FACTORS AFFECTING CUSTOMER’S BEHAVIOR AND INTENTION ON PURCHASE INTENTION THROUGH BRAND RETAIL’S SOCIAL MEDIA

ABSTRACT


INTRODUCTION
Social media users in Indonesia now reach 160 million people (Datareportal 2020). This has an effect on the marketing model that has begun to shift from conventional marketing to marketing through social media. Social media platforms are rapidly changing the way consumers interact with products and retail companies (Kao et al. 2016;Sidharta et al. 2021). Retail brands use technology to reach consumers, and almost all have a form of presence on social media. Several studies have shown that buyers use social media to make purchases online while interacting with companies (Haslehurst et al. 2016). Consumer engagement is getting higher and more and more retail companies are using social media to communicate with consumers (Assimakopoulos et al. 2017). According to Lin et al. (2014) and Van Doorn (2010), consumers' involvement in retail brands can be defined as the representation of their continuous intention to use social media to interact or purchase products and services.
Social media has influenced various aspects of consumer behavior including the level of consumer awareness of a product (Mangold and Faulds 2009), internet usage patterns (Laroche et al. 2012;Ross et al. 2009), as well as information obtained on behavior after purchase, for In particular, there are four gaps found in this study, namely: First, at least consistent research on consumer interactions with retail brands through Facebook (Tsai and Men 2013;Yadav et al. 2013), as well as the potential impact of such interactions on purchasing behavior (Bianchi and Andrews 2018) in Indonesia. Second, in general research on social media is conducted on cognitive variables (e.g. perceived usefulness) as drivers of intention to engage in social media, and few studies consider affective variables (e.g. enjoyment), relational variables (peer communication), and the influence of social media credibility on purchase intentions (Bianchi and Andrews 2018). Third, there has never been any research on the relationship between consumer attitudes towards engagement with retail brands and consumers' intention to purchase retail products through Facebook (Bianchi and Andrews 2018). Fourth, according to Wang, et al. (2019), when consumer expectations of products correspond to reality, their purchasing intentions will also increase. Hence, this study included a new variable of confirmation expectations as a moderator of consumer attitudes towards consumers' intentions to buy retail products.

METHOD
The model hypothesis was tested through an online survey conducted on consumers. The study's population consisted of Facebook users in Indonesia who had followed the retail brand's Facebook page and worked in the office. The questionnaire was developed in Indonesian to ensure respondents' understanding of the question. The survey will be tested ahead of time with an initial sample of 30 consumers in Indonesia, which is expected to result in small changes to the wording for some questions to improve meaning. After that, an online survey will be sent to several participants from a database of random Facebook users in Indonesia. Participants received a message presenting the purpose of the study and a survey link.
The questionnaire was distributed in the time span between January 2021 and April 2021. The respondents of this study are the general public who work in offices, this can be seen from 195 respondents consisting of 67 working as executives/managers, 34 working as teachers, and the remaining 96 respondents are office workers/cashiers/administration. Hence, the respondents represent office workers in Indonesia.
The survey began with a section that asked about social media usage and the main activities respondents did on Facebook related to the brand. The following sections provide items that measure the variables indicated in the model. The questionnaire uses the steps available from the previous literature.
All steps used in this study were taken or adapted from previous studies and measured on the Likert scale with a scale of 1 -5 (1 = strongly disagree once and 5 = strongly agree with it). The measurement of the dependent variable, the intention to buy through social media, was measured by three questions adapted from Lin et al. (2014). The measurement of intent to engage with retail brands through social media is measured by three questions adapted from Lin et al. (2014) and Yang (2010). The measurement of attitudes in involving with retail brands through social media is by 3 questions adapted from Yang (2010). To measure usage perception, the 3 questions were adapted from Venkatesh et al. (2003). Meanwhile, to measure credibility, the researchers used questions adapted from Colliander and Dahlén (2011). The measurement of enjoyment was adapted from Lin et al. (2014) which was measured by three questions. For peer communication measurements adapted from Wang et al. (2012) which were measured by four questions. And the last is the measurement of confirmation expectations adapted from Wang et al. (2019).
SPSS is used to test validity using confirmatory factor analysis for validity tests, namely by looking at the Kaiser-Msyer-Olkin measure of sampling (KMO) values and measures of sampling adequacy anti image matrix correlation (MSA). In the validity test with the analysis of variable factors perception of usability, compatibility, enjoyment, credibility on social media, peer communication, sikap towards engagement with retail brands through social media, intention to buy an item through social media, and expectation of confirmation of all statements are declared valid, provided that the KMO and MSA (anti-image correlation) values > 0.500 with 1 component matrix. Likewise, the reliability test of all variable statements is declared reliable with the provision that the alpha cronbach > 0.6 (Sekaran and Bougie 2016). Thus, after analyzing the pre-test results of 30 respondents, all statements were declared valid and reliable to be used as questionnaires in this study, namely 29 questions.
Meanwhile, structural equation modeling (SEM) uses Lisrel 8.80 to test the proposed model and hypothesis path.

RESULTS AND DISCUSSION
From the results of descriptive statistics, an overview of the sample of social media behavior is obtained (see Table 1). Although these characteristics do not allow generalizing the data to a broader population, the goal is to reach a sample of participants who use social media and to examine their attitudes and intentions to engage with retail brands through social media. It was found that 100 percent of respondents had used Facebook at least once. The vast majority (87%) of respondents visited a facebook page less than equal to one (≤ 2) hours per week. All respondents have visited or liked a retail brand's Facebook page at least once. Lisrel 8.80 is used for single measurement models and confirmatory factor analysis (CFA). Prior to the CFA analysis, the measurement of the construct validity test in this study showed that all indicators were acceptable / valid on all variables with a factor loading value (>0.50), and a t-value greater than the t-table (1.96) at a significance level of 5%. While the construct reliability test, it is said to meet the reliability requirements if the CR (Construct Reliability) value is above 0.60 and the VE (Variance Extracted) value is above 0.50 (Hair et al., 2018). The results of the scale reliability test from the measurement revealed that the contruk had met the reliability requirements.
Based on the analysis of model suitability (table 2) from the first group to the seventh group, all test results obtained a good match including Chi Square, ECVI, AIC and CAIC, Fit Index. There are results in the form of close fit, namely RMSEA and in the form of marginal fit, namely Critical N and Goodness of Fit. Thus, it is concluded that the match of the entire model has met the requirements. All constructs undergo factor analysis. This procedure illustrates that there is no single factor or common factor that accounts for most of the variance in independent and dependent variables, this indicates that there is no bias problem in the data. Structural tests are also carried out to be able to find out the value of R² in each equation which serves to see how far the independent variable is able to explain the dependent variable. Based on the SEM analysis, the results were obtained, namely the first, the attitude variable towards engagement with retail brands through social media (AT) was influenced by the variables Perception of Usability (PU), Compatibility (CO), Enjoyment (EN), Credibility (CR), and Peer Communication (PC) with an R value of 2 of 0.92. Thus it can be interpreted that 92% of the attitudes towards engagement with retail brands through social media (AT) can be explained by Perception of Usability (PU), Compatibility (CO), Enjoyment (EN), Credibility (CR), and Peer Communication (PC), while the remaining 8% is explained by other variables outside the variables in the study.
The second analysis is that the Intention to engage with retail brands through social media (IN) variable is influenced by the Attitude variable towards engagement with retail brands through social media (AT) with an R2 value of 0.82. This can be interpreted to mean that 82% of the intention to engage with retail brands through social media (IN) can be explained by the variable Attitudes towards engagement with retail brands through social media (AT), while the remaining 18% can be explained by other variables.
The third analysis is that the variable Intention to buy through social media (INP) is influenced together by the variable Attitude towards engagement with retail brands through social media (AT), Intention to engage with retail brands through social media (IN), and the moderation role of Confirmation of expectations (ATEC) with an R2 value of 0.68. This can be interpreted to mean that 68% of the variants of Intention to buy through social media (INP) can be explained by the variable Attitude towards engagement with retail brands through social media (AT), Intention to engage with retail brands through social media (IN), and the role of moderation from Confirmation of expectations (ATEC) while the remaining 32% can be explained by other variables outside this study.
The results of the study are as depicted in the T-Value diagram (see figure 2). The hypothesis is accepted if the T-Value value>1.96 and not accepted if the T-Value<1.96. As shown in Table 2, the results show H1, H2, H4, H5 and H9 have a value of t-value<1.96 so the hypothesis is not supported. In addition, the results of the study also showed that H3, H6, H7, and H8 had a t-value value of > 1.96 so that the hypothesis tested was supported by research data (see table 3). The data supports the hypothesis The findings confirm three significant antecedent variables of attitudes that are indirectly related to consumer engagement intentions with retail brands: peer enjoyment, credibility and communication.

Figure 2. T-Value Diagram Path
The results showed that pleasure is the most powerful positive antecedent of attitudes. This suggests that a consumer's intention to engage with retail brands through social media increases when he or she feels enjoyed using social media (Davis et al. 1992;Lin and Lu 2011). These results support previous studies showing that people have a stronger motivation to use technology if they feel intense pleasure while doing so (Lin and Lu 2011). These findings suggest that enjoyment is important for attitudes towards social media use, this condition is sufficient to generate a positive attitude towards engagement with retail brands through social media.
Furthermore, the researchers discovered that credibility on social media (Facebook) was found to be quite strongly related to the attitude to engage with retail brands through their Facebook pages. These findings support previous research linking the credibility of social media to consumer interactions with brands through social media (Eastlick, Lotz, and Warrington 2006;Gefen et al. 2003;Pavlou and Fygenson 2006). However, in this study, credibility negatively influenced attitudes to engage with retail brands through their Facebook pages. We tried to deepen qualitatively, namely the data collected naturally (Tamaschke 1996) to get the whole picture. The object of this study was 5 people who were office workers, Facebook users and were willing to be interview objects according to the sampling requirements according to Sugiyono (2017). From the interview results, it was found that all subjects stated that they knew that the retail brand's facebook page had good credibility but they had a reluctance to engage with the retail brand because they saw that the facebook page was less attractive and unpleasant to follow. This is likely due to office workers in Indonesia who only have little time to use Facebook, this makes the attitude to want to engage with the social media of retail brands even smaller because of their priority in social media. These results have influenced the measurement of social media credibility (Li and Zhang 2018). From these arguments, it can be said that such credibility does not positively affect the attitude to engage with the social media of retail brands.
The results showed that peer communication has a strong relationship with attitudes. These results support previous research that has shown that peer communication has a strong impact on attitudes towards shopping orientation (Lueg et al. 2006) and consumer decision making (Smith et al. 2005). However, the results of this study show that consumers' intention to engage with retail brands through social media is negatively influenced by peer communication. This research was also carried out qualitatively, namely the collection of interview data naturally (Tamaschke 1996), this was done to get a complete picture of the research object. Qualitative interviews were conducted on 5 office workers who were Facebook users and were willing to be interview objects according to the sampling requirements according to Sugiyono (2017). From the interview results, it was found that all respondents admitted that they communicated with their peers about the retail brand's Facebook page. The thing that is communicated with peers is mostly negative content from the retail brand's facebook page because it is considered interesting to talk about so it makes an attitude not to want to engage with the retail brand's facebook page. The limited time for office workers to access social media is also a factor that they only talk about interesting things so that it affects the attitude to engage with the retail brand page so low.
Based on the research findings, the perception of usefulness has no significant effect on attitudes. The influence of the perception of usability does not sufficiently support the proposition of the TAM. Consistent with some previous studies (Bianchi and Andrews 2018; Venkatesh et al. 2003), these findings suggest that the perception of the usefulness of retail brand social media sites does not have a significant impact on the attitude of engaging with retail brands through social media. Although consumers in Indonesia consider that the retail brand facebook page is very helpful to them, it does not affect their attitude to engage with the retail brand facebook page. This may happen because for office workers there are difficulties and limited time to access social media during working hours. This is evidenced by the extensive use of social media on a daily basis.
In this study, it was also found that compatibility did not have a significant effect on attitudes. This shows that the compatibility of retail brand social media sites does not have a significant impact on the attitude of engaging with retail brands through social media. This unexpected finding implies that on the one hand, although the Facebook page of a retail brand is considered by Indonesian consumers to be in accordance with their lifestyle and suits them, it does not imply an attitude towards engagement with retail brands through social media. This may also be due to limited time to access social media during working hours. The ineffection of compatibility was also found in previous studies including the research of Ramdani et al. (2009) which found compatibility to be an insignificant factor in the application of corporate systems. In the same case Low et al. (2011) also found insignificant compatibility impacts in cloud computing deployments.
The results of the study also found that consumers' attitudes to engage in retail brand social media have a positive effect on the purchase intention of retail products. This is in accordance with the theory of planned behavior (SDGs), which states purchase intentions are largely determined by consumer attitudes (Bhattacherjee 2000;George 2004;Phau and Teah 2009). This also supports previous research which stated that the more positive the attitude towards a product, the higher the purchase intention will be (Bianchi and Andrews 2018; Chiu and Leng 2016). The results of the study also revealed that the attitude to engage with retail brands' social media towards product purchase intentions has a greater direct effect value (1.02) than the indirect effect value (0.36) through the mediation variable of the intention to engage with retail brand social media. So that the attitude to get involved will tend to directly encourage consumers' intention to buy products.
This study also confirms previous research on the topic of attitude relationships towards intention to engage with retail brand facebook pages (Bianchi and Andrews 2018) namely predicting attitudes and intentions to engage with retail brands through social media. The findings reveal that consumers who have a more positive attitude towards engagement with retail brands through social media, will tend to engage more with the retail brand's Facebook page.
The study also revealed that the intention to engage with the social media pages of retail brands positively affects consumers' buying intentions towards products through social media pages. This is in accordance with the results of research by Bianchi & Andrews (2018). This corroborates the literature of planned behaviour theory by proving that consumer attitudes influence the purchase intention of retail brand products through social media.
Confirmation expectations are the perception that users' expectations match the reality of using social media services. If individual expectations match the actual performance of social media services, it will increase user confidence increasing purchase intentions (Wang et al. 2019). Consumers form satisfaction, or influence, based on their level of confirmation and expectations on which their confirmation is based (Bhattacherjee 2001). Bhattacherjee (2001) also revealed that confirmation expectations are cognitive beliefs (the degree to which user expectations of system use are realized during actual use) derived from previous uses. It affects subsequent use through satisfaction and the construction of intentions. The intention will always change from the initial acceptance to the next or not (Bhattacherjee 2001). Thestudy showed an insignificant influence on the variability of confirmation expectations moderating consumer attitudes towards purchase intentions towards retail products. We also conducted qualitative deepening by collecting data naturally (Tamaschke 1996) to obtain a complete picture of respondents' answers. This qualitative research was conducted on 5 office workers who are Facebook users and are willing to be interview objects according to sampling requirements according to Sugiyono (2017). The interview results stated that respondents were right to have confirmed expectations for the retail brand's facebook page but that did not make them want to buy back similar or other products beca use they would see their needs according to their priorities. They also stated that their expectations could also change over the assessments made during their use of the products they purchased. And the results of qualitative interviews also found that Most of their confirmation expectations dropped in line with the use that caused discontent. This shows that even though expectations are confirmed, it does not necessarily affect the attitude of consumers to buy products through the retail brand's Facebook page. This is in accordance with the research.

CONCLUSION
In general, the results of this study suggest that the suitability or suitability of social media pages may no longer be of interest to consumers. Therefore, to ensure consumer engagement and long-term purchasing behavior requires strategies that emphasize visitor enjoyment, providing engaging and appropriate content, as well as positive social relationships between users within the retail brand's Facebook social media pages. Overall, the findings of this study contribute to the growth of knowledge about consumer engagement on social media platforms and provide theoretical and managerial implications that may help scholars and practitioners in the future.
This study has some limitations, the first, the hnaya sample involves consumers who work in the office because they are considered potential online shoppers. In developing countries, such as Indonesia, the demographic segment of office workers tends to have access to digital technology. Therefore, they can provide more relevant information for research, thereby addressing challenges in reliable data collection for the retail market. New researchers can conduct similar studies using samples taken from more populations or on other demographic segments. Second, the study focused on only one social media site: Facebook. Although Facebook is the world's leading social media site, subsequent studies should consider other social media sites where retail companies and brands are also present there, such as YouTube, Twitter and Instagram, to gain a broader understanding of consumer-brand engagement through social media.