To examine our research model, we applied PLS (Partial least squares) as the method of analysis. In this research, we used SmartPLS to help us analyze R2 (R-squared) and Path Coefficient to validate our research hypothesis.
20
Using the PLS algorithm attained an explanation for the variance in advertisement performance (R2
=34%), which is qualified because an R2 value larger than 33% can be considered a medium explained ability. Moreover, using the Bootstrapping method to calculate the path coefficient, we found that there were several paths that reached significance levels (t > 1.96, p<0.05). The relations among each factors are detailed below.
Figure 4. Results of research model 1
Table 7. Results of research model 1
Hypothesis Path β P Values Result
H1A Vividness -> Like 0.057 0.167
H1B Vividness -> Comment 0.063 0.09 Support
H1C Vividness -> Share -0.069 0.088 Support
H2A Interactivity -> Like -0.054 0.115
H2B Interactivity -> Comment 0.011 0.677
H2C Interactivity -> Share 0.083 0.005 Support
H3A Richness -> Like 0.064 0.089 Support
H3B Richness -> Comment -0.097 0.022
H3C Richness -> Share -0.032 0.415
H4A Emotiveness -> Like 0.092 0.01 Support
H4B Emotiveness -> Comment 0.073 0.008 Support
H4C Emotiveness -> Share 0.006 0.856
H5A Like -> Link 0.414 0 Support
H5B Comment -> Link -0.039 0.759
H5C Share -> Link 0.336 0 Support
21
Figure 5. Results of research model 2
Table 8. Results of research model 2
Hypothesis Path β P Values Result
H6A Vividness -> Link -0.096 0.008
H6B Interactivity -> Link 0.034 0.346
H6C Richness -> Link -0.105 0
H6D Emotiveness -> Link 0.076 0.021 Support
4.2 Discussions and implications:
Based on the results, we obtained several findings. At first, we expected that when we used content with various elements, which is considered vivid, we could hold the customers’ attention and induce increased resonance and even customer engagement behavior. However, the analytic results do not show the consequence that we expected; thus, vividness is not a significant influence factor on on number of likes and shares, but just only does on number of comments. Ferran (2013) indicated that content with images receives a higher number of likes and comments than does content with videos or links; he thus asserted that images are easier than videos to invoke customers’ feeling and
opinions to react in few seconds. Hence, we fail to support H1A and H1C, but support H1B.
Although U&G theory asserts that one of the reasons why people use social media is to obtain
22
information, the results show that content with deep and rich information does positively affect only the number of likes (β=0.064, p<0.10). In addition, richness negatively affected the number of comments (β=-0.097, p<0.01); thus, we assume that content with rich information may effectively specify and automatically answer the question in customers’ mind, thereby reducing the intention of commenting.
Interactivity is not a significant factor regarding likes (β=-0.054, p>0.05) or comments (β=0.011, p>0.05). These results show that the intensity of interactivity does not positively affect the number of likes and comments. Cvijikj (2013) also supported this perspective and even negative
relationships, though we found a positive correlation among interactivity and the frequency of shares (β=0.083, p<0.01), thus supporting H3C. Because most brand fan pages would like to host an activity or contest to induce their users to interact with the brand and receive prizes for increasing exposure, sharing the post is considered part of the activity, and number of shares will thus increase throughout the activity.
We found that emotiveness positively affects the number of likes (β=0.092, p<0.01) and the number of comments (β=0.073, p<0.01), which supports H4A and H4B. Emotions are useful in computer-mediated communication because communication with emotions can enrich the communication itself and convey the sender’s emotions or feelings (Derks et al., 2008). Hence, we can understand that content with strong emotion would increase interactivity with users.
Regarding resonance, we can see that the numbers of likes (β=0.414, p<0.001) and shares (β=0.336, p<0.001) have a strongly positive correlation with customer engagement. The act of liking can be considered an intensity agreement and inclination, which is closely related to satisfaction. We know that when a customer is intensely satisfied, the possibility of a revisit or a repurchase increases;
hence, engaged customers are not only satisfied with the brand but plan to keep using it and following it in the future (Smith, 2013). In contrast, due to the attributes of social media, when a company receives likes and comments from customers, the posts will appear in the customer’s dynamic profile and be seen by their friends, which effectively increases the possibility of the post be seen and clicked (Williams et al., 2012) and results in the more direct sharing of the posts on their profile, thereby supporting H5A and H5C. However, the number of comments (β=-0.097, p >
0.05) is not a significant influence factor on customer engagement, so we fail to support H5B.
Finally, in exploring the relationship between dimensions and customer engagement, we have some important findings. First, the results showed that vividness negatively impacted customer
engagement (β=-0.096, p <0.001); this was not our expectation, so we assume that a possible cause of the negative relationship may be the use of too many elements in a limited space, which would annoy and interfere with the customer reading information, thereby and reducing their intention. As we know from collecting data from business magazine brands, having so much content with rich
23
information related to products may be a reason for the reduction in the customers’ intention to click a link because studies have shown that customers generally have negative attitudes toward
advertising. However, emotiveness can positively affect customer engagement (β=0.076, p <0.01), thereby supporting H6D. For instance, using the interrogative method can increase the customer’s curiosity, and the imperative method can attract the customer’s attention, so these are both effective methods for inducing the customer to go to a particular website.