• Tidak ada hasil yang ditemukan

LAPORAN AKHIR PENELITIAN BERBASIS H-INDEX (PHI) RANDOM UTILITY MAXIMIZATION WITH INCLUSION LATENT MOTIVATIONS TO FORECAST DEMAND ON SHIFTING TO PUBLIC TRANSPORT MODE Tim Penelitian

N/A
N/A
Protected

Academic year: 2022

Membagikan "LAPORAN AKHIR PENELITIAN BERBASIS H-INDEX (PHI) RANDOM UTILITY MAXIMIZATION WITH INCLUSION LATENT MOTIVATIONS TO FORECAST DEMAND ON SHIFTING TO PUBLIC TRANSPORT MODE Tim Penelitian"

Copied!
59
0
0

Teks penuh

(1)

LAPORAN AKHIR

PENELITIAN BERBASIS H-INDEX (PHI)

RANDOM UTILITY MAXIMIZATION WITH INCLUSION LATENT MOTIVATIONS TO FORECAST DEMAND ON SHIFTING TO PUBLIC

TRANSPORT MODE

Tim Penelitian

Dr. Eng. Ir. Sugiarto, ST., M.Eng., IPM 198104102006041003 Ketua Peneliti Prof. Dr. Ir. Sofyan, M.Sc.Eng., IPU 195905121987021001 Anggota Peneliti Prof. Dr. Ashfa, S.T., M.T. 197302152000031001 Anggota Peneliti

Dibiayai Oleh:

Universitas Syiah Kuala,

Kementerian Pendidikan, Kebudayaan, Riset dan Teknologi, Sesuai dengan Surat Perjanjian Penugasan

Pelaksanaan Penelitian H-Indeks Tahun Anggaran 2021 Nomor: 169/UN11/SPK/PNBP/2021 Tanggal 19 Februari 2021

FAKULTAS TEKNIK UNIVERSITAS SYIAH KUALA

OKTOBER 2021

(2)
(3)

ii

ABSTRACT

Public Transport is commonly recognized as a valid transport policy to mitigate cars and motorcycles traffic in both developed and emerging countries. Sustaining the demand for public transport is crucial to any investigations for developing public transport. In such a case, the public transport provider needs to ensure the demand for public transport to meet the minimum passenger (load factor) to maintain the implementation of public mode. This proposal deals with the demand forecast and the motivations of private mode usage on switching to public transport mode that has been promoted and operated by the Aceh government as widely known as the Trans Koetaradja bus system. The specific goal of this proposal is to investigate motivations for individual modal splits of private mode (cars and motorcycles) in the context of promoting new bus reform of Trans Koetaradja in Banda Aceh. The Stated Choice (SC) experiment will be designed to obtain significant variables on five types of latent motivations including perceived quality of service (PSQ), social interaction, concern for the city environment, concern to traffic congestion, and believes of effectiveness the Trans Koetaradja policy implementation. The data was collected along the busiest Trans Koetaradja corridors and private mode usage who lives at surrounding targeted corridors. The two-stages demand modeling approaches will be conducted in this study to formulate the demand function of the modal split. First, the completed structural equation modeling (SEM) is used to converts latent motivation to continuous data sets. Then, the discrete choice based on random utility maximization is integrated with latent motivations.

Finding from the empirical result discloses that the determinants factors of selecting motorcycle as representative daily mode in Banda Aceh consist of travel destination, travel time, vehicle operating costs, gender, education level, motorcycle ownership, and driver’s license ownership. Further findings reveal that the probability of modal split among mode choice set obtained using binary logit are 70%, 30%, for Motorcycle and BRT-lite, respectively, and the derived model is considered very good based on discrepancy among observed and predicted value about 0.136 and 0.367 for Motorcycle and BRT-lite, respectively. Empirical findings from the confirmatory factor analysis have explicitly investigate the relationship between motivation factors and their attributes/indicators. The first results of the measurement model show that mandatory activities have the highest significant correlation with the attribute of liking to drive private vehicles (car/motorcycle), while non-mandatory activities have the most significant sign congestion increasing within the city due to highly autos dependency. The second model is the relationship amongst motivation variables. The results depict that as for mandatory activities appear that linkage with private vehicles significantly affects private vehicle problem (congestion and environment deterioration), while in non-mandatory activities the quality of service of current bus system has significantly affects satisfaction on using public transport.

Keywords: Public transport, demand forecast, motivation, modal split, mandatory, non- mandatory, private vehicles, confirmatory analysis.

(4)

iii

PREFACE

This report is a final report of the Penelitian Berbasis H-Index (PHI) funded by Universitas Syiah Kuala, Kementerian Pendidikan, Kebudayaan, Riset, dan Teknologi, Sesuai dengan Surat Perjanjian Penugasan Pelaksanaan Penelitian H-Indeks Tahun Anggaran 2021 Nomor:

169/UN11/SPK/PNBP/2021 Tanggal 19 Februari 2021. This research progress report current achievements of the research in accordance with the content of the research proposal approved for its financing.

The empirical results from this research reports can be used as the insight specifically studies about for the feasibility studies of the new Trans Koetardja public transport system, whether the agreement will result in a good modal shift in the future was conjectured. The critical demand forecast, and modal split estimation consider it an essential policy variable to ensure the sustainability of Trans Koetaradja operation. The methodological framework drawn from this study could serve as an important step as a planning tool for the transportation administrators, particularly for the planning of a new bus system. As public transport is a vital for successful urban bus system planning and operation. Thereby, such a transport policy would gain efficient policy implementation.

Finally, the authors would like to express their deepest gratitude to the LPPM Universitas Syiah Kuala for financially supported this study. All remaining oversight in this study in own.

Banda Aceh, 24th October 2021

Principal Investigator

(5)

iv

TABLE OF CONTENTS

CONFIRMATION PAGE ... i

ABSTRACT ... ii

PREFACE ... iii

TABLE OF CONTENTS ... iv

LIST OF TABLES ... v

LIST OF FIGURES ... vi

CHAPTER 1. INTRODUCTION ... 7

CHAPTER 2. LITERATURE REVIEW ... 9

2.1 Role of Travel Behaviors in Travel Demand Model ... 9

2.2 Conceptual Model of Hybrid Choice Model ... 10

CHAPTER 3. RESEARCH GOALS AND CONTRIBUTIONS ... 13

3.1 Research Goals ... 13

3.2 Research Urgency and Contribution... 13

CHAPTER 4. METHODOLOGY ... 15

4.1 Data Collection ... 16

4.2 Data Processing and Model Calibration ... 17

CHAPTER 5. RESULTS AND OUTPUTS ... 19

5.1 Modeling Travel Mode for Motorcycle and BRT-lite ... 19

5.1.1 Modeling Specifications ... 19

5.1.2 Sampling and sample size ... 20

5.1.3 Survey Procedure ... 20

5.1.4 Analysis Procedure and Data Setting ... 21

5.1.5 Results and Discussions ... 22

5.2 Modeling Motivation Factors Affecting Daily Mode Choice ... 24

5.2.1 Data and Distributions ... 24

5.2.2 Modeling Results and Discussion ... 26

5.3 Outputs ... 30

CHAPTER 6. CONCLUSIONS ... 21

REFERENCES ... 22

Appendix 1. Documentation of research activities ... 25

Appendix 2. Acceptance letter of manuscript ... 29

Appendix 4. Accepted Manuscript for Scopus Proceedings and Draft of Journal. ... 31

(6)

v

LIST OF TABLES

Table 5.1 Mode choice scenario with SP technique ... 21 Table 5.2 Explanatory variable setting ... 21 Table 5.3 Correlation result among explanatory variable of mode choice ... 22 Table 5.4 Calibrated parameters of mode choice (based outcome MC) using binary logit 23 Table 5.5 Level of model accuracy ... 24 Table 5.6. Cut-off value GoF ... 24 Table 5.7. Summary of responses of Linkage to Private Vehicles (Mandatory Activities) 24 Table 5.8. Summary of responses of Private Vehicle Problems for Mandatory Activities . 25 Table 5.9. Summary of responses of Quality Trans Koetaradja for Mandatory Activities . 25 Table 5.10. Summary of responses of Satisfaction with Trans Koetaradja for Mandatory Activities ... 25 Table 5.11. Summary of responses of Linkage to Private Vehicles (Non-Mandatory Activities) ... 25 Table 5.12. Summary of responses of Private Vehicle Problems (Non-Mandatory Activities) ... 26 Table 5.13. Summary of responses of Quality Trans Koetaradja for Non-Mandatory Activities ... 26 Table 5.14. Summary of responses of Satisfaction with Trans Koetaradja for Non-Mandatory Activities ... 26 Table 5.15. Measurement Model of CFA for Mandatory and Non-Mandatory Activities . 28

(7)

vi

LIST OF FIGURES

Figure 2.1 Conceptualized of hybrid discrete choice (modify from Sugiarto et al., 2017b)10

Figure 4.1 Research flowchart ... 15

Figure 4.2 Study area and Trans Koetaradja Corridors ... 16

Figure 4.3 the Framework of Hybrid Model for Demand Forecast Model ... 17

Figure 5. 1 Path Diagram (Mandatory Activities) ... 27

Figure 5. 2 Path Diagram (Non-Mandatory Activities) ... 27

(8)

7

CHAPTER 1. INTRODUCTION

Modern motorized society faces several externalities due to traffic congestion such as unwarranted travel times, air pollution, unnecessary energy consumption, and driver frustration in Banda Aceh (Saleh et al., 2017). Facing increasing roadway congestion and deterioration of city environment (emission and noise), the modal shift from private vehicles to public transportation becomes a foremost answer available to urban planners and transport policy makers (Barthelemy, 2016). Studies have suggested solutions to promote modal shift, involving improving service quality and accommodating perceptions of motivation to switch (Mugion et al., 2018; Redman et al., 2013). Furthermore, the term of social interaction is also found to play a significant positive role in promoting this shift (Abou-Zeid et al., 2013).

Like many other rapidly growing cities in Indonesia, the city center of Banda Aceh has not run-away from the influences of impecunious public transport services such as insufficient quality of services, lack safety for passengers, and inefficiency capacity (Irvansyah et al., 2020; Safitri et al., 2020). The current public transportation system is complicated in coping with the growing spatial and demographics of the city. Studies have shown that bus reform policy such as BRT has appeared as a cost-effective public transport alternative with noteworthy possible for Jakarta (Ernst., 2005; Susilo et al., 2007). Public transport is commonly recognized as a valid transport policy to mitigate cars and motorcycles traffic in both developed and emerging countries. Understanding the motivations of individual preference on modal split could serve potential demand and tool for promoting new public transport system.

This proposal deals with the demand forecast and the motivations of private mode usage on switching to public transport mode that has been promoted and operated by the Aceh government as widely known as the Trans Koetaradja bus system. The specific goal of this proposal is to investigate motivations for individual modal splits of private mode (cars and motorcycles) in the context of promoting new bus reform of Trans Koetaradja in Banda Aceh. The Stated Choice (SC) (Sugiarto et al., 2017a) experiment will be designed to obtain significant variables on five types of latent motivations including perceived quality of service (PSQ), social interaction, concern for the city environment, concern to traffic

(9)

8 congestion, and believes of effectiveness the Trans Koetaradja policy implementation. The data will be collected along the busiest Trans Koetaradja corridors and private mode usage who lives at surrounding targeted corridors. The two-stages demand modeling (Sugiarto et al., 2017a; Sugiarto et al., 2020) approaches will be implemented in this study to formulate the demand function of the modal split. The two-stage approach or known as hybrid formulation of discrete choice model has been used and proved as valid tool to forecast and evaluate demand to evaluate the feasibility of transport policy of Road Pricing in Jakarta, a capital of Indonesia (for more detail information see Sugiarto et al., 2017a; 2017b; 2020).

Two-stage modeling proposed in this proposal is used to demand forecast and evaluate the motivations of modal split for specific transport policy known as new Trans Koetaradja bus reform proposed by government. The two-stage modeling approach consist of (1) the completed structural equation modeling (SEM) will be used to converts latent motivation to continuous data sets; (2) the discrete choice based on random utility maximization will be integrated with latent motivations. This proposed model could be able to explain the decisions of each user on the current mode and Trans Koetaradja bus system and determine the motivation for switching their mode. These two-stage approaches could be used for the feasibility studies of the new Trans Koetardja public transport system, whether the agreement will result in a good modal shift in the future was conjectured. The critical demand forecast, and modal split estimation consider it an essential policy variable to ensure the sustainability of Trans Koetaradja operation. The methodological framework drawn from this study could serve as an important step as a planning tool for the transportation administrators, particularly for the planning of a new bus system.

(10)

9

CHAPTER 2. LITERATURE REVIEW

2.1 Role of Travel Behaviors in Travel Demand Model

Dealing with an increasing roadway congestion and deterioration of city environment (emission and noise), the modal shift from private vehicles to public transportation becomes a foremost answer available to urban planners and transport policy makers (Barthelemy, 2016). Studies have suggested solutions to promote modal shift, involving improving service quality and accommodating perceptions of motivation to switch (Mugion et al., 2018;

Redman et al., 2013). Furthermore, the term of social interaction is also found to play a significant positive role in promoting this shift (Abou-Zeid et al., 2013).

While previous model predicted demand for travel has mainly been related to predict and analysis travel demand based on a traditional random utility maximization framework to forecast the travel demand. That is, their models have considered only objective or measurable attributes (observed variables) from the choice alternatives and socio- demographic attributes as explanatory variables. This despite that fact that it has been well recognized in recent years that attitudes and perceptions also influence individual choice behavior in travel demand forecasting (see for example Bolduc et al., (2008); Yanez et al., (2010); Bierlaire et al., (2010); Raveu et al., (2010); Sugiarto et al., (2017a; 2017b; 2020).

Treating psychological determinants (unobserved variables) as latent variables in models is an advanced step in travel behavior analysis. Although unobservable, these determinants can be measured by indicators and explained by other observable elements (Bollen, 1989), and may then be used to enhance the explanations of modeling travel behaviors (Ben-Akiva et al., 2002).

Several initial works attempt to integrate the conventional model with the latent variables to obtain more robust model with more efficient empirical findings. For instances, Morikawa et al. (2002) on mode choice with the latent variables of sensitivity to comfort and convenience, or the study by Ramming (2002) on route choice with the latent variable of knowledge of the transportation system. This approach is determined to allow more realistic

(11)

10 interpretation of travel behaviors. In relation to transport demand and policy analysis, some psychological determinants, such as problem awareness and policy beliefs, were found to have direct impacts on the acceptability of policy measures (Sugiarto et al., 2020; Sofyan et al. 2019). Other studies related to the car dependency (Beirao and Cabral, 2007), transport policy satisfaction (Zhao et al., 2013) and environment concerns (Atasoy et al., 2013; Kim et al., 2017). The effects of social interaction, represented by the proportion of the reference group choosing an alternative, can be included directly in the utility function of choices (Kim et al., 2017) or indirectly in the class membership function (Maness and Cirillo, 2016).

2.2 Conceptual Model of Hybrid Choice Model

Note that during the last decade a breed of “hybrid model” has been proposed for allowing individual’s attitudes and perceptions to be incorporated through latent variables in a standard discrete choice setting (see Walker & Ben-Akiva, 2002; Yanez et al., 2010). Two approaches to hybrid model are now widely available, one based on the sequential approach (see Yanez et al., 2010; Sugiarto et al., 2017a; 2017b; 2020) and the other a simultaneous framework (see for instances Bolduc et al., 2008; Raveu et al., 2010).

Figure 2.1 Conceptualized of hybrid discrete choice (modify from Sugiarto et al., 2017b)

The second approach offers efficient and consistent estimators of parameters but has been used less because of it is more complex and computationally cumbersome due to the multidimensional integral in the joint distribution. Sugiarto et al., (2017b) concluded that

Tangible Variables, x

Psychological Perceptions, y

Latent Variables, η

Utility Function, U

Stated Preference Choice, y*

Latent Variable Model (MIMIC Model)

Discrete Choice Model (DCM)

Structural equations Measurement equations

Latent (intangible) variables Observed (tangible) variables

(12)

11 the simultaneous method of discrete choice model with integration of latent variables has restricted number of latent variables (typically three or fewer latent variables), increasing number of latent variables, the computational complexity may rise exponentially.

The general framework of the hybrid discrete choice has two-stage model consist of two sequential modeling components: 1) a latent variables model using the covariance-based structural equation model (CB-SEM); and 2) discrete choice model. A detailed discussion of each component is given in the following sections, and it illustrates in Figure 2.1. The upper model (a latent variables model) consists of the covariance-based structural equation model (CB-SEM). This model is generally used to delve more deeply into behavioral information related to respondents’ psychological perceptions. The multiple-indicators multiple-causes (MIMIC) model is one of complete CB-SEM that used to capture both observed (causes) and unobserved (indicators) variables. This is a multivariate analysis as originally proposed by Joreskog & Goldberger (1975). the MIMIC model consists of a structural equation model and a measurement model, respectively given by:

z , and (1)

where yi is a vector of observable psychological indicators variables, zi is a vector of explanatory variables that cause hi, , and are matrices of unknown parameters to be estimated, and the terms zi and i are vector of measurement errors.

The lower part model of discrete choice model is drawn from binary response model (BRM).

The BRM offers convenient framework to analysis the situation of choice between two alternatives-whether the individual takes an action or does not, or chooses between one of two elemental choices (Greene and Hensher, 2010). Formally, the structural model consists of a latent equation below:

, (2)

With the observed binary outcomes are 1 > 0 , otherwise 0. Where hi and xi are vectors of the intangible and tangible exogenous variables, respectively. Moreover, g’ and

(13)

12 b’ are vectors of the unknown parameters, and represents an individual observation across samples. The model is simply completed by assuming that the latent errors eihas a univariate standard normal distribution (IID) and estimation of structural equation using a simple probit link identifies the structural treatment effect g’ and b’. Then, the individual contribution to the likelihood function can be drawn as:

| ∏ = [1 Φ ] x ∏ = Φ , (3)

where 1 is the indicator function for a respondent, with 1 if a respondent i chooses outcome 0, otherwise 1 and Φ(.) represents a standard normal cumulative density function. The unknown parameter in equation (3) can be estimated using maximum log- likelihood estimation.

(14)

13

CHAPTER 3. RESEARCH GOALS AND CONTRIBUTIONS

3.1 Research Goals

This research allows us to formulate the robust methodological approaches for to examine whether the agreement of Trans Koetaradja implementation will result in a sufficient demand and modal shift to sustain future operation. The specific goal of this proposal is to investigate motivations for individual modal splits of private mode (cars and motorcycles) in the context of promoting new bus reform of Trans Koetaradja in Banda Aceh. The goal of this research consists of several objectives as devised mainly to:

1) formulate multivariate model using the structural equation model (SEM) to examine significant factors that motivated individual to shift their mode from private and public mode.

2) formulate two-stage model for demand forecast of modal split by considering latent motivations determinants, and.

3) concluded an empirical model that the methodological framework can be used as a public transport planning tool for the government/department of transport.

3.2 Research Urgency and Contribution

The demand forecast is crucially essential aspects to ensure the passengers of the bus system meet the minimum demand to maintain the sustainability/efficiency of public transport implementation. Recent studies have been alerted that the perceptions could manifest latent motivations that significantly affecting individual choice behavior on shifting/switching their mode from private to public transport. Few studies have thought the latent motivation can be simultaneously examined in transport demand forecasting, and perhaps this study is an initial one in Indonesia. By jointly examine both observed demand forecasting and the motivations to modal shift, allows us to make the significant following contributions:

1) We propose a comprehensive demand model that integrates unobserved variable (latent motivations) in demand forecast and identify the behavioral adoption by examining determinant factors of switching modes.

(15)

14 2) We propose a methodological framework drawn from this study that could serve as an important step as a planning tool for the transportation administrators, particularly for the planning of a new bus system.

3) We will produce scientific paper in which would be published in Scopus indexed conference and reputable international journal

(16)

15

CHAPTER 4. METHODOLOGY

Aiming to achieve the goals of this research, a sequence of steps of research in building model of demand forecast and modal split using a hybrid discrete choice approach is illustrated in the flowchart shown in Figure 3.1, and these processes are discussed in the following sub-sections.

Figure 4.1 Research flowchart

NO

YES

NO YES

Start

Research Background, Problem Statement & Review Existing

Works/Methodology

Defining Variables, Indicators and Designing Questionnaire Form using Stated Choice 1. Observed respondent's socio-demographics

2. Observed respondent's mobility and travel attributes.

3. Perceptions (psychological intension) of modal shifting motivations

Data Processing: Tabulation, Coding and Testing and Distributions

Variables Setting

&

Two-stages Discrete

Establish Model & Testing Final Model

Analysis & Interpretasion using establishing model

Conclusions, Recomendations and Future Works

Finish

1. Map of study area

2. Map of Corridors of Trans Koetaradja 3. Existing related works

Testing Goodness of Fits (GoF) Model

GFI (>90%) RMSE; CFI (< 5%)

Sig. T (<10%) latent variables model using the covariance-based structural

equation model (CB-SEM)

Hybrid Discrete Choice (Integrated Latent Motivations into Random

Untility Maximization)

Testing Goodness of Fits

(GoF) Model

(17)

16 4.1 Data Collection

This research proposal will be involved several students, namely bachelors (2-4 students), master (2 students) and started in February the years of 2021. Data processing will be carried out at the Transportation Laboratory, Syiah Kuala University. The object of research is private and public mode users who uses Trans Koetaradja bus or people who live nearby corridors. Data collection will be carried out along the busiest corridors of the bus system.

The area of study and maps of Trans Koetaradja corridors can be seen in Figure 4.2.

Figure 4.2 Study area and Trans Koetaradja Corridors

Stratified Random Sampling based Stated Choice (Sugiarto et al., 2017a) method is adopted and will be used for data collection in this research. Surveys are conducted to collect data on household characteristics, travel behavior (travel behaviors) and perceptions of indicators related to the motivations on switching travel mode on five types of latent motivations including perceived quality of service (PSQ), social interaction, concern for the city environment, concern to traffic congestion problem, and believes of effectiveness the Trans Koetaradja policy implementation. These four variables will be used to upper model using

(18)

17 the MIMIC modeling framework as described in equation (1). This study used 15 indicators related to four psychological motivations. The minimum number of samples is 10 x 15 indicators or 150 samples. According to Hair (2019) if the model estimation will be used is the maximum likelihood estimator (MLE) as mentioned in equation (3), the minimum valid data required for MLE is 200-250 samples. Therefore, it will take 400 samples to avoid the bias of the sample distributions.

Figure 4.3 the Framework of Hybrid Model for Demand Forecast Model

4.2 Data Processing and Model Calibration

The data will be analyzed and calibrated using the MIMIC method whose parameters will be estimated using the LISREL 10.2 econometric software for the upper model (see equation 1). The use of this method aims to test a model hypothesis and find out the vector of indicator variables (y) connected by a latent variable (h) with the covariate (x). The MIMIC model consists of two equations which are exploratory relationships between (x) and (h) and confirmatory relations between (y) and (h) as in equation (1) as described in section 2. The

Random Utility Maximization

(RUM), ( ) Socio-demographics variables Mobility and travel attributes. Mode Choice and Shifting Choice sets

Mode Choice Sets

Indicators/Perceptions of Latent motivations

( ):

Latent Variables of Motivations

( ) Percieved

Quality of Service (PSQ) ( )

Social Interaction

( )

Concern to Traffic Congestion

( ) Concern for

the City Environmen

( )

Effectiveness the Trans Koetaradja policy ( )

(19)

18 lower model will be analyzed and calibrated using hybrid discrete choice framework, and the regression parameters will be estimated GAUSS econometric version 3.2.32. The structural relationship among service encounter and global satisfaction in SEM is hyphothesized in Figure 4.3.

(20)

19

CHAPTER 5. RESULTS AND OUTPUTS

In this chapter will be described the results of research that has been conducted until the progress report need to be submitted. The result that will be presented in this section mainly the result of the paper which we have submitted and accepted in the international conference indexed by Scopus.

5.1 Modeling Travel Mode for Motorcycle and BRT-lite 5.1.1 Modeling Specifications

The proposed analysis of the mode choice of transportation is based on disaggregated data characterizing individual travel behavior. The disaggregate model treats each person independently, thus providing a broader prediction of behavior. The binary logit has been widely used in predicting travel demand and its contributing factors. These models relate individual characteristics to more effective mode choice behavior and provide more accurate estimates of the elasticity of demand. In this study, we adopted the model formulation simple binary logit to predict modal split and its relevant factors for motorcycle and BRT-lite.

Assumed that, an individual i asked to choose one of two options (k), i.e., motorcycle and BRT-lite for their travel mode during distributed questionnaire as a representative mode.

Then, the formulation of the utility function (U) and probability (P) of selecting mode choice can be drawn as:

Uk Χkk βKTΧk εk (4)

Pik exp(βKTΧik)

K= exp(βTΧik)

(5)

where X is input variable or explanatory variables, β represents unknown parameters which need to calibrated using the input data and ε is systematic disturbance which is to be identical and independent (IID) and assumed to be type II Gumbel distributed. Model selection is conducted based on the Goodness of Fit model and the statistically significant of the estimated parameters. As for the goodness of fit, the coefficient of determination of rho- squared and the log-likelihood ratio (LR) test is conducted based on chi-squared distribution.

Both LR test and significant parameters are considered to have significant error less than or equal of 5% error level.

(21)

20 5.1.2 Sampling and sample size

In this study, the determination of the number of samples was taken by taking the average of the calculation using Krejcie and Morgan formula and Slovin formula. The Krejcie and Morgan and Slovin formula can be drawn, respectively as:

n = X2N

d2 N− + X2 (6)

n = N

+ N Ԑ2 (7)

where n is number of samples, N is total population, P is proportion of population (0.5), d is degree of accuracy (5%), χ2 is the value of the chi-square used in the analysis, and ε is error level (error rate) generally used 1%, 5% and 10%. By employing equation (3) and (4) the total of sample requires in this study is about 400 samples. Using self-reported experience online survey, a total of 470 samples has collected and 400 samples are valid and used in this paper.

5.1.3 Survey Procedure

Stated preference (SP) is a method to find out the opinion of respondents in facing various choices offered through online-based survey. The survey with the SP method has several conveniences, one of which is as a strong basis for respondents' statements (travelers) regarding what they will feel in using real alternative modes of transportation such as vehicle operating costs, travel time, and other attributes. In this study, the questionnaire was distributed online in the form of a google form with a choice of answers according to events in the field (before Covid-19). The criteria for respondents in this study are respondents who live in the city of Banda Aceh who are 17 years old and have their income (working) and has experienced both modes of transportation, i.e., motorcycle, and Bus Trans Koetaradja.

In this case, prior to selecting their mode, we introduced the BRT-lite asked their willingness to choose and shift to BRT-lite from motorcycle mode. Samples were taken using probability sampling techniques, which are sampling techniques that provide equal opportunities for each member of the population. This technique is a type of random sampling, which is a random sampling technique. The respondents were also given a scenario for alternative

(22)

21 mode choices by giving an illustration of what advantages Bus Rapid Transit (BRT) has over other modes as the public mode offered. The following is a mode choice scenario table with stated preference techniques.

Table 5.1 Mode choice scenario with SP technique

Service Motorcycle Existing Bus

(Trans Koetaradja)

Proposed new system (BRT-Lite) Cost > Rp. 150 / Km Government subsidies

(free)

Government subsidies (free) Comfort Comfortable, flexible There are 5 corridors There are 5 corridors

Headway - 15 minutes 5-10 minutes

Lanes Join other vehicles Join other vehicles Has a special lane

Table 5.2 Explanatory variable setting

Variable Setting

data Codes setting used in the utility function of the model Mode choice Binary 1 (BRT-

lite) 0 (MC) Travel

destination distance

Ordinal 1 2 3 4 5 -

<4 Km 4-6.9 Km 7-9.9 Km 10-12.9 Km

> 13

Km -

Traveling time Ordinal

1 2 3 4 5 6

<10 min.

10-19 Min.

20-29 Min.

30-39 Min.

40-49 Min.

> 50 Min.

Vehicle

operating costs Binary 1 0 - - - -

Rp. 150 Rp. 0 - - - -

Gender Binary 1 0 - - - -

male female - - - -

Level of

education Ordinal

1 2 3 4 - -

Primary

school Diploma Bachelor Post

graduate - -

Ownership of

motorbikes Continue 1 2 3 >4 0 -

Ownership of

driver’s license Binary 1 0 - - - -

Yes Not - - - -

5.1.4 Analysis Procedure and Data Setting

All data listed in the questionnaire were coded for several levels according to the Likert scale to obtain a comprehensive evaluation of each attribute as describes in Table 5.2. Some variables are categorized into two levels, such as the mode of transportation, i.e. (MC = 0 and BRT-Lite = 1), driver’s license ownership, i.e. (Yes = 1 and No = 0), while other

(23)

22 variables for 6 levels, such as occupation and monthly income. Several other variables use concepts such as almost every day, 3-4 weeks, 1-2 weeks, <1 time a week, and never.

Therefore, for analysis purposes, the levels for the same concept are combined into one level or converted into a dummy form by taking the mean value of each attribute then making it i.e., <3 = 1 the rest = 0. This is because the software handles each level differently, so it does not reflect the actual situation.

Table 5.3 Correlation result among explanatory variable of mode choice

Parameters Coefficient of

correlation P-Value

Travel time to the bus stop (X1) 0.215 0.095

Purpose of the trip (X2) 0.774 0.697

Travel destination distance (X3) -5,862 0.000

Origin zone district (X4) 0.195 0.018

Destination zone district (X5) 0.081 0.334

Travel time with representative modes (X6) -0.351 0.031

Vehicle operating costs (X7) 14,623 0.000

Gender (X8) 0.975 0.060

Age (X9) -0.351 0.317

Education level (X10) -0.486 0.045

Occupation (X11) -0.317 0.061

Monthly income (X12) -0.096 0.528

Motorcycle ownership (X13) 0.252 0.238

Driver’s license Ownership (X14) 0.732 0.139

5.1.5 Results and Discussions

Table 5.3 summarizes the significance of each attribute that has been tested using bivariate correlation. The P-value helps to decide whether there is a correlation between the dependent and independent variables or not. The lower the P-value means that the higher the level of correlation among the variables. Variable with a P-value less than 0.05 were significantly correlated, while those greater than 0.05 were considered insignificant. Therefore, there are seven significant attributes with the mode choice set in this study, namely the distance of travel destination, travel time, vehicle operating costs, gender, latest education level, motorcycle ownership, and driver’s license ownership.

(24)

23 Table 5.4 Calibrated parameters of mode choice (based outcome MC) using binary logit

Variables Coefficient P-Value

Parameters:

Intercepts -7,114 0.000

Travel Destination distance (X3) 2,278 0.001

Travel time (X6) 0.473 0.000

Vehicle operating costs (X7)) 2,715 0.000

Gender (X8) 0.648 0.035

Level of education (X10) 0.409 0.006

Ownership of motorcycle (X13) 0.324 0.018 Ownership of driver’s license (X14) 0.709 0.039 Summary of Statistics:

Sample (N) 400

χ2 test (sig) 76.58

(0.000)

Chi-Squared 159,895

Rho-Squared 0.326

All attributes that were significantly correlated with the selected mode by respondents were identified, all uncorrelated or ineligible attributes were excluded, and the analysis was repeated for attributes that met the requirements only for higher accuracy, and the results of this analysis are presented in Table 4. According to the results of logistic regression analysis that has been carried out several times to obtain attributes that are significantly correlated with the mode choice set of the MC and BRT-lite. The utility function of the MC can be drawn from Table 5.4 as:

UMC = -7,114 + 2,278 (X3) + 0,473 (X6) + 2,715 (X7)) + 0,648 (X8) +

0,409 (X10) + 0,324 (X13) + 0,709 (X14) (8)

According to the utility function in equation (7), the contributing factors of selecting the MC as representative mode are travel destination, travel time, vehicle operating costs, male gender, education level, motorcycle ownership, driver’s license as it shown in Table 5.4.

The model goodness of fit model has shown in the summary of statistic in the bottom of Table 5.4.

(25)

24 Table 5.5 Level of model accuracy

Mode Observation Estimate Discrepancy

Motorcycle (MC) 70 81 0.136

Bus Rapid Transit-Lite (BRT-Lite) 30 19 0.367

5.2 Modeling Motivation Factors Affecting Daily Mode Choice 5.2.1 Data and Distributions

Data was collected using the Revealed Preference (RP) data with total of 400 respondents who are divided evenly into 2 groups, for mandatory and non-mandatory activities by distributing questionnaires using Google Form. Questionnaires were distributed to respondents with the target of 17 years old, had used a private vehicle and Trans Koetaradja Bus domiciled in Banda Aceh City and its surrounding areas. In this study, respondents were asked about indicators of linkage to private vehicles, factor of private vehicle problems, quality of Trans Koetaradja and its satisfaction. The questionnaire was prepared using a Likert scale with a scale of 1-4 from very negative to very positive and strongly disagree to strongly agree. The method used is Confirmatory Factor Analysis (CFA) executed using Analysis Moment of Structures. There are several fit indexes and cut-offs to test whether a model can be accepted or rejected. The Cut-off for Goodness of Fit (GoF) can be seen in Table 5.6.

Table 5.6. Cut-off value GoF

Table 5.7. Summary of responses of Linkage to Private Vehicles (Mandatory Activities) Indicators Factors of Linkage to Private

Vehicles (LPV)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree LPV1 Private vehicles are indispensable in

daily activities 4.0% 4.5% 41.5% 50.0%

LPV2 Likes to drive private vehicles 3.5% 4.5% 51.5% 40.5%

LPV3 Prefer traveling personal space 3.0% 12.5% 59.0% 25.5%

GoF Index Cut-off Value

Goodness of Fit Index (GFI) ≥0,90

Adjusted Goodness of Fit Index (AGFI) ≥0,90

Comparative Fit Index (CFI) ≥0,90

The Root Mean Square Error of Approximation

(RMSEA) ≤0,80

(26)

25 Table 5.8. Summary of responses of Private Vehicle Problems for Mandatory Activities Indicators of Factor Private Vehicle Problems

(PVP)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree PVP1 The number of accidents is increasing 1.5% 7.5% 47.0% 44.0%

PVP2 Congestion is increasing 3.0% 8.5% 59.0% 29.5%

PVP3 Emissions and noise are increasing 2.5% 4.5% 48.0% 45.0%

PVP4 Driving a private vehicle is at risk of

being more unsafe 7.0% 31.0% 44.5% 17.5%

Table 5.9. Summary of responses of Quality Trans Koetaradja for Mandatory Activities Indicators Factor of Quality Trans Koetaradja

(QTK)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree QTK1 Satisfactory bus stop facilities 6.5% 39.0% 47.0% 7.5%

QTK2 Cleanliness and Comfort 0.5% 14.5% 65.5% 19.5%

QTK3 The officers serve friendly and provide a

sense of security 2.0% 11.0% 68.0% 19.0%

Table 5.10. Summary of responses of Satisfaction with Trans Koetaradja for Mandatory Activities

Indicators Factors of Satisfaction with Trans Koetaradja (STK)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree STK1 The bus is needed in daily activities 5.0% 27.0% 48.5% 19.5%

STK2 The distance between work/home is not

far from the bus stop 7.5% 44.5% 41.5% 6.5%

STK3 The waiting time at the bus stop doesn't

take long time 8.5% 51.0% 34.0% 6.5%

STK4 The bus conditions that are not

crowded/full during peak hours 11.0% 48.0% 30.5% 10.5%

Table 5.11. Summary of responses of Linkage to Private Vehicles (Non-Mandatory Activities)

Indicators Factors of Linkage to Private Vehicles (LPV)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree LPV1 Private vehicles are indispensable in

daily activities 4,0% 13,0% 41,5% 41,5%

LPV2 Likes to drive private vehicles 3,0% 12,5% 46,5% 38,0%

LPV3 Prefer traveling personal space 4,5% 14,5% 53,5% 27,5%

(27)

26 Table 5.12. Summary of responses of Private Vehicle Problems (Non-Mandatory

Activities) Indicators of Factor Private Vehicle Problems

(PVP)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree PVP1 The number of accidents is increasing 0,5% 16,5% 39,5% 43,5%

PVP2 Congestion is increasing 0,5% 13,5% 55,0% 31,0%

PVP3 Emissions and noise are increasing 2,5% 12,0% 49,0% 36,5%

PVP4 Driving a private vehicle is at risk of

being more unsafe 8,5% 31,5% 47,0% 13,0%

Table 5.13. Summary of responses of Quality Trans Koetaradja for Non-Mandatory Activities

Indicators Factor of Quality Trans Koetaradja (QTK)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree QTK1 Satisfactory bus stop facilities 9,0% 28,5% 48,0% 14,5%

QTK2 Cleanliness and Comfort on the bus 1,5% 16,0% 61,5% 21,0%

QTK3 The officers serve friendly and provide a

sense of security 4,0% 10,5% 66,5% 19,0%

Table 5.14. Summary of responses of Satisfaction with Trans Koetaradja for Non- Mandatory Activities

5.2.2 Modeling Results and Discussion

In this section, the analysis using the Confirmatory Factor Analysis (CFA) model is used to investigate and determine the relationship between indicators and their latent variables, as well as the relationship between latent variables. Figure 5.1 shows that the LPV latent variable, the LPV1 indicator as to the standard variable then the LPV2 indicator has a coefficient value of 0,996 which is automatically fulfilled to 1,00 and becomes an indicator

Indicators Factors of Satisfaction with Trans Koetaradja (STK)

Responses (Percentage of Responses) Strongly

Disagree Disagree Agree Strongly agree STK1 The bus is needed in daily activities 4,5% 18,5% 54,0% 23,0%

STK2 The distance between work/home is not

far from the bus stop 10,0% 32,5% 49,0% 8,5%

STK3 The waiting time at the bus stop doesn't

take long time 12,0% 47,5% 36,0% 4,5%

STK4 The bus conditions that are not

crowded/full during peak hours 12,0% 44,0% 32,5% 11,5%

(28)

27 that affects the LPV latent. The indicator that affects the PVP latent variable is the PVP3 indicator which has the largest coefficient value of 1,08. The QTK3 indicator is an indicator that affects the latent variable QTK with the largest coefficient value of 1,09. In the STK variable, the indicator that affects is the STK4 indicator with the largest coefficient value, namely 1,05. The fit model test results showed that the Goodness of Fit values are as follows:

GFI is 0,932 > 0,900 (good fit), AGFI is 0,903 > 0,900 (good fit), CFI is 0,978 > 0,900 (good fit) and RMSEA is 0,044 < 0,080 (good fit).

Figure 5. 1 Path Diagram (Mandatory Activities)

Figure 5. 2 Path Diagram (Non-Mandatory Activities)

(29)

28 Figure 5.2 shows that the LPV latent variable, the indicator that affects is the LPV2 indicator with the largest coefficient value of 1,05. The PVP3 indicator is an indicator that affects the latent PVP variable with the largest coefficient value, which is 1,002. The indicator that affects the latent variable QTK is the QTK3 indicator with a coefficient value of 1,06. In the STK variable, the indicator that affects is the STK4 indicator with the largest coefficient value, namely 1,05. The fit model test results showed that the Goodness of Fit values are as follows: GFI is 0,939 > 0,900 (good fit), AGFI is 0,914> 0,900 (good fit), CFI is 0,985 >

0,900 (good fit) and RMSEA is 0,036 < 0,080 (good fit).

Table 5.15. Measurement Model of CFA for Mandatory and Non-Mandatory Activities Laten

Variables Indicators

Mandatory Activities Non-Mandatory Activities Estimate t-value Estimate t-value LPV

LPV1 LPV2 LPV3

1,000 0,996 0,728

14,717 10,514

1,000 1,055 0,903

12,193 11,175

PVP

PVP1 PVP2 PVP3 PVP4

1,000 1,035 1,077 1,054

10,977 11,488 9,199

1,000 0,945 1,002 0,924

13,108 12,425 10,049 QTK

QTK1 QTK2 QTK3

1,000 1,051 1,092

10,025 10,034

1,000 1,024 1,061

9,421 9,426 STK

STK1 STK2 STK3 STK4

1,000 0,901 0,940 1,048

8,979 9,240 9,217

1,000 0,968 0,934 1,047

10,279 10,491 10,324

Table 5.15 shows the based on the results obtained for mandatory activities, the highest t- value is in the latent variable Linkage to Private Vehicle (LPV) indicator “Likes to drive private vehicles (LPV2)” with the significant value of 14,717 > 1,96. Then in the latent variable Private Vehicle Problems (PVC) indicator “Emissions and noise are increasing (PVP3)” with the significant value of 11,488 > 1,96. Then, the latent variable Linkage to Private Vehicle (LPV) indicator “Prefer traveling personal space (LPV3)” with the significant value of 10,514 > 1,96. Meanwhile on non-mandatory activities based on the results obtained, the highest t-value is in the latent variable Private Vehicle Problems (PVP) indicator “Congestion is increasing (PVP2)” with the significant value of 13,108 > 1,96.

Then with a t-value of 12,425 > 1,96 found on the indicator “Driving a private vehicle is at

(30)

29 risk of being more unsafe (PVP4)”. The last with the third highest t-value is in latent variable Linkage to Private Vehicle (LPV) indicator “Likes to drive private vehicles (LPV2)” with the significant value of 12,193 > 1,96.

Figure 5. 3 The Results of Relationship Between Latent Variables and Their Coefficients for Mandatory Activities

Figure 5. 4 The Results of Relationship Between Latent Variables and Their Coefficients for Non-Mandatory Activities

0,388 0,306 0,524 LPV

PVP

QTK

STK

0,464 0,254 0,315 LPV

PVP

QTK

STK

(31)

30 Figures 5.3 and 5.4 show the relationship between the latent variables that is all in positive correlation. It can be concluded from the two images that, the latent variable linkage private vehicle (LPV) affects the latent private vehicle problem (PVP), which means increasing the height of private vehicles, the awareness of problems caused by vehicle linkages will also increase. The latent variable of private vehicles problem (PVP) affects the latent quality of bus (QTK) which means increasing awareness about the problem of private vehicles, the quality of bus that affects the community for mode selection, if the quality of bus is good then the satisfaction with bus (STK) is also good because the quality of bus greatly affects latent satisfaction with the bus.

The findings and discussions in previous section concludes that as for mandatory activity modeling calibration results using CFA with the highest significant on the variable latent linkage private vehicle (LPV) indicator Likes to drive private vehicles (LPV2). Further investigation depicts that as for non-mandatory activities obtained the highest significant on latent variable private vehicle problems (PVP) indicator congestion is increasing (PVP2).

To sum up the empirical findings from this study have explicitly investigate the relationship between motivation factors and their attributes/indicators. The first results of the measurement model show that mandatory activities have the highest significant correlation with the attribute of liking to drive private vehicles (car/motorcycle), while non-mandatory activities have the most significant sign congestion increasing within the city due to highly autos dependency. The second model is the relationship amongst motivation variables. The results depict that as for mandatory activities appear that linkage with private vehicles significantly affects private vehicle problem (congestion and environment deterioration), while in non-mandatory activities the quality of service of current bus system has significantly affects satisfaction on using public transport.

5.3 Outputs

The output of research till this progress report is listed as below:

1. Accepted and presented paper, entitles “Choice Model and Influencing Factors of the Travel Mode for Motorcycle and BRT-lite in Banda Aceh, Indonesia”. This paper presented on Computer Science, Physics, and Engineering AIP Part 1 and accepted to be published in AIP conference series (indexed by Scopus).

(32)

31 2. Drafting manuscript entitle “Exploring Mode Choice Model for Motorcycle Transport with Latent Motivation Variables” and will be submitted to Journal of Communications - Scientific Letters of the University of Žilina.

(33)

21

CHAPTER 6. CONCLUSIONS

Findings from the first analysis “Modeling Travel Mode for Motorcycle and BRT-lite”

summarizes as follows:

a. The determinants factors of selecting motorcycle as representative daily mode in Banda Aceh consist of travel destination, travel time, vehicle operating costs, gender, education level, motorcycle ownership, and driver’s license ownership.

b. The probability of modal split among mode choice set obtained using binary logit are 70%, 30%, for Motorcycle and BRT-lite, respectively.

c. The derived model is considered very good based on discrepancy among observed and predicted value about 0.136 and 0.367 for Motorcycle and BRT- lite, respectively.

Finding from the second part analysis “Modeling Motivation Factors Affecting Daily Mode Choice” listed several conclusions:

a. As for mandatory activity modeling calibration results using CFA with the highest significant on the variable latent linkage private vehicle (LPV) indicator Likes to drive private vehicles (LPV2).

b. Further investigation depicts that as for non-mandatory activities obtained the highest significant on latent variable private vehicle problems (PVP) indicator congestion is increasing (PVP2).

c. The measurement model show that mandatory activities have the highest significant correlation with the attribute of liking to drive private vehicles (car/motorcycle), while non-mandatory activities have the most significant sign congestion increasing within the city due to highly autos dependency.

d. Further finding depict that as for mandatory activities appear that linkage with private vehicles significantly affects private vehicle problem (congestion and environment deterioration), while in non-mandatory activities the quality of service of current bus system has significantly affects satisfaction on using public transport.

(34)

22

REFERENCES

Abou-Zeid, M., Schmöcker, J.D., Belgiawan, P.F., et al., 2013. Mass effects and mobility decisions. Transp. Lett. 5 (3), pp.115–130.

Atasoy, B., Glerum, A., Bierlaire, M. 2013. Attitudes towards mode choice in Switzerland.

disP–The Panning Review, 49 (2), pp.101–117.

Barthelemy, M. 2016. A global take on congestion in urban areas. Environment and Planning B: Planning and Design, 43(5), pp.800–804.

Bierlaire, M., Hurtubia, R., Flötteröd, G., 2010. Analysis of implicit choice set generation using a constrained Multinomial Logit Model. J. Transp. Res. Rec. 2175, pp. 92–97.

Beirão, G., Sarsfield Cabral, J.A. 2007. Understanding attitudes towards public transport and private car: A qualitative study. Transport Policy, 14(6), pp. 478–489.

Bolduc, D., Boucher, N., Álvarez-Daziano, R. 2008. Hybrid choice modelling of new technologies for car choice in Canada. J. Transp. Res. Rec. 2082, pp. 63–71

Bollen, K.A. 1989. Structural equations with latent variables. John Wiley & Sons, New York, USA.

Ernst, P.J. 2005. Initiating Bus Rapid Transit in Jakarta, Indonesia. Journal of the Transportation Research Record, 1903, pp. 20-26.

Greene, H. W., and Henser, A. D. (2010). Modeling ordered choices: A primer, Cambridge University Press, Cambridge, U.K.

Irvansyah, R. Sugiarto, S. Achmad, A. Fahlevi, H. 2020. Analysis of the Trans Koetaradja bus services considering latent variables of bus line services. IOP Conference Series:

Materials Science and Engineering,917, pp. 012036.

Irza, M., Sugiarto, S., Saleh, S.M. 2020. Mode choice analysis among motorcycle and Trans Koetaradja urban bus and its contributing factors using revealed preference (RP) data.

IOP Conference Series: Materials Science and Engineering, article in press.

Joreskog, G.K., Goldberger, S.A., 1975. Estimation of a model with multiple indicators and multiple causes of a single latent variable. J. Am. Stat. Assoc. 70(351), pp.631–639.

Maness, M., Cirillo, C., Dugundji, E.R. 2015. Generalized behavioral framework for choice models of social influence: Behavioral and data concerns in travel behavior. Journal of Transport Geography, 46, 137–150.

(35)

23 Maness, M., Cirillo, C. 2016. An indirect latent informational conformity social influence choice model: Formulation and case study. Transportation Research Part B:

Methodological, 93, 75–101.

Mugion, R.G., Toni, M., Raharjo, H., et al., 2018. Does the service quality of urban public transport enhance sustainable mobility? J. Cleaner Prod. 174, pp.1566–1587.

Morikawa, T., Ben-Akiva, M., McFadden, D., 2002. Discrete choice models incorporating revealed preferences and psychometric data. Econometric Models in Marketing, 16, pp.

27–53.

Ramming, M.S. 2002. Network Knowledge and Route Choice. Doctoral Dissertation, Massachusetts Institute of Technology, USA.

Raveau, S., Álvarez-Daziano, R., Yáñez, M.F., Bolduc, D., Ortúzar, J.de D., 2010.

Sequential and simultaneous estimation of Hybrid Discrete Choice models– some new findings. J. Transp. Res. Rec. 2156, pp. 131–139.

Redman, L., Friman, M., Gärling, T., et al., 2013. Quality attributes of public transport that attract car users: A research review. Transp. Policy 25, pp.119–127.

Safitri, E. Sugiarto, S. Anggraini, R. Achmad, A. Fahlevi, H. 2020. Analysis on the effect of socio-economic and travel attributes to perceptions of the Trans Koetaradja quality of services. IOP Conference Series: Materials Science and Engineering, 917, pp. 012036.

Saleh, S.M., Sugiarto, S., Hilal, A., Ariansyah, D. 2017. A study on the traffic impact of the road corridors due to flyover construction at Surabaya intersection, Banda Aceh of Indonesia. AIP Conference Proceedings. AIP Conf. Proc. 1903, pp. 060005.

Saleh, S.M. Sugiarto, S. Anggraini, R. 2019a. Analysis on public’s response toward bus reform policy in Indonesia considering latent variables. Open Transportation Journal, 13(1), pp. 17-24.

Saleh, S.M. Sugiarto, S. Salmannur, A. 2019b. Attitudinal dataset for mediating the effects of public acceptance on bus reform scheme in a developing country context. Data in Brief, 25, pp. 104035.

Sugiarto, S. Miwa, T. Sato, H. Morikawa, T. 2017a. Explaining differences in acceptance determinants toward congestion charging policies in Indonesia and Japan. Journal of Urban Planning and Development, 143(2), pp. 04016033.

Sugiarto, S. Miwa, T. Morikawa, T. 2017b. Inclusion of latent constructs in utilitarian resource allocation model for analyzing revenue spending options in congestion charging policy. Transportation Research Part A: Policy and Practice, 103, pp. 36-53.

(36)

24 Sugiarto, S. Saleh, S.M. Anggraini, R. Merfazi, M. 2019. Investigating public perceptions and its implication toward Trans Koetaradja policy considering latent motivation. IOP Conference Series: Materials Science and Engineering, 523(1), pp. 012036.

Sugiarto, S. Miwa, T. Morikawa, T. 2020. The tendency of public's attitudes to evaluate urban congestion charging policy in Asian megacity perspective: Case a study in Jakarta, Indonesia. Case Studies on Transport Policy, 8(1), pp. 143-152.

Susilo, Y.O., T.B. Joewono, W. Santosa, D. Parikesit. 2007. A reflection of motorization and public transport in Jakarta Metropolitan Area. IATSS Research, 31(1), pp.59–68.

Walker, J., Ben-Akiva, M., 2002. Generalized random utility model. Math. Soc. Sci. 43, 303–343.

Yanez, M.F., Raveau, S., Ortuzar, J.de D. 2010. Inclusion of latent variables in Mixed Logit models: modelling and forecasting. Transp. Res. Part A, 44, pp.744–753.

Zhao, L.N., Wang, W., Hu, X.J., Ji, Y.J. 2013. The importance of residents’ attitude towards service quality in travel choice of public transit. Procedia – Social and Behavioral Sciences, 96, pp. 218–230.

(37)

25

Appendix 1. Documentation of research activities

o Review Metode statistik dan ekonometrik terutama tentang pemodelan pemilan moda dengan menggunakan model-model pemilihan diskrit.

o Pendalaman penggunaan software Stata 16 dan Nlogit 5 untuk pemodelan pemilahan moda.

(38)

26 Pengumpulan data primer yang dibantu oleh mahasiswa yang terlibat dalam penelitian ini.

(39)

27 Surveyor/mahasiswa yang terlibat penelitian sedang melaksanakan interview/survey data

primer penelitian.

(40)

28 Menganalisis data dan mengkalibrasi parameter model dengan software Nlogit 5 dengan

menggunakan model binomial logit.

(41)

29

Appendix 2. Acceptance letter of manuscript

(42)

30

(43)

31

Appendix 4. Accepted Manuscript for Scopus Proceedings and Draft of

Journal.

(44)

Choice Model and Influencing Factors of the Travel Mode for Motorcycle and BRT-lite in Banda Aceh, Indonesia

Sugiarto Sugiarto

1,*)

, Miftahul Jannah Huta Barat

1)

, Sofyan M. Saleh

1)

, Ashfa Achmad

2)

, Irham Iskandar

3)

1Department of Civil Engineering, Universitas Syiah Kuala, Darussalam, Banda Aceh 23111, Indonesia

2Department of Architecture and Planning, Universitas Syiah Kuala, Banda Aceh 23111, Indonesia

3Regional Development Planning Agency, BAPPEDA Aceh, Banda Aceh, Indonesia, Banda Aceh 23111, Indonesia

1Corresponding author: sugiarto@unsyiah.ac.id

1mjhb1606@gmail.com; sofyan.saleh@unsyiah.ac.id; 2ashfa.achmad@unsyiah.ac.id

3irham.iskandar84@gmail.com

Abstract. The City of Banda Aceh continues to experience rapid development both in terms of economy, infrastructure, and transportation which has resulted in increased autos mobilities. This increased mobility has an impact on the increasing tendency to use motorcycles which harm the environment for instances traffic congestion, air pollution, noise, and accidents. The Government of Aceh has implemented a policy called Bus Trans Koetaradja with the concept of Bus Rapid Transit-Lite (BRT-lite), but due to inaccuracy and lack of exclusive lanes, it has caused people to be reluctant to shift from private mode to the urban Bus. Therefore, this aiming at investigates the determinants of shifting travel mode choice from motorcycle to BRT-lite based travel survey conducted in Banda Aceh using the contexts of Stated Preference. Due to pandemic Covid-19, the data was collected using online-based questionnaire survey using Google form. A total of 400 data is valid and used for the analysis of travel mode choice using Binary Logistic approach. The results of the binary logistic regression analysis reveal that determinants of selecting motorcycle as a representative daily mode used is the distance to make a trip, travel time within origin-destination, and vehicle operating costs. The probability of selecting travel modes is 70% and 30% for motorcycle and BRT-lite, respectively.

Keywords: BRT-lite, Motorcycle, Mode choice, Binary logistic, Banda Aceh.

INTRODUCTION

The development of a country is always in line with the development of its transportation services due to its significant role in improving the country's social and economic. In developing countries, the use of motorcycles continues to increase every year which significantly manifested congestion-related problem such economic loss, increasing travel time, and additional fuel consumption [1]. One of promising strategy to reduce the effects of motorcycle use is to promote the use of public transportation such as initiating Bus Rapid Transit especially in Indonesia [2] and Malaysia [3].

In case of Banda Aceh city, the Trans Koetaradja, a type of urban bus reform, it has been reported as an affective policy that able to reduce congestion within city center [4] but people to be reluctant to shift their mode to Trans Koetaradja due to inflexibility route [5]. Furthermore, the current Trans Koetaradja is operated within type C of public transport. That is operated within mixed traffic and there is no exclusive lane assigned for this bus system. Therefore, the Government has considered to improve the current bus into type B of public transport system. Within type B public transport system, the bus will have an exclusive lane and it refers to BRT-line [6].

Referensi

Dokumen terkait

Masing-masing konveyor digerakan oleh 1 motor arus searah tipe CSD80- D Penyearah Terkontrol dengan topologi penyearah gelombang penuh 1 fasa setengah terkontrol berhasil

Gambar 4.27 merupakan peta kendali nilai efisiensi Pembakaran pada Boiler tanggal 15 Februari – 31 Maret 2010, gambar tersebut memberikan informasi bahwa pada

Pada percobaan ini peran pNP-palmitat sebagai substrat dapat digantikan oleh substrat lain yang bias digunakan untuk menentukan aktivitas dari enzim lipase, misalnya olive oil..

Kerangka pemanfaatan hasil yang diharapkan dilakukan oleh institusi terkait di daerah ialah : (1) Mengenalkan penggunnaan Feromon Exi pada budidaya bawang merah, dan (2)

bahwa prinsip nilai-nilai moral tidak hanya didapatkan oleh peserta didik melalui sosialisasi atau pelajaran di sekolah, tetapi juga melalui interaksi sosial mereka

Each website users have different characteristics. The development of Website Personalization Based on Adaptive Hypermedia System is one of the ways to make website

Rencana Pembangunan Jangka Panjang Daerah Provinsi Bengkulu Tahun 2OA?2A25 yang selanjutnya disingkat RPJPD adalah dokumen perencanaan daerah untuk periode 2O (dua

Hasil analisis ragam menunjukkan perlakuan penambahan kompos baru campuran jerami padi dan kiapu pada media sisa pemberian periode tanam pertama berpengaruh nyata terhadap