Determinants for Bus Stop Performance in Penang Island

Less dependency on public transport has caused a high reliance on private vehicles. This is due to several factors including a comfortable aspect, safety, and inconvenient infrastructure and facilities at the bus stop. This paper attempts to identify the significant determinants affecting the bus stop performance by evaluating 123 bus stops based on the developed indicators within maximum walking distance to transit stations along the proposed alignment of Light Rail Transit (LRT) stations. The indicators were adapted from Transport Assessment Indicators and suited with local guidelines which were subsequently developed into Transport Assessment Checklist. 13 indicators in the audit checklist were significantly reduced to five factors containing 12 significant indicators (factor loadings greater than 0.5) through Factor Analysis and Principal Components Analysis (PCA). These indicators were identified to give an effect on bus stop performance. The findings can help solve the low dependency on public transport by emphasizing the significant determinants at the bus stop for future improvement.


Introduction
The rapid development and technology in these recent years have caused people to be highly dependent on private automobiles especially cars [1]. The alarming rate of private vehicles triggers concern for its adverse consequences in terms of congestion and pollution [2][3]. The Malaysia Automotive Association (MAA) has reported that up to June 2017, the total number of vehicles exceeds 28 million with more than 13 million registered cars and 12 million and more for motorcycles [4]. Even though demand in cars is positively contributing to automobile development and industry, it depletes the environmental health and sustainability by causing air pollution due to various carbon remittances [5].
According to the recent report from the Malaysian Automotive Association [6], in 2017, Penang was reported to have 1,126,618 registered private motor vehicles. Despite its small area, Penang owned more vehicles compared to other countries. Furthermore, Penang's increasing population rate also contributed to the increasing demand for private vehicle ownership. Adding to that, the average daily traffic (ADT) was reported high with more than 60 000 vehicles per day exceeding ADT in other countries [7]. The increment of private vehicle usage was claimed to be associated with poor public transport systems [8] which cause people to shift to private vehicles. Rohani et al [9] listed the external and internal factors affecting bus ridership which includes design of the route, service schedules, and frequency, reliability, accessibility, and parking availability. Furthermore, Bernal [10] claimed that integration with other types of user facilities will also give a comfortable travel experience. Chakour and Eluru [11] stated that an equitable public transport service and accessibility can produce an efficient public transportation system. Lack of facilities and infrastructure will cause users to feel uncomfortable and insecure. Hence, shifting to private vehicles will be the first option. An infrastructure that connects to the bus stop such as, the pedestrian walkway is also significant to facilitate and ease the movement of users from their origins to  [12]. Hess [13] added that the decision to ride public transit is affected by good access. In other words, the built environment is vital to give comfort to the users and encourage transit use [14][15]. A study made by Eboli and Mazulla [16] found that bus stop furniture was statistically significant towards users' satisfaction on bus service. Furthermore, the well-equipped bus stop will give an effect on the efficiency of transit and traffic operation as well [17]. Previous studies often touch on reliability services [18][19][20] and poor bus service without discussing much on the detail of infrastructure and facilities that also affect the bus performance. Thus, this study came out with the micro-aspects of design that influence interest to access the bus stop.

Data Collection
The data collection was conducted at 123 bus stops (Table 1) detected within the 800-meter walking distance from the proposed Light Rail Transit (LRT) stations ( Figure 1). There are 19 LRT stations (exclude 8 stations on the 3 future reclaimed islands) on the island proposed in Penang Transport Master Plan (PTMP) [21]. The bus stops were evaluated based on the Transport Assessment Indicators Checklist created based on the literature review of transport assessment guidelines and active mode criteria for walkability and cycling. These indicators were filtered based on the most highlighted criteria by previous scholars and subsequently adapted with the local guidelines. Table 2 shows the final indicators of the Transport Assessment Indicators Checklist. The bus stops were evaluated with three categories; Available and following the guideline (/); Available and not following the guideline (X); Not available/not existed (0). The evaluation was given a standardized coding; (/=3); (X=2); (0=1) for data computation in SPSS. Figure 2 shows the procedure of data collection.  [25], [26] Bus stop is located close (15meters) to a pedestrian crossing [25], [26] Bus stop is provided between 2 to 5 lighting sources [25], [26] Provision of bus laybys (if provided) with a minimum width of 4.0meters and 12.0 meters length is not placed within 60 meters from road intersection/junction [25], [26]

Statistical Analysis
In this case, factor analysis was performed using IBM Statistic SPSS 22 to reduce the number of indicators and detect structures in the relationships between the indicators. The overall 13 indicators for evaluating bus stop performance were extracted by the maximum likelihood method and varimax rotated was implied.

Identifying significant determinants using Factor Analysis
By following the requirements for factor analysis, the indicators had been reduced from 13 to 10 as the eliminated indicators were identified to be not significant due to the low value of communalities. Based on Table 3, the Kaiser-Meyer-Olkin (KMO) measure of adequacy for indicator affecting bus stop's performance was 0.640, trespassing the minimum threshold value of 0.6 [28] which means the indicators are adequate to be further analysed. Bartlett's Test of Sphericity shows significant value (χ 2 (78) = 311.657, p-value = 0.000) indicating the correlation matrix is significantly different from an identity matrix, which makes the correlations between all indicators are zero. The significance level is small enough to reject the null hypothesis. .000 Nevertheless, the indicators should possess the value of communalities greater than 0.4. From the results in Table 4, indicator PT42; "Parking area is prohibited near stops and stations by providing "no parking" signage" does not influence the bus stop performance since the communalities value is lower than the threshold value (0.295). In other words, 29.5% of the variance 'Parking area is prohibited near stops and stations by providing "no parking" signage' is accounted for. Thus, indicator PT42 was removed for further analysis. PT39; "shelters are well-maintained in good condition" possessed the highest value of communalities (0.817) which means this indicator correlates with the bus stop performance and retained for further analysis.
Based on Table 5, the indicators with a value of more than 0.5 are categorized as indicators with high correlations. Five components (factors) were generated with an eigenvalue of > 1 as showed in Table 7. This explained 63% of the variance in the data. Thus, Table 6 summarized the factors with the respective indicators.    Table 7, these components show a high correlation compared to the other indicators which means these indicators are significant for bus stop performance. This is related to the study made by Fitzpatrick et al [29] where they stated that curbside elements have an impact on comfort, convenience and safety of the bus stop. Apart from that, proximity, bus shelters and time table information were among the demands made by the users for bus service quality improvement [9]. Previous studies mostly claimed that physical attributes such as accessibility, cleanliness, seat and space [30,31] significantly affected the satisfaction level of passengers regarding public bus ridership. The location of a bus stop near to the road intersection was mathematically proven to have a relationship with traffic delay as it involves the bus stopping and bus flow [32].

Conclusion
From the factor analysis, 13 indicators were reduced to 12 indicators consisting: PT31, PT32, PT33, PT34, PT35, PT36, PT37, PT38, PT39, PT40, PT41, and PT43. The indicators formed five components which are proximity and vicinity, well-maintained shelters, provision of curbs, bus information and facilities and location near road intersection were detected to have an impact on bus performance based on the satisfy the value of factor loadings and eigenvalues of principle component analysis. These indicators can be a reference for urban practitioners and transport engineers to improve the condition of the bus stop in the future. A well-equipped and good condition of the bus stop will attract and convince people to shift to public transport. Therefore, bus stop infrastructure and facilities should be given more attention to encourage dependency on public transport.