Mode Choice. Xxxxxxxx et al. (2017) presented individuals with various scenarios and asked them to choose the car they would use for their commute based on the characteristics of their current commutes. A vehicle choice model which includes three options is estimated: continue use a regular car; buy and shift to a privately-owned AV, or shift to SAV. Five latent variables were identified based on factor analysis: technology interest, environmental concern, enjoy driving, public transit attitude, and pro-AV sentiments. Only three of these factors played a significant role in estimating the choice decision: enjoy driving, environmental concern, and pro-AV attitude. The effects of the attitudinal variables were very significant and could be influenced by educational campaigns. Xxxxxxxxxx et al. (2017) asked respondents to choose the most preferable vehicle option to purchase among the four described alternatives: non-automated gasoline vehicle, non-automated electric vehicle, automated gasoline vehicle, and automated electric vehicle. Young adults, well- educated and tech-savvy respondents, those with high annual VMT and those who have long- distance work trips, are found to be more willing to choose automated and electric automated options. Table 4 presents a summary of findings focusing on mode choice. Table 4 Summary of Literature on Mode Choice Findings Detail Reference Regular Cars Positive Impact Escorting trip Xxxxxxxx et al. (2017) Age, Seniors Xxxxxxxxxx et al. (2017) Licensed drivers Enjoy driving Xxxxxxxx et al. (2017) Do not telecommute Shabanpour et al. (2017) Negative Impact Education Experienced accidents Residential location, downtown Residential location, suburban (work in the city) Parking price Xxxxxxxx et al. (2017) PAVs Positive Impact Distance traveled (Travel time) Xxxxxxxx et al. (2017), Pro-AV attitude Experienced accidents Trip cost (If trip cost PAV < Regular car) HH income Education Negative Impact Purchase price (If Purchase price > regular car) Trip cost (If trip cost PAV > Regular car) Xxxxxxxxxx et al. (2017) Age, seniors Xxxxxxxx et al. (2017) SAVs Positive Impact Distance traveled (Travel time) Pro-AV attitude Environmental concern Negative Impact Subscription cost Age, seniors Commute frequency Number of young children in HH Trip cost (If trip cost SAV < Regular car)
Mode Choice. For mode choice, two of the most commonly used models are the multinomial logit and nested- logit models (Xxx-Xxxxx and Xxxxxx, 1985). Similar to logit-based destination models, a multinomial logit mode-choice model hypothesizes that the probability for an individual to choose a particular mode for a trip of certain purpose depends on the ratio of the mode’s utility, expressed as a function of the mode’s characteristics (e.g., availability at the origin TAZ, travel time, cost, and convenience for the trip purpose), to the sum of utilities of all competing modes. The other commonly used mode-choice model, nested logit model, hypothesizes a nested structure, in which mode choice alternatives that share similarities are pooled together. The process of choosing a mode for a particular trip purpose is represented as a multistep decision. The probability of choosing an alternative within its nest of similar alternatives is given by the ratio of the mode’s utility to the sum of utilities of all alternative modes within the same nest. The probability of choosing a nest against other nests depends on the ratio of the nest’s utility, which is expressed as a composite of utilities of all alternatives within the same nest, to the sum of composite utilities of all nests. In current travel demand modeling practice, the use of either a multinomial or nested logit model is considered acceptable practice in all regions (VDOT, 2014). However, because transportation mode choices do exist in nested structures, the use of nested logit models is the most common practice. According to NCHRP Synthesis 514, 14 states were identified as using nested logit model, compared to only two states using multinomial logit (NASEM, 2017).
Mode Choice. The Missoula Model includes a mode choice component that separates the person trip tables into the drive alone, shared ride (i.e., carpool), transit (walk access and drive access), and non-motorized (bicycle and walk) modes. Information about transit routes and the quality of bicycle and pedestrian facilities provides important input to the mode choice model. The mode choice model also considers trip lengths produced by the gravity model, resulting in sensitivity to higher density and mixed-use areas. Such areas will produce shorter trips that are more likely to be made using non-motorized modes. The Missoula Model mode choice is a nested logit model and no modifications will be conducted to the structure of this model. The 2010 Missoula mode choice component was calibrated to reproduce observed mode shares. The observed mode share for transit is based on the number of boardings from Mountain Line’s Automatic Passenger Counts (APC) data whereas the non-motorized shares were obtained from the 2000 CTPP. No observed data or data from the Census were available during the 2014 model update; however, 2010 CTPP data are available now and the CTPP data for 2012–2016 was released toward the end of 2018. LSA will review these two CTPP datasets to develop the mode share targets for the home-based work trips in the model as CTPP data are only available for work trips. Pivot point analysis similar to previous model updates will be conducted to develop mode share targets for the other trip purposes in the model. The 2018 average daily transit boarding will be obtained from Mountain Line and will be used for the transit mode share calibration. The percentage distribution of transit trips by trip purpose will be based on latest CTPP data and distributions in the 2014 and 2010 models. LSA anticipates modifications to alternative specific constants during the mode choice calibration. Other attributes such as mode choice coefficients, value of times, and any of the cost variables will be reviewed and updated accordingly during the calibration effort.
Mode Choice. The Missoula Model includes a mode choice component that separates the person trip tables into the drive alone, shared ride (i.e., carpool), transit (walk access and drive access), and non-motorized (bicycle and walk) modes. The Missoula mode choice model is a nested logit model. LSA will update mode choice parameters of the model as the MPO does not anticipate any changes to the structure of the mode choice model. The observed mode share for transit will be based on the number of boardings from Mountain Line’s Automatic Passenger Counts (APC) data, whereas the non-motorized shares will be obtained from the most recent CTPP data. LSA will review CTPP and 2017 National Household Travel Survey (NHTS) datasets to develop the mode share targets for the home-based work trips in the model as CTPP data are only available for work trips. Pivot point analysis similar to previous model updates will be conducted to develop mode share targets for the other trip purposes in the model. LSA will make modifications to alternative specific constants during the mode choice calibration. Other attributes such as mode choice coefficients, value of times, and any of the cost variables will be reviewed and updated accordingly during the calibration effort.
