscedasticity Test (Park test)
A good regression model does not occur heteroscedastically by looking at the significance value > 0.05. Park Test results can be seen in table 6. Based on the table above, it shows that the significance (sig) of Brand Ambassador (X1)=0.251; (sig) Website Quality (X2)=0.146; and (sig) Electronic Word of Mouth (X3)=0.264 where the significance of the three variables is more than 0.05, so the model is free from heteroscedasticity.

Coefficient of Determination
The coefficient of determination aims to determine how much the ability of all independent variables to explain the variation of the dependent variable (Tnius, 2018). The results of the coefficient of determination are shown in Table 7. Based on Table 7, it can be concluded that the value of the coefficient of determination is 0.579. So, it can be stated that 57.9% of variation in dependent variable is determined by the independent variables in this resaerch with moderate category, the remaining 42.1% is determined by other factors.

F-test (simultaneous effect)
F-test is used with the aim of testing whether the independent variables on the dependent variable have a significant effect simultaneously (Sugiyono, 2017). The basis for making the decision, if the significance value of Sig. smaller than 0.05, then the independent variable has a simultaneous effect. The results of the simultaneous hypothesis test (F-test) are described in the Table 8. Based on table 8, it can be seen that the calculated F value is 57.947 which indicates that the calculated F value is greater than the F table value of 2.68, and has a p-value (sig.) 0.000 <0.05, it can be concluded that the test Simultaneous hypothesis (Test F) of Brand Ambassador, Website Quality and E-WOM variables on the dependent variable, namely Purchase Decision, has a significant effect.

T-test (partial effect)
Hypothesis testing using (t-test) aims to determine the effect of each independent variable partially (alone). In this case, hypothesis testing will be carried out using a significance level of 0.05 (α=5%) and a t-table of 1.657. The t-test results can be seen in Table 9.  Table 9 by observing the row, column t and sig. can be explained as follows. The significance value of the brand ambassador variable is 0.041<0.05 and tcount is 2.063>1.657. The result has a beta value of 0.112 or 11.2%. This means that H1 is accepted. So it can be stated that the brand ambassador has a significant effect on the purchase decision. The significance value of the website quality variable is 0.000<0.05 and the t-count is 3.947>1.657. The result has a beta value of 0.163 or 16.3%. This means that H2 is accepted. So it can be stated that website quality has a significant effect on the purchase decision. In line of that result, the significance value of the e-WOM variable is 0.000<0.05 and the t-count is 5.136>1.657. The result has a beta value of 0.179 or 17.9%. This means that H3 is accepted. So it can be stated that e-WOM has a significant effect on the purchase decision.
A brand ambassador is someone who describes the best image of a product in an advertisement (Rozalia et al., 2020). Brand ambassadors can be judged as advertising attractiveness and able to influence consumer preferences, because brand ambassadors can be a trendsetter for the products sold by the company. Brand ambassador is one of the strategies used by companies in the product marketing process with the aim of increasing sales to the wider community. This study shows that there is a positive or direct and significant influence between the brand ambassador variable and the purchase decision. This illustrates that the more capable the brand ambassador is in describing the best image of e-commerce when doing promotions or Shopee advertisements, it will have an impact that more and more consumers are determined to make purchase decisions on Shopee e-commerce platform. In this study the brand ambassador variable is able to influence purchase decision. This study supports the research results of Nurhasanah et al., (2021); Nadia and Aulia (2020); and Dewi et al. (2020).
Website quality is a seller's effort in displaying products and information that aims to attract buyers in making a purchase decision. In this study, the results showed that there was a positive or unidirectional and significant influence between the website quality variable and the purchase decision. This illustrates that the better the quality of the website displayed by Shopee's e-commerce platform, both in terms of operation, quality of information, and services, the more consumers will make purchase decisions on Shopee's e-commerce platform. In this study, the website quality variable is able to affect a purchase decision. This research supports the research results of Solihin and Zuhdi (2021); Ghafiki and Setyorini (2017); and Safuan and Khuzaini (2018). E-WOM is an activity to disseminate information about products, brands, services, or companies that are considered important as sources of information in influencing human behavior through internet media. E-WOM is carried out through social media networks so that consumers can experience more effective and practical communication that can positively influence on consumers' purchasing decisions. This study shows there is a positive or direct and significant influence between the E-WOM variable and the purchase decision. This illustrates that if there are more positive reviews and recommendations from Shopee customers on the internet and social media. It will have an impact on more and more consumers who make purchase decisions on Shopee e-commerce platform. In this study, the E-WOM variable is able to influence purchase decisions. This research supports the research results of Pristiwasa and Widodo (2019) Based on this research, the results also show that the Brand Ambassador, Website Quality, and E-WOM variables have a significant and simultaneous effect on Purchase Decisions at Shopee e-commerce platform. This illustrates that if the Brand Ambassador, Website Quality, and E-WOM variables can be optimized and further improved, it will have an impact that more consumers will make purchase decisions on Shopee ecommerce platform.

Conclusions
Based on the results of this research and discussions, it can be concluded that brand ambassador has a positive and significant effect on purchase decision. It illustrates that the better Shopee e-commerce platform in choosing celebrities as brand ambassadors correctly, the better both the promotions carried out by the brand ambassadors which will later have an impact on more who will make purchase decisions. Website quality also has a positive and significant effect on purchase decisions. This illustrates that the better the Shopee e-commerce design and features, the more consumers will be interested in making purchase decisions. Furthermore, E-WOM has a positive and significant effect on purchase decisions. This illustrates that the more positive reviews and recommendations come from Shopee customers which is posted on the internet, the more stable consumers will be in making purchase decisions on Shopee e-commerce platform. Further than that, brand ambassador, website quality, and E-WOM have affected on purchase decision simultaneously. This illustrates that the better and the right brand ambassador selection, the better the Shopee e-commerce design and features, and the more positive reviews and recommendations come from Shopee customers on the internet, the more consumers will make purchase decisions in the Shopee e-commerce platform.

Recommendations
Based on the results of these research conclusions, the researcher tries to provide some suggestions or recommendations for Shopee. Shopee is expected to be able to optimize and further improve in choosing a good and appropriate brand ambassador, in order to attract the attention of consumers with the aim of influencing consumer enthusiasm in making purchasing decisions. In addition, it is expected that Shopee can optimize and improve the elements of use in the operation of the quality of its website, so that Shopee consumers can more easily operate it, improve the appearance or improve the quality of the website design. On the other hand, Shopee is also expected to optimize and improve space for customers who have executed purchasing decisions on Shopee e-commerce platform in providing their reviews or reviews, so their customers and prospective buyers can interact with other Shopee users. Therefore, the intensity of interactions can be carried out to provide information for potential Shopee customers before making a purchase decision. As well as for further researchers are expected to expand the topic of research or add other variables related to the topic of consumer purchase decisions.

Research Limitations
Based on the researcher's direct experience in the research process, there are some limitations experienced and can be a number of factors that can be given more attention to further researchers in further perfecting their research because this research itself certainly has shortcomings that need to be improved continuously in the future research. The limitations in this study, among others is the number of respondents who were only 125 people, of course, is still not enough to describe the actual situation. Besides that, the research subject is only focused on the student segment where there are still various other segments that can be used as research subjects. In the process of collecting data, the information provided by the respondents through the questionnaires sometimes does not show the respondent's actual opinions. This can happen because sometimes the thoughts, assumptions, and understandings of each respondent are different, as well as other factors, such as the honesty factor in filling out the respondent's opinion in the questionnaire.