SOCIAL MEDIA MARKETING AND PURCHASE INTENTION: EVIDENCE FROM KOSOVO

The main objective of this study was to explore the relationship of social media marketing and purchase intention among consumers in Kosovo as a transition economy. The study used a structured questionnaire to collect the primary data and collected a sample of 334 responses from participants in Kosovo, using google forms as an online survey tool. In order to test the proposed research model, structural equation modelling was used. The hypotheses and the mediating effects of brand awareness and brand engagement were tested using the bootstrapping method. The findings revealed a positive indirect impact of social media marketing on purchase intention. They also showed that brand engagement fully mediates the relationship between social media marketing and purchase intention, whereas brand awareness does not mediate the relationship between social media marketing and purchase intention. This study contributes to the literature by providing insights of the impact of social media marketing on brand awareness, brand engagement, and purchase intention in transition economy such as Kosovo. On a practical level, research results provide companies with customer insights relevant to designing effective social media marketing strategies for their potential prospects in transition economies.


INTRODUCTION
The topic of digital marketing has been researched a lot recently, but still there is little research on the impact of social media marketing in emerging markets in general, and especially in Kosovo market. Even though there are some challenges in emerging markets with the adoption process of social media marketing, still the adoption process is gaining momentum because of the benefits it provides to both companies and businesses (Olotewo, 2016;Tamilmani et al.,2018). In this line, concerning the slow adoption process of social media in emerging markets Coleman, et al., (2016) outlined technological infrastructure, country regulations and the cultural values of these emerging economies as some important challenges.
More specifically, the main aim of this research is to explore the impact of social media marketing, brand awareness, and brand engagement on purchase intention. In addition, the research tries to analyse whether brand awareness and customer engagement mediate the relationship between social media marketing and purchase intention.
The paper introduces the research problem. The subsequent section reviews the relevant literature regarding social media marketing, brand awareness, brand engagement and purchase intention. Next, the methodology and empirical findings are shown, and finally discussion and conclusion are presented.

LITERATURE REVIEW
From a company perspective the proliferation of social media marketing helps them reach their target customers using any of social media platforms. Businesses perform numerous marketing activities thanks to social media platforms. They can create profiles for their brands, offer online services, provide information about their products as well as any promotional strategy (Breitsohl et al., 2015).
This kind of interaction provides companies with customer insights in order to assess customer's journey in different stages of the adoption process. From a consumer's perspective, the use of these digital technologies offers customers many benefits, such as: useful information, efficiency, comparing products and services, comparing prices, convenience, variety of products, etc. (Tiago & Verissimo, 2014). Companies in the past used to engage their customers through many forms of traditional marketing communication tools including television, radio, direct marketing, telemarketing to disseminate information about their products or services. Consumer behaviour has changed a lot concerning technology adoption, and as a result consumers nowadays prefer to use internet for getting and sharing information, getting feedback and reviewing products and services. Therefore, companies are pressured in order to interact with their customers, and henceforth becoming competitive, they need to change the way they used to interact and communicate with their targets and fit in a new digital media environment (Järvinen et. al, 2012).
Social media platforms seem to enhance consumer interaction more proactively, and companies should use social media marketing to communicate with their prospects. (Tiago & Verissimo, 2014). The popularity of social media is rising every day. People can communicate with each other about anything that interests them in this digital environment. More importantly, people talk and share experiences about products and services they buy and consume (Gray & Fox, 2018). This makes social media marketing a very valuable tool for companies to promote their marketing offers.
Social media has become a very popular place for finding information, since people post information, videos, photos, and other posts using different social media, including Facebook, Instagram, LinkedIn, Pinterest, etc. Organizations can use different social media marketing strategies, a passive approach focusing on using social media as a source of customer insights, and as a source of customer engagement (Constantinides, 2014).
Based on the above literature review a conceptual framework is proposed with purchase intention as a dependent variable, and brand awareness, customer engagement and social media marketing as independent variables. Therefore, the hypothesized relationship between these variables is depicted in Fig. 1.

Social Media Marketing
Information technologies have changed the way people and companies communicate and interact. Nowadays consumers use social media platforms to interact and share information with others. The increased number of social users has forced companies to find new ways to communicate with their target markets. As a result of social media expansion made companies to be more active with digital marketing activities for their marketing purposes (Çizmeci, & Ercan, 2015).
Organizations use many social media platforms, and the most common used are Facebook, Twitter, Instagram and YouTube (Cole et al., 2017). This kind of digital marketing strategy enables companies build customer relationships (Patrutiu- Baltes (2016). A lot of studies point out the impact that social media marketing has on brand awareness (Seo, & Park, 2018;Ansari, et al., 2019), and on brand engagement (Holliman & Rowley, 2014;Meire et. al, 2019), and its impact on purchase intention. Social media impacts the purchasing intention and enhances experience sharing (Hajli, 2013). Therefore, the following hypotheses are postulated as follows: H1a: Social media marketing has a significant direct positive impact on brand awareness H1b: Social media marketing has a significant direct positive impact on brand engagement H1c: Social media marketing has a significant direct positive impact on purchase intention

Brand Awareness
Companies use their brands to distinguish their products and services from other competitors. Brands embody in themselves new names, symbols or logos for new products or services (Keller 2008). Branding is all about differentiating products and services from other offers in a given market (Kotler & Keller 2016). Brand awareness is the ability of consumers to recognize that a certain brand belongs to a certain product category (Aaker, 1996). Brand awareness helps consumers decrease the time spent during the search process for a product (Verbeke et al., 2005).  Aljumah, et al., (2021) also support the correlation between social media marketing and brand awareness. Therefore, based on the above we propose the following hypothesis: H2a: Brand awareness has a significant direct positive impact on purchase intention H2b: Brand awareness mediates the relationship between SMM and purchase intention

Brand engagement
There are different definitions concerning the comprises customer engagement. Many scholars define customer engagement based on psychological perspective and organizational perspective (Hollebeek, 2011). The former is linked with the value perceived by the customer during interaction with the brand (Gambetti, et al., 2012;Brodie, et al., 2013;Bowden, 2009) and the later represents the interaction of consumers with the product and organization (Kumar, et al., 2010;Verhoef, et al., 2010). The rapid diffusion of social media communication influenced the process of interaction between people, the way they communicate and the way they interact with one another (Budden, & Jones, 2011;Tiago & Veríssimo, 2014). Many studies point out that customer engagement leads to purchase intention. For example, studies by Husnain and Toor, (2017) ;Meire et. al, (2019); Chen and Xu, (2019), and Pansari, and Kumar, 2017) reveal that brand engagement moderates the relationship between SMM and consumer purchase intention. Therefore, the following hypotheses are postulated as follows: H3a: Brand engagement has a significant direct positive impact on purchase intention H3b: Brand engagement mediates the relationship between SMM and purchase intention 4. METHODOLOGY

Sample and data collection
The data were collected from a sample of 334 responses from participants in Kosovo, using google forms as an online survey tool. The study tried also to reveal the mediating effect of both brand awareness and brand engagement on the relationship between social media marketing and purchase intention in Kosovo. A convenience sampling technique was used to collect the data. The data were collected from March to May 2021.

Instrument
This study used a structured questionnaire designed by the authors to investigate the effect of social media marketing on brand awareness, brand engagement, and purchase intention. Respondents were asked to respond the structured questionnaire with a 5-point Likert scale, which was distributed to them through online google forms. The first part of the questionnaire provided demographic data of respondents and the second part customer perceptions toward the impact of SMM on BA, BE, and PI. Table 1 shows the respondents' characteristics. Of the respondents, 121 were males (36.2%), and 213 were females (63.8%). Concerning age, 99 (29.6%) were under 20. Most respondents were 21 to 30 years of age, or 177 (53.0%) of the sample, 45 (13.5%) were 31 to 40, 7 (2.1%) were 41 to 50, and 6 (1.8%) were more than 51. Regarding respondents' social media experience 4 (1.2%) were with 1 to 2 years of social media experience, 45(13.5%) with 3 to 4 years of social media experience, 82 (24.6%) with 5 to 6 years of social media experience, and majority of respondents 203 (60.8%) were with more than 7 years of social media experience. More than half of the respondents were college students 154 (46.1%), 102 (30.5%) working full-time, 52 (15.6%) working part-time and going to college, 26 (7.8%) working part-time only. Table 1 Demographic profile of respondents

Data analysis and results
A structural equation modelling (SEM) is used to test observant and latent variables. A two-step strategy for assessing the measurement model and the structural equation modelling was used. The SEM technique was used to evaluate the measurement model (validity and reliability of the measures) and to estimate the structural model. A bootstrapping method was used for testing the significance of the path coefficients and the factor loadings (Hair et al. 2016).

Measurement model, reliability and validity
To calculate scale reliability of the four constructs, Cronbach's alpha was used. The proposed threshold of 0.70 or greater reliability coefficient indicates a good reliability (Hair et al., 2014). As it can be seen in Table 2, Cronbach alpha for all constructs is above 0.7 recommended threshold (Ursachi et al., 2015).
This study used SEM to analyse the proposed research model. Further, SEM was used to test hypotheses and validate the conceptual model. The SEM technique evaluates R 2 , β, and t-values (Zeqiri et al., 2020) to assess the model fit. Therefore, SEM technique is used to evaluate the measurement model and the structural model. The obtained results from CFA showed that chi-square/df was 2.071 indicating satisfactory fit; a ratio of five or less is a good fit (Wheaton et al., 1977), CFI (0.971), GFI (0.949), AGFI (0.919), NFI (0.945), TLI (0.960), and RMSEA (0.057) all denote an adequate fit of the model. In order to test the normality of the data, all VIF values are used to check if the model is biased. VIF values greater than 3.3 indicate a collinearity problem (Kock, 2015). Thus, the results denote that the model has a normal fit of the data (see Table 2).

Convergent validity
Convergent validity examines how close items relate to each other in a construct (Becker et al., 2013). Convergent validity is usually tested by assessing the loadings, average variance extracted (AVE) and the composite reliability (Gholami et al. 2013). Cronbach's alpha for social media marketing construct is 0.726, 0.850 for brand awareness, 0.800 for brand engagement, and 0.807 for purchase intention. In addition, composite reliability was assessed showing that all constructs are above the proposed threshold of 0.70. The results in table 2 reveal that all four constructs had reliability coefficients greater than the proposed threshold of 0.70. Convergent validity is reached if the factor loading exceeds the 0.7 threshold and the average variance extracted (AVE) exceeds 0.5 (Fornell and Larcker, 1981).

Discriminant validity
Discriminant validity examines how items in one construct are different statistically from items in other constructs (Zeqiri, 2020). As it can be seen in table 2, the AVE is greater than 0.5 for all constructs, except Social MM which is 0.477. Therefore, achieving discriminant and convergent validity, the square root of the AVE should be greater than the correlations among the constructs (see Table 2).

Structural Model Estimation
As shown in Figure 2, the R 2 (R Square) for the proposed research model is estimated that purchase intention variance is explained by 75 percent, brand awareness is explained by 71 percent of its variance, and brand engagement is explained by 58 percent of its variance by the model.   Table 5 shows the results of path coefficients of the constructs. The results in table 5 indicate that social media marketing positively affected brand awareness with path coefficient = 0.975, t-value = 10.391, p < 0.000). Thus, H1a, is supported. H1b is also supported, social media marketing positively affects brand engagement with path coefficient = 0.997, t-value = 10.119. The results reveal that social media marketing does not have any significant impact on purchase intention, with path coefficient = 0.190, t-value = 0.400, p < 0.409). Thus, h1c is not supported. The results in table 5 also show that brand awareness does not have any effect on purchase intention, with path coefficient = 0.071, tvalue = 0.486, p < 0.627). Thus, h2a is not supported. Finally, the results show that brand engagement has a significant positive effect on purchase intention, with path coefficient = 0.0739, t-value = 6.688, p < 0.000). Thus, h3a is supported.

Mediation effect of brand awareness and brand engagement
A mediation effect analysis was used to find out if there were any mediation effect of brand awareness and brand engagement on the relationship between social media marketing and purchase intention. A mediation effect occurs when the direct path between the independent variable and dependent variable decreases when the mediator variable is used in the model. A full mediation effect occurs when there is not a direct effect of independent variable to dependent variable, and the effect is only reached through the use of the mediator variable. The significance of the mediation was measured using Amos bootstrapping technique to reveal the mediating effect of brand awareness and brand engagement on purchase intention. The results were based on factor loading and p-values. Source: Author's calculation based on results Table 6 reveals that brand awareness showed no mediation effect on the relationship between social media marketing and purchase intention. Thus, H2a is not supported. Brand engagement mediates fully the relationship between social media marketing and purchase intention as seen from results in Table 6, with path coefficient 0.737 and p-value= 0.000. Thus, supporting H3b, because the direct effect of SMM on PI is not significant, and its full effect is reached through the mediation of BE.

DISCUSSION AND CONCLUSION
Social media platforms have become a current mean of communication where consumers express their purchasing behaviour, their experiences, their likes and preferences concerning particular brands. Social media facilitates consumer interaction and communication with other users. In today's world more and more consumers are present in social media platforms, which urges companies to make efforts to reach their clients in social media platforms. This presents a significant opportunity for companies to reach large consumer groups with much less efforts. For this reason, many companies have created their websites and are present with their profiles on social media platforms and try to engage consumers providing them online contents about their products, and services. This research has been conducted to explore the effect of social media marketing on purchase intention in Kosovo by deploying a model consisting of three independent variables related to social media marketing, brand awareness, and brand engagement, and purchase intention as a dependent variable.
The results from the study reveal that social media marketing had a positive and a significant effect on brand awareness. This finding corresponds with many other previous studies that investigated the impact of social media marketing on brand awareness, such as, findings from Seo and Park, (2018), Ansari, et al., (2019), Jamali and Khan, (2018), Ezenwafor, et al., (2021), Aljumah, et al., (2021 and other studies. The findings also revealed that social media marketing was positively related with brand engagement. These findings are supported by other studies including Kumar, et al., (2010), andVerhoef, et al., (2010). In addition, the findings did not show any positive direct impact of social media marketing in purchase intention. This finding justifies the idea that companies need to be vigilant in their content marketing since providing information or making efforts for brand awareness does not necessarily lead to purchase intention (Baltes, 2015). This emphasizes the importance of content marketing strategies to be deployed in order to persuade consumers to buy products or services.
Moreover, the findings also revealed that brand awareness did not show any positive and significant impact on purchase intention. On the other hand, brand engagement proved to be positively and significantly related with purchase intention, corresponding with other studies such as Husnain andToor, (2017), Meire et. al., (2019), Chen and Xu (2019), Pansari, and Kumar (2017), and other studies.
Our findings also revealed the mediation effect that brand engagement has on the relationship between social media marketing and purchase intention. Moreover, brand awareness did not mediate the relationship between social media marketing and purchase intention.
Our study is unique in that it contributes to research of social media marketing in emerging markets. The study offers customer insights concerning consumer behaviour regarding the influence that social media marketing has on purchase intention. The study provides some interesting facts that even though social media marketing enhances brand awareness but does not necessarily mean that consumers being aware of certain brands would buy them at the end unless they are engaged or linked with the brand. Companies need to try to engage consumers since the results revealed that brand engagement leads to purchase intention.
The findings also provide organizations with some insights how to deploy digital marketing strategies in emerging markets in order to reach consumers to promote their products and to engage them to purchase behaviour (Hollebeek, & Brodie, 2014;Cheung, Pires, & Rosenberger, 2020). Despite this study's contributions to literature, it has several limitations. First, we used a convenience sample. Second, the sample size is from only one country, which makes the results very difficult to generalize in other countries.
Future research is recommended to focus on cross national analysis by performing multi group analysis to check similarities and differences about consumers' perceptions in different countries concerning the effect that social media marketing has on purchase intention.