Mediating Effect of Customer Engagement on the Relations between eWOM, Environmental Concern, and Intention to Purchase Second-hand Clothing among College Students in Yogyakarta

Research aims: This study examines the direct and indirect effects of eWOM and environmental concerns on second-hand clothing purchase intention. However, on the intervening variable, this study used customer engagement. Design/Methodology/Approach: This quantitative research distributed questionnaires to 222 college students aged 18-24 years who were interested in and were familiar with the concept of using second-hand clothes, who were also lived in Yogyakarta. The questionnaire results were then processed utilizing the structural equation modeling-partial least square (SEM-PLS) technique. Research findings: This study’s findings showed that consumer engagement had a full mediating impact between eWOM and intention in buying second-hand clothes. However, it only partially affected the relationship between environmental concern and the intention to purchase used clothes. Theoretical contribution/Originality: Previous studies have investigated the relationship between eWOM and purchase intention through customer engagement as a mediating variable, but little literature involves environmental concerns in the model. Another contribution is the findings that revealed that eWOM could not influence students' purchase intention in second-hand clothes without the involvement of customer engagement. Practitioner/Policy implication: The author suggests that second-hand clothes sellers and non-profit organizations could increase young people's involvement in making environmentally friendly consumption in the form of buying used clothes by increasing consumer engagement. Research limitation/Implication: This research's limitations include the limited variables studied and the characteristics of respondents who only focused on generation Z, especially students in Yogyakarta.


Introduction
In recent years, there is plenty of evidence to suggest that increased concern for the environment has also improved consumer interest in using second-hand goods (Paço et al., 2021;Yrjölä, Hokkanen, & Saarijärvi, 2021).
Along with worsening global warming, natural destruction, and an abundance of waste, especially clothing waste, many people feel it is essential for the world's population to consume more sustainable alternative products. The World Economic Forum (WEF) in 2020, together with the COVID-19 lockdown issue, states that sustainable consumption and production are the primary agenda for world business stakeholders (Kim & Kim, 2020). Used goods are considered more environmentally friendly because they do not require new resources in their production and have a longer life span to slow down the increased rate of the waste. Hur (2020) mentioned that environmental concern is a determinant variable that affects the willingness of consumers to purchase ecologically friendly products, especially second-hand ones. Otherwise, several past studies stated different facts that environmental concern is not a key determining factor of second-hand fashion products purchase intention because customers evaluate other features, such as aesthetics and economic factors (Gullstrand Edbring, Lehner, & Mont, 2016;Wei, Ang, & Jancenelle, 2018;Yan, Bae, & Xu, 2015). Due to variations in the findings of earlier studies, the author is interested in using the environmental concern factor as one of the antecedent variables of second-hand clothes purchase intention. With this study's results, it is hoped that second-hand fashion business actors and non-profit organizations that promote sustainable lifestyle, especially those whose target market is young people in Yogyakarta, could get a deeper understanding of whether the environmental concern aspect can be used to promote the purchase of more sustainable fashion products or not.
Moreover, the trend of fashion thrifting is not only developing in western countries, but the younger generation of Indonesia has also begun to be interested in wearing secondhand clothes. In this case, many studies have suggested that electronic words of mouth (eWOM) are the driving factor for purchase intention. Today's consumer behavior trend also trusts the reviews they read on the internet before making a purchase decision (Mohammad, Quoquab, & Mohamed Sadom, 2020). On the other hand, Srivastava and Sivaramakrishnan (2020) argue that eWOM solely is not strong enough to influence consumer purchase intention in fashion products. Hence, brands need to develop continuous interactions with consumers or customer engagement to boost their purchasing likelihood and make consumers more driven and loyal to the product (Belyakov, 2020). Due to variances in research results regarding the effect of eWOM, environmental concern, and customer engagement on second-hand fashion purchase intention, it is hoped that this research can provide a comprehensive view of these areas to stakeholders concerned with environmentally friendly fashion consumption. In the academic context, it is expected that this study also can provide an overview of driving factors influencing second-hand fashion products purchase intention in the context of young people in Yogyakarta.
However, in this study, E-WOM is defined as the online reviews conducted by internet users related to the purchase of second-hand fashion products. According to Mohammad et al. (2020), eWOM significantly influenced consumer attitudes and customer engagement. The effect of eWOM on attitudes (Al Halbusi & Tehseen, 2018;Ladhari & Michaud, 2015) and the relationship of eWOM on consumer engagement (Fang et al., 2017) have been widely studied in various previous contexts. Besides, marketing users, especially in the fashion sector, use many efforts to increase eWOM to increase their product sales. It is in line with Anastasiei and Dospinescu's (2019) research, which revealed that eWOM is an antecedent factor influencing one's purchase intention and purchasing decisions. Because of the many supporting theories, this study examines the effect of eWOM on consumer attitudes, consumer interactions, and purchase intention toward second-hand fashion products, as follows:

Environmental Concern
Concern about environmental issues refers to the level of consumer awareness of the natural damage problems and willingness to be part of the solution to these problems and efforts to conserve nature (Chuah et al., 2020). Regarding this, the consumption of second-hand clothing is considered a form of consumer concern with environmental conservation efforts because these products are considered could save resource consumption, slow down the increase rate in fashion waste, and create a circular consumption climate in the industry (Tunn et al., 2019). Hur (2020) states that the factors that motivate someone to buy second-hand products include economic motivation, selfexpressive, hedonic, concern for the environment, and the desire to contribute to social values. According to Howell (2013), when an individual is concerned about specific issues and has good internal motivation, he will be more enthusiastic and engaged in a behavior for a prolonged period. In line with the findings of Jägel et al. (2012), someone will have a positive attitude, and there will be a desire to purchase second-hand fashion when they have insight into the natural environment, want to minimize consumption, which is bad for the environment and minimize the carbon footprint on their consumed products. Based on these reasons and findings, some hypotheses formed are as follows: H3: Environmental concern positively influences customer engagement H4: Environmental concern positively influences second-hand clothes purchase intention

Customer Engagement
Customer engagement is a mechanism for adding customer value to a company, either through direct or indirect contributions (Pansari & Kumar, 2017). In this study, what is meant by customer engagement is the development of good communication between the brand and its consumers regarding second-hand fashion products on social media, the occurrence of consumer involvement in responding to promotions given by the company, and indeed the existence of consumer feedback for the company so that it has a better performance. Concerning this, discussions about customer engagement have become a focal point in various studies. In addition, customer engagement is a behavioral manifestation of consumers that not only exceeds the desire to buy but also has an emotional or motivational state related to involvement in a brand or product (Verma, 2021). According to Dwivedi (2015), customer engagements are a positive consumer perspective with dimensions of vigor, absorption, and dedication. Besides, Vale and Fernandes (2018) said that customer engagement is a hierarchical process in the consumption, contribution, and creation phases. Further, De Villiers (2015) generally classified consumer involvement in a product into two categories: passive and active engagement. Consumer engagement is said to be passive when they are unaware or ignorant of a brand. On the other hand, consumers could be said to have active engagement when they have high cognitive, affective, and behavioral levels towards an offer. Moreover, high consumer involvement can be reflected in a deep emotional connection so that consumers are willing to develop and sustain their relationships with the offering (Zainol et al., 2016).
In Fernandes and Esteves's (2016) research, the customer engagement variable was associated with trust, co-creation, and satisfaction variables. Moreover, Ahn and Back (2018) and Prentice et al. (2019) found that the customer engagement variable had a significant influence on consumer purchase intention. This opinion is strengthened by the research of Islam, Rahman, and Hollebeek (2018), which revealed that customer intentions were influenced by the social interaction level carried out both online and offline. Therefore, when consumers have intense involvement with the activities of a brand or product, it can be assumed that they tend to have purchase intentions and make purchases in the future, especially to purchase products with environmentally friendly values (Joshi & Srivastava, 2019). Thus, the researcher posited: H5: Customer engagement positively influences second-hand clothes purchase intention.

H7: Customer engagement fully mediates the relationship between environmental concern and second-hand clothes purchase intention
Based on previous research findings, which uncovered that eWOM, environmental concerns, and customer engagement are the driving factors for purchase intention in second-hand fashion products, the researcher formulated a research framework as shown in Figure 1.

Sampling
The questionnaires in the study were distributed to a total of 222 respondents. Respondents in this study were selected purposively by determining criteria, including individuals aged 18-24 years, domiciled in Yogyakarta, familiar, and interested in used clothes. The questionnaires were distributed from one respondent to others who were considered interested in and knowledgeable about fashion-thrifting activities. In this study, definitions from many previous studies were identified as operational definitions. Pre-validated scales and a pilot survey were employed in this study, so the author only used valid and reliable questionnaire items. However, some modifications were made by writing to adjust the existing questionnaire statement items to the respondent's context and circumstances. The data obtained were then processed utilizing the SMART-PLS 3 software. The demographic characteristics of the sample are shown in Table 1. The majority of respondents were female, 59%, and the other 41% were males. However, most respondents' monthly fashion product purchases were around 82.4%, in the range of IDR 0-500,000 and only about 17.6% spent in the range of IDR 500,000 to 1,000,000.

Scales
This study involved two exogenous variables, eWOM and environmental concern, one intervening variable/endogenous variable, customer engagement, and one other endogenous variable, second-hand purchase intention. Items in each variable were taken from various previous studies, and researchers only used questionnaire items relevant to the investigation and were also valid. A four-item Likert scale was employed to measure all constructs: (1) completely disagree -(4) completely agree. Table 2 displays the questionnaire items used in this study and the adapted sources of the previous research.  (2020) Before buying second-hand clothes, I explored online product review sites (e.g., reviews on Shopee or social media such as TikTok, Instagram, Facebook, etc.) to study other customers' evaluations. I believe the reviews of social media users or other buyers who have previously had the experience of buying second-hand clothes. I often browse other customers' online product evaluations of used clothing to determine what brands or retailers make a good impression on others. I always read reviews of other consumer goods online to ensure that I purchase the right clothing. I often gather information from online customer product reviews before I purchase clothing.

Environmental Concern
Buying second-hand clothes is a form of sustainable consumption.

Data Analysis Methods
This study examined the research model utilizing the partial least squares (PLS) technique and the SmartPLS 3.0 software. PLS is a multivariate approach that minimizes the variance in endogenous variables that is unexplained . Furthermore, PLS is capable of handling complex models with numerous linkages (Hair et al., 2015). Hair et al. (2015) proposed a two-stage analytical process for examining the measurement model's validity and reliability, followed by the structural model's in-sample prediction power and out-of-sample prediction capacity (Tan et al., 2017). In the evaluation stage of the measurement model, also known as the inner model, the first thing that needs to be tested is the reliability test or its internal consistency. If the value of loading is exceeding 0.7, the composite reliability is regarded to be reliable. Similarly, if Cronbach's alpha is greater than 0.6, this study demonstrates adequate internal consistency dependability. The second stage is convergent validity testing, which determines the degree to which the measures are positively linked with alternative measures of the same construct (Hair et al., 2015). Convergent validity is carried out by reviewing the values of the outer loading and the Average Variance Extracted (AVE). If the value of the outer loading is greater than 0.7 and the AVE value is greater than 0.5, it can be concluded that the construct explains more than half of the variance of its indicators or the convergent validity is acceptable. The next stage is discriminant validity testing, namely assessing the extent to which a construct is truly distinct from other constructs by empirical standards, by looking at the value of the HTMT and Fornell Larcker indicators. At the structural model's evaluation stage, the indication that should be examined is the value of R 2 . After testing the inner and outer models to ensure they comply with the rule of thumb, a testing hypothesis can be developed.

Results and Discussion
From the analysis results, output could be seen that all constructs with reflexive indicator loading factors generated value > 0.70, meaning that all items constructs in this study were valid. Likewise, the Average Variance Extracted (AVE) value produced by all reflexive constructs had a value above 0.5 so that it met the convergent and reliability requirements. Besides, all constructs' Cronbach's Alpha values were very good, which was above 0.7, so it could be concluded that all indicators of reflexive constructs were reliable or met the reliability test. The reliability test using the composite reliability value also gave good results because all reflexive constructs had values above 0.7. Matters related to the validity and reliability test are presented in Table 3.
There are two main steps to test the discriminant validity: Fornell-Larcker Criterion and the heterotrait-monotrait ratio of correlations (HTMT) (Ab Hamid, Sami, & Mohmad Sidek, 2017). In this study, the discriminant validity assessment using Fornel and Larcker was carried out by comparing the square root of each AVE value on the construct displayed diagonally with the correlation coefficient (off-diagonal). Overall, Fornell and Larcker's results showed that discriminant validity in this study did not exist, and all the constructs acceptable for this calculation also supported discriminant validity between each construct. On the other hand, Henseler, Ringle, and Sarstedt (2015) revealed that the heterotrait-monotrait ratio is more reliable than the Fornell Larcker test. In some cases, the Fornell Larker criterion test failed to identify problems with discriminant validity. In heterotrait-monotrait testing, a construct is said to meet discriminant validity if its value is below the conservative threshold of 0.85. It can be seen in Table 4 that there was no lack of discriminant validity in this study, both from the Fornell and Larcker test and the heterotrait-monotrait ratio of correlations results.  The research results' structural model was assessed by looking at the R 2 value, path coefficients, and t-values generated from the bootstrapping procedure with 5,000 resample. The R 2 value results are exhibited in Table 5, showing 0.478 for customer engagement and 0.440 for the purchase intention variable. It indicated that 47.8% and 44% of the customer engagement and purchase intention variance could be explained by their exogenous variables, respectively. For research focusing on marketing issues, the endogenous variable R 2 value close to 0.5 can be classified into the moderate to weak category (Hair, Ringle, & Sarstedt, 2013). The research hypothesis testing was conducted by looking at the path coefficient value and t-value shown in Table 6. The results showed that the direct path between eWOM and customer engagement had a positive correlation (β = 0.410) and significant (t = 7.323), supporting Hypothesis 1. Meanwhile, hypothesis 2, namely the positive effect of eWOM on second-hand clothes purchase intention, was not supported because the correlation results were positive (β = 0.15), but the effect was not significant (t = 0.193). Besides, the direct path between environmental concern and customer engagement had a positive correlation (β = 0.380) and significant (t = 7.554), supporting Hypothesis 3. Likewise, with hypothesis 4 about a positive relationship between environmental concern and secondhand clothes purchase intention, the result was supported because the correlation coefficient results showed a positive relationship (β = 0.140) and significant (t = 1.965).
The meaning of a positive relationship is that the exogenous and endogenous variables have a unidirectional relationship. In other words, if the respondent has a great environmental concern, the purchase intention will also increase and vice versa. Hypothesis 5, related to a positive and significant relationship between customer engagement and purchase intention, was also supported by a positive correlation value (β = 0.560) and a significant relationship (t = 8.659).
In line with hypothesis 6, the customer engagement variable fully mediated the relationship between eWOM and purchase intention toward second-hand clothes. To test the form of media for customer engagement, whether full, partial, or full mediation, the researcher needs to see the direct effect and the indirect effect values. Full mediation occurs when exogenous variables cannot significantly influence endogenous variables without going through the mediator variable (Hair et al., 2013). In this study, it could be seen that the direct effect value between eWOM and purchase intention was not significant, but the indirect effect by involving intervening variables became significant. This study's results indicated that the direct effect between eWOM and second-hand purchase intention had a positive relationship (β = 0.015) but not significant (t = 0.193), while the indirect effect had a positive (β = 0.230) and significant (t = 5.580).
On the other hand, hypothesis 7 in this study was rejected because the customer engagement variable only partially mediated the relationship between environmental concern and purchase intention variables. Partial mediation is a condition when the mediating variable only affects partially, not all the relationships between exogenous variables and endogenous variables, which are proved from the significant direct effect relationship, and the indirect effect also shows significant results (Hair et al., 2013). In this research, the direct effect between environmental concern and second-hand purchase intention was positive (β = 0.140) and significant (t = 1.965), and the indirect effect also had positive (β = 0.213) correlation and statistically significant (t = 5.738).

Conclusion
This study uncovered that eWOM had a direct positive effect but did not significantly affect the purchase intention toward second-hand clothes. This finding does not agree with the previous research results stating that eWOM is an antecedent with a significant effect on attitudes, purchase intention, and consumer behavior on purchasing sustainable fashion products (Bedard & Tolmie, 2018;Mohammad et al., 2020). Besides, positive reviews by other internet users and good information circulating on digital platforms have proven unable to make the younger generation more interested in purchasing secondhand clothes. The information about second-hand products available on the internet was not strong enough to motivate respondents to have the desire to buy second-hand clothing products. Regarding this, the practical implications that thrift fashion businesses can utilize are that marketers need to engage in active engagement with their audiences on social media. With a well-developed customer engagement, consumers are motivated to keep finding out about used clothes products, regularly want to keep updated about thrifting activities and make them enjoy their spare time by doing thrifting. In other words, the effect of eWOM will only influence purchase intention if there is intervention from the customer engagement variable.
Meanwhile, the young generation's environmental concern significantly affected purchase intention in second-hand clothing products. Respondents had the awareness to make more environmentally friendly consumption, one of which was manifested in thrifting. This study's respondents have already comprehended that using used clothes is a form of their behaviors and efforts to create environmentally friendly behavior. College students have also begun implementing conscious consumption efforts that do not harm the environment. The presence of this customer engagement variable further strengthens purchase intention. It aligns with Yan et al.'s (2015) study results that the construct of environmentalism is one of the predictor variables of shopping behavior at second-hand clothing stores. From a marketer's point of view, the author advises that second-hand clothes sellers can make efforts to enhance consumer attention to be more environmentally friendly by often publishing entertaining facts about how buying secondhand clothes may help protect the environment. Clothing retailers can also increase the perceived environmental knowledge, particularly environmental concerns, by building cognitive and affective aspects; a real example is conducting broad public campaigns regarding environmental damage problems (Rausch & Kopplin, 2021).
From the respondent data processing results, it is evident that customer engagement fully mediated the relationship between eWOM and purchase intention. Meanwhile, in the relationship between environmental concerns, customer engagement had a partial mediating effect. Hence, purchasing second-hand fashion products might result from their eWOM experiences, a form of their contribution to the environmental conservation movement, and engagement of the product. In this case, Joshi and Srivastava (2019) revealed that when individuals engage with a brand and have environmental concerns, they will be encouraged to solve environmental problems unless they experience cognitive dissonance. Thus, this study suggests that marketers need to improve customer engagement, for instance, by performing promotions that stimulate excitement and emotional attachments towards pre-loved apparel.
Nevertheless, this research's limitations include the limited variables studied and the respondents' characteristics who only focused on generation Z, especially students in Yogyakarta. Further research can be developed using other variables, which are also thought to be driving factors of purchase intention toward second-hand and sustainable fashion products, such as economic motivation, perceived hygienic risk, channel familiarity, style orientation, and other variables. This study's results are also likely to be different if the respondents who filled out the questionnaire come from other age groups and different economic classes. Understanding second-hand fashion products' buying behavior patterns will also be more complete if the next researcher takes the purchase decision variable and the loyalty to use pre-loved clothes as the dependent variable. Furthermore, the researcher suggests that the following researchers can use qualitative methods, such as in-depth interviews, observation, and forum group discussion, to explore things that motivate informants to buy used clothes and other more sustainable fashion choices. The things that make consumers not want to buy second-hand clothes are also worth researching so that fashion businesses can minimize the barrier to consuming consumers in Indonesia.