4. Research method
5.4 Hypothesis testing
To test the hypotheses, a hierarchical multiple regression analysis was performed to test the ability of recognition, reciprocal benefits, and network exposure to predict QS app usage continuation intention, after controlling for gender. The results of this analysis are presented in
variable (gender), while also adding the moderating variables recognition, reciprocal benefits, and network exposure.
Model 1 was not statistically significant, F(1,180)=0.22; p=.883 and explained 0% of variance in QS app usage continuation intention. Subsequently, the moderators were added in Model 2, increasing the total variance explained by the model to 20.2%, indicating that 20.2%
of the variance in QS app usage continuation intention is explained by the predictor variables, after controlling for gender (R2 Change=.202; F(3,177)=14.892; p<.001). In Model 2, one out of four predictor variables was statistically significant. Network exposure was statistically significant with p=.001 and =.298, indicating that when app users’ network exposure increases by one, their QS app usage continuation intention increases by .298.
Table 4
Hierarchical Regression Model of QS App Usage Continuation Intention
R R2 R2 Change B SE t
Model 1 .011 .000
Gender .032 .215 .011 .147
Model 2 .449 .202*** .202
Gender .063 .194 .022 .325
Recognition .158 .113 .159 1.399
Reciprocal benefits
.040 .105 .040 .382
Network exposure
.304 .092 .298** 3.297
Note. N = 182. Statistical significance: *p <.05 **p <.01 ***p<.001
Hereafter, a regression-based process analysis was conducted to test the effect of the independent variables recognition, reciprocal benefits, and network exposure on the dependent
variable QS app usage continuation intention, and furthermore, to perform simple moderation analyses measuring the effects of the personality types (Hayes, 2018). As the control variable gender had no significant effect in the hierarchical regression (Table 4), it was excluded from a further analysis. All independent and moderating variables of the model were mean-centered for the subsequent moderated regression analysis.
Table 5 shows the regression model measuring the direct effect of recognition on QS app usage continuation intention, under the moderation of openness to experience. The p-value of recognition is highly significant (p<.001), indicating that recognition has a positive effect on QS app usage continuation intention. Yet, the interaction effect between recognition and openness to experience is not significant (p>.05). Furthermore, as R2=.150, 15% of the change in QS app usage continuation intention is explained by this model. The model is generally significant, as p<.001.
Table 5
Regression Model of Recognition (IV) and QS App Usage Continuation Intention (DV) with Moderating Effect of Openness to Experience
Coefficient SE t p
Intercept i1 3.993 .103 38.926 .000
Recognition (X) c1 .377 .069 5.472 .000
Openness to experience (W) c2 .062 .108 .575 .566
Openness to
experience*recognition (XW)
c3 -.013 .059 -.219 .827
Note. N = 182. R2=.150 p < .001 F(3,178)=10.474
Table 6 summarizes the results of the regression model testing the direct relationship
moderating effect of agreeableness and conscientiousness. The model is significant with a p value of p<.001 and an R2 value of .159, implying that 16% of the change in the outcome variable is explained by this model. With a p-value of p<.001, reciprocal benefits as an independent variable has a significant effect on the outcome variable. The interaction effect between reciprocal benefits and agreeableness (XW) is non-significant (p>.05). Furthermore, the interaction effect between reciprocal benefits and conscientiousness (XZ) is significant (=-.164, p<.01), indicating that a moderation effect is taking place. This moderation effect is negative (=-.164), signifying that based on this dataset, perceived reciprocal benefits do not result in a higher QS app usage continuation intention for highly conscientious individuals, but contrarily, decrease the usage continuation intention for people with this personality trait.
Contrarily, for users with low levels of conscientiousness, perceived reciprocal benefits do lead to a higher QS app usage continuation intention. Table 9 in Appendix 7 confirms the significant negative moderating effect and indicates that the effect of the focal predictor depends on the value of the moderator.
Table 6
Regression Model of Reciprocal Benefits (IV) and QS App Usage Continuation Intention (DV) with Moderating Effect of Agreeableness and Conscientiousness
Coefficient SE t p
Intercept i1 3.984 .103 38.620 .000
Reciprocal benefits (X) c1 .332 .070 4.769 .000
Agreeableness (W) c2 .052 .111 .464 .644
Reciprocal benefits*agreeableness (XW) c3 .041 .057 .730 .466
Conscientiousness (Z) c4 .124 .092 1.343 .181
Reciprocal benefits*conscientiousness (XZ) c5 -.164 .060 -2.753 .007
Note. N = 182. R2=.159 p < .001 F(5,176)=6.677
Lastly, the direct relationship between network exposure and QS app usage continuation under moderation of extraversion and neuroticism has been tested. Overall, this model is significant with p<.001. As R2 =.203, 20% of the change in QS app usage continuation intention is explained by this model. Network exposure is the only significant variable of this model, with p<.001. Neither the interaction effect between network exposure and extraversion, nor the interaction effect between network exposure and neuroticism is significant (p>.05). The moderation interaction graphs for the five moderators can be found in Appendix 8.
Table 7
Regression Model of Network Exposure (IV) and QS App Usage Continuation Intention (DV) with Moderating Effect of Extraversion and Neuroticism
Coefficient SE t p
Intercept i1 4.003 .100 39.879 .000
Network exposure (X) c1 .404 .071 5.687 .000
Extraversion (W) c2 .150 .077 1.938 .054
Network exposure*extraversion (XW) c3 -.054 .050 -1.078 .283
Neuroticism (Z) c4 -.006 .097 -.062 .951
Network exposure*neuroticism (XZ) c5 .003 .074 .034 .973
Note. N = 182. R2=.203 p < .001 F(5,176)=8.990
5.4.1 Hypotheses H1 and H1a
H1 There is a positive relationship between recognition and QS app usage continuation intention.
Table 5 indicates that recognition has a significant positive effect on QS app usage continuation intention (rec=.337, p<.001). Thus, this hypothesis is accepted.
H1a The relationship between recognition and QS app usage continuation intention is positively moderated by openness to experience.
Table 5 indicates that a moderating effect of the personality trait openness to experience on the relationship between recognition and QS app usage continuation intention is not taking place (rec x open=-.013, p>.05) Therefore, this hypothesis is rejected.
5.4.2 Hypotheses H2, H2a, and H2b
H2 There is a positive relationship between perceived reciprocal benefits and QS app usage continuation intention.
The results in Table 6 indicate that there is indeed a significant positive relationship between reciprocal benefits and the dependent variable (reci=.332, p<.001). The hypothesis is therefore accepted.
H2a The relationship between perceived reciprocal benefits and QS app usage continuation intention is positively moderated by agreeableness.
H2b The relationship between perceived reciprocal benefits and QS app usage continuation intention is positively moderated by conscientiousness.
Table 6 shows a significant negative moderating effect of conscientiousness on the relationship between reciprocal benefits and QS app usage continuation intention (reci x cons =-.164, p<.01). As a positive moderating effect of conscientiousness on this relationship was hypothesized, H2b is rejected. With regards to the moderating effect of agreeableness, no significant relationship has been detected based on the results of the regression analysis (reci x agre=.041, p>.05). Thus, both hypotheses are rejected.
5.4.3 Hypotheses H3, H3a, and H3b
H3 There is a positive relationship between network exposure and QS app usage continuation intention.
Table 7 shows a significant positive relationship between network exposure and QS app usage continuation intention (net=.404, p<.001). Hypothesis 3 is therefore accepted.
H3a The relationship between network exposure and QS app usage continuation intention is positively moderated by extraversion.
H3b The relationship between network exposure and QS app usage continuation intention is positively moderated by neuroticism.
As indicated in Table 7, neither extraversion, nor neuroticism has a significant moderating effect on the relationship between network exposure and QS app usage continuation intention (net x extr=-.054, p>.05; net x neur=.003, p>.05). Thus, both hypotheses are rejected.