Trust and Perceived Risk toward Actual Online Purchasing: Online Purchasing Intention as Mediating Variable

This research aims to test the influence of trust and perceived risk on Actual Online Purchasing (AOP) either directly or through Online Purchase Intention (OPI) as a mediating variable. Purposive Random Sampling used in this research with 100 respondents who have searched product through the E-Commerce website (B2C & C2C). This research uses Structural Equation Modelling (SEM) model with SmartPLS 3 program. The results of the test showed that Security practice and Privacy Concern have a negative effect on Perceived Risk, which subsequently Perceived Risk has a negative effect on OPI. However, OPI has no significant effect on AOP. Between the two of trust antecedents, the EWOM and Perceived Reputation only Perceived Reputation has a significant influence on the Trust. From the four constructs in this study (AOP, OPI, Trust and Perceived Risk), the Trust Model is the most powerful predictive model compared to other models such as Perceived Risk, OPI, and AOP.


Introduction
Recently the development of technology is influenced by many things like lifestyle, how to shop and get information, socialize or transact. These change ultimately encourages many companies to change their marketing strategies by using the Internet or digital marketing. Business Model by utilizing the Internet network where the website is used as a medium to conduct business transactions called e-commerce. There are some benefits gained by businessmen by utilizing e-commerce websites; the expansion of marketing or market share without having to open many branches or add many employees, the profit can eventually lower the operational cost, so that companies can sell cheaper and consumers are more interested in purchasing the product. The challenge for companies selling their products through ecommerce is how to make appropriate policies to build customer interest and trust to transact online so that the company can achieve competitive advantage by considering competition in the e-commerce sector is now increasing. IDEA (Indonesian E-Commerce Association) stated that the number of official e-Commerce websites increased from 37 by the end of the year 2017 to 44 in June 2018. Meanwhile APJII, survey stated that there is the number of Internet user increasing sharply from 28.552% over 20 years (1998-2017), where the most accessible services frequently are chatting services, social media, search engines, see pictures, view videos, download videos and images, articles, upload files, email, buy items, register and sell goods. As presented here the activities of selling and buying goods are at the end, It means that Internet utilization for online buying and selling transactions is still relatively low and most users access e-commerce sites only to view or survey prices on the e-commerce website.
Lack of online buying and selling transactions in Indonesia can be caused by several factors that affect online transactions such as risk factors and trusts. Koufaris & Hampton-sosa (2004) stated that the consumer confidence of e-commerce is one of the key factors for online buying and selling activities. Kim (2008) stated that perceived risk factors and beliefs have a strong influence on the buying interest that leads to purchasing decisions. The belief has a negative impact on the perceived risk in buying online and the belief positively affects the online purchasing intention (Anwar & Adidarma, 2016). In recent online consumer behavior studies, it was stated that risk and trust are important psychological factors and need to consider for online marketers in establishing its business strategy. Regarding the factors that can increase interest in online purchase, it still needs to study and research more. The results of this research are expected to be utilized for business owners in analyzing the factors affecting the consumer interest in online transactions, as well as additional insight into formulating and implementing business strategies to increase online purchase interest.

Relationship of Trust and Its Antecedent with Online Purchase Intention
Trust is one of the elements needed in business transactions, especially in online transactions that are full of uncertainty. E-commerce companies that have a good reputation will usually gain consumer trust. Firmansyah & Suwarno, (2017) found that perceived reputation positively affects the trust of Internet users in Indonesia to e-commerce systems. In addition, many consumers first check the testimonials of buyers who have shopped on the website, as well as through the media and other online communities to increase the confidence in the transaction. (Tian, Wong, & Pang, 2018) said that retail online reputation can increase trust that can ultimately increase online purchase interest, but that reputation does not directly influence the purchase interest. Choon & Corresponding, (2010) stated that trust in online positively affects online purchase intention, concerning Internet coverage, transparency, and accessibility has given new meaning to the WOM concept done conventionally become electronically based, and that is what makes marketers more interested in engaging in E-WOM activities (Kozinets et al., 2010). To exchange product information, customers start using various types of the web such as Customer review sites (testimonials), weblog, social networking sites, online discussion forums, etc. (Lee, Park, & Han, 2008). Thus, more positive information related to E-Commerce spread over the Internet network can increase the confidence of consumers to buy products through E-Commerce, EWOM positively influences the trust and with the trust can increase the consumers purchase interest through E-Commerce (Puspitasari et al., 2018). Based on the explanation above, the proposed hypothesis is as follows: H1A: Perceived Reputation has positive effect on Trust H1B: EWOM has positive effect on Trust H1C: Trust has a positive effect on OPI (Online Purchase Intention)

Relationship of Perceived Risk and its antecedent with Online Purchase Intention
McCole, Ramsey, & Williams, (2010)stated that the risk is related to how a consumer has the confidence to involve technology in the various jobs undertaken by consumers including in terms of finding the required products and purchasing the product. Perceived risk and trust have relationships with each other, where the higher the perceived risk the lower the trust in online purchase (D'Alessandro, Girardi, & Tiangsoongnern, 2012), which can ultimately affect consumer interest (Pappas, 2016). Security practice is an online store's ability to control and maintain the security of consumer transactions (Park & Kim, 2006). Doolin & Zealand, (2005) stated that perceived risk is related to security practice and privacy concern, where the higher ability of the company in maintaining the security of the customer's online transactions then the consumer perceived risk will be lower. Security practice and privacy concern applied by the company is crucial considering conducting online transactions will be required to provide personal data and with credit or debit card numbers, so that data security should be well maintained to prevent data or credit card breach. Based on the explanation above, the proposed hypothesis is as follows: H2A: Security Practice has negative effect on Perceived Risk H2B: Privacy Concern has negative effect on Perceived Risk H2C: Perceived Risk has negative effect on the Trust H2D: Perceived Risk has negative effect on Online Purchase Intention

Relationship between Trust, Perceived Risk, Online Purchase Intention and Actual Online Purchase
The development of technology-supported with the extended Internet network in Indonesia created a lot of consumers who choose online shopping as one of their shopping media. In which online shopping offers many benefits like they can save their time in choosing products, consumers can compare many variations of goods from different online shop at different prices. The decision of the purchase is an act of selection of various alternatives owned by the consumer, where the decision-making process is initiated from the introduction of problems that are then solved through the purchase of several products (Tjiptono, 2010). While purchase interest is a tendency for respondents to act before purchasing decisions are implemented. Online marketers need to consider that policies promoting trust and reducing risk are important means to increase purchases. In particular, the use of multichannel will reduce the perceived risk. There is an influence between the trust disposition and the risk aversion together against the online purchase intention (Muhibuddin & Untarini, 2014). The results of the study by (Carlos Roca, Jose Garcia, & Jose de la Vega, 2009) stated that perceived trust is an important issue in the online trading system when the company enhances the security of online trading systems can increase trust and consequently consumers will tend to use online services. Online purchase intention affects purchasing behavior (Puspitasari et al., 2018). Intentions have been determined as a prominent predictor of actual behavior for online shopping (Lim et al., 2016). Based on the explanation above, the proposed hypothesis is as follows: H3A: Online Purchase Intention has positive effect on Actual Online Purchase H3B: Trust affects Actual Online Purchase through Online Purchase Intention as a mediating variable H3C: Perceived Risk has a negative effect on Actual Online Purchasing through Online Purchase Intention

Research Methods
This research used quantitative survey method to examine consumer behavior in making online purchasing. Population in this study is the whole E-Commerce user. Sampling is based on purposive random sampling method. The consideration of respondents as a sample is E-Commerce B2C or C2C users and have made transactions in E-Commerce. The number of samples in this study was 100 respondents which mostly are students of the University of Technology Yogyakarta. The data were obtained by using questionnaire.
Perceived Reputation is a consumer belief that organizations selling goods or services will be honest and caring to consumers. Trust is the willingness of consumers to rely on sellers and to take purchase actions even if the seller can easily harm the consumer. Security Practice is a practice in ensuring security as an online store's ability to control and security for data transactions. Privacy concern is a part of data privacy, which involves the right to mandate personal privacy about the storage, provision to third parties, and display information relating to yourself through the Internet. Perceived risk is a way for consumers to perceive possible losses due to online purchase decisions because of the uncertainty of the decision. Online Purchase Intention is a mental activity that precedes it before a decision to buy. Actual Purchase Intention is a concept in purchasing behavior where consumers decide to act/do something and in this case make a purchase electronically. Measurements of Privacy Concern, Perceived Risk and Online Purchase Intention refer to the research by Kim (2008), while Actual Online Purchasing measurement refers to the study by Puspitasari et al. (2018).

Figure 1. Research Conceptual Framework
Researcher tested the hypotheses using SmartPLS3 program. Outer Model Test (measurement model) was done to specify the relationship between latent / construct variables and indicators. Measurement model to test the validity was done by looking at the AVE value and loading factor values. In confirmatory research, the minimum limit of loading factor that is considered feasible is 0.5 (Ghozali, 2014). Meanwhile, to test reliability by seeing the value of Cronbach's Alpha and Composite Reliability with each value must be above 0.7 (Chin, 1998

Results of Validity and Reliability Test
Following  (Ghozali, 2014). The results of the test in Table 2 indicates that there are no indicators which have loading factor below 0.5 so that it can be inferred that the construct has fulfilled good convergent validity.

Structural Model Test
This structural model is to evaluate the dependent variable used R-Square values which the higher the value indicates the predictive model used better. Below is the Goodness of Fit test results: it shows the antecedent model of Trust has the most excellent predictive ability than any other model in this study. The higher value of R-Square indicates that the model is better (Chin, 1998).
To test the hypothesis in the structural model, it can be seen with the line Coefficient values and P-values or T-Statistic of each path. The following is the results of coefficient path test:  The coefficient beta values and the value of T-statistics and significance (P-Value) indicates that the 1A hypothesis stated that Perceived Reputation significantly affects positively on trust is supported (β = 0.257, P-Value = 0.025). Unlike the 1B hypothesis that EWOM had significant positive effect on Trusts is unsupported (β = 0.058, P-Value = 0.553). The 1C hypothesis is unsupported (β = 0.083, P-Value = 0.487) indicates that in this case, the Trust on the E-Commerce website does not affect online purchase interest.
Concerning the results of the antecedent of perceived risk test is supported. 2A and 2B hypothesis indicating that security practice significantly has negative effect on perceived risk (β =-0.641, P-Value = 0.000) and privacy concern negatively affects toward perceived risk (β =-0.275, P-Value = 0.000). The 2C hypothesis stated that perceived risk negatively affects the trust is also supported (β =-0.576, P-Value = 0.000), the 2D hypothesis is also supported which states that perceived risk negatively affects online purchase interest (β =-0.699, P-Value = 0.000).
The 3A hypothesis test shows that OPI had positive effect on the AOP (β =-0.026, P-Value = 0.801). In relation with the role of OPI in the dissemination of the trust (hypothesis 3B) and perceived risk (hypothesis 3C) are not supported as indicated by the Beta and P-value respectively β =-0.002, P-Value = 0.893 for hypothesis 3B and β =-0.02, P-Value = 0.800 for 3C hypothesis.

Relationship between Trust and its Antecedent with Online Purchase Intention
Related to trust antecedent; EWOM and reputation perception, it shows that only a reputation perception has positive and significant influence on trust. It means that the E-Commerce website with a good reputation can improve consumer confidence and EWOM has no effect on the trust. The issues related to trust in online purchase behavior are important because in the process of online transactions there is no direct interaction and communication. EWOM in the E-Commerce website in this study is a review (online review) or buyer testimonial related to products sold in the E-Commerce website. Based on the results of the study of the reviews from other consumers, it does not necessarily make consumers or prospective buyers trust. It possibly occurs since there is no social capital bonding among the consumers in e-commerce website since they don't know one another and a belief that the information conveyed by the commentator is honestly given information. In addition, there is usually no detailed identity associated with the review provider so that the credibility can be doubtful. This insignificant outcome with the influence of EWOM on the trust can also be caused by the limitation in this research that does not distinguish E-Commerce model is B2B and C2C (respondents in this research are respondents who have transacted on the B2B and C2C E-Commerce website) in which the E-Commerce model C2C allows website users to double the role of sellers and buyers that allow the occurrence of cheating in providing information. Dharmaadi & Supangkat (2014)stated that online review can lead to uncertainty caused by the possibility of manipulation of data from the person giving the review so that there are still a lot of consumers who have not fully trusted E-Commerce website.

Relationship between Perceived Risk and its antecedent with Online Purchase Intention
As presented in the structural test model, it shows the online purchase intention influenced by the perception of the risk to the E-commerce website, which the risk perception antecedent is formed by the determination of privacy concern and security practice, each of which has a negative influence and significant to risk perception, but the effect of security practice on perceived risk is greater than privacy concern. Model of perceived risk antecedent shows that the E-Commerce Website that considers the security and privacy of consumers can make consumers feel safer or feel a small risk in transaction so that it can eventually increase purchase interest through the E-Commerce website. It can be interpreted that consumers consider risks in building confidence to make online purchases through the E-Commerce website, but in this study having an interest is not enough to explain whether consumers will decide to purchase through the website. This can be explained because it is likely that many consumers use the E-Commerce website only to select and compare products and prices, but they ultimately make purchases on the offline store or COD (Cash on Delivery) system to minimize the risks incurred by consumers.

Relationship between Trust, Perceived Risk, Online Purchase Intention and Actual Online Purchase
Insignificant results regarding the influence of Online Purchase Intention toward Actual Online Purchase can be caused by limitations in research in which researchers do not distinguish the behavior of purchases on the pure E-Commerce website as in the purchase of digital products or partial E-Commerce websites such as on the purchase of products that usually have a physical form.

Conclusion
Based on the results on the previous discussion, it can be concluded that the most powerful predictive model is a model of perceived risk antecedent influence on online purchase interest either directly or through perceived risk. Moreover, the results of hypothesis test between hypotheses 1B, 2A, 2B, 2C and 2D were supported, while the hypotheses of 1C, 1D, 3A, 3B and 3C were not supported. From these results, it can be stated that marketers in the E-Commerce business need to consider and ensure the security of both consumer privacy and the online transaction process. With a sense of security, consumers can reduce the perceived risk of increasing online purchase interest. By reducing consumers ' perceived risks it can also increase trust or consumer confidence in E-Commerce websites. Based on the R-Square value, it can be concluded that the Level of Goodness of Fit from several endogenous variable models which are OPI, Trust and Perceived Risk is categorized as moderate because it has a score between 0.33 -0.67 While the level of Goodness of fit of the AOP model is considered weak because it is smaller than 0,33 (Chin, 1998).

Suggestion
About this study results, researcher suggests for the further study to separate purchase behaviors based on online shop type such as online fashion, electronic or general, in which the category of electronic products can be considered to have greater risk compared to fashion products because there is a possibility that electronic products are more risky to suffer damage in the process of goods shipping or goods does not function properly, so it is possible for consumer's online purchase behavior may be different depending on the category of products sold on the website. The results of this research is not able to propose a model that can predict the behavior of online purchases, so that for the further researcher is advised to include other variables of marketing strategies such as sales promotions that can be in the form of discounts, vouchers or other short-term incentive forms, as well as differentiating online purchase behaviors between men and women. Further researchers can distinguish online purchase behavior on pure E-Commerce types with Partial E-Commerce. On the pure E-Commerce products sold are digital products that are only sold through online channels, so consumers who already have the interest to buy may continue the actual purchase process. Unlike Partial E-Commerce in which products sold are products that have physical form and consumers have an alternative to making purchase transactions both online and offline channels. The further researchers can compare the online purchase behavior through B2C and C2C E-Commerce website because the credibility of the two models can differ each other in which the credibility is related to the reputation and the level of consumer confidence in the E-Commerce website.