• No results found

In order to test the effects of the six motivations on the intention to use scooter

sharing, a hierarchical multiple regression was performed. In the first model of the regression, only the significant covariates were included to test their effect on the dependent variable.

Out of all of the recorded demographics of the survey, only the statistically significant

nominal variables were chosen. The heightened attention given to the nominal variables came from the purpose of preserving the linearity assumption, which is essential for regression analysis. Therefore, the nominal variables which deemed important as covariates and had more than three categories were dummy coded for the regression. To assess the importance of the nominal variables Household_Structure and Profession in the intention to use scooter sharing, two separate One-Way ANOVA tests were conducted. In both cases, no significant effect was found between the groups of people for each of the variables (p > .05). The only significant difference was noticed for the Household_Structure group of people living in student dorms, compared to people living with their parents, in a Tukey post-hoc test (p <

.05). Even so, given the small sample of individuals living with their parents (N = 14), the findings were dismissed. Therefore, Household_Structure and Profession were excluded from the regression analysis. Controlling for covariates, the first model of the regression included the following demographic factors: gender, age, the existence of scooter sharing in their city [yes/no], level of education, monthly income, the possession of a driving license [yes/no] and previous usage of scooter sharing [yes/no]. In the final model, all of the

dependent (motivation) variables were included in order to test their effects on the intention to use scooter sharing. These motivations, as previously operationalized, fall into three categories: instrumental, social-hedonic, and normative.

For the scale items of the dependent variables, skewness and kurtosis were evaluated. Most of the scale items were normally distributed with a few exceptions of

skewness with values between -1 and 2 (Alt_Instrumental_7, Social_Hedonic_1). Given that the hierarchical multiple regression is not so heavily dependent on the assumption of normal distribution and no extreme cases were found in the scale items (Rencher & Christensen, 2012), the variables were not transformed before computing the scale variables. This decision was taken in order to preserve a certain variance between respondents, especially in the context of a relatively small sample size. Furthermore, to ensure the normal distribution of model residuals, standardized coefficients for the regression residual and regression predicted values were plotted, with no drastic deviations from the normal probability plot, values lying between -3.29 and 3.29 (Field, 2017). Additionally, the absence of multicollinearity was tested by calculating the VIF values, which lied within the normal range

(EconMot, Tolerance = .72, VIF = 1.38; PracMot, Tolerance = .46, VIF = 2.16; HedonicMot, Tolerance = .45, VIF = 2.20; SustMot, Tolerance = .63, VIF = 1.58; ConfMot, Tolerance = .54, VIF = 1.86).

In the first step of the hierarchical multiple regression, the predictors were statistically significant F (7, 187) = 16.69, p < .001, explaining 36.1% of the variance (adjusted R2 = .361) in the intention to use. In the final model, the total explained variance was 61.2%

(adjusted R2 = .612); F (12, 182) = 26.55, p < .001, with all the motivations being accounted for. The introduction of Economic, Practical, Hedonic, Sustainability and Conformity

motivations explained an additional 25.2% variance in the intention to use (R2 change = .252;

F (5, 182) = 25.22, p < .001). In the final model, five out of twelve predictor variables were statistically significant. The practical motivations scored the highest Standardized Beta value (β = .43, p < .001), implying that for each one increase in IntentionToUse, PracMot will increase by .43 standard deviations. Table 4.1 shows all the coefficients at the two steps of the Hierarchical Regression.

Table 4.1. Hierarchical Regression Model of Intention to use Scooter Sharing

R R2 R2 Change B SE β t

Step 1 .62 .39***

Gender .12 .18 .04 .67

Age -.02 .01 -.10 -1.59

CityHasScooter .25 .21 .07 1.17

EducationLvl -.09 .08 -.07 -1.16

MonthlyIncome .02 .03 .03 .56

DrivingLicense .40 .23 .11 1.73

UsedSharing 1.40 .19 .50*** 7.40

Step 2 .80 .64*** .25***

Gender .18 .14 .06 1.29

Age -.02 .01 -.08 -1.64

CityHasScooter .02 .17 .01 .11

EducationLvl -.05 .06 -.04 -.82

MonthlyIncome .03 .02 .06 1.32

DrivingLicense .60 .18 .16** 3.24

UsedSharing .69 .17 .25*** 4.11

EconMot .20 .05 .20*** 3.74

PracMot .42 .06 .43*** 6.49

HedonicMot .21 .08 .18** 2.74

SustMot -.12 .06 -.10 -1.80

ConfMot -.01 .06 -.01 -.18

Note.

*p < 0.05.

**p < 0.01.

***p < 0.001.

4.1 Instrumental Motivator

By running the Hierarchical Regression, each of the hypotheses of this study was gradually tested. The first Instrumental motivation, namely Economic Motivation, had a significant positive role (β = .20, p < .001) on the dependent variable, intention to use scooter sharing. Therefore, H1, “The users’ intention to use powered seated scooter sharing is

positively influenced by the perceived economic gain over traditional means of

transportation.” is supported. Besides the effect of Economic Motivation, the analysis shows that the perceived practicability of scooter sharing also plays a role in the intention to use.

The variable PracMot was found to have the most significant effect (β = .43, p < .001) on the dependent variable, intention to use scooter sharing. Therefore, H2 “The users’ intention to use powered seated scooter sharing is positively influenced by the perceived practicability of such service, namely travel time and availability.” is also supported.

Furthermore, to understand the importance of the scooter availability (as an element of the practical motivation) an additional question was added, calculating the importance of scooter proximity (“When using scooter sharing, it is always important to have a scooter close to my location.”). While this question wasn’t used in the regression analysis, it had many respondents that rated proximity as highly important (M = 5.99; SD = 1.27). The variance in proximity between different cities does not seem to be significant for the current sample, with a Levene’s test showing equal variances between groups, F (3, 191) = 4.81, p <

0.05. Even so, the proximity variable consists of only one element, most likely being very week in terms of internal validity

4.2 Social-Hedonic Motivator

In the construction of the Social-Hedonic Motivator, the theory was used for uniting the two concepts. The constructed scale for the perceived social advantages did not score sufficiently in the correlation analysis, resulting in a poor Chronbach's alpha value (α = .46).

Given the unreliability of the scale, the Social Motivation was not accounted for in the Hierarchical Regression, leaving H3 unanswered. The latter part of the motivator, Hedonic Motivation, was found to significantly positively influence the participants’ intention to use scooter sharing (β = .18, p < .01). This finding leads to a part of the Social-Hedonic

Motivator being effective, with H4 “The users’ intention to use powered seated scooter sharing is positively influenced by the perceived hedonic value of such service” confirmed.

Additionally, the 88 participants who never tried scooter sharing (M = 4.55; SD = 1.24) were compared to the 107 participants who did report using scooter sharing (M = 5.80; SD = 0.90).

By running an Independent Samples T-test, a significant difference in the scores of hedonism was found for the two groups, t (156) = -7.92, p < 0.001.

4.3 Normative Motivator

Out of the three motivator categories, the Normative Motivators are the only ones that did not get any significant contribution to the individuals’ intention to use scooter sharing.

While it was close to being significant, the Sustainability Motivation still has a statistically insignificant contribution to the dependent variable (β = .10, p > .005). Therefore, H5

“Sustainability concerns positively influence the users’ intention to use powered seated scooter sharing.” is rejected. The last variable, conformity motivation, scored the poorest out of the motivators, with a high insignificant coefficient (β = .01, p > .005). Hence, the last hypothesis, H6 “The users’ intention to use powered seated scooter sharing is positively influenced by their conformity level.” is also rejected.

Table 4.2. Summary of the hypotheses and their outcomes

H1 The users’ intention to use powered seated scooter sharing is

positively influenced by the perceived economic gain over traditional means of transportation.

Supported

H2 The users’ intention to use powered seated scooter sharing is

positively influenced by the perceived practicability of such service, namely travel time and availability.

Supported

H3 The users’ intention to use powered seated scooter sharing is positively influenced by the perceived social advantages.

Dismissed

H4 The users’ intention to use powered seated scooter sharing is

positively influenced by the perceived hedonic value of such service.

Supported

H5 Sustainability concerns positively influence the users’ intention to use powered seated scooter sharing.

Rejected

H6 The users’ intention to use powered seated scooter sharing is positively influenced by their conformity level.

Rejected

GERELATEERDE DOCUMENTEN