• No results found

Determinants of private car ownership and choice as work travel mode: a binomial logistic

5.6 Travel mode preferences and choice considerations

5.6.1 Determinants of private car ownership and choice as work travel mode: a binomial logistic

The private-car was the main work travel mode among 21% of workers interviewed. This section combines the choice considerations discussed above with other socio-economic and spatial variables to examine how these factors determine car ownership and choice as work travel mode using a binary logistic regression model. The analysis of private car use as work travel mode also reflect car ownership among households of the workers. In view of this, variables representing attributes of the worker’s households including marital status, family size and income are included in the analysis. A summary of the variables in used in the analysis is provided in table 5.4.

Table 5.4: Variables in the binary logistic regression of private car use (N= 1158)

Variable name Type and coding

Dependent variable

Private car use Categorical, coded 1-Yes 0-No

Independent variables

Income-group18 Categorical, coded: 1- low (earnings below 25th percentile income;

2- middle (earnings between the 26th percentile and 75th percentile

income; and 3- high (earnings above 75thth percentile)

Education Categorical coded: 1-Tertiary; 0-basic & secondary Marital status Categorical, coded: 1-couple; 0-Single

Family size Scale variable— Nominal values Distance of

residence to CBD

Scale variable—of actual road distances from the home to the CBD.

Residential zone of residence

Categorical coded: 1-Historical-core; 2-Inner-Suburb; and 3-Outer- suburb

Average home-work distance

Scale variable—of actual road distances from the home to the work place averaged for all working members in the household

Predictable Categorical, code 1-important, 0-not-important Affordable Categorical, code 1-important, 0-not-important Comfort Categorical, code 1-important, 0-not-important Privacy Categorical, code 1-important, 0-not-important Flexibility Categorical, code 1-important, 0-not-important

Quicker travel Categorical, code 1-important, 0-not-important

The average distance from the place of residence to the CBD was 7km (SD = 2.85) while that of the home to the place of work was 4.5km (SD = 2.99). A hierarchical logistic regression model was specified in which the predictors were entered in a systematic manner in the order presented in table 5.4. Five models were initially formulated using this approach. The extent to which the addition of each predictor improved the model fit was assessed using the Omnibus Tests of Model Co-efficient19. Based on the contribution of each predictor variable to the improvement of the model, a final model was specified to explain the determinants of private car use as work travel mode. Estimates of model co-efficient are presented in table 5.5.

Table 5.5: Determinants of private-car ownership and choice as work travel mode

Predictors b(SE)

95% C.I. for EXP(B) Lower Odds Ratio Upper Low-income 0b Middle-income 2.038 (0.61) ** 2.32 7.673 25.384 High Income 3.368(0.618) *** 8.649 29.029 97.434 Education of household-head 1.491(0.187) *** 3.08 4.441 6.404 Marital status -0.158(0.225) 0.549 0.854 1.327 Family size 0.181(0.054) ** 1.079 1.199 1.332 Distance of home to CBD -0.110(0.048) * 0.815 0.895 0.983 Historical-core residence 0b Inner-suburb residence 0.7(0.303) * 1.111 2.014 3.652

Outer suburb residence 1.245 (0.369) ** 1.686 3.473 7.155

Predictable -0.456 (0.402) 0.288 0.634 1.392 Affordable -0.542 (0.253) ** 0.354 0.582 0.956 Comfort 0.265 (0.426) 0.565 1.304 3.007 Privacy 1.568 (0.255) *** 2.911 4.797 7.906 Flexibility -0.867 (0.39) * 0.196 0.42 0.904 Quicker travel 1.385 (0.426) ** 1.735 3.996 9.203 Constant -5.677 (0.743) *** 0.003

Note: R2 = 0.49 (Nagelkerke), 0.33 (Cox & Snell) *p < 0.05, **p < 0.01 ***p < 0.001 bThis parameter is set to zero because it is redundant

Results of the binary logistic regression showed that all the predictor variables included in the model, except marital status and the level of importance attached to comfort, had statistically significant effect on the choice of private-car as work travel mode. The likelihood of car ownership and use increases with higher household income. High income households (earning monthly income of GH¢ 2,050 and above) have odds of owning and commuting to work on a private car, 29 times higher than low income households. In the case of middle-income

19 The initial hierarchical model specification included two models in which interaction terms were specified

between (i) distance of residence to CBD and Urban-zone of residence; and (ii) marital status and family size. These did not yield improvement in model fit and were therefore excluded from the final model specification.

households (i.e. earning incomes between GH¢50 and GH¢2000), the odds are relatively smaller, although nearly eight times higher compared to low-income households. Moreover, being tertiary-educated was associated with odds of owning and commuting in a private car to work four times higher than those with low-levels of education controlling for other factors. Households with relatively larger relatively larger family size would own a private car if they could afford it. Consequently, individual workers within such households have odds of commuting to work in a private car, 1.2 times higher than those with smaller family sizes controlling for other factors.

The analysis further shows that as distance between the place of residence and the CBD decreases, the odds of a worker’s household owning a car decreases. Consequently, workers living in the suburban neighbourhoods of the metropolis were more likely to commute to work in a private car than those in the historical-core neighbourhoods. Specifically, worker’s having residence in the inner and outer suburban locations of the metropolis have odds of commuting to work by private car two and 3.5 times higher than those residing in the historical-core neighbourhoods.

The final part of the analysis show that the importance worker’s attach to affordability, privacy, flexibility and the length of time spent in commuting to work had significant effect on private car choice. Whereas greater importance to affordability reduced the likelihood of car- ownership and hence the likelihood of individuals traveling by this mode, workers who attached greater importance to privacy were nearly five times more likely to choose the private, controlling for other factors. Similarly, workers who attached importance to flexibility in travel have odds of choosing to commute to work in a private car four times higher than those who did not.

In summary, the analyses show that socio-economic characteristics including income levels, educational attainment and family size; spatial attributes including distance between the place of residence and the CBD, urban-zone of residence (i.e. historical-core or suburban); as well as the level of importance attached to affordability, privacy, flexibility and time spent travelling determine private car ownership and choice as work travel mode. Together, these variables account for nearly 50% of the explained variance in private car ownership and use as work travel mode.

5.6.2 Determinants of choice between walking and motorized transport as work