英语论文网

he variance inflation factor (VIF) and tolerance values. There is no high correlation between the independent variables in the regression as all condition indices and VIF fall bellow the threshold values.

Lastly, there was a check for outliers (i.e cases falling at the outer ranges of the distribution). A Threshold of 3 standard deviations was used, which is appropriate for our sample size of 88 to identify outliers. All observations outside this range (3o+) were considered outliers and were duly dropped from the regression.

The coefficient of determination (R2) for the regression is .215 and .306 for perceived usefulness and perceives ease of use respectively, indicating that 21.5% and 30.6% of the variation in perceived usefulness and ease of use are explained by system risk. The p-value of .000 for both usefulness and ease of use indicates that the higher the system risk in tourism organization towards e-commerce, the less e-commerce is perceived to be useful and the less to be easy to use. Both hypotheses H1 and H2 are then accepted.

The coefficient of determination (R2) for this regression is .138, indicating that 13.8% of the variation in intention to transact is explained system risk. The p-value of .000 indicates that the higher the system risk in tourism organization towards e-commerce, the less the organization intention to transact online. Hypothesis 3 is accepted.

The relationship between system risk on one hand and intention to use e-commerce on the other were mediated by perceived usefulness and ease of use. The hierarchical multiple regression model was employed to analyze the indirect relationships hypothesized (Abrams, 1999). The predictor variables (i.e. the independent variables and the intervening variables) were entered into the model in different stages. The hierarchical regression is employed so that the increase in R2 corresponding to the inclusion of each category of predictor variables and the inclusion of each category of predictor variables and the unique variance in e-commerce adoption explained by the predictor categorized could be examined. The R2 for all sets can be analyzed into increments in the proportion of Y variance due to addition of each new set of predictor variables to those higher in the hierarchy. These increments in R2 are squared multiple-partial correlation coefficients. The following general hierarchical model equation for four sets in alphabetical hierarchical order was adopted from Cohen and Cohen (1983):

A two model hierarchical regression is adopted in these analyses as suggested above to investigate the mediating effect of perceived usefulness and perceived ease of use between system risk in one hand, and intention to use e-commerce in the other hand, resulting to test our hypotheses H4 and H5.

As we can see from the table above, the beta coefficient for model 1 is significantly higher than of the model 2 in for both variables (i.e. perceived usefulness and perceived ease of use). Coupled also with the increase in R2 in both variables explain the mediating effect of usefulness and ease of use on the relationship between system risk and intention to use e-commerce. There is therefore an indirect negative relationsh本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。