No significant differences was indeed discovered to possess availableness, amount of usage, sign in frequency, intercourse and you will years, whereas Bonferroni modification pairwise evaluations of estimated marginal setting unveiled brand new significant difference amongst the next parameters
Facebook are utilized from the a majority of participants (64.5 per cent) via both Desktop computer and you can smart phone; over sixty per cent signed on to Facebook relaxed; spent you to definitely (58.5 percent) to a few (22.8 per cent) era for every join; and more than 72 per cent up-to-date the reputation no less than out-of once a week. The brand new try incorporated a small almost all lady (54.8 per cent); and also the cultural teams precisely illustrated brand new ethnicity of your own West Cape, and additionally mostly black colored (thirty-five.dos percent) and you may coloured (36.4 %) ethnic communities (Statistics SA, 2012).
Dining table II also offers a complete breakdown of the usage of features and you can demographics regarding Millennials respondents which use Myspace
As stated, the new respondents’ behavioral emotions to your Twitter advertising are determined by the 9-goods bills for each of the steps effect profile (Dining tables III and you can IV).
Cronbach’s ? was 0.843 for the Facebook advertising intention-to-purchase scale (Table III) and 0.742 for the Facebook advertising purchase scale (Table IV), which indicated good internal consistencies. A non-parametric one-sample bi-nominal standardised test was utilised to determine if there was a significant difference. The test showed that for both of the nine-item scales, there was a significant difference at p < 0.001 and p < 0.05, with the exception of one item in the intention-to-purchase scale. Pearson’s correlation coefficient analysis (Tables III and IV) showed a positive medium (r > 0.3) to strong (r > 0.5) relationship between a majority of the variables for the intention-to-purchase and purchase measurement scales, but there was weak positive correlation between a minority of the variables, especially in terms of the negatively reversed variables that were recoded.
The new GLM ANOVA, because chatted about during the previous text message, was applied due to the fact data contains yet another number of findings needless to say independent variables, and that is seen from the large standard errors (an example of here is the reasonable number of respondents you to signed onto Twitter at least once thirty day period). Van Schalkwyk blued inloggen (2012) discloses the GLM requires that it under consideration and you can “normalises” the results. Tables V and you can VI reveal the outcome when it comes to Wald ?2 shot, which is in line with the Bonferroni modification pairwise post hoc test among projected marginal form.
The Wald ?2 test revealed that there was a significant difference at p < 0.001 for intention-to-purchase (M=2.94, SD=0.805) because of Facebook advertising.
Log on duration (p < 0.001): respondents who logged on for ?1 hour (M=2.82, SE=0.033) resulted in lower intention-to-purchase levels in comparison to those who logged on for two hours (M=2.98, SE=0.039).
Profile update incidence (p < 0.001): respondents who updated their Facebook status daily (M=3.06, SE=0.041) resulted in greater intention-to-purchase compared to those who updated once a week (M=2.93, SE=0.044), two to four times a month (M=2.81, SE=0.050) and once a month (M=2.81, SE=0.042); those who updated their Facebook status two to four times a week (M=2.98, SE=0.043) showed an increase in intention-to-purchase compared to those who updated it two to four times a month (M=2.81, SE=0.050) and once a month (M=2.81, SE=0.042).
Ethnic group (p < 0.001): white respondents (M=2.79, SE=0.041) exhibited lower intention-to-purchase levels than black (M=3.01, SE=0.035) and coloured (M=2.96, SE=0.037) respondents.
The Wald ?2 test disclosed that there was a significant difference at p < 0.001 for purchase (M=2.94, SD=0.656), which was caused by Facebook advertising. No significant differences were found for access, length of usage, log on frequency, age, gender and race; however, Bonferroni correction pairwise comparisons of estimated ongst the following variables.