“The impact of social commerce on the purchase intentions of Millennials using Facebook”

This study examined the impact of social commerce on the purchase intentions of Millennials who use Facebook by exploring how social commerce constructs influence consumer trust. A quantitative research approach was used and data were collected via an ‘online’ survey. The target population was 386 young adults aged 25 to 34 residing in KwaZulu-Natal, South Africa, referred to as Millennials, and comprises the largest group of users on Facebook. Convenience sampling, namely snowball sampling, was used to target participants. It was found that social commerce constructs significantly influence trust, which positively influences consumer purchasing decisions. The re-sults of the study showed that trust explained 68% of the variance in purchasing intentions. Since trust is an integral and vital component of social commerce, the role of social commerce constructs and social support is to build trust in the ‘online’ context and consumers’ intention to buy. This suggests that businesses should monitor the quality and content of the engagements around their brands on social media, as information sharing in social commerce has a significant impact on consumer decisions, i


INTRODUCTION
Social commerce is a new category of e-commerce that has emerged due to the increased acceptance and use of social media (Al-tit et al., 2020). It is an internet-based commercial application leveraging Web 2.0 and social media technologies to influence consumer purchasing intentions, preferences, and decisions . Social commerce constructs are tools generated from social commerce such as ratings and reviews, recommendations and referrals, forums and communities, which have been responsible for altering consumer experiences and perceptions towards purchasing (Dashti et al., 2016;Hajli et al., 2017b). Through social commerce, consumers are steered away from ineffective individual decisions towards collective decision making with greater effectiveness (Dashti et al., 2016).
Facebook is the most popular social media platform among South Africans, with consumers aged 25 to 34 being the largest users (Kemp, 2019;Qwerty, 2017). These consumers are referred to as Millennials and are an attractive market segment due to their disposable income, technologically savvy, and open to new ways of communicating (Jacobsen & Barnes, 2017).
Only a few studies have looked at social media explicitly for its ability to influence purchase intention, and even fewer have looked at

LITERATURE REVIEW AND HYPOTHESES
Social commerce constructs are "tools derived from social commerce and include ratings and reviews, recommendations, referrals, forums and communities" (Dashti et al., 2016;. Social commerce constructs were defined by Al-Tit et al. (2020) as Web 2.0 platforms utilized in social media that allow consumers to generate and share their own content, including their opinions and experiences. Consumers are increasingly relying upon content generated by other consumers when making purchases (Chen, 2017). Dashti et al. (2016) stated that consumers highly value and depend upon social commerce constructs to make purchase decisions. Consumers utilize these social commerce constructs for the aforementioned social activities, resulting in a higher level of trust and intention to purchase (Hajli, 2015).
According to Hajli (2015cited in Lal, 2017, social interactions provided through social commerce increases the level of consumer trust and reduces perceived risks, which increases the likelihood of consumer purchasing. This is consistent with several researchers who have argued that social commerce constructs are supporting and increasing trust, which has resulted in increased purchase intentions among consumers (Shanmugam et al., 2016(Shanmugam et al., , 2018; Al-tit et al., 2020).
It has been asserted by several researchers that consumer trust is one of the most significant aspects in determining purchase intention in an 'online' environment (Hajli et al., 2017a;Che et al., 2017;Manzoor et al., 2020). Consumers regard social commerce platforms as being more trustworthy than traditional promotional features (Schivinski & Dabroski, 2016), since they allow for direct networking, immediate feedback, and more authentic material provided by consumers through social commerce frameworks (Manzoor et al., 2020). Through interactions provided by the social commerce constructs, consumers are enabled to produce and share content, offer advice and utilise other consumers' information which is providing 'online' social support. Hence, social commerce constructs have a significant impact on social support, which ultimately builds consumers trust, thereby impacting purchase intentions (Shanmugam, 2016;Hidayatulloh, 2018).
Ratings are numerical assessments of the calibre of products and services (Hajli et al., 2017a), whereas reviews are product evaluations written by peers and posted on a business's website or a third-party platform (Nambisan, 2002 cited in Hajli et al., 2017a). Ratings and reviews have shaped social commerce information sharing, providing a wealth of information about products and services to other potential consumers. Research conducted by the Spiegel Centre of Research (2017) found that 95% of consumers read 'online' reviews prior to making purchases, while 97% of consumers stated that reading other consumers reviews influenced their purchasing decisions.
In social commerce, referrals and recommendations play a critical role in increasing information exchange (Hajli et al., 2017b). Consumers are more reliant on the experiences of other consumers, such as referrals and product recommendations, in the 'online' setting because they cannot directly experience the goods or services (Seneal & Nantel, 2004 cited in Hajli et al., 2017b). This is in accordance with the findings of a Nielsen Believe Survey (2017), which found that 92 percent of consumers believe recommendations from friends, while just 33% trust 'online' marketing. Recommendations arise when individuals visit a product webpage, based on the assumption that the consumer is interested in a product and endorses it to others. Social media has allowed consumers to express their thoughts about products and services, as well as company procedures. This has increased the propensity and volume of social commerce, highlighted by the impact of socially influenced purchasing.
Forums and communities are two kinds of social platforms created by 'online' businesses to allow consumers to share and obtain information, as well as participate in group activities (Hajli et al., 2017b). Hettiarachci (2017) suggested that businesses should create 'online' forums and communities to enhance communication channels with consumers, allowing them to participate in group discussions and share information about products and services.
The emergence of social networking sites has been responsible for revolutionising channels of information exchange, and within the context of social commerce, users frequently utilise social commerce constructs in their information-seeking and exchange processes ( To further explore the relationships alluded to above, several hypotheses have been postulated with respect to the sample of Millennials in South Africa: H1: There is a significant positive relationship between ratings and reviews (as elements of social commerce), on the purchase intentions of Millennials who use Facebook.
H2: There is a significant positive relationship between referrals and recommendations (as elements of social commerce), on the purchase intentions of Millennials who use Facebook.
H3: There is a significant positive relationship between forums and communities (as elements of social commerce), on the purchase intentions of Millennials who use Facebook.
H4: Trust mediates the relationship between ratings and reviews and the purchase intentions of Millennials when using Facebook.
H5: Trust mediates the relationship between referrals and recommendations and the purchase intentions of Millennials when using Facebook.
H6: Trust mediates the relationship between forums and communities and the purchase intentions of Millennials when using Facebook.
H7: There is a significant positive relationship between trust in the social networking site and the purchase intentions of Millennials when using Facebook.
H8: Structural capital positively influences the ratings and reviews by Millennials on Facebook.
H9: Relational capital positively influences the referrals and recommendations by Millennials on Facebook.
H10: Cognitive capital positively influences the forums and communities by Millennials on Facebook.

RESEARCH METHODOLOGY
A quantitative research approach was used, and data were collected via an 'online' survey, guided by its advantages (Hair et al., 2003). The target population was young adults between the ages of 25 to 34 years residing in KwaZulu-Natal, who are referred to as Millennials and comprise the largest group of users on Facebook (Jocobsen & Barnes, 2017). Since a greater percentage of Millennials in KwaZulu-Natal will have disposable income compared to those in other provinces (Stone, 2018), KwaZulu-Natal was selected as the location of the study as income impacts purchase intentions (Imelia & Ruswanti, 2017).
Non-probability sampling, particularly snowball sampling, where the sample increases as the study proceeds (Pandey & Pandey, 2015), was adopted in this study. The initial participant was selected based on convenience, and this participant was requested to recommend individuals within similar demographics and 'online' purchasing behaviour. Hence, through social networks, the researcher was able to reach participants who would otherwise be challenging to recognise. In addition to its effectiveness, snowball sampling is also efficient, as it assists the researcher in finding the right persons with a minimum amount of time, money, and effort (Cohen & Arieli, 2011).
An 'online' calculator (Raosoft, 2004) was utilized to ascertain the study's sample size based on Krejice and Morgan's recommendations (1970). The primary data collection instrument was a structured questionnaire. An 'online' survey was selected as a data collection mechanism since it requires less time and effort to manage; it is cost-effective and has minimal risk for bias. The data was analyzed primarily by using descriptive and inferential statistical techniques.
The social commerce constructs were measured using questions based on a five-point Likert scale with responses ranging from 'strongly agree' to 'strongly disagree'. The questions attempted to understand why a respondent made purchases using information obtained from social commerce constructs. Responses to these items were treated as interval data (Sekaran & Bougie, 2016).

RESULTS
Although 386 respondents were targeted, 309 responses met the conditions for inclusion in the data analysis. This translated to an 80% response rate, which is generally satisfactory for web administered questionnaires (Saunders & Lewis, 2016).
Respondents in the 29-32-year category were in the majority (39.5%), followed by those who were between the ages of 24 and 28 (30.4%) and 33 and 35 (30.1%). The majority (54%) were female respondents. Figure 1 shows that a significant number of respondents had been on Facebook for between 3 -9 hours in the month prior to participating in the survey, accounting for 37.5% of the sample [p < .001]. There was an equal representation of respondents who spent less than 3 hours on Facebook (23.3%) and those who spent between 9 and 14 hours (24.9%) [p < .001]. A smaller proportion of the sample spent more than 14 hours of their time on Facebook in the previous month (14.2%) [p < .001].
Confirmatory factor analysis was applied to the measurement portion of the model to measure reliability and validity and to test the fitness of the model to the data. The factor loadings are reported in Table 1 through Table 4 for the social commerce constructs, social capital constructs, trust and purchase intentions, respectively. In addition, the results also report Composite Teliability (CR), Average Variance Extracted (AVE), and Maximum Shared Square Variance (MSV). Reliability is attained when CR > .7 and all loadings >.5. Convergent validity is attained when CR > AVE and AVE > .5. Discriminant validity is attained when AVE > MSV. The results in Table 5 through Table 10 indicate that reliability and convergent validity were attained for all the constructs; however, discriminant validity could not be achieved for social capital and trust. For social capital, the average variance extracted (AVE) value of 0.624 was less than the maximum shared square variance (MSV) values of 0.656 for trust, an AVE of 0.678 was reported, which is greater than the MSV of 0.712, confirming that discriminant validity could not be achieved for both social capital and trust.
Both the initial structural model and the modified model were evaluated against four criteria: the chi-square (χ2) likelihood ratio statistic, the Source: Primary data.

Measured variables Loadings
Ratings and Reviews q6.1 Using ratings and reviews has been a good experience for me.
0.646 q6.2 I recommend using ratings and reviews on Facebook.
0.746 q6.3 I use ratings and reviews for acquiring product information over Facebook.
0.793 q6.4 I browse through ratings and reviews before making any purchase decisions over Facebook.
0.775 q6.5 I use ratings and reviews to make purchases over Facebook.
0.763 q6.6 I contribute to ratings and reviews.

Measured variables Loadings
Structural Capital The one-sample chi-square test is used to evaluate whether any misfit observed between the patterns in the sample data and patterns in the population is likely to be due to sampling error. The value of the ratio of χ2 to df was 3.72 (p < .001) did not fit the desired criteria for the initial model. The IFI, CFI, and RMSEA values also did not meet the de-sired criteria, indicating that the initial model did not fit the data. However, improved values for the fit indices were observed with the modified model, as shown in Table 5, indicating that the modified model exhibits an improved fit than the first. It, therefore, follows that the modified measurement model was accepted.
The modified measurement model, which was accepted, was converted to a structural equation model represented by the path diagram shown in Figure 2. The structural equation model was also evaluated for fit against the same criteria used   for the measurement model yielding the results shown in Table 6. The results in Table 6 show that the model has an acceptable fit as all the values of the fit indices meet the criteria for model fit. Hence the structural model was accepted.  Table 7.

Measured variables Loadings
As can be seen from Table 7, all the composite variables had Cronbach's alpha values greater than 0.7, indicating that all of them were reliable. Hypotheses H4 to H6 for the mediating factors were tested separately, and the results are summarized in Table 9. For each of these, the effect of the IV (SCC) on the DV (PI) was reduced when the mediator (TRUST) was included in the model. This means that there is partial mediation in each case. Based on the t-test of significance, hypotheses H4 through H6 are not rejected.

DISCUSSION
The findings of this study confirmed a positive relationship between ratings and reviews and the purchase intentions of Millennials. These findings are consistent with prior research conducted within the discipline of social commerce (Lal,  ence of reviews on purchase intentions is explained in terms of the value of user-generated content to communicate effective information about products and services for the benefit of other potential consumers (Hajli, 2015). On the other hand, ratings immediately communicate the perceived value of the standard of a product or service (Hettiarachci, 2017). The influence of ratings and reviews has been recognised by some brands which are open to requesting consumers to share their reviews about their products and services on social media, to leverage the role of existing consumers acting as active communicators and content creators who create, engage and contribute to product and service ratings and reviews (Onete et al., 2017). Increased engagement with ratings and reviews is recognised as being responsible for communicating more social cues to other actors on social networks to support their purchase intentions (Hajli et al., 2017b).
It was also ascertained that there is a positive relationship between referrals and recommendations  The study also confirmed that trust mediates the relationship between ratings and reviews and consumer purchase intentions. These findings imply that as consumers engage with the ratings and reviews that other consumers post on Facebook, they trust these ratings and reviews as they reflect the experiences of other consumers like them. Consequently, the trust generated will have a positive influence on their purchase intentions.
Furthermore, the study confirmed that trust mediates the relationship between referrals and rec-ommendations on consumer purchase intentions. This implies that referrals and recommendations on Facebook provide social cues that appear as endorsements, therefore positively impacting trust in the recommended products and services. The trust generated will consequently have a positive influence on consumer purchase intentions in social commerce.
Additionally, the results of the study confirmed that trust mediated the relationship between forums and communities and consumer purchase intentions. The structure provided by forums and communities supports systems of common meanings among members of the 'online' communities. The knowledge sharing that follows generates trust in the individuals sharing the information and provides informational support that allows consumers to trust their purchase decisions. These findings show that the three dimensions of social commerce constructs considered in this study influence consumer trust, and consumer trust, in turn, influences consumers' purchase decisions.
The research findings also confirmed that structural capital positively influences consumers' information and knowledge sharing behavior of ratings and reviews. These findings are in line with those of earlier studies such as Ghahtarani et al. (2020), who concluded that interactions are an important aspect of structural capital. These researchers argued that social interaction through social commerce constructs is a collaborative way to access information resources that sees an increase in familiarity, creating a structural capital between the parties. Hence, relationships influence the willingness of customers to buy. Lee's (2017) distinction of the two dimensions of structural capital expands the concept firstly to referent ties which motivate the consumption of products and services used by social referents such as friends and family. It follows then that reviews by referent ties are likely to be held at high value by potential customers due to the intimate social relationships they possess. Secondly, majority ties motivate decisions that are supported by the pursuit of a sense of belonging to the mainstream, which can be achieved by consuming products and services that are enjoyed by the majority and getting involved in engagements and discussions taking place in the mainstream (Lee, 2017 Ghahtarani et al. (2020) consider these social motives as assets created and applied through social relationships, making it plausible to conclude that relational capital can positively affect consumer information and knowledge sharing behavior when making referrals and passing recommendations due to the perceived benefit of the potential assets that can be drawn from such interactions.
It was postulated that cognitive capital positively influences consumer information and knowledge sharing behavior on forums and communities. Forums and communities are social commerce constructs that allow consumers to share information and perspectives, create content and engaging with others. Cognitive capital, therefore, results in a structure that supports systems of common meanings among groups, incentivising knowledge sharing over forums and communities, since individuals belonging to those groups share a common understanding among each other and share collective goals (Tang et al., 2016;Ghahtarani et al., 2020, p. 196). As such, engagement in the group activities of the forums and communities will work towards achieving the group's common goals.

CONCLUSION
This study aimed to examine the impact of social commerce on the purchase intentions of Millennials who use Facebook by exploring how social commerce constructs influence consumer trust. A quantitative research approach was used, and data was collected via an 'online' survey. The study sheds light on the role of social commerce constructs in the purchase intentions of Millenials using Facebook by exploring the relationships between the Social Network Theory, Social Cognitive Theory, and Social Capital theory in social commerce.
It was ascertained that, in contrast to traditional physical store environments, social commerce allows more open links amongst consumers as they can engage amongst themselves and share information through user-generated content. The implications for social commerce are that the information shared over social commerce constructs in more cohesive networks is likely to have a stronger influence on the consumers' decisions than in less cohesive networks. Therefore, this justifies recommending that businesses should monitor the quality and content of the engagements around their brands on social media as information sharing in social commerce has a significant impact on influencing consumers' decisions, especially purchase intentions. In addition, businesses should embrace social commerce constructs and encourage existing customers to post reviews, recommendations, ratings and refer other consumers. User-generated content can help establish reciprocal interconnections with and amongst consumers in cohesive networks that can strongly influence customer behaviour and positively impact purchase intentions.