4.1.3.1 Brand and service trust (TRU)
Among 5 items of Brand and service trust dimension that mean value are between 3.06 and 4.03. Besides, this value is not so different among TRU1-prestigious enterprises, TRU2-famous author and TRU4-copyright from 3.654 of the average mean value. Three items are appreciated in user’s mind. The mean value of TRU3-high quality is on somewhat “neutral” (3.06), it means that users are not interesting in this factor. The standard deviation of TRU1, TRU2 and TRU4 are in the interval of 0.809 and 0.843 which demonstrates that respondents have similar perception about the items. However, standard deviation of TRU3 is rather high with 1.019 which reflects that the respondents’ opinions …show more content…
While Corrected Item-Total Correlation of TRU1, TRU2 and TRU4 are between 0.571 and 0.635; which meet the standard of greater than 0.5. In addition, Cronbach's alpha if Item Deleted of these items also satisfy standard of less than overall Cronbach's alpha (0.689). On the contrary, the Corrected Item-Total Correlation of TRU3 is relative low (0.178) which indicates it do not measure the same thing the rest TRU. It might lead us to remove this item. Actually, removal of TRU3 leads a significant improvement in Cronbach's alpha which increases the overall total from 0.689 to 0.827. It appears to be useful and contributes to the overall reliability of TRU. Hence, after screening, the measurement items in TRU are still maintained from 4 to 3 items: TRUi (i=1, 2, 4) and the correlation among these items is rather high (0.676; 0.715 and 0.664, respectively). It indicates that there is a positive correlation between one item and the rest of TRU (see in Table …show more content…
The overall reliability coefficient of each dimension is between 0.827 and 0.901, which is very high and indicates strong internal consistency among items of each dimension. Essentially, this means that respondents who tended to select high or low scores for one item also tended to select high or low scores for the others (as in Descriptive analysis all items get high scores). Therefore, knowing the score for one item would enable to predict with some accuracy the possible scores for the other items. Furthermore, the item-to-total correlation coefficient of the appropriate items after screening in each dimension was between 0.627 and 0.840, which also met the standard of greater than 0.5. These results indicate that the reliability of all dimensions attained an excellent degree of consistency. Consequently, we have screened from 32 to 24 items of independent variables and 4 to 3 items of dependent