Can Product Quality Improve Purchase Decisions in E-Commerce and Social Media through Customer Loyalty and Trust?

How to Cite: Nofrizal, Sucherly, Juju,


INTRODUCTION
Customers' purchasing behavior patterns have changed in the last few years.It is due to the advancement of Internet quality and the ease of getting smartphones as hardware and software in Indonesia.According to data from the Indonesian Internet Service Providers Association/ Asosiasi Penyelenggara Jasa Internet Indonesia (APJII) (Dewi, 2022), Internet use reached 77% of Indonesia's total population in 2022 due to the push from the COVID-19 pandemic.It also added around 35 million new Internet users (Hasanat et al., 2020).This change also appears in the customers' behaviors.They have started to get used to using e-commerce and social media as a platform to buy a product when they previously often uses offline methods (Chen et al., 2022).E-commerce is a new concept that processes purchasing and selling products or services using the Internet network (Tian, 2018).E-commerce is a component of e-business.The scope is broader and includes customer service, job openings, and working with business partners in addition to commerce.
According to Von Helversen et al. ( 2018), the study of online purchase decisions increases by 17,5% every year, with e-commerce (Businessto-Customer (B2C)) globally accounting for 8,7% of the world's retail industry.Some factors in customers' online shopping decisions are individual attributes, demography, geography, and availability of technology and the Internet (Yousefi et al., 2023).In Indonesia, the e-commerce platform with the best delivery performance is Tokopedia, with a value of 52%.It is followed by Shopee with a value of 45%, Lazada with a value of 43%, OLX with a value of 40%, and Bukalapak with a value of 20% (Mulyana, 2022).Meanwhile, the e-commerce with the highest number of website visitors is Tokopedia, with 157 visitors in Indonesia and landmark Asia in the first quarter of 2022 (CNBC Indonesia, 2022).
Meanwhile, according to Kemp (2021), active users on social media in Indonesia reached 191 people in 2022.It had more than 12,35% growth compared to the prior year, with the most popular social media platform being WhatsApp at 88,7%.Based on these data, it is evident that the market share of e-commerce and social media in Indonesia has a big potential that start-ups can capture to offer their products to customers in Indonesia.
Before customers decide on an online purchase, they will look for more information online on social media and websites and assess the advantages of other products (Sharma et al., 2021;Kotler et al, 2019).So, product quality encourages customers to decide on purchases on e-commerce networks (Yang et al., 2022).Products with good review quality in e-commerce will improve the quality of these products (Liu et al., 2018).Brand trust also makes customers decide more about online purchases (Qalati et al., 2021).Trust is also a key factor that determines the success of an e-commerce platform, which is why online dealers have to do it (Kim & Jones, 2009).Trust is a prerequisite in e-commerce and other online transactions (Mumu et al., 2022;Amin et al., 2015).For example, loyalty and trust are one of the sources of purchase decisions in Tokopedia (Wardhana et al., 2021).
According to Hride et al. (2022), there is a relationship between trust and customer loyalty in the Facebook-based e-commerce industry in Bangladesh.If organizations want to have a large number of loyal customers in e-commerce, they need to provide social and financial benefits to their customers (Arghashi et al., 2021;Azizan & Yusr, 2019).Loyalty can be built by building long-term relationships, providing incentives, managing customer complaints, and encouraging customer feedback (Barreda et al., 2020).It is also suggested that social media marketing significantly affects e-commerce customer loyalty and satisfaction of Indonesian e-commerce SMEs (Suharto et al., 2022;Nofrizal et al.,2022).
Trust and convenience have a significant positive effect on the purchase decisions of e-commerce customers in Indonesia (Fadhilla & Farmania, 2017).The purchase decision is the most important thing.It is the end of a series of purchase processes (Kotler & Keller, 2016).After various stimuli have been carried out, customers will choose to shop in e-commerce according to their final choice (Suleman et al., 2020).It is said that purchase decisions are closely related to customer behavior.Customer behavior in purchase decisions exists because, at this stage, the decisionmaking process is carried out by customers (Kotler & Keller, 2012).
Based on several previous studies, no one has studied why customers are loyal and decide to purchase fashion products sold on e-commerce and social media, especially using service quality, trust, loyalty, and purchase decisions.So, the research is essential for the fashion product industry in e-commerce and social media.Moreover, it can develop marketing strategies for e-commerce, especially fashion products.
Online purchase decisions, according to Torabi (2022), are influenced by the completeness of information and risk-taking by using an online media approach.Online purchase decisions are also influenced by the discounts offered on the website (Kim & Tanford, 2021).From an administrative perspective, the choice means picking one option from various options.Customers choose products to fulfill their wants when they make purchases (Shah et al., 2019).Moreover, the quality of information influences decisions to purchase products online (Lin et al., 2019).Furthermore, purchase decisions are a component of customer behavior.It is defined as the study of how people, communities, and organizations pick, acquire, and utilize how the items, services, concepts, or firsthand accounts to meet their needs and wishes (Kotler et al., 2018).
Integrating theories of customer behavior increases the knowledge of how people utilize the Internet to make purchases of products and services.The research expands on and evaluates earlier studies on the variables influencing decision-making for online purchases.Therefore, the research is important as reference material for start-ups worldwide and in Indonesia to find out why customers in Indonesia decide to buy fashion products through e-commerce and social media.Hence, innovative marketing strategies can be arranged in accordance with customers' purchase behavior in e-commerce and social media.The factors seen in the research are product quality, trust, customer loyalty, and purchase decisions.
In the conceptual model and hypotheses development in the research, customers' trust in intermediaries and online suppliers is directly correlated with the trust factor for online commerce.Hence, it is a very important factor in making travel purchase decisions (Pop et al., 2022).According to Abidin and Triono (2020), trust in online retailers is the readiness to accept flaws in online transactions based on optimistic expectations for the future conduct of online retailers.For example, trust affects purchase decisions for frozen fish products (Mareta et al., 2021).It is found that the larger the customers' trust level is, the greater customers' buying intention, trust, and purchase decisions will be (Chandra et al., 2019).It is necessary to have sufficient trust in the party providing the financial information and other personal data when customers place purchases online and execute financial transactions (Eggert, 2006).Making decisions about online purchases involves combining information to evaluate two or more possible actions and selecting the one that is most strongly correlated with the buyer's characteristics, the vendor or service's performance, the website's usability, the buyer's attitude at the time of purchase, the intention to make an online purchase, and decision-making.Trust has a significant advantage when customers buy insurance products because business insurance is based on trust (Amron, 2018).
In previous research, trust has a favorable and substantial impact on purchase decisions of online purchases made by users of the Ride Inc. website (Mahkota et al., 2014).It is also supported by Wijaya and Saparso (2019) that customer trust has a favorable and considerable impact on e-commerce purchase decisions.Trust also affects purchase choices at the Mikaylaku Online Shop (Solihin, 2020).Furthermore, it has an impact on decisions to buy original products (Pratama & Santoso, 2018).According to Giampietri et al. (2018), trust and short food supply chain purchase decisions in Italy have a positive and substantial association.Hence, the first hypothesis is as follows.
H1 : Trust influences the decision to buy fashion products via social media and e-commerce.
Customers' evaluation of the apparent discrepancy between expectations and the actual performance of the product is reflected in their level of trust.It is a post-purchase assessment when the chosen option meets or surpasses the customers' expectations (Sumarwan & Tjiptonon, 2019).According to Cunningham and De Meyer-Heydenrych (2021), three factors shape one's trust in others: ability, benevolence, and integrity.There is a direct connection between trust to customer loyalty on the Internet (Muhtarom et al., 2022).According to Islam et al. (2021), trust has a big and favorable impact on customer loyalty in hospitals.Meanwhile, trust also has an impact on what customers decide to buy from an online coffee business (Ibrahim et al., 2021).Greater trust will result in greater loyalty of Koreans to the news (Nelson & Kim, 2021).The following hypothesis is put forth.H2 : Trust impacts customer loyalty for fashion products in social media and e-commerce.
The research on customer loyalty sometimes refers to brand loyalty and typically focuses on customer devotion to physical objects.Customer relevance to the selection and usage of a product is referred to as customer loyalty in general.This linking measure gives an overview of how the product will be used in the long run and future.It may also indicate a very tiny chance that a customer will switch to using a competitor's product if it changes in price, quality, and other product attributes (Tuti & Sulistia, 2022;Peppers & Rogers, 2022).It is revealed that commitment to virtual purchases has a favorable effect and connection (Chiang et al., 2018).According to research by Hála et al. (2022), there is a significant influence of loyalty program membership on the purchase behavior of customers at the supermarket in general.There is an influence of loyalty on purchase decisions (Ferrao & Alturas, 2018).Similarly, a relationship exists between loyalty and purchase decisions in drug stores (Supiyandi, Hastjarjo, & Slamet, 2022).There is an influence of loyalty on purchase decisions of traditional fashion products (Harahap et al., 2021).Then the hypothesis proposed is as follows.
H3 : There is an influence of customer loyalty on purchase decisions for fashion products in e-commerce and social media.
According to Sun et al. (2022), product quality is defined as a customer's complete assessment of the product and service quality level.However, what criteria customers use to evaluate a product's performance is the fundamental problem.Product quality is covered in part by the product policy.As stated by Chinomona et al. (2013), product quality affects customer trust in purchasing electronic device products in Africa.Product quality has an impact on public trust (Pranowo, 2019).Meanwhile, according to Panigrahi et al. (2018), there is an effect of product quality on trust in the life insurance industry.Product quality and customer trust are also related in South Korean full-service versus low-cost operators (Han et al., 2019).The following hypothesis is proposed.H4 : Trust in fashion products in e-commerce and social media is impacted by product quality.
A product's capacity to perform its functions is a measure of its quality (Kartajaya et al., 2021).It encompasses the product's overall strength, dependability, accuracy, simplicity of use, maintenance, and other desirable qualities.It indicates that the functionality of a product, including its general design, dependability, accuracy, simplicity of use and maintenance, and other essential characteristics, is directly correlated with its quality.In other words, product quality is a collection of traits and qualities of products and services that can satisfy the demands.It is a combined understanding of a product's dependability, accuracy, convenience, maintenance, and other traits.Even if the product category is the same, the products sold by each business entity will differ and need to have features that set them apart from rival items (Budiantoro, 2022).Product quality has an impact on Happy Garden restaurant's ability to retain customers for coffee (Putro, 2014).There is also an effect of the product quality of Clear shampoo on customer loyalty (Rayuwati et al., 2022).Similarly, according to Sihombing (2022), product quality affects customer loyalty in B2B manufacturing relationships.Product quality affects customer loyalty in Protected Designation of Origin (PDO) items (Fandos & Flavián, 2006) and the motorbike sector (Amri, 2022).The following hypothesis is proposed.
H5 : Product quality impacts customer loyalty for fashion products in e-commerce and social media.
The results of previous research mention that in online purchases of Simpati prepaid cards for students in Jakarta, there is an indirect relationship between product quality and customer loyalty (Setyo et al., 2021).Moreover, product quality influences customer loyalty via trust in Tokopedia (Wijaya & Saparso, 2019).The next hypothesis is as follows.
H6 : Trust in fashion items in e-commerce and social media has an impact on product quality in purchase decisions.
Customer loyalty functions as a moderator variable.It means that the level of customer loyalty can influence the extent to which product quality influences purchasing decisions (Walsh et al., 2008).High-loyal customers tend to stick with a particular brand even though the product quality may not always meet their expectations fully (Yoon et al., 2013).Conversely, less loyal customers may be more prone to brand change if they perceive better product quality to be available elsewhere (Nyadzayo & Khajehzadeh, 2016).According to Apriyanti et al. (2022), there is substantial and favorable impact on product quality on purchase decisions through customer loyalty in Cardinal fashion products.So, the hypothesis is proposed as follows.
H7 : There is an influence of product quality on purchase decisions through customer loyalty to fashion products in e-commerce and social media.
According to research results from Hong and Cho (2011), there is a relationship between trust and purchase choices through customer loyalty in B2C e-marketplaces.In addition, there is a large relationship between trust to interest in purchasing food products through loyalty (Calvo Porral & Levy-Mangin, 2016).Hence, the hypothesis proposed is as follows.
H8 : There is a relationship between trust and purchase choices through customer loyalty for fashion products in e-commerce and social media.
According to Pratama and Santoso (2018), the varying level of customers' trust in Stuck original products has an impact on how they decide what to buy.Similarly, the research on customer trust in honey products in Wamena Bina Marga shows that product quality impacts purchase decisions (Wahyuni et al., 2021).The following hypotheses are proposed.
H9 : There is an effect of product quality on purchase decisions through trust in fashion products in e-commerce and social media.
H10 : Product quality has an impact on purchase decisions through trust and loyalty to fashion products in e-commerce and social media.
The research aims to examine the reasons customers buy fashion products through e-commerce and social media using product quality, trust, and loyalty and find the most dominant gender in purchasing fashion products in e-commerce and media social.Therefore, the research results will be useful for e-commerce and social media platforms for developing applications that align with customers' expectations.Besides that, they can also be taken into consideration for product providers (manufacturers) to prioritize product quality, trust, and care for loyal customers.In addition, the results are also useful for the government in policymaking to provide consumers with a sense of security when shopping online.Unfortunately, many consumers are victims of fraud when shopping on e-commerce.Thus, consumers are not deceived when making transactions.With the policies made, e-commerce platforms can be more careful and responsible in running their business.

METHODS
The research applies a quantitative method.There are two data collection techniques with questionnaires and literature studies.First, the research uses primary and secondary data.Operational variables use 4 variables with 17 indicators from several previous studies.Purchase decisions recognize needs and want, search for information, and evaluate alternatives, purchase decisions, and post-purchase behavior (Kotler et al., 2019).Then, customer loyalty primary and secondary includes repurchasing, always liking the brand, recommending, and consuming habitually (Kotler et al., 2018).Moreover, trust has a guarantee of satisfaction, guarantee of replacement, attention, and disclosure of information (McKnight et al., 2002).Last, product quality consists of product performance, specifications, appearance, and durability (Tjiptono, 2008;Kotler & Keller, 2016).
Second, the research object is customers who have bought fashion products in e-commerce and on social media in Indonesia.According to Hair, Jr. et al. (2017), a minimum sample size ranging from 100 to 150 is recommended.Data are collected online with Google Forms.Online form links have been spread in various ways (Facebook, Instagram, WhatsApp).The sampling method also uses purposive sampling with age criteria from 17 to over 40 years because this age group uses social media the most (Kemp, 2021).The number of samples is as many as 244 customers.
The analysis technique uses data processing and presentation with SmartPLS 4.0 application and stages of Structure Equation Modelling (SEM).SEM modeling is used as the data processing technique in the research (Cepeda-Carrion et al., 2018).Then, convergent and discriminant validity uses the Heterotrait-Monotrait Ratio of Correlations (HTMT) validity test with a loading factor of more than 0,7 (Adam, 2018).The reliability test utilizes integral reliability.The composite dependability value may be used to calculate the reliability of a construct using reflexive indicators.
Composite dependability gauges a construct's true level of reliability.If a construct's composite reliability value is more significant than 0,07, it is deemed dependable (Hamid & Anwar, 2019).Additionally, structural model assessment (inner model) is a structural model that specifies the connection between latent variables and describes that connection in terms of the substantive theory of inquiry is referred to as an inner model.R-square (R 2 ) is used to assess the structural model, with R 2 values of 0,75, 0,50, and 0,25 for strong, moderate, and weak models, respectively (Ghozali, 2018).It implies that the prediction and suggested research models are better when the R 2 value is higher.
A global optimization criterion may be identified via the fit model, specifically, PLS path modeling, to assess the Goodness of Fit with the Golf index.In developing a model fit index or GoF, several criteria can be used to evaluate the quality of the fit between the theoretical model and empirical data.According to the SmartPLS website (SmartPLS, 2023), there are two criteria for structural equation analysis.First, Standardized Root Mean Square Residual (SRMR) is a measure to measure the average prediction error in the model.The SRMR value should be less than 0,10 or 0,08.The smaller the SRMR value is, the better the fit of the model is with the data.Second, Normed Fit Index (NFI) describes how well the model improves fit compared to the simple baseline model.The NFI value should be greater than 0,9.The closer the NFI value is to 1 is, the better the model fits.years old, followed by 24−27 years old.The result proves that it is dominated by young people.Moreover, female respondents dominate with 65%.In a month, the respondents mostly shop one to three times.Furthermore, after the descriptive analysis of the respondents, the next step is data analysis with Structural Equation Modeling (SEM) using the SmartPLS 4.0 program.It evaluates the measurement model with three criteria: composite reliability, convergent validity, and discriminant validity.If a reflexive measure's correlation with the measured concept is more than or equal to 0,70, it is high.Table 2 shows the results of the analysis measurement model.

RESULTS AND DISCUSSIONS
Table 2 shows that every construct loading value is more than 0,70, demonstrating that this model satisfies the requirements for excellent convergent validity.Next, utilizing discriminant validity ensures that each variable is unique from several factors.When compared to other loading values on other latent variables, a latent variable's loading value must be the highest as excellent discriminant validity characterizes the model.The outcomes of the discriminant following validity tests show that several loading factor values for each latent variable indicator still have a loading factor value that is not the biggest compared to the loading factor related to other latent variables.It is because certain latent variables are strongly linked with other constructs; not all latent variables have good discriminant validity.
Next, the Average Variation Extracted (AVE) score and the reliability score for each construct can also shed light on the validity and reliability criteria.The construct is regarded as good if the value is 0,70 and the AVE is more than 0,50.The AVE values listed in Table 2 show that all structures meet the dependability standards.All statements show that all AVE values are above 0.50 with the result of a purchase decision getting a value of 0,618.The value of product quality is 0,75.Then, trust and consumer loyalty have 0,761 and 0,601, respectively.
In the next stage, testing the structural model (inner model) refers to evaluating the relationship between constructs in the research model through significance tests of structural path parameters (see Figure 1) and R2 (see Table 3).The R2 is evaluated for the dependent construct to indicate how well the dependent construct can be explained by the independent constructs in the model.
The R 2 is interpreted from the value of endogenous variables as the structural model's capacity for prediction.It is to see whether specific external latent factors significantly impact endogenous latent variables.Their effects can be explained by changes in the value of R 2 .It is considered robust, moderate, and weak, with R2 values of 0,75, 0,50, and 0,25, respectively (Ghozali, 2018).It suggests that the more the R2 value is, the higher the quality of the suggested research model and prediction model will be.Therefore, analyzing R2 for each latent dependent variable in the model with PLS should be considered first.The outcome of the R2 estimate using SmartPLS is shown in Table 3.According to Table 3, the R2 value of customer loyalty is 0,525.It suggests that product quality and trust impact 52% of customer loyalty.Trust has an R2 value of 0,631.Then, product quality impact trust with R2 value of 63%.Meanwhile, the purchase decision has R2 value of 0,572.So, these results prove that product quality, trust, and consumer loyalty by 57% encourage purchase decisions in e-commerce and social media.The rest is influenced by other variables not examined in the research.
Next, the GoF index is created.According to the description on SmartPLS's website, the following restrictions or criteria for model fit must be met for the model to pass.First, the SRMR value should be 0,10 or 0,08.Second, the RMS Theta value is 0,102.Last, the NFI value is between 0 and 1 or, preferably, close to that value.Based on Table 4, the NFI value is 0,832, and the SRMR value is 0,064, which is less than 0,102.However, it is close to the best value, 1.So, the model in the research is a fit.
Using the SmartPLS 4.0 tool, the hypotheses are tested.The t-test is used to test the significance of structural path parameter coefficients.The results are in Tables 5 and 6, indicating that the relationship between the constructs is statistically significant.Hypotheses are tested using the outcomes of the internal 0,631 0,572 0,525 model (structural model) test, represented by the outputs of t-statistics, R 2 , and parameter coefficients.The relationship between p-values, t-statistics, and constructs can help determine if a hypothesis is valid or invalid.The results of the bootstrapping procedure may show these values.T-statistic > 1.96 with a p-value < 0,05 (5%) significance cutoff and a positive beta coefficient are the general criterion applied in this analysis.Table 5 shows the analysis of the data.
According to Table 5, it is known that the p-value is 0,000 < 0,05.So, trust is positively and significantly influenced by purchase decisions for fashion products in social media and e-commerce.H1 is accepted.The results are in line with the previous studies.Trust has a considerable and favorable impact on purchase decisions on Ride Inc websites (Mahkota et al., 2014).Trust affects customers' purchase of fast food in the UAE (Hanaysha, 2022).Customer trust has a favorable and large impact on purchase decisions in the case study of the Giant hypermarket (Meliana et al., 2013).Additionally, trust also affects purchase choices at the Mikaylaku online shop (Solihin, 2020).Purchase decisions are influenced by trust on supplier (Gao et al., 2005).In food supply chains in Italy, it has a favorable and strong association between trust and purchase decisions (Giampietri et al., 2018).Customers can be persuaded to purchase products via trust (Laini et al., 2022).
The research results and existing references support the importance of building customer trust.It also proves that customers consider the reputation of the e-commerce and social media accounts before purchasing fashion products on e-commerce and social media.If the reputation level of customer trust with comments, reviews, and followers is high, the level of customer trust to decide on a purchase is also higher.Hence, it can accelerate purchase decisions.
It is clear from Table 5 that there is a positive and substantial impact of trust on customer loyalty because the p-value is 0,000 < 0,05.So, H2 is accepted.These results also support previous research conducted by Ribbink et al. (2004), stating a direct relationship between trust and customer loyalty on the Internet.Customer loyalty can be formed through trust in customers who choose entertainment venues (Milman & Tasci, 2022).Trust also has a positive and significant effect on customer loyalty in hospitals (Martínez et al., 2013).Moreover, according to Kassim and Asiah Abdullah (2010), trust influences purchase decisions in e-commerce in Malaysia and Qatar.Trust also affects customer loyalty in PT Pandu Dana Utama Express in Manado City (Soegoto, 2013).Trust is why customers become loyal to Philips lighting products in Indonesia (Budi, 2021).
The research results and existing references show a close relationship between trust and customer loyalty.It also proves that e-commerce and social media have been able to generate customer loyalty by increasing trust when purchasing fashion products, which has been doubtful in the past decade.In addition, the requirements for loyal customers to e-commerce are if producers and e-commerce can fulfill all promised things to customers.
Customer loyalty affects purchase decisions for fashion products in e-commerce and social media positively and substantially.The p-value is 0,000 < 0,05, so H3 is accepted.These results also support previous research conducted by Kim et al. (2004) regarding a positive relationship between loyalty to online purchases.Similarly, a positive relationship exists between loyal customers and purchase decisions at Shopee in Indonesia (Supiyandi et al., 2022).There is also a significant influence of loyalty program membership on the customers' purchase behavior at the supermarket in general (Meyer-Waarden, 2008).Loyalty also affects purchase decisions on virtual travel (Sanchez-Franco & Rondan-Cataluña, 2010), drug stores (Stern & Hammond, 2004), and traditional products and services (Martinelli et al., 2015).Last, the decision to buy products on social media is based on trust and customer loyalty (Ferrao & Alturas, 2018).
The results of the research and references show that loyalty is one of the causes of purchase decisions for fashion products on e-commerce and social media.E-commerce and social media have succeeded in building customer loyalty as a sales medium through the service quality of existing e-commerce applications.Examples are Shopee, Tokopedia, Blibli, Tokopedia, and others.
There is a positive and significant correlation of product quality on trust in fashion products in e-commerce and social media because the p-value is 0,008 < 0,05.Hence, H4 is accepted.These results also support previous research conducted.Product quality affects trust in electronic device products in Africa (Chinomona et al., 2013) and generates trust through video streaming on e-commerce (Wang et al., 2022).The effectiveness of a product has an impact on community trust (Pranowo, 2019).In addition, product quality has an impact on trust in the life insurance sector (Panigrahi et al., 2018).According to Han et al. (2019), product quality and customer trust are related to full service versus low-cost carriers in South Korea.Product quality can make customers believe when purchasing regional agricultural products (Sun et al., 2022).
Research results and several references prove that with good product quality, customers have more trust in every product offered.Therefore, it is essential for fashion manufacturers to prioritize product quality when offering their products to customers.Then, various promotional media used will be accepted by customers.Moreover, fashion producers must strengthen the brand of the products so that customers will decide to buy wherever the product is offline or online.
Based on Table 5, product quality has a positive and significant effect on customer loyalty for fashion products in e-commerce and social media.The p-value is 0,005 < 0,05.Hence, H5 is accepted.These results also support previous research conducted.For example, there is an effect of product quality on customer loyalty at Happy Garden restaurants (Putro, 2014).Product quality can also make customers loyal to e-commerce in Indonesia (Putra et al., 2022).Similarly, there is product quality affects customer loyalty to Clear shampoo (Kurriwati, 2015), B2B manufacturing (Setyo et al., 2021), PDO products (Fandos & Flavián, 2006), PT Sido Muncul (Hermawan, 2011), and Jember's restaurant business (Musmedi & Prasodjo, 2018).
Research results and references show a close relationship between product quality and customer loyalty.This finding also proves that the fashion product industry sold on e-commerce and social media has been able to make customers loyal to the products consumed.The product also follows the quality promised to customers, so they are sure of the quality.Therefore, even though they do not have the opportunity to try on clothes in the online market, they buy them.Moreover, the product return facility for unmatched size, color, and others is also a strong reason why customers are loyal.
After analyzing the direct effect, the research indirectly discusses each variable's influence.It is clear from Table 6 that there is a positive and significant influence on product quality on customer loyalty through trust in fashion products in e-commerce and social media.The p-value is 0,009 < 0,05.So, H6 is accepted.These results also support previous research conducted.For example, product quality influences customer loyalty through trust in Tokopedia (Wijaya & Saparso, 2019).
The research results added with references have proven that the sale of fashion products through e-commerce and social media has succeeded in making customers trust and decide to buy.It is due to good product quality.Good product quality will increase the level of customer trust in purchasing fashion products.Therefore, product quality and trust cannot be separated as a way of making purchase decisions, especially in e-commerce and social media.
Next, product quality positively and substantially impacts purchase decisions through customer loyalty for fashion products in e-commerce and social media.The p-value is 0,002 < 0,05.Therefore, H7 is accepted.These results also support previous research conducted by Apriyanti et al. (2022) that there is positive and significant effect of product quality on purchase decisions through customer loyalty to Cardinal fashion products.
The results added with references prove the importance of loyalty to strengthen purchase decisions by improving product quality.If the product quality is not good and the e-commerce platform is good, purchase decisions will not occur.Conversely, if customers are loyal to the product, but the platform is not good, the purchase decision will still occur.
H8 is supported because the p-value is 0,014 < 0,05.Trust has a positive and substantial effect on purchase decisions through customer loyalty for fashion products in e-commerce and social media.Additionally, these findings are consistent with an earlier study (Hong & Cho, 2011) that there is an influence of trust on purchase decisions through customer loyalty in B2C e-marketplaces.Similarly, according to Calvo Porral and Levy-Mangin (2016), there is a large relationship between trust to interest in The research results indicate that trust can encourage customers to make purchase decisions in e-commerce and social media.It also proves the level of honesty on the part of e-commerce and social media producers in offering their products.In conclusion, e-commerce and social media have succeeded in building customer trust and loyalty.
H9 is supported since it can be inferred from Table 6 that the p-value is 0,007 < 0,05.Product quality positively and substantially impacts purchase decisions through trust in fashion products in e-commerce and social media.Additionally, these findings are consistent with earlier studies that there is a relationship between product quality and purchase choices due to the erratic level of customer trust in Stuck's original products (Pratama & Santoso, 2018).The result supports the opinion of Wahyuni et al. (2021) on the analysis of customer trust in Wamena Bina Marga Honey Products.It shows that product quality has an impact on purchasing decisions.
Research results and references show the importance of product quality and trust in encouraging customers to make purchase decisions on e-commerce and social media.It also proves that customer trust will arise if the product quality matches what is displayed on the platform.Then, if customers believe it, a purchase decision will occur.These results also prove that product quality and trust have been carried out properly by e-commerce and social media.
H10 is accepted because the p-value is 0,013 < 0,05 indicating a large and favorable impact of product quality on purchase decisions through customer loyalty and belief in fashion products in e-commerce and social media.These results indicate that these four factors cannot be separated from one another because they have a close relationship with product quality, trust, and customer loyalty to make product purchase decisions on e-commerce and social media.It also proves that e-commerce marketing media and social media have convinced customers to buy fashion products online and not offline.So, it needs to be considered by all lines to improve the quality of services for e-commerce and social media platforms.
Based on the results and discussion, the research helps the fashion product industry, which is sold online through e-commerce and social media, to become material for preparing marketing strategies using approaches.It can improve product quality, assurance, and honesty to increase customer trust so that customers will grow a sense of loyalty to products sold through these media.In addition, e-commerce and social media can add and develop programs/ facilities to increase customer trust and loyalty.The results also add managerial implications in the field of management science, especially variables to improve online product purchase decisions through product quality, trust, and loyalty which have been used for offline products.
The research also tests two groups based on gender between women and men.Partial Least Squares-Multigroup Analysis (PLS-MGA) is carried out to determine the value of the coefficient and significance.Based on Table 7, the influence of product quality on customer loyalty on women is greater than men.It has a significant effect.Then, the influence of customer loyalty on purchase decision is greater for women than men.Similarly, trust on purchase decision has a greater influence on women than men.However, the influence of product quality on trust is greater in men than women.The result proves that the role of women when buying products online is very large.

CONCLUSIONS
Using sales methods with social networking and e-commerce is beneficial to increase trust.The customers feel the certainty that the product purchased follows what is displayed on e-commerce and social media, such as price, size, materials used, and color.In addition, the existence of product return facilities when the product does not meet customer expectations also makes customers feel safe.The positive impact is also on the effect of customer loyalty by using this sales method.The customers feel that their expectations are fulfilled if the social media and e-commerce accounts have a high level of trust from other customers by looking at the seller's information based on comments from buyers and recommending these social media or e-commerce accounts to other customers.The results prove that customer loyalty can be formed through online sales, which was previously considered impossible before.
E-commerce and social media can also encourage customers to make purchase decisions after customers have observed product quality in detail by looking at trusted e-commerce and social media platforms.Then, trust emerges.So, the impact of trust is

Relationship
Difference of Path Coefficient (Women-Men)

P-Value Conclusion
Product Quality  Customer Loyalty -0,403 0,023 Accepted Product Quality  Trust 0,532 0,002 Accepted Customer Loyalty  Purchase Decision -0,721 0,001 Accepted Trust  Purchase Decision -0,098 0,209 Rejected that customers feel loyal and decide to make purchases online.Manufacturers and providers of e-commerce platforms can offer product quality after building trust in customers by telling every product offered honestly and guaranteeing that the product can be returned if it is not in line with customers' expectations.After that, the manufacturers and providers of e-commerce platforms can ask customers to provide comments on each product received to increase customer trust.The analysis findings demonstrate how product quality affects customer loyalty and purchase behavior.Then, there is also the influence of loyalty on purchase decisions.Moreover, product quality and trust have an impact on purchase decisions through customer loyalty.
Furthermore, based on the results of PLS-MGA, women are the most dominant consumers and often make fashion purchases.It proves that women already believe in product quality.So, they are loyal to e-commerce and social media compared to men who still lack confidence in the quality of fashion products offered.
The research limitation includes the number of samples qualified.It is still small and limited.Then, the number of variables is small, so the number of research findings is also limited.Therefore, further research can use a more significant number of samples and new variables.It can add more to the findings.
Based on Table 1, it is known that Shopee occupies the top position as the platform for customers, with a total of 36,6%.It is followed by Instagram and Tokopedia.Based on age, it is dominated by 21−23 Table 1 Respondents' Profile in the Research

Table 2 The
Results of the Measurement ModelFigure1The Tested Structural Model

Table 3 The
Results of R-Square Values

Table 5
Bootstrapping PLS-SEM for Direct Effects

Table 6
Bootstrapping PLS-SEM for Indirect Effects