Chapter IV Data Analysis
4.4 Test of Hypotheses
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
68
A correlation analysis is conducted to examine the relationship between purchasing behavior under the influence of beauty blogs and three types of blog posts. From observing and comparing the Pearson Correlation score, the hypothesis can be verified or rejected.
Hypothesis 5a: The relationship between blog readers’ attitudes toward beauty bloggers posting sample trial articles and purchasing behavior will be positively associated with the level of STBP WOM credibility.
Hypothesis 5b: The relationship between blog readers’ attitudes toward beauty bloggers posting product placement articles and purchasing behavior will be positively associated with the level of PPBP WOM credibility.
In order to study if WOM credibility in both STBP and PPBP moderates the effect of blog readers’ attitude on purchasing behavior, regression analysis was applied to examine whether there was interaction effect. To test H5a, “attitude toward STBP” as independent variable and “STBP WOM credibility” as moderator variable are standardized and multiplied to get an interaction term. For H5b, “attitude toward PPBP” as independent variable and
“PPBP WOM credibility” as moderator variable are standardized and multiplied to get an interaction term. For both H4a and H4b, Moderated Multiple Regression with two main effects (Independent and Moderator variables) entered into the 1st block and the interaction term entered as the 2nd block.
4.4 Test of Hypotheses
As shown In Table 4-7, the correlation analysis suggests significant relationship between blog readers’ satisfaction toward beauty blog content and interest-arousing in products, services, or brands.
‧
Table 4-7. Correlation Analysis on Blog Readers’ Satisfaction toward Beauty Blog Content and Interest-arousing in Products, Services, or Brands
Interest-arousing in products, services or brands
Blog readers’ satisfaction toward beauty blog content
Pearson Correlation .399**
Sig. (2-tailed) .000
N 578
**. Correlation is significant at the 0.01 level (2-tailed).
*p <.05; **p <.01, ***P<.001
Observing the result from hierarchical regression analysis (see Table 4-8), it is found age (β = -.11, p<.05) and level of education (β = .11, p<.05) showed significant influence on interest-arousing while only demographics were input. After frequency of beauty blog
reading was added and showed great significance (β = .26, p<.001), only level of education remains its significance (β = .14, p<.05). This indicates a positive relationship that the more time one spent on reading beauty blogs, the more likely one would be interested in products, services, or brands. Furthermore, in model three, information collecting also shows strong significance (β = .39, p<.001), while frequency of beauty blog reading still remains great significance (β = .16, p<.001), and age (β = -.09, p<.05) as well as education (β = .10, p<.01) show significance. Participants with either younger age or higher education will spend more time on reading beauty blogs, and hence involve more in information collecting. Finally, content satisfaction also performs great significance (β = .19, p<.001) on readers’ interest-arousing in products, services, or brands., reflecting higher level of satisfaction along with more information collecting and higher frequency in blog reading will lead to more interest.
Notice the significance of education has increased as more variables were entered, suggesting positive relationships with frequency of blog reading, information collecting, content
satisfaction, and readers’ interest-arousing in products, services, or brands.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
70
Table 4-8. Hierarchical Regression Analysis on Blog Readers’ Satisfaction toward Beauty Blog Content and Interest-arousing in Products, Services, or Brands (Standardized regression coefficients)
Model 1 Model 2 Model 3 Model 4
Block 1: Demographics
Age -.11* -.10 -.09* -.09
Education .11* .14* .10** .11**
Monthly Income -.06 -.03 -.00 .01
Incremental R2 (%) 7.31
Block 2:
Frequency of beauty blog reading .26*** .16*** .16***
Incremental R2 (%) 24.12
Block 3:
Beauty-related Information
Collecting .39*** .27***
Incremental R2 (%) 60.15
Block 4:
Content Satisfaction .19***
Incremental R2 (%) 64.93
Note: N=578. Entries for all models are final standardized regression coefficients β
*p <.05; **p <.01, ***P<.001
From the results shown in Table 4-7 and 4-8, hypothesis 1 “Blog readers’ satisfaction toward beauty blog content is positively related to interest-arousing in products, services or brands” is accepted.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
71
The result of correlation analysis (Table 4-9) shows that blog readers’ satisfaction toward beauty bog content is significantly correlated with purchasing behavior under the influence of beauty blog WOM.
Table 4-9. Correlation Analysis on Blog Readers’ Satisfaction toward Beauty Blog Content and Purchasing Behavior
Purchasing Behavior Blog readers’ satisfaction toward
beauty blog content
Pearson Correlation .428**
Sig. (2-tailed) .000
N 578
**. Correlation is significant at the 0.01 level (2-tailed).
*p <.05; **p <.01, ***P<.001
The result of hierarchical regression is shown as Table 4-10, in which demographics did not show much significance, except monthly income that show higher significance in model 1 (β = -.14, p<.01) yet diminished after other variables were inserted. Frequency of beauty blog reading performed great significance on purchasing behavior (β = .27, p<.001);
its significance remained after more variables were input. For beauty-related information collecting, it is also significantly associated with purchasing behavior (β = .42, p<.001).
Content satisfaction also showed significance (β = .19, p<.05) on purchasing behavior, indicating higher level of content satisfaction along with more information collecting as well as higher frequency in blog reading will be more positively associated with purchasing behavior.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
72
Table 4-10. Hierarchical Regression on Blog Readers’ Satisfaction toward Beauty Blog Content and Purchasing Behavior (Standardized regression coefficients)
Model 1 Model 2 Model 3 Model 4
Block 1: Demographics
Age -.04 .-02 -.02 -.14
Education .06 .09* .05 .06
Monthly Income -.14** -.10* -.09 -.07
Incremental R2 (%) 10.42
Block 2:
Frequency of beauty blog reading .27*** .15*** .16***
Incremental R2 (%) 35.63
Block 3:
Beauty-related Information
Collecting .42*** .29***
Incremental R2 (%) 93.87
Block 4:
Content Satisfaction .19*
Incremental R2 (%) 101.15
Note: N=578. Entries for all models are final standardized regression coefficients β
*p <.05; **p <.01, ***P<.001
From the results shown in Table 4-9 and 4-10, hypothesis 2 “Blog readers’
satisfaction toward beauty blog content is positively related to purchasing behavior” is accepted.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
73
As shown in Table 4-11, blog readers’ satisfaction toward beauty blog content is significantly correlated with WOM Referral.
Table 4-11. Correlation Analysis on Blog Readers’ Satisfaction toward Beauty Blog Content and WOM Referral
WOM Referral Blog readers’ satisfaction toward
beauty blog content
Pearson Correlation .364**
Sig. (2-tailed) .000
N 584
**. Correlation is significant at the 0.01 level (2-tailed).
*p <.05; **p <.01, ***P<.001
In Table 4-12 (as shown below), demographics have no significant influence on the WOM Referral. Nevertheless, the frequency of beauty blog reading shows great significance (β = .31, p<.01) on WOM Referral, with its significance remained after more variables being input. This indicates a positive relationship that the more time one spent on reading beauty blogs, the more likely one would make WOM referrals to others. Furthermore, information collecting also shows strong significance (β = .38, p<.01), which shows higher involvement in information collecting will cause more WOM Referral. Finally, content satisfaction also performs significance (β = .13, p<.05), reflecting higher level of satisfaction along with more information collecting and higher frequency in blog reading will lead to more WOM Referral.
Table 4-12. Hierarchical Regression Analysis on Blog Readers’ Satisfaction toward Beauty Blog Content and WOM Referral (Standardized regression coefficients)
Model 1 Model 2 Model 3 Model 4
Block 1: Demographics
Age -.019 .003 .004 .005
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
74
Education -.034 .004 -.031 -.024
Monthly Income -.070 -.033 -.011 -.002
Incremental R2 (%) 2.38
Block 2:
Frequency of beauty blog reading .309*** .205*** .206***
Incremental R2 (%) 26.94
Block 3:
Beauty-related Information
Collecting .381*** .296***
Incremental R2 (%) 63.01
Block 4:
Content Satisfaction .127*
Incremental R2 (%) 65.35
Note: N=573. Entries for all models are final standardized regression coefficients β
*p <.05; **p <.01, ***P<.001
From the results shown in Table 4-11 and 4-12, hypothesis 3 “Blog readers’
satisfaction toward beauty blog content is positively related to the conduction of WOM Referral” is accepted.
As shown in Table 4-13, it is evident that WOM Credibility in both Pure Sharing Blog Post and Sample Trial Blog Post is significantly correlated with purchasing behavior, while WOM Credibility in Product Placement showed no significance. With the correlation coefficient higher than WOM Credibility in STBP, WOM Credibility in PSBP reflected higher correlation with the dependent variable—purchasing behavior under the influence of beauty blogs. Therefore, hypothesis 4 “The correlation between WOM Credibility of Pure
‧
Sharing Blog Post and purchasing behavior under the influence of beauty blogs will be higher than other two types of blog posts” is accepted.
Table 4-13. Correlation analysis on WOM Credibility and Purchasing Behavior.
WOM I have been influenced by
beauty blogs and performed purchasing behavior.
Pearson Correlation .364** .276** .059
Sig. (2-tailed) .000 .000 .162
N 578 576 568
**. Correlation is significant at the 0.01 level (2-tailed).
*p <.05; **p <.01, ***P<.001
From the regression analysis on interaction effect of attitudes toward Sample Trial Blog Post and purchasing behavior (Table 4-14), the main effects were examined significant in both model 1 (Attitudes toward SPBP: β=.14, P<0.01, WOM credibility in STBP: β=.22, P<0.001) and model 2 (Attitudes toward SPBP: β=.15, P<0.01, WOM credibility in STBP:
β=.23, P<0.001). The positive coefficients suggest positive relationships with the dependent variable; in other words, if one agrees more on STBP, one will be more likely to have
purchasing behavior under the influence of beauty blogs. As well, if one feels WOM in STBP more credible, one will tend to perform purchasing behavior.
Yet, no interaction effect was found, which indicates the level of WOM credibility in Sample Trial Blog Post does not moderate the effect of readers’ attitudes on their purchasing behavior. Hence, hypothesis 5a “The relationship between blog readers’ attitudes toward
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
76
beauty bloggers posting sample trial articles and purchasing behavior will be positively associated with the level of STBP WOM credibility” is rejected.
Table 4-14. Regression Analysis on Interaction Effect of Attitudes toward STBP and Purchasing Behavior (Standardized regression coefficients)
Model 1 Model 2
Block 1:
Attitudes toward Sample Trial Blog Post .14** .15**
WOM credibility in Sample Trial Blog Post .22*** .23***
Incremental R2 (%) 33.09
Block 2: Interaction
Interaction term .07
(Attitudes toward Sample Trial Blog Post x WOM credibility in Sample Trial Blog Post)
Incremental R2 (%) 34.80
N=591. Entries for all models are final standardized regression coefficients β
*p <.05; **p <.01, ***P<.001
As to Product Placement Blog Post, the result of regression analysis on interaction effect of attitudes toward PPBP and purchasing behavior H5b is shown in Table 4-15, in which the Interaction effect is observed. Before adding the interaction term, both attitudes toward Product Placement Blog Post (β=-.13, P<0.01) and WOM credibility in Product Placement Blog Post (β=.13, P<0.01) have significance. However, by adding interaction terms, not only the significance of attitudes toward Product Placement Blog Post (β=-.19, P<0.001) has enhanced but also the WOM credibility in Product Placement Blog Post (β=.19, P<0.001) on purchasing behavior shows higher significance. Meanwhile, interaction effect shows strong significance with β=.19, P<0.001, implying the level of WOM credibility in
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
77
Product Placement Blog Post significantly moderate the influence of readers’ attitudes on purchasing behavior. As the level of credibility in Product Placement Blog Post increases, the impact of readers’ attitudes towards beauty bloggers posting Product Placement Blog Post on purchasing behavior decreases.
Table 4-15. Regression Analysis on Interaction Effect of Attitudes toward PPBP and Purchasing Behavior (Standardized regression coefficients)
Model 1 Model 2
Block 1:
Attitudes toward Product Placement Blog Post -.13** -.19***
WOM credibility in Product Placement Blog Post .13** .19***
Incremental R2 (%) 5.67
Block 2: Interaction
Interaction term .19***
(Attitudes toward Product Placement Blog Post x WOM credibility in Product Placement Blog Post)
Incremental R2 (%) 17.47
N=582. Entries for all models are final standardized regression coefficients β
*p <.05; **p <.01, ***P<.001
From the interaction plot shown as Figure 4-1, it is indicated that readers holding lower perception of credibility in Product Placement Blog Post will be more influenced by attitude toward the bloggers’ posting product placement articles when purchasing.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
78
Figure 4-2. Plotting Interaction Effect (Scale ranges only partially displayed on Y-axis)
Concluding from the results of regression analysis on interaction effect shown in Table 4-15 and the interaction plotting shown in Figure 4-1, there do show interaction effect, indicating the level of PPBP WOM does moderating the relationship between blog readers’
attitudes toward beauty bloggers posting product placement articles and purchasing behavior.
Nevertheless, instead of positively influencing the relationship, a negative result is observed.
In conclusion, hypothesis 5b “The relationship between blog readers’ attitudes toward beauty bloggers posting product placement articles and purchasing behavior will be positively associated with the level of PPBP WOM credibility” is as well rejected.