Passengers' perceptions and effects of bus-holding strategy using automatic vehicle location technology

Journal of Advanced Transportation, Fall, 2009 by Shinya Hanaoka, Fayyaz Mahmood Qadir

[FIGURE 1 OMITTED]

Analysis of Survey Responses with Respect to Features of an AVL System

Major findings from the attitudinal survey are to be discussed with respect to passengers' perceived waiting-time, car ownership, and length of route using period as the groupings analysis. The reasons for selecting these three factors out of socio-economic and trip information are discussed below:

* Passengers' perceived waiting-times at bus stops have a direct relation with existing bus operation and they would choose a satisfaction level based upon their level of anxiety;

* Car ownership might be a general factor useful to separate captive and choice passengers who may have different perceptions;

* Length of passengers' route using period describes satisfaction and perception based upon past experience of using this route which might lead the researchers to critical service attributes.

Tables 1 and 2 show the ranking of service attributes and perception of satisfaction level of passengers in different perceived waiting-time groups. Table 3 and Figure 2 show the ranking of service attributes and perception of satisfaction level according to passengers' car ownership status. Similarly, Tables 4 and 5 show the ranking of service attributes and perception of satisfaction level with respect to the length of time period during which passengers used this route. The statistical tests were applied to determine whether there is any similarity or difference among various groups of perceived waiting-time and length of route using period for different attributes. For this purpose, a nonparametric Kruskal-Wallis test was applied which tests the null hypothesis that multiple independent groups come from the same population. The reason why this test was selected because the collected data for passengers' satisfaction is ordinal, however, non-normal distribution and invalidity in mean as a valid estimate are properties of the ordinal data. Therefore, Kruskal-Wallis test measures the variances in ordinal data by rank.

For Table 2, it was found that all waiting-time groups were statistically significant ([chi square]=10.38, [alpha] =0.034, df= 4) for punctuality of arrival and departure. The result for real-time information of bus schedule showed that all waiting-time groups were statistically significant ([chi square]=8.93, [alpha] =0.063, df= 4) except between group 2 (mean rank value = 93.52) and group 3 (90.46) so now groups 2 and 3 of this attribute can be combined together into one group. Test results showed that there was no significant difference ([chi square]= 5.90, [alpha] =0.207, df = 4) among all waiting-time groups for safety in travel, while looking at mean rank values, it was decided to combine groups 1 (105.80) and 2 (103.07) together, and groups 3 (85.00) and 4 (84.95) together. Similarly, announcement of next stop was not significantly different ([chi square] = 1.243, [alpha] = 0.871, df = 4) so groups 1 (95.83), 2 (97.27) and 3 (98.79) can be combined together. Coordination of schedule with other buses was not significantly different ([chi square] = 7.70, [alpha] =0.103, df= 4) among all waiting-time groups, but there is significance among all other groups except groups 1 (95.50) and 2 (93.42) so these two can be combined into one group.