THE INFLUENCE OF SITE QUALITY AND TRUST ON REPURCHASE INTEREST OF TOKOPEDIA THORUGH PERCEIVED USEFULNESS

Purpose: this study aims to examine the relationship between site quality variables and trust in Tokopedia consumer repurchase interest in Surabaya through the mediating variable of perceived usefulness. Design/methodology/approach: This is a quantitative study and the data will be analysed through partial Least Square (PLS). Findings: 5 out of 7 hypotheses cannot be proved and supported by this study. Research limitations/implications: The respondent of this study are 95 Tokopedia user. Practical implications: Only 2 out of 7 hypotheses can be proved and supported. Paper type: This study is categorised as a case study.


I. INTRODUCTION
In today's society, there are more and more choices for consumers to choose where to shop from and without special reasons, according to (Venkatesh, Thong and Xu, 2012). they will switch places to shop, both online and offline. Al-Maghrabi et al. (2011) argued that finding new customers will cost five more times than retaining the existing customers. Therefore, understanding the factors that can encourage consumer loyalty to certain shopping site is a priority.
E-shopping is a combination of product information searching and purchasing activities. Therefore, the measurement of actual purchases will narrow the assessment of e-shopping activities because gathering product information is essential in this activity (Kim, Mirusmonov and Lee, 2010).
This research is based on technology acceptance model (TAM) which is a representation of Theory of Reason Action (TRA) and Expectation Confirmation Theory (ECT) TAM discusses user acceptance of information systems that will help researchers explain the factor that influences user acceptance of computer technology and analysing user behaviour on a large scale in a population that applies computer technology (Holden and Karsh, 2010). According to Marangunić and Granić, (2014), one of the main factors is perceived usefulness in which will influence the attitude formation.
On the other hand, according to Udo, Bagchi and Kirs, (2010), ECT helps predict consumer behaviour in the time period of before, during and after the repurchase of the products and/or services. In the ECT approach, consumers will define repurchase interests by determining whether the product/service meets their initial expectations. Comparison between perceived benefits and real expectations of consumers will determine consumers' decision to make a repeat purchase.
Continuous intentions depend on cognitive beliefs about perceived usefulness-which will make consumers more willing to use new information systems (Brown, Dennis and Venkatesh, 2010). Perceived usefulness is a strong predictive factor in TAM and in continuance intentions-when combined with ECT-so it

A. Previous Studies
Research on the relationship between the site quality, trust, perceived usefulness and interest in repurchase has been conducted before. Baskara and Sukaatmadja, (2016), examined the effect of online trust, perceived enjoyment on shopping satisfaction and repurchase intention, which produce the statement that trust has a significant and positive influence on repurchase intentions of the consumers. In addition, (Basyar and Sanaji, 2016) argue that the perception of benefits has a major influence on online repurchase intentions. Al-Maghrabi et al., (2011) stated that site quality influences trust and perceived benefits, trust also has a positive and significant influence on perceived benefits, while perceived benefits also have the same influence on repurchase intentions. According to Saidani, Lusiana and Aditya, (2019), site quality and consumer's trust have a positive influence on repurchase interests. The effect of each variable used on consumers' repurchase interest, directly or indirectly. This study is categorised as a cross-sectional study, which the data sourced from the sampling that happened only once. Because the variables are analysed together and did not go through the control and/or manipulation process by the researcher, this study can be categorised as ex-post-facto research (Kerlinger and Lee, 2010); (Seniati et al., 2011). The conceptual framework of this study is depicted in Figure 1.

C. Population and Sample
The population of this study are the consumer of Tokopedia in Surabaya and the sample is a part of the population mentioned. Sampling technique is adapted from (Joseph F. Hair et al., 2014) on the minimal sample of multivariate analysis, which is five times of the parameter used (19 indicators). In total, there is 95 respondent used in this study. The respondent of this study must fulfil the requirements set by the researcher, which are: 1) is a consumer of Tokopedia and have shopped at least once; 2) Aged 18 or above; 3) currently living in Surabaya.
Operational Definition of Research Variable X1 : Site quality The characteristics of an online shop platform wanted by the customers X2 : Trust The expectation of the other party will not be disappointing and will give what has been promised M : Perceived Usefulness The belief that by using a certain technology, the performance of a person's work will improve Y : Repurchase Interest A decision making process done by the consumer after purchasing a product proposed by the seller or needed.

D. Data Analysis Method
The data will be analysed through Partial Least Square (PLS) technique. The outer model will be analysed through convergent and discriminant validity from its indicators and composite reliability for the block indicators. The inner model will be evaluated through the percentage of the variance-which can be seen through the R2 (R squared) of latent dependent constructs. The stability of this estimate can also be evaluated through the bootstrapping procedure (Ghozali, 2018).

A. Reliability and Validity
In order to be accepted as reliable and valid, the correlation of latent variable and its indicators must have the minimum value of convergent validity, which is >0.7. From the research, it can be concluded that the convergent validity value of each indicator is greater than 0.7, which means each indicator are reliable and valid to be used in this research.
The discriminant value needed for each variable is outer loading > 0.7. The same with the convergent validity before, all of the variables passed with the score depicted in Table 1. To determine the composite reliability if the construct has the value >0.8, then it can be said that a construct is highly reliable. From the research, the value of the composite reliability of each construct exceeds 0.8, which means each variable are suitable to be used to analyse its influence on the dependent variable. The reliability test with Cronbach's Alpha will give a good result if the value is above or equals 0.6. All of the variables in this research scored above the minimum in this reliability test.

B. Hypotheses Testing
Hypotheses testing through PLS is carried out in two stages, which are: 1) measuring directly the influence of independent variables on the dependent; 2) measuring the influence of independent variables on the dependent variable through the mediator, perceived usefulness. The criteria for the hypotheses to be supported by the research with 5% significance level (α) are the t-count of each hypothesis must exceed the t-table, 1.96. If the t-count did not exceed the determined minimum value, the hypothesis cannot be supported and therefore rejected. Here is the output (PLS bootstrapping) of said hypotheses: