High-Speed Rail in the US—Mode Choice Decision and Impact of COVID-19

Abstract

While high-speed rail (HSR) has achieved success in major cities in Europe and Asia, it is a new phenomenon in the US, and few studies on HSR in the US are available, especially from the users’ perspective. This study aims to fill the research gap by investigating the mode choice behavior in the Los Angeles and San Francisco corridor where HSR may soon become a feasible option. The impact of COVID-19 was also examined with regard to how people view modes of domestic travel and how their view may change. The geographic locations of travelers and the possible HSR characteristics in the US were also explored. Survey data of US travelers was collected on MTurk, which was analyzed using logistics regression and Two-Way MANOVA. The results indicated that convenience in transport, travel frequency, gender, mobility issues, income, and total travel time were determinants in the choice between HSR and air service, while travel frequency and total travel time were important in the choice between HSR and car transport. Most US travelers changed their views following COVID-19 in terms of domestic travel and exhibited a higher intention to travel by train and HSR. Geographic patterns were identified, such as people in the southern US were the most knowledgeable of HSR and had the greatest intention to use HSR, while people in the northeast exhibited the lowest intention. The findings indicate potential interest in HSR among US travelers, and offer much-needed empirical evidence for the potential success of HSR in the US.

1. Introduction

High-speed rail (HSR) is broadly defined as a railway system with an average commercial speed of 155 miles/h [1]. Since it began operating in Japan in the 1960s, HSR has enjoyed a global expansion. Today, around 35,000 miles of HSR lines are in operation globally, transporting over three billion passengers a year [2]. Despite the successful international experience, HSR has achieved little progress in the US. Currently, Amtrak Acela Express serving the Northeast Corridor (NEC) is arguably the only HSR in the US. While the train can reach a maximum speed of 150 miles/h on some sections, the average speed remains below the HSR criterion due to infrastructure constraints [3].

Likely due to low HSR penetration in the US, the literature that addresses HSR in the US is almost exclusively at the policy level, focusing primarily on the advantages and disadvantages of HSR development in the country. Opponents of HSR often cite low population density, long distances between major cities, and inadequate investment in rail infrastructure as concerns for the economic viability of HSR in the US [4]. The strong car and airplane culture in the country, driven partially by the prevalence of individualism and antistatism is also argued to hinder HSR acceptance in the US [5,6]. In recent years, however, there has been renewed discussion of HSR in the US. The decline in car ownership among young Americans, coupled with the trend of relocation to urban areas where travelers of different ages can easily find mobility alternatives, may re-stimulate the interest in HSR [7]. Supporters of HSR also see promising opportunities in the US to shift from polluted cars and airplanes to cleaner HSR, which promotes sustainability of the transport system [6]. With the introduction of HSR in the US, traffic congestion in busy corridors is likely to decrease, the dependency on foreign oil would be reduced, the environment could be further protected, and a more balanced, multimodal transport system could be developed with improved service quality and efficiency [6,8]. This long-term vision in transportation has set the stage for meaningful development of HSR in the US. Encouraging progress has already been made, thanks to recent political and financial commitments to accelerate HSR development. The HSR project in California, for example, has drawn large investments, contributing significantly to HSR progress and the local economy [9,10,11].

While most research on HSR in the US has been conducted at the policy level, few studies discuss the user’s choice toward HSR. The current literature has identified factors that underlie the choice of HSR in matured HSR countries, citing travel time and convenience, among others, as the main motivators in the short- and medium-haul markets [12,13,14,15]. Only one early study has investigated HSR in the US from the passengers’ perspective, but the focus was on the behavioral intention to use HSR rather than on the choice of HSR over alternative transport modes [16]. In the US, the transport market is characterized by well-established car and air services, and travelers’ selection of HSR should be understood in the context of intermodal choice, where travelers evaluate all possible transport options. Research focusing on intermodal choice involving HSR is, therefore, needed in the US to not only fill the research gap but to also provide empirical evidence for HSR development in busy transport corridors.

Another meaningful topic in the research on HSR is the impact of COVID-19. Unlike in some countries where HSR is a well-accepted travel option, HSR has only received limited attention in the US. The initial stage of HSR development in the US means market entry of HSR could be affected by COVID-19, a public health crisis that fundamentally changed domestic transport and travel behaviors. An important question to ask is whether COVID-19 presents an opportunity for the success of HSR. Specifically, do travelers in the US view HSR differently given their COVID-19 experience, and have the travel- and HSR-related characteristics in the US changed in the post-pandemic era. This study aimed to examine the intermodal choice involving HSR in the US market and understand how demographic and COVID-19 factors influenced the knowledge and intentions toward HSR in the US. Two research questions were used to guide the study: (1) What demographic, travel, and HSR factors are important for travelers to choose HSR over air and car transport in high-demand markets in the US if HSR becomes a feasible travel option. (2) Do travelers view domestic travel differently following the COVID-19 pandemic, and if so, do their new views and their habitual residence affect their travel habits, knowledge of HSR, likelihood to use trains, and intention to use HSR in the post-pandemic era.

2. Literature Review

2.1. HSR in the US

Since the first HSR operation in 1964, HSR has become a global phenomenon, with rapid development, mostly in major cities in Europe and Asia. China has led HSR development, accounting for nearly two-thirds of the world’s HSR lines [17]. While the discussion of HSR in the US can be traced back to the 1960s, it has yet to materialize into meaningful progress. At the time of writing this article, Amtrak’s Acela Express connecting Washington DC and Boston, with an hourly speed close to 155 mph on some sections of the rail lines, remains the only rail service in the US that is the closest to the functioning HSR service [4]. In fact, HSR has long been a controversial topic in the US due to its unique characteristics. HSR performs the best on short- and medium-distance routes, which typically go through densely populated, high-demand economic centers. Opponents of HSR argue that metropolitan areas of this type are uncommon in the US, as cities typically grow in a sprawling pattern in the country [7]. Another common reason for not favoring HSR is economic viability. Some question the substantial investment in infrastructures for HSR development and maintenance, which often make cost-covering difficult [18]. Social norms may also play a role in HSR stagnation in the US. Long embracing car and air travel, likely due to the cultural influence, US travelers may simply lack the motivation to seek alternative travel options [5]. Recent years, however, have seen new opportunities for HSR in the US, driven primarily by rapid urbanization, changing mobility habit, and environmental benefits of HSR. On-going urbanization in the US is a result of urban population expansion, a phenomenon that has been witnessed in many countries around the world [19]. From 2001 to 2011, the US experienced a growth rate of 11% in urbanization, with the south and west of the country leading population growth [20]. Continuous growth in population, economic activities, and inter-city mobility provide a foundation for HSR success, which, in turn, can reshape the urban transportation systems [21]. The past decade in the US has also seen changing mobility patterns, mostly in young and senior populations. These two demographic groups have shown greater interest in settling in urban areas where they can depend less on cars and use public transport more often [7], which could make HSR a preferred option for these people. Economic development, especially at the regional level, may also drive the development of HSR in the US. A positive relationship between the tourist industry and HSR has been documented. Studies suggest that the central location of the HSR station is an enabling factor for the promotion and growth of the tourist market, and HSR-air cooperation can encourage the arrival of foreign tourists [22]. For busy markets like SF-LA, where hospitality and tourism represent an important source of revenue for local businesses and government, there can be a high incentive to develop HSR. Finally, the electric-powered HSR is considered a less polluting way to travel compared to cars and airplanes. The environmental benefits of HSR align with the long-term goal of establishing a sustainable transport system in the US, which can further promote HSR development.

Clear research gaps can be seen regarding HSR in the US. First, prior studies of HSR in the US focused almost exclusively on the economic viability, challenges, and opportunities of developing HSR, with few studies considering the perspective of US travelers. The success of HSR depends not only on HSR infrastructures and operations but, more importantly, on public acceptance and the decision to use HSR. Prior studies have recognized the value of engaging the public opinion at the initial stage of HSR development and demonstrated that well-implemented public participation can save costs for HSR development [23]. Therefore, research at a micro level focusing on individual US travelers and their interest and choice regarding HSR is needed to fill the research gap. Second, unlike many other countries, transportation in the US is supported primarily by cars (for short distances) and airplanes (for long distances). The mode decision of travelers must be examined when multiple travel options, including HSR, are available, and the factors that are important in their intermodal choice need to be considered, especially in high-demand markets like San Francisco (SA) and Los Angeles (LA) where different travel options are available. While Gehrt [16] investigated HSR use in the US, the study was conducted when policy and the economic environment for HSR development was vastly different. The study focused only on the intention to use HSR, which ignored the fact that mode decision is based on an evaluation of all available transport modes. To our knowledge, no prior study has examined travelers’ intermodal choices involving HSR in the US. The findings of this study will provide much-needed empirical evidence for the viability of HSR service in the country. Moreover, with the renewed discussion of HSR in the US, the potential impact of COVID-19 should be examined for its effect on views about domestic travel, including the use of new transport modes like HSR. For example, has the COVID-19 experience led to an increased interest and intention toward HSR, especially when the geographic factor is considered? Empirical research on the impact of COVID-19 and the use of HSR in the US is needed to answer this important question.

2.2. The Choice toward HSR

Much has been studied regarding air–rail competition and coping strategies in the changing transport markets [24,25,26]. At the micro level, passenger behaviors in the HSR context have been frequently examined, focusing primarily on the choice between HSR and air transport (FSCs) [12,27] and HSR and private cars [28], with findings indicating that travel time, travel cost, convenience, safety, and demographic characteristics, among other factors, were important in passengers’ intermodal choices. In this study, respondents were given a scenario of traveling between LA and SF, a busy corridor that will potentially be served by HSR. As HSR remains a new phenomenon in the US, this study selected relevant factors in the context of mode choice based on the HSR literature. Specifically, this study examined whether total travel time, convenience in transport, travel frequency, mobility issues, and traveler demographics would influence travelers’ mode choice of air, car, and HSR in the LA-SF market. The remainder of Section 2.2 justifies the factor selection for this study.

Travel time is a key motivator to choose one transport mode over the other [29]. Studies showed that travel time was an important factor in increasing HSR ridership [30] and affecting mode shift toward HSR [31]. In this study, the medium-distance LA-SF corridor was used as the travel scenario, based on which respondents were asked to choose from air service (1.5 h airport-to-airport time), HSR (less than 3 h station-to-station time), and car (about 6 h) for the trip. While time spent in the vehicle is important, it is only reasonable to also consider ground access and egress when the impact of travel time is examined (total travel time). Fu et al. [32] argued that, compared to air service, HSR enjoys an advantage in terms of total travel time because airline passengers often need a much longer time for airport procedures, which can increase the total time these passengers spend on the entire trip. Furthermore, the centrally located rail stations and easy ground access can reduce the total travel time of HSR, bringing benefits to HSR passengers [32]. Studies in Europe have shown that total travel time affected the market share of HSR and significantly influenced passengers’ choice of HSR [12,15]. Compared to station-to-station time, total travel time can better reflect the time-saving benefit of HSR. Considering total travel time can be particularly important in the US given its geographic characteristics [33]. Noticeably, when total travel time is considered, the three transport modes, especially HSR and air service, offer competitive travel time for passenger evaluation. This study, thus, included total travel time as a predictor of the intermodal choice in the LA-SF market.

HSR is typically located near the city center with well-connected surface transport, allowing quick access to the train station [32]. The convenience of using HSR can increase passenger satisfaction and influence the decision to choose HSR [34]. Studies showed that travelers may be more concerned with convenience, reliability, and door-to-door time than station-to-station time in making the decision to use HSR [35]. Similar findings were obtained in the US, suggesting that the growth in ridership of Amtrak may be partially driven by convenience in train transportation [7]. The convenience in using train transport stands in sharp contrast to the mixed experience in air travel. As air travel has been increasingly affected by inconvenient airport access, traffic congestion, and tightened security at airports, airlines may lose passengers to other transport modes including HSR [7]. This study, thus, added convenience in transport to the predictive model.

Traveler demographics and travel characteristics were added to the predicting model given their possible influence on the mode choice decision. Studies show that demographic characteristics such as gender, income, and education can be determinants of the choice of HSR [34,36,37]. Some studies indicate social inequalities that shape the uneven use of HSR. For example, Dobruszkes et al. [38] suggested that HSR passengers were predominantly male, higher income, highly educated, and belonging to higher social occupational groups. Gender appears to be a particularly relevant factor in the selection of transport mode. Female travelers, compared to their male counterparts, are more likely to use public transport including rail and HSR [39,40]. Travel frequency is another likely determinant of intermodal choice, as previous studies indicated the significant impact of bus ride frequency on mode choice behaviors [41]. However, the impact of travel frequency on the mode decision involving HSR remains uninvestigated, making it an interesting predictor for the mode choice behavior in this study. A factor that can be important in HSR use in the US is the mobility of travelers, given that one in four of adults in the United States have some type of disability [42]. It is likely that travelers who are disabled or have limited mobility may prefer certain types of transport modes. Recent studies showed that disabled people are more likely to use a bus and taxi, whereas non-disabled people use rail more often [43]. Thus, mobility issues were added as a predictor in the model.

2.3. HSR and the Impact of COVID-19

The COVID-19 pandemic has had a profound impact on the transport industry. While much attention has been given to the economic impact of the pandemic, there has been increasing recognition of the behavioral change in transportation during COVID-19, including travel avoidance [44], bus users’ compliance to COVID-19 measures [45], mask-wearing behaviors onboard airplanes [46], and transport mode shifts [47,48]. A noticeable behavioral change is mode shift from public transport (airplane and train) to private transport mode (private car and taxi), due to reduced operation of public transport during COVID-19 as well as avoidance of public settings like airports and train stations to minimize the risk of infection. While this change has been universally witnessed, it remains unclear how long it would sustain in the post-pandemic era. Recent studies have suggested that lasting COVID effects can be expected [49], and travelers may have varied responses to COVID effects in the long term due to different levels of perceived health risks [50]. For travelers, this means on-going adjustment in their view of transportation and search for ways to adapt to the new normality in the post-pandemic era. It is meaningful to monitor behavioral changes in the post-pandemic era, as people may be more capable of adapting their travel behaviors to new environments than usually expected following COVID-19 [51].

The issues of COVID-19 impact, measures to reduce transmission, and the role of technology in controlling COVID-19 have been investigated [52,53]. On the micro level, a few studies have explored the impact of COVID-19 on travel behaviors in railway transportation, including HSR. Changes in travel behaviors were observed when comparing demographic characteristics, mode choice, and travel purpose between pre- and post-COVID-19 phrases in China [54]. Aghabayk et al. [55] focused on the measure of social distancing during the pandemic and the changes in public perception of crowding on the railways before and after COVID-19. The results indicated a strong impact of crowding levels on perceived comfort. Further, it showed that passengers perceived more disutility when traveling during the pandemic compared to the pre-pandemic era.

Such understanding is particularly important in the US given the magnitude of the COVID impact on travelers, compared to other countries. Noticeably, HSR in the US has gained renewed attention in recent years, during which time the country has experienced and emerged from the COVID-19 pandemic. It is, thus, important to consider the impact of the pandemic on the study of HSR in the US. Given the significant change in travel and mode choice behaviors due to COVID-19 [47], it is reasonable to assume that the catastrophic impact of the pandemic may have changed the public view of what transport mode to use for domestic travel. It is possible that travelers view HSR more favorably following COVID-19 given its convenience and controllable health risks compared to other transport modes, especially air travel. In other words, there could be a wider public acceptance of HSR in the post-pandemic era because of the impact of COVID-19. Yet, the relationship between COVID-19 and HSR in the US remains unexamined, especially from the user perspective, which is a research gap this study aims to bridge.

3. Methods

3.1. Sampling and Data Collection

This study adopted a survey design and a convenience sampling strategy to collect online data from Amazon Mechanical Turk (MTurk). MTurk provides an online platform for researchers to post various tasks, including surveys, for registered members to complete voluntarily in exchange for payment. Before the survey questionnaire was posted online, the Institutional Review Board (IRB) reviewed and approved the survey instrument to ensure compliance with research ethics requirements. Given the COVID-19 situation, an online survey provided a feasible and efficient method of data collection. Three qualification checks were applied to ensure data quality from survey participants: (1) participants must have successfully completed at least 100 tasks on MTurk, (2) participants must have received over 98% approval rates, and (3) participants in the pilot study were not eligible to participate in the main survey. Data were collected in June 2022, which was considered an ideal time for data collection for the purpose of this study. After years of limited progress in HSR development, there has been discussion of HSR development in the busy corridors across the country. The timeline for the HSR development coincides with the country’s experience and re-emergence from the COVID-19 pandemic. It is reasonable to assume that the pandemic has changed how people view and use transport modes in the US, and this change is likely to sustain for some time in the future. It is also likely that travelers have developed different views of HSR following the COVID-19. The survey data (n = 1033) collected at this time allowed the researcher to better capture travelers’ opinions toward this new transport mode in the US.

3.2. Survey Design and Survey Questionnaire

A survey design was best suited for this study for two reasons. First, as the study focused on intermodal choices and mode use intentions, a survey design enabled the researcher to reach large numbers of respondents and directly interact with them to collect relevant information. Second, given the COVID-19 impact and lack of existing data on HSR at the time of conducting this study, a survey design provided a feasible and efficient way to generate empirical data. Multiple measures were taken to ensure the quality of the survey design, including (1) engaging in an extensive literature review to ensure the literature support the survey questions and measurement validity, (2) adding a qualification check of participants using the tools in MTurk to improve data quality, (3) providing a real-life scenario in the survey to facilitate correct understanding of the survey questions, and (4) conducting a pilot study on a small group of participants to obtain preliminary responses and make necessary revisions before the large-scale survey was performed.

A survey questionnaire containing four major sections was developed for data collection. The first two sections collected information of participants’ demographics and travel experiences. Section 3 measures factors that can be important for the choice of HSR. Likert scale questions were developed for respondents’ evaluation of these factors, from 1 (strongly disagree) to 5 (strongly agree). Most scale items were adopted from validated scales in the literature to increase the validity of the measurement [56,57,58,59,60]. Section 4 collects data of mode choice behaviors by providing a future scenario of traveling from LA to SF, for which respondents would be asked to choose air, HSR, or car for the trip. The use of the LA-SF scenario in the survey considered high travel demand and potential HSR operation in this market, which presents a real-life scenario that respondents would find easy to understand. In addition, the mid-distance trip (around 350 miles) in this corridor makes air, HSR, and car competitive transport modes, allowing participants to realistically assess the mode choice behavior if HSR becomes a viable choice. The survey scale, mode choice scenario, and mode choice questions are provided in Appendix A. The survey respondents’ profile is shown in Appendix B.

3.3. Treatment of Data

This study aimed to answer two research questions about travelers in the US: (1) What factors were important in the intermodal choice for a domestic trip in a highly competitive market following the introduction of HSR? (2) What is the impact of COVID-19 on HSR use, considering the geographic locations of travelers? Logistic regression was performed to answer the first question, focusing on the effect of seven demographic, travel, and HSR factors (gender, age, income, travel frequency, mobility issues, total travel time, and convenience in transport) on the intermodal decision in the LA-SF market. Two sets of analysis were conducted, including (1) multinomial logistic regression (MLR) to investigate travelers’ choice from air, HSR, and car transport and (2) binary logit regression (BLR) analyses between air and HSR and between car and HSR to verify the first analysis due to the small sample size for some mode choices. To answer the second question, a Two-Way MANOVA was performed to identify the effect of two independent variables (IVs)—view change on mode use for domestic travel following COVID-19 (View_Change) and geographic location of participants (Geo_Location)—on four travel- and HSR- related dependent variables (DVs) including knowledge of HSR, travel habit, likelihood of using trains, and the intention to use HSR in the post-pandemic era. Both main and interaction effects of the two IVs were identified.

4. Results

4.1. Logistic Regression Analysis—Multinomial and Binary

Multinomial logistic regression was conducted to exam whether gender, age, income, mobility issues, travel frequency, total travel time, and convenience in transport would affect the choice of air, car, and HSR transport in the LA-SF market. As gender, age, income, and travel frequency were ordinal or categorical variables with more than two levels, dummy coding was performed to make these variables suitable for logistic regression analysis. An important assumption of logistic regression is the absence of multicollinearity among the explanatory variables. All the variance inflation factor (VIF) values were less than 3 (

Table 1. Multinomial Logistic Regression—Choice of air, HSR, and car in the LA-SF market.

	HSR n = 681	Air n = 243	Car n = 101	VIF
Model Factor	Odds Ratio	Odds Ratio	Odds Ratio
Gender	R			<3
Female		0.562 **	1.11
Others		7.173 × 10−8	6.082 × 10−8
Age	R			<3
21–30		0.923	0.561
31–40		1.355	0.544
41–50		0.815	0.407
51–60		0.949	0.292
>60		1.544	0.930
Income	R			<3
USD 25,001–50,000		2.021 **	0.662
USD 50,001–75,000		1.544	0.705
USD 75,001–100,000		2.743 **	0.949
USD 100,001–125,000		2.131	1.32
>USD 125,000		1.08	0.379
Travel Frequency	R			<3
Once		0.694	0.939
2–3 times		0.506 **	0.644
4–5 times		0.455 **	0.331 **
>5 times		0.385 **	0.521
Mobility Issue	R	0.498 ***	0.747	<3
Total Travel Time	R	0.143 ***	0.189 **	<3
Convenience	R	1.992 **	0.944	<3
Model Assessment
2LL	1564.951 ***
Pearson χ2	2036.271 (p = 0.302)
Nagalkerke R2	0.174
Classification a	95.30%	17.30%	3%

Note: *** = p < 0.001, and ** = p < 0.05; R = Reference Category; ^a: Overall Classification Accuracy = 67.7%.

), indicating minimum concern of multicollinearity. Regarding sample size, Hair et al. [61] recommended a sample of 400 or more for logistic regression analysis, with at least 10 observations per category of the dependent variable (DV), which was followed in this study. Both the total sample size and observations for each category of the DV (n = 243, n = 101, and n = 681 for the choice of air, car, and HSR, respectively) met the sample size requirement.

Table 1 shows the results of the MLR analysis. Most respondents (681, or 66.4%) selected HSR as a preferred mode of travel from LA to SF, suggesting that HSR was the most popular travel option in this study. Model estimation showed that the final model containing all the predictors represented a significant improvement in model fit over the null model (X² (38) = 156.82, p < 0.001). The test of goodness of fit yielded similar results, further suggesting that the model fit the data adequately (X² (2004) = 2036.271, p = 0.302). For the purpose of investigating the mode preference in this study, HSR was used as the reference group (represented as R in Table 1) against which the other two modes were compared.

The comparison between air travel and HSR indicated that six factors (total travel time, convenience in transport, gender, income, travel frequency, and mobility issues) were significant predictors. For these factors, the interpretation was based on the sign of the coefficient (negative or positive) and the value of odds ratio. Annual income (USD 25,001–50,000 and USD 75,001–100,000) and convenience in transport had positive coefficients. With their odds ratios, the interpretation was that, in generally, participants who earned moderate annual incomes (USD 25,001–50,000 and USD75,001–100,000) had greater odds of selecting air travel (versus HSR) by factors of 2.021 and 2.743, compared to those who earned low incomes (less than USD 25,000). When transport convenience increased, the participants were more likely to choose air travel over HSR. Gender, mobility issues, travel frequency (travel annually twice or more), and total travel time had negative coefficients. The interpretation was that female participants’ odds of choosing air travel (versus HSR) was smaller by a factor of 0.562 compared to that of male participants; participants who traveled 2–3 times, 4–5 times, and more than 5 times annually had smaller odds of selecting air travel (versus HSR) by factors of 0.506, 0.455, and 0.385, compared to that of participants who traveled less than once a year; participants with mobility issues had smaller odds of selecting air travel (versus HSR) by a factor of 0.498, compared to those without mobility issues; and when total travel time increased, participants were less likely to select air travel over HSR. Only two factors—total travel time and travel frequency (4–5 times annually)—were found to be significant in the choice between car travel and HSR, both with negative coefficients. This indicated that participants were more likely to choose HSR over car travel when total travel time increased; in general, participants who traveled 4–5 times annually had smaller odds of choosing car travel (versus HSR) by a factor of 0.331 compared to those traveled less than once a year.

The prediction accuracy of the model was further quantified using the classification statistics. While the model had an overall prediction accuracy of 67.7%, the predictive ability of the choice of air (17.3%) and car (3%) travel was low. This may be related to the relatively small sample size, especially for the car mode. As the survey also collected separate data for the choice between air travel and HSR and between car travel and HSR, the data, with a larger sample size for the car and air choices, were then used to estimate two Binary Logistic Regression (BLR) models to verify the key predictors identified in the MLR. The results are shown in

Table 2. Binary Logistic Regression—choice between air and HSR and choice between car and HSR.

	Air (370) vs. HSR (654)	Car (201) vs. HSR (823)
Model Factor	(Odds Ratio)	(Odds Ratio)
Gender
Female	0.739 **	NS
Others	NS	NS
Age
21–30	NS	NS
31–40	1.448 **	NS
41–50	NS	NS
51–60	NS	NS
>60	NS	NS
Income
USD 25,001–50,000	NS	NS
USD 50,001–75,000	NS	NS
USD 75,001–100,000	NS	NS
USD 100,001–125,000	NS	NS
>USD125,000
Travel Frequency
Once	NS	2.094 ***
2–3 times	NS	NS
4–5 times	NS	NS
>5 times	NS	NS
Mobility Issue	0.270 ***	NS
Total Travel Time	0.141 ***	0.128 ***
Convenience	2.093 **	NS
Model fit measurement
2LL	1189.188(Δ 41.343)	911.967(Δ 20.224)
Hosmer and Lemeshow X2	0.060 (Δ 0.009)	0.739 (Δ 0.602)
Nagelkerk R2	0.187 (Δ 0.048)	0.151 (Δ 0.029)
Classification a	69.4% (Δ 3.9%)	80.8%

Note: Δ = Change from base model; NS = Not Significant; ***: p < 0.001, **: p < 0.05; ^a: Group Classification—82.4% and 46.5% for HSR and air, 97.9% and 10.4% for HSR and car.

. The two BLR analyses showed that total travel time, convenience in transport, gender, mobility issues, and age (31–40 years old) were significant in the choice between air travel and HSR, while total travel time and travel frequency (once a year) were significant in the choice between car travel and HSR. While the BLR analysis identified age (31–40 years old) as an additional significant factor in the choice between air travel and HSR, the results generally supported the MLR findings that total travel time, convenience in transport, gender, travel frequency, and mobility issues were important predictors for travelers to choose from air travel, HSR, and car travel in the LA-SF market. Classification accuracy was 69.4% for the air-HSR choice (82.4% for HSR and 46.5% for air) and 80.8% for car-HSR choice (97.9% for HSR and 10.4% for car), representing improvement in predictive accuracy from the MLR analysis.

4.2. Two-Way MANOVA

The goal of the Two-Way MANOVA was to identify the impact of COVID-19 and the geographic location of respondents on a number of HSR and travel characteristics. For the purpose of this analysis, the impact of COVID-19 was defined as whether COVID-19 has changed the view of what transport modes to use for domestic travel, or View_Change (travelers choose between change or not change). The geographic location of respondents (Geo_Location) was represented using the four geographical regions defined by the US Census (Northeast, Midwest, South, and West). A Two-Way MANOVA was performed to test whether four HSR and travel characteristics (dependent variables, or DVs), namely knowledge level of HSR, travel habit, likelihood of using trains, and behavioral intention to use HSR following the pandemic, differed across the levels of the two independent variables (View_Change and Geo_Location). Noticeably, 697 respondents (67%) reported that COVID-19 has changed their view of what transport mode to use for domestic travel, demonstrating a great impact of COVID-19 on domestic travel and potential mode shift following the COVID-19 crisis.

For the MANOVA analysis, a sample size of 20 or more was recommended for each level of IV, which was satisfied in this study [61]. The four DVs showed moderate correlations, with Pearson’s r between 0.218 and 0.634, indicating minimal concern of multicollinearity. Box’s Test of Covariance matrices was statistically significant, indicating that equal variance at the multivariate level was not satisfied. Due to the partial violation of the assumptions, Pillai’s Trace was used for the interpretation of multivariate test results, given its robustness to departures from assumptions.

Table 3. Two-Way MANOVA—multivariate, main effect, and interaction effect.

	F	df1	df2	p	η2
Box’s Test of Covariance matrices	6.418	70	284,497.98	***	-
Multivariate Test—Pillai’s Trace
Geo	2.937	12	2943	***	0.012
View	40.933	4	979	***	0.143
Interaction	1.785	12	2934	**	0.007
Univariate Test (Main Effect)
Geo—KN	5.986	3		***	0.018
Geo—LU	4.188	3		**	0.013
Geo—HA	4.372	3		**	0.013
Geo—BI	8.426	3		***	0.025
View—KN	162.178	1		***	0.142
View—LU	56.523	1		***	0.054
View—HA	12.675	1		***	0.013
View—BI	28.332	1		***	0.028
Geo*View (Interaction Effect)
View*Geo—KN	3.523	3		**	0.011
View*Geo—LU	1.889	3		NS
View*Geo—HA	2.125	3		NS
View*Geo—BI	1.599	3		NS

Notes: KN = Knowledge of HSR; LU = Likelihood to use trains after COVID-19; HA: Travel habit; BI = Intention to use HSR after COVID-19; NS = Not Significant; *** = p < 0.001, ** = p <0.05.

shows the results.

The Pillai’s Trace was significant at p < 0.001. Therefore, the four DVs, when considered together, differed significantly across the levels of the two IVs—View_Change and Geo_Location (Pillai’s Trace = 0.143, F (4, 979) = 40.933, p < 0.001, partial η² = 0.143, observed value = 1.00, and Pillai’s Trace = 0.036, F (12, 2943) = 2.937, p < 0.001, partial η² = 0.012, observed value = 0.992). View_Change had a large effect on the linear combination of the DVs, while Geo_Location had a small effect, as indicated by the effect size. There was a significant interaction effect of View_Change and Geo_Location on the combination of the DVs (Pillai’s Trace = 0.022, F (12, 2943) = 1.785, p = 0.045, partial η² = 0.007, observed value = 0.892), although the effect size was marginal. At the univariate level, both IVs demonstrated significant main effects on the individual DVs (when the DVs were tested separately). Thus, the respondents who changed their view on transport mode due to COVID-19 differed significantly in HSR knowledge, travel habit, likelihood to use trains, and behavioral intention to travel by HSR, compared to respondents whose view on mode use was not affected by COVID-19. The Mean Estimates (

Table 4. Mean Estimates—univariate test and interaction effect.

Univariate Test		Geo				View
		NE	MW	SO	WE	Yes	No
	KN	5.629	6.11	6.483	6.251	7.135	5.102
	LU	5.679	6.284	6.470	6.430	6.888	5.544
	HA	3.280	3.430	3.423	3.433	3.452	3.331
	BI	3.262	3.508	3.510	3.524	3.562	3.340
Interaction Effect a		Yes	No
View and Geo on KN	NE	7.032	4.226
	MW	6.777	5.443
	SO	7.392	5.575
	WE	7.339	5.163

^a: Only significant interaction effect was included; NE = Northeast; MW = Midwest; SO = South; WE = West.

) further showed that those reporting changed views on domestic travel due to COVID-19 scored higher on all the DVs, indicating that this traveler group had more HSR knowledge and higher intention to use both trains and HSR in the post-pandemic era.

Similarly, respondents from the four geographical regions of the US (Northeast, Midwest, South, and West) differed significantly in HSR knowledge, travel habit, likelihood to use trains, and the intention to use HSR. Close examination of the results revealed a pattern indicating generally higher scores from South and West regions compared to lower scores from Northeast and Midwest regions. To further identify which pair of the geographic regions produced a significant difference, a Post Hoc Analysis was performed.

Table 5. Post Hoc Analysis—Geo_Location with mean difference.

	NE/MW	NE/SO	NE/WE	MW/SO	SO/WE	MW/WE
KN	−0.0626	−0.6848 **	−5.966	−0.6222 **	0.0883	−0.5340
LU	−0.3291	−0.7976 **	−0.7295 **	−0.4685	0.0681	−0.4004
HA	−0.1031	−0.1156 **	−0.1502 **	−0.0125	−0.0346	−0.0471
BI	−0.1887 **	−0.2146 ***	−0.2623 ***	−0.0260	−0.0477	−0.0737

Notes: NE = Northeast; MW = Midwest; SO = South; WE = West. *** = p < 0.001, ** = p < 0.05.

shows the pairs of regions that differed significantly, supported by mean differences. Consistent with the patterns observed in the univariate tests, the Northeast region (which had the lowest mean scores on all the DVs in the univariate test) differed significantly from the South and West regions (the two regions that had the highest mean scores in the univariate test), while the other pairs of the geographic regions showed no significantly different mean scores. This suggests that travelers coming from the Northeast region of the country have significantly lower level of knowledge of HSR, likelihood to use trains, and the intention to use HSR following COVID-19, compared to travelers from the South and West regions of the country.

Of the four DVs, only knowledge of HSR was significantly affected by the interaction effect of View_Change and Geo_Location. This suggests the level of HSR knowledge of the respondents were determined by the combined effect of the IVs, rather than any of them individually. In other words, the effect of View_Change (travelers changed their views on transport mode due to COVID-19) on the knowledge level of HSR depends on the geographic locations of the travelers. Respondents who changed their views on what mode to use for domestic trips and lived in the South region of the country had the highest knowledge level of HSR, whereas those coming from the Northeast region with unchanged views of mode use for domestic travel (views not affected by COVID-19) had the lowest level of knowledge of HSR.

5. Discussion

More males than females participated in the survey, and more respondents were White adults than people from other ethnic groups. The participants were generally younger, received more education, and earned less income than the national average [62,63]. Most of them traveled three times or less each year, primarily for personal purposes (leisure, vacation, and visiting family and friends). Surprisingly, 80% of the respondents traveled by train at least once over the past five years, and over half of them traveled two times or more, which differed from the general perception of lacking rail experience in the US. Noticeably, half of the respondents reported having mobility issues, either concerning themselves or their families, which aligned with the government statistics of mobility issues in the country [42]. Respondents demonstrated a strong recognition of the impact of COVID-19, with nearly 70% of them claiming that COVID-19 changed their view of what transport mode to use for domestic travel. Interestingly, respondents obtained HSR information mostly from social media, followed by sources of family, friends, and co-workers. Only one-third of the respondents received HSR information from national/international news or government agencies, and 9% indicated they did not receive any information about HSR. This suggests that HSR information is not widely available in the US, especially from formal government sources.

Responding to the scenario of choosing from air services, HSR, and car-based transport to travel between LA and SF, two-thirds of the respondents preferred to use HSR, indicating potential interest in HSR service in the US. All factors except for age were found to be significant in the intermodal choice, and more factors were found to impact on HSR-air choice than HSR-car choice, likely indicating a stronger competition between air and HSR in the high-demand market. Convenience in transport and travel frequency were two major predictors in the choice between air transport and HSR. The importance of the convenience factor is supported by the HSR literature [35]. This suggested that travelers in mega metropolitan areas like LA and SF value the central location of HSR stations, easy and quick access to HSR facilities, and flexible, well-connected public transport to the train station, which can become an important motivator in the choice between HSR and air services. Travel frequency was another significant factor, indicating that higher travel frequency (twice or more) was associated with a greater likelihood of choosing HSR over air. The finding could be related to the convenience of using HSR, which makes HSR a reasonable choice when travel frequency increases. Gender was another important factor in the intermodal choice, with female travelers being more likely to choose HSR over air services in the LA-SF market. The finding was consistent with gender differences in mode choice identified in previous studies, suggesting that gender-based differences do exist in mode choice, and females exhibited greater preference for HSR compared to males [39,64]. Mobility issues were also a significant predictor of intermodal choice. The findings indicated that travelers with mobility issues were more likely to choose HSR over air transport, which is not surprising due to the user-friendly nature of train transport. With easy access to the train station, simple station procedures, and spacious train cabins to freely move about, HSR can provide greater accessibility and simplicity over air transport, driving the mode choice of travelers with special needs. Finally, travelers’ decision between air travel and HSR was affected by total travel time, as supported by the literature [12,15,32]. The findings of this study suggest that travelers in the US value the total time saving of HSR, which can drive the decision to choose HSR, especially on short- and medium-haul routes.

The choice between car transport and HSR was affected only by travel frequency and total travel time, making the comparison relatively straightforward. Only one category of travel frequency (4–5 times annually) significantly affected intermodal choice. This suggested that, while travel frequency influenced the choice between air travel, HSR, and car travel, it had less impact on the choice between car travel and HSR than on the choice between air travel and HSR. Total travel time was the other significant predictor of the choice between car travel and HSR, which was expected. With speed acceleration (less than three hours from station to station on the LA-SF route) and convenient locations, HSR can provide greater time-saving benefits than cars (six hours in driving), which can drive the mode shift from car travel to HSR. It is worth noting that the existing HSR literature focuses primarily on HSR-air competition instead of HSR-car competition, a research gap this study can bridge. In this study, fewer factors affected car-HSR choices than air-HSR choices. This may be due to some similarities between car and HSR transportation. Both can move large amounts of passengers with relatively high operational flexibility (e.g., service frequency) and user convenience (e.g., easy access to service). The major difference between the two lies in travel time. While traditional trains and cars do not differ substantially in travel time, HSR can run much faster and, thus, enjoys a clear advantage. This study found travel time to be significant in the choice between cars and HSR, reflecting travelers’ emphasis on time-saving benefits. As such, car-HSR competition can be intensified when HSR starts operating in the US, with travel time being a major determinant. Overall, the Logistic Regression Analysis showed that travelers would focus on different factors when choosing from air travel, HSR, and car travel from LA to SA. While the decision-making between HSR and car travel was relatively straightforward, the choice between air travel and HSR was affected by multiple factors, indicating potential strong competition between air services and HSR when HSR services enters the LA-SF market.

The survey revealed a great impact of COVID-19 on the perception of domestic travel, with over two thirds of the respondents reporting a change in view of what transport mode to use as a result of COVID-19. This implies opportunities for new transport modes such as HSR to gain success in the US market. The view change, together with the geographic location of travelers, significantly influenced the travelers’ HSR knowledge, travel habit, likelihood to travel by train, and the intention to use HSR in the post-pandemic era. With respect to view change, the findings showed that the respondents who had changed their view on mode use due to COVID-19 were more knowledgeable of HSR, had different travel habits, were more likely to travel by train, and had higher intentions to use HSR compared to respondents whose view of mode use was not affected by COVID-19. The findings were important and timely as they connected COVID-19, travel behaviors, and HSR to provide empirical evidence of how travelers’ perceptions and intentions toward HSR in the US can be reshaped by COVID-19. It is likely that many travelers in the US feel strongly about the impact of COVID-19 and they have changed their view of what mode to use for domestic travel. These people are generally more curious and open-minded regarding HSR, as demonstrated by their greater knowledge level and intention toward HSR. Thus, this new traveler segment is likely to be a strong supporter of HSR in the US, as they may perceive HSR as a safer and more suitable transport mode to meet their travel need domestically in the post-pandemic era.

Concerning geographic location, the findings revealed clear geographic patterns regarding the travel and HSR characteristics across the four geographic regions (West, Midwest, South, and Northeast, as defined by the US Census). While travelers in the four regions demonstrated similar travel habits, the Northeast and Midwest regions generally scored lower in HSR knowledge, likelihood to use trains, and the intention to use HSR, compared to the South and West regions. Major differences appeared between the Northeast, South, and West regions. Travelers from the Northeast region exhibited the lowest scores among all travel and HSR characteristics in this study. Specifically, its scores on the likelihood of using trains after COVID-19 and intention to use HSR were significantly lower than those from the South and West regions. It appeared that, while the Northeast corridor operates the fastest rail system in the US, travelers in this region had less HSR enthusiasm than travelers in the South and West regions. The finding may be partially explained by the more rapid population growth in the South and West compared to Northeast and Midwest over the years, which could increase the interest and intention regarding HSR [20]. The finding may also be explained by the mixed experience of the train ridership with Amtrak in NEC. Currently, train service in some sections of the Amtrak’s Acela line can reach 150 miles/h, arguably making it the only HSR in the country. The average speed of Acela, however, is generally around 79 mph due to infrastructure constrains [7]. With the low average speed, the time benefit of using HSR diminishes, which could contribute to misunderstanding and lower interest toward HSR. The HSR project in the West region, on the other hand, has been built and promoted with a much-improved speed (217 mph), allowing for travel between LA to SF in just under three hours. This may have successfully enhanced HSR enthusiasm and anticipation in this region.

As indicated by the interaction effect between view change and geographic location, travelers’ knowledge level of HSR can be best explained by combining the two factors instead of analyzing them individually. Simply, whether the view change in mode use affects the knowledge level of HSR depends on where the traveler comes from (geographic location). The finding indicates that, among travelers whose view of mode use has been changed by COVID-19, those from the South region reported the highest level of HSR knowledge, while those from the Midwest region showed the lowest level of HSR knowledge. This was consistent with the greater intention to use HSR in the South region, as identified earlier, indicating that travelers who have greater intention to use HSR would exhibit greater interest in learning about HSR. Among respondents whose views were not affected by COVID-19, respondents from the South region remained the most knowledgeable of HSR while those from the Northeast region were the least knowledgeable. It appeared that travelers from the South region have the highest level of HSR knowledge in the country. Unlike travel habits, likelihood to travel by train, and intention to use HSR, the forming of HSR knowledge appears to be more complicated and influenced by combination of different factors.

6. Conclusions

With the renewed discussion, HSR has been given new opportunities in the US. This study investigated HSR from the traveler’s perspective, mainly focusing on travelers’ choice among air transport, HSR, and car travel in the highly competitive LA-SF corridor. The impact of COVID-19 on travel and HSR characteristics was also explored, especially in terms of the travelers’ geographic locations. There were two major findings from the analyses of logistic regression and Two-Way MANOVA: (1) Convenience in transport, travel frequency, gender, mobility issues, income, and total travel time were key determinants for the choice between air travel, HSR, and car travel in the LA-SF market, though they affected choice in different ways due to the specific mode characteristics. Convenience in transport and travel frequency were among the major factors in the decision between air travel and HSR, while the choice between car travel and HSR was mostly influenced by travel frequency and total travel time. (2) Most travelers have changed their view about which transport mode to use for domestic travel as a result of COVID-19, and they were more likely to travel by train and had a greater intention to use HSR in the post-pandemic era. In addition, travelers from the Northeast region had significantly less intention to use either train or HSR compared to travelers from southern or western US. Finally, neither change in view nor geographic location could individually affect the travelers’ knowledge level of HSR; rather, the knowledge level is determined by both factors. Travelers from southern US reported the highest level of HSR knowledge, while travelers from the Northeast and Midwest were least knowledgeable of HSR.

This study contributes to the HSR literature in two important ways. While the choice between HSR and other transport modes has been frequently studied in other countries, little research has been conducted on mode choice involving HSR in the US. Most of the literature on HSR in the US is exclusively focused on policy and economic aspects, and the perspective of travelers is rarely considered. Thus, the findings in this study on the key determinants of choice between air transport, HSR, and car transport can fill this research gap. The findings also reveal significant associations between COVID-19, HSR use, and geographic patterns in the US. To the best of our knowledge, this is the first study to examine HSR use in the broad context of COVID-19 in the US. The finding of the impact of COVID-19 on the potential mode shift, together with the geographic patterns of HSR intentions, significantly enhanced the understanding of the interest and intention toward HSR, especially given the long-time debate of whether HSR is a suitable transport mode in the US.

The findings of this study can provide implications for intermodal competition in the US. Strong competition following the market entry of HSR has been observed worldwide. The HSR impact on air transport is particularly large as HSR attracts passengers away from airlines, reducing airline operations and market shares. While an impact of this magnitude in the US is not likely to occur soon, given the early stage of HSR development, the continuous development of HSR, coupled with HSR’s strengths in travel time, service frequency, and convenient service, has the potential to generate strong market competition, especially with air and cars. This study has several policy implications. First and foremost, the findings indicate a strong preference toward HSR over air and cars in the busy transport markets, suggesting favorable attitudes and public acceptance of HSR in the US. This new insight is valuable given limited HSR research from the average US traveler’s perspective. With the empirical evidence generated by this study, the US government can better understand public acceptance of HSR and be more confident in HSR’s strategic decisions and long-term success. Specifically, the respondents cited convenience and total travel time as significant factors for their mode choice, which can inform strategy formation in the competitive transport market. An important implication is that HSR should be promoted primarily based on user convenience (e.g., quick access and simplified station procedure) and shortened total travel time (as a result of convenience location and speed acceleration) to increase public acceptance of HSR. Second, understanding which demographic segments are more likely to choose HSR is important for HSR policies in the US. The findings on the gender and mobility factors imply that efforts should be made to promote HSR to female travelers and travelers with mobility issues, to broaden the customer base. Third, the findings reveal that COVID-19 has changed the view of many travelers in the US regarding their transport mode for domestic travel, and travelers now have a greater intention to use HSR. The post-pandemic views offer a potential opportunity for successful market entry and greater public acceptance of HSR in the US. To leverage this opportunity, the government and HSR providers should promote HSR as a safe, hassle-free, and less crowded travel option to accommodate travelers’ mode preference in the post-pandemic era. Finally, the geographic patterns identified in this study indicate an uneven interest and intention to use HSR across the US. Therefore, policies should be directed to promote HSR especially in the northeast region where interest and intention to use HSR were low. Reliable information, especially from government sources, is essential for increasing the interest in HSR. This is particularly relevant given the limited information of HSR that is available to the general public in the US.

This study has some limitations. The convenience sampling method and the cross-sectional nature of the survey design may limit the generalizability of the findings, especially to HSR users outside the US. Also, survey data collected from MTurk are typically skewed toward younger, more educated, and lower income participants. This study is also limited in using participants’ geographical origins to define groups for comparison. Particularly, the reliance on self-reported geographical information may create certain biases. For example, the participant may interpret “geographical origin” differently and fail to provide the needed information. In addition, the cross-sectional survey design only collects data at a single time. As such, the study cannot provide insights into the change in population over time. While this study used the four geographical zones developed by the US Census Bureau to group participants, it should be noted that other ways to create geographic groups, especially using existing data, may provide a better understanding of the geographic patterns in the use of HSR. Factor selection in model estimation is another limitation. The survey design and questionnaire can only cover some determinants of intermodal choice in the US. Likely, other factors such as comfort and cost may also influence the choice of HSR. Acknowledging this limitation, the researcher focused on estimating a model with factors selected from multiple categories to achieve greater coverage while keeping the number of predictors manageable. While the survey cannot include all possible influencing factors, the balance between demographic, travel, and HSR factors can provide useful insights into intermodal choice in the LA-SF market.

Future research can build and expand on the findings of this study. Since the study model only covered a limited number of predictors, more factors, especially comfort and the price of HSR, can be added to the model to enhance the overall findings. Given the cross-sectional and self-reported characteristics of the survey design, which can limit the study’s findings, future research can utilize existing data to gain further insights into HSR use in the US. Government data, if available, can be used to define demographic segments for comparison. Particularly, compiling and analyzing existing data sources from various channels can help identify patterns and trends of HSR in data. The results can be used to verify the findings of the present study. In addition, future research can adopt different analytical approaches to gain insights into potential HSR passengers in the US. A structural equation modeling technique, which is based on well-established theories and used to examine the relationship between latent variables, can provide in-depth understanding especially when the factor under investigation is intangible in nature, such as the intention to use HSR. As HSR is a new phenomenon in the US, and the literature from the travelers’ perspective is limited, the findings of this study can provide a meaningful starting point for researchers and travelers to re-think the preferred modes of domestic travel in the post-pandemic era.