Research On Activity-Travel Decision Behavior From The Perspective Of Heterogeneity

With the rapid development of economy in China,its urbanization is growing faster than ever before,whilst mobility demand within urban area increases exponentially as a result of the enhancement of daily activities.Furthermore,people‘s demand for private vehicles also grows significantly.However,the current road capacity cannot effectively satisfy these mobility demands since the urban geographical area is rather limited.All the above-mentioned issues contribute to the congestion and chaos of urban transportation system,as well as the level of environment pollution,which will further exert a negative influence on the residents‘ quality of life and sustainable development of the whole society.In order to keep these problems from getting worse,one of the most durable and effective measures corresponds to leading people‘s activity-travel behavior to become more beneficial to the road,transportation and city.To guarantee the feasibility of this measure,it is of great importance to get a comprehensive understanding of individual activity-travel decision making process.Only by realizing the critical factors that have significant influences on these behavioral decisions would it be possible to design specific travel demand management measures and strategies.Notably,individual‘s activity-travel decision behavior is closely associated with heterogeneity.On the one hand,objective socio-economic characteristics and alternative attributes cannot make a reasonable explanation on the choice behavior,whereas the latent variables capturing differences in individual attitude,perception,and preference reserved in the behavioral ―black box‖ are expected to provide a deeper insight into the choice behavior.On the other hand,decision behavior of different population groups might be influenced by completely distinctive factors,and even the same effecting factor might play different roles across population groups.Taking all the above arguments into account,this dissertation research investigates several important points in the analysis of activity-travel decision behavior that are strongly related to heterogeneity,which are found to be rather limited in the transportation research arena of our country.The empirical applications on the Household Travel Survey data are expected to gain more knowledge about the activity-travel decision-making process,and provide some valuable suggestions to the development,implementation,and evaluation of Travel Demand Management strategies.Specifically speaking,the main research contents and conclusions are summarized as follows:(1)The latent psychological factors reflecting individual preference heterogeneity are directly incorporated into an integrated Structural Equation Model(SEM)to simultaneously analyze five activity-travel behavioral dimensions,including three activity time allocation decisions(i.e.,subsistence,maintenance,and recreation)and two commuting choices(i.e.,commuting mode and departure time).The two mode-specific preference factors are confirmed to have significant effects on the five choice dimensions,demonstrating that the incorporation of attitudes and perceptions enhances the behavioral representation by providing insight into the choice process underlying the ―black box‖.The pro-car factor is found to have a positive effect on the maintenance activity duration and a negative effect on subsistence activity duration,while those with greater preference for bus spend less time on recreation activity.Besides,both the preference factors have positive influences on the choice of corresponding mode.The inter-dependencies between multiple activity-travel choices are also explicitly recognized,which confirms the importance of integrating various behavioral dimensions as a ―bundle‖.The choices of which mode to choose and how long to spend on recreation activities,are obviously conditional on the subsistence activity duration.The commute mode choice determines,to some extent,the engagement in maintenance activity.Due to ―time poverty‖,individuals tend to schedule daily activities according to the sequence as subsistence,maintenance,and recreation based on their priority.(3)The MNP-kernel ICLV model is applied to empirically explore commuting mode choice,which is estimated by the maximum approximate composite marginal likelihood(MACML)approach.It is found that commuters with different socio-economic characteristics have comparable attitudes and perceptions with various transport modes,which are confirmed to not only significantly influence choice of specific mode,but have an importance effect on the other modes.For instance,public transit aversers are not willing to use this mode,and they are more likely to commute by car.Those with positive attitudes towards electric bike have a greater propensity to choose it,while bus and car are less likely to be their commuting modes.(4)In addition to directly using latent variables to account for heterogeneity,a three-stage approach,i.e.,factor extraction-segmentation-multi-group SEM(MSEM),is designed to explore the commonalities and diversities with respect to differences on continuous activity-travel durations across segments.As being suggested by their labels,segments of bus addicts,car individualists,and e-bike enthusiasts show a positive preference for bus,private car,and electric bike,respectively;while negatively evaluate the other two modes.Multiple descriptive comparisons confirm that each segment also possesses unique socio-demographics and time-allocation pattern.Finally,results of the MSEM find that time allocated to subsistence,maintenance and recreation activities,as well as the total travel time across distinctive segments are influenced by different factors,or by the same factor with varying effect magnitudes.This directly validates the unique mechanism of activity-travel time allocation with respect to each segment.(5)Another important segmentation approach,latent class model(LCM),is applied to classify the sample travelers on a probabilistic basis,and analyze the discrete commuting mode.In addition to the objective travel time and cost,four mode-specific attributes including levels of comfort,safety,convenience,and cost as perceived and evaluated by respondents are also incorporated in the choice model.Three segments with heterogeneous preferences of various modes are derived.E-bike enthusiasts have a strong baseline inclination for electric bike,with comfort and safety being the most and least important considerations for them,respectively,when making commuting mode decision.Car individualists pay the most attention to comfort,while they are less sensitive to cost due to their financial flexibility.Bus addicts place high value on three out of the four perceived attributes,except for the perceived cost.Such outcome might be understood as that bus addicts have no choice but to use bus,and they cannot do anything about this.On the other hand,as illustrated by the coefficient of travel time,this segment is time-insensitive,and they are expected to enjoy the journey by bus,which might be considered as the reason for their concern about comfort,convenience,and safety.Both e-bike enthusiasts and car individualists are time-sensitive as reflected by the negative coefficients of travel time.(6)The above two aspects reflecting heterogeneity,i.e.,latent variable and segmentation,are simultaneously utilized to build an advanced model integrating the expanded theory of planned behavior(TPB)and the LCM,with aim to comprehensively investigate the heterogeneity within commuting mode decisions.A cognition-intention-behavior hierarchy in the choice model enhances the behavioral representation from the perspective of latent variable,while at the same time a set of mode-specific habits are employed as segmentation covariates in the class model.This integrated model demonstrates that for commuters with diverse psychological profile(here,mode use habits),the same perception factor plays a relatively different role in their commuting mode choice.More importantly,the mechanism of using decision regarding different modes is significantly different,which should be identified on the basis of traveler‘s personal psychological features.For individuals with strong habits of using electric bike or car,commuting by driving a car or riding a bus are more likely to be a habitual behavior under a certain situation.Comparatively,for habitual bus users,commuting by driving or riding may be,to a greater extent,the decision with a deliberate process.This section finalizes the moderating effect that habit plays in the mode choice decicion-making process.The main contribution of this dissertation is that it systematically investigates the heterogeneity of people‘s activity-travel decisions.Correspondingly,the framework to analyze mobility decisions is improved from two aspects,i.e.,the inclusion of latent variables and segmentation,based on which a set of discrete and continuous behaviors are modeled.Specifically speaking,(1)Heterogeneity is further elaborated,i.e.,preference heterogeneity and response heterogeneity,based on which a comprehensive activity-travel analysis framework is proposed.(2)Multiple discrete-continuous activity-travel decisions are treated as a highly-integrated bundle.By incorporating the influences from preference heterogeneity associated with latent variables,the correlations between activities,as well as between activity and travel are investigated.(3)Latent variables reflecting preference heterogeneity is directly incorporated into the discrete choice model,by which the ICLV model is obtained to analyze the discrete mode choice.(4)Posterior segmentation is conducted based on the combination of latent variables.Thus,the heterogeneous responses with respect to activity-travel time allocation decisions are revealed using a three-step approach.(5)The TPB and LCM are integrated,where latent psychological factors are included as variables in both choice model and class model.The mechanism regarding transport modes using decisions of each segment is finally identified.