How does Public Transit Serve Post-Secondary Students in Toronto? A Utility- based Analysis of Accessibility by Transit for Discretionary Trips

doi:10.21203/rs.3.rs-3112752/v1

Download PDF

Research Article

How does Public Transit Serve Post-Secondary Students in Toronto? A Utility- based Analysis of Accessibility by Transit for Discretionary Trips

https://doi.org/10.21203/rs.3.rs-3112752/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 24 Feb, 2024

Read the published version in Transportation →

You are reading this latest preprint version

Post-secondary students are a segment of the population whose activity-travel behaviour is not well understood. In particular, there is a relative dearth of studies that have examined the determinants of behaviours related to the participation of out-of-home activities among post-secondary students. This study uses data from a web-based survey administered to students attending universities in Toronto to examine the determinants of the location choices of these individuals when using transit to participate in discretionary activities. Additionally, count- and utility-based measures of accessibility by transit for university students in Toronto are calculated and compared. The specification of the location choice model offers insights into the determinants of location choice decisions made by university students for discretionary trips and highlight the impacts of transit level-of-service and land use attributes on location choice decisions. Moreover, the findings suggest that the impacts of these attributes can differ based on socio-demographic characteristics. The comparison of count- and utility-based accessibility measures underscore the shortcomings of the former, which stem from the treatment of all opportunities as equally attractive. The results of this study aim to contribute to the literature by offering insights into an aspect of the activity-travel behaviour of post-secondary students that has received relatively little attention. Similarly, the results of the study can be used to help inform planning decisions by shedding light on the activity-travel behaviour of a segment of the population that has typically been underrepresented in traditional household travel surveys.

post-secondary students

transit accessibility

location choice

discretionary trips

utility-based accessibility

The activity-travel behaviour of post-secondary (i.e., college and university) students has received increased attention in the literature in recent years. Compared to the general population, post-secondary students tend to have relatively unique travel needs and display different travel patterns (Akar et al., 2013; Limanond et al., 2011; Taylor & Mitra, 2021). In particular, post-secondary students tend to make more trips and use public transit and active modes more frequently than the general population (Wang et al., 2012; Whalen et al., 2013). Given that post-secondary students tend to be young adults, investigations into the activity-travel behaviour of these individuals have partially been motived by the potential for the habits formed at this stage of their lives to influence their future travel behaviour (Nash & Mitra, 2019; Polzin et al., 2014). Besides, studies of this nature have also been motivated by the potential for this segment of the population to represent a relatively large percentage of trip-makers as well as the potential for post-secondary institutions to influence travel demand and patterns in the surrounding region (Eom et al., 2009; Garikapati et al., 2016; Hossain et al., 2022; Khattak et al., 2011; Zhan et al., 2016). Given its potential short- and long-term implications for the transportation system, it is essential to understand the activity-travel behaviour of post-secondary students.

Previous studies have attributed differences in the activity-travel patterns of post-secondary students and the general population to the tendency for the former to live and work in areas that are denser and contain a greater mix of land uses (Devi et al., 2019; Wang et al., 2012). However, these differences may also stem from the relatively flexible and irregular nature of the daily activity schedules of post-secondary students, which tends to vary from one day to the next. Traditionally, operational travel demand models have accounted for the travel patterns of post-secondary students by treating post-secondary institutions as special generators; however, this approach is insufficient for several reasons. First, these models tend to be developed based on data obtained through household travel surveys, where post-secondary students tend to be underrepresented (Mitra & Nash, 2019; Wang et al., 2012). Second, this approach places a greater emphasis on non-discretionary trips (e.g., trips to campus or to return home) than on discretionary trips, which tend to be more flexible and can contribute positively to physical and emotional well-being (C. Chen & Mokhtarian, 2006; Dharmowijoyo et al., 2018). Third, traditional trip-based models cannot analyze the effects of activity schedules, due to the assumption that trips are made independently of one another (Castiglione et al., 2015).

This study uses data from a travel survey of university students in Toronto to explore the determinants of their location choice decisions – an aspect of activity-travel behaviour that has received relatively little attention in the literature. Specifically, the data are used to develop a location choice model for post-secondary students using public transit to participate in discretionary activities. The focus on these activities stems from the relative control that individuals have over the timing and locations of discretionary activities, the influence of discretionary trips on travel demand, and the tendency for studies on post-secondary students to focus on commuting trips (Coutts et al., 2018; Nguyen-Phuoc et al., 2018; Ortúzar & Willumsen, 2011; Taylor & Mitra, 2021). Using the location choice model, utility-based measures of accessibility by transit for post-secondary students are derived; these values are then compared to traditional count-based measures to highlight how accessibility by transit can be over-stated by traditional measures. In a large city such as Toronto, public transit plays an important role in helping post-secondary students fulfill their mobility needs, and accessibility by transit has the potential to influence the well-being of students (Cooper et al., 1992; De Vos et al., 2013; Kamruzzaman et al., 2011; Taylor & Mitra, 2021).

The goals of this study are: 1) to offer insights into the factors influencing the location choice decisions for discretionary activities among university students, 2) compare and contrast the findings derived from utility- and count-based accessibility measures, and 3) explore spatial variations in accessibility by transit for discretionary activities among post-secondary students in Toronto. The findings of this study can help shed light on the determinants of the location choice decisions of post-secondary students, which can influence mode choice decisions. Additionally, the results of this study offer insights into the strengths and shortcomings associated with using utility-based measures of accessibility. Overall, the findings of this study can help inform transportation planning decisions by offering further insights into the activity-travel behaviour of post-secondary students – a segment of the population that tends to be underrepresented in traditional household travel surveys.

The remainder of the paper is structured as follows: first a review of relevant prior studies is presented. Next, information about the datasets used in this study is provided. Then, the methods applied in this study are summarized. Finally, the key findings of the study are presented and discussed.

Literature Review

In the literature, many studies on the activity-travel behaviour of post-secondary students tend to focus on travel behaviour. Moreover, these studies predominantly focus on mode choice decisions for commuting trips or trip rates (Taylor & Mitra, 2021). Studies on the former typically find that travel times are negatively associated with the probability of choosing a mode, which is consistent with expectations (Akar et al., 2012; Danaf et al., 2014; Nguyen-Phuoc et al., 2018). However, the findings of Whalen et al. (2013) suggest that certain students may derive positive utility from travelling by car and bicycle; although this is somewhat counterintuitive, the enjoyment of travel has also been reported in other studies (Diana, 2008; Ory & Mokhtarian, 2005). Besides, mode choices have also been found to be affected by gender, with female students typically being found to be less likely to bike than their male counterparts (Akar et al., 2012; Delmelle & Delmelle, 2012; Zhou, 2012). There is also evidence that the living situation (Delmelle & Delmelle, 2012; Whalen et al., 2013) and reason(s) for choosing the current home location (Zhou, 2012) can influence the mode choices of post-secondary students. Additionally, the attributes of the built environment, including the coverage of transit stops, coverage of sidewalks, and topography, have also been found to influence the use of transit and active modes (Akar et al., 2012; Rodrı́guez & Joo, 2004; Whalen et al., 2013; Zhan et al., 2016). For a summary of the factors influencing mode choice decisions among post-secondary students, see Aghaabbasi et al. (2020).

Aside from mode choice decisions, certain studies have tried to jointly model the decision to use a given mode and the frequency with which it is used. An early example is the work of Wang et al. (2012), who used a zero-inflated Poisson model to model the number of trips made by students at Old Dominion University, in addition to the number of trips they made using an automobile and active mode. The results highlighted that undergraduate students and students who were enrolled on a full-time basis were more likely to make more trips, while the opposite was true for students who lived further from campus. Taking a similar approach, Daisy et al. (2018) estimated a zero-inflated negative binomial model to examine the determinants of the number of trips made by automobile, transit, and active modes at Dalhousie University. The authors found that students who lived further from campus were less likely to use transit, while students earning less that $15,000 annually and who lived at their current residence for less than a year were more likely to use active modes.

In addition to more traditional modes of travel, there are a small number of studies that have modelled the use of emerging mobility services among post-secondary students. One example is the work of Aghaabbasi et al. (2020), who developed a Random Forest model and applied Bayesian network analysis to study the adoption and frequency of ride-sourcing use for commuting trips at Universiti Teknologi Malaysia. The results of the analysis highlighted the influence of age on the decision to use ride-sourcing. Hossain et al. (2022) utilized a more traditional approach in their study of the determinants of the frequency with which post-secondary students in the Greater Toronto and Hamilton Area used various modes for commuting trips. Based on the development of five zero-inflated ordered probit models, the results highlighted the influence of latent attitudinal factors on the frequency with which a given mode was used.

In contrast to the effort dedicated to studying travel behaviour, there is a relative dearth of studies focusing on the activity component of post-secondary students' activity-travel behaviour. Moreover, studies that have explored the activity component tend to focus on activity participation. For example, Eom et al. (2009) used data from a survey conducted at North Carolina State University to examine the amount of time allocated to various activities on a given day. They found that attending classes, having meals, studying, and recreation were among the most common activities and that activity profiles differed between graduate and undergraduate students. Additionally, the authors developed an activity transition matrix to gain insights into the sequencing of activities. In contrast, Chen (2012) applied a similar methodology using data from a survey of university students in Virginia but did not find evidence that activity profiles differed across students based on gender, age, or student status.

There is also a dearth of studies that have explored the determinants of the location choices of post-secondary students, which is an important component of activity-based travel demand models. One example of such a study is Garikapati et al. (2016), who developed a location choice model as part of their framework for modelling university travel demand. Using data from a survey of students attending Arizona State University, the multinomial logit (MNL) model was used to model the location choice decisions of the students; however, this model was primarily used to identify the origin of trips made to campus. A similar approach was applied by Eom et al. (2010) to model the location choices of students among the buildings located on the campus of North Carolina State University. As part of their model, the feasible choice set for a given student was determined based on the corresponding activity.

This study aims to contribute to the literature by offering insights into the location choice decisions of university students using public transit to participate in discretionary activities. Additionally, the findings of this study also shed light on the factors influencing the accessibility by public transit for university students. Despite the potential benefits of socialization and participation in discretionary activities on the well-being of students (Cooper et al., 1992), relatively little work has been done to explore the factors influencing where these activities occur. Besides, the number of studies focusing on accessibility by transit among post-secondary students in the literature appears to be limited. Consequently, the results of this study can help inform planning and policy decisions by shedding light on the activity-travel behaviour of a segment of the population that tends to be underrepresented in traditional data collection methods. Moreover, this information can help transit agencies plan routes that serve university students and universities by identifying the types of locations that students favour. This can help make transit a more attractive option for students, which has the potential benefit of helping transit use become a habit that persists during the later stages of the students' lives (Nash & Mitra, 2019; Schwanen et al., 2012).

Data for Empirical Investigation

The primary source of data for this study was the information collected through StudentMoveTO (SMTO) – a web-based survey of students attending the four universities in Toronto (University of Toronto, Ryerson University, York University, and OCAD University). The survey was conducted in 2015, with invitations to participate in the survey being sent to all students attending the four universities. As shown in Fig. 1 the Glendon and Keele campuses of York University and the Scarborough campus of the University of Toronto (U of T) are located outside downtown Toronto. A total of 15,226 students completed SMTO, representing a response rate of 8.3% (StudentMoveTO, 2018). As part of SMTO, respondents were asked to provide various information regarding their socio-demographic attributes, primary campus, home location, mobility tool ownership, attitudes, and use of various modes of travel over the past month (Nahal & Mitra, 2018). Additionally, respondents were asked to provide a one-day travel diary outlining the locations they visited on the survey day and the mode they used to travel from one location to the next.

As shown in Table 1, roughly two-thirds of respondents were female, while almost three-quarters of respondents were undergraduate students. As expected, the majority (72.7%) of respondents were under 25, although it is notable that 12.8% of respondents were over 29. Based on a comparison with the most recent iteration of the regional household travel survey, SMTO respondents displayed similar driver's license ownership rates and higher rates of transit pass ownership than the general population (Data Management Group, 2018). The latter could stem from the local transit agency, the Toronto Transit Commission (TTC), offering discounted monthly passes to post-secondary students. Besides, most respondents (76.8%) indicated they had used public transit at least once in the past month.

Table 1

Key summary statistics of the sample
Attribute	Category	% of Respondents
Gender	Male	32.1%
	Female	67.0%
	Another gender identity	0.9%
Age group	15 to 19 years old	26.7%
	20 to 24 years old	46.0%
	25 to 29 years old	14.6%
	30 to 34 years old	6.0%
	35 to 39 years old	2.4%
	40 years old and above	4.4%
University affiliation	OCAD University	3.1%
	Ryerson University	19.3%
	University of Toronto	54.0%
	York University	23.6%
Student status	Undergraduate	74.3%
	Graduate	23.4%
	Other	2.3%
Mobility tool ownership	Driver's license	60.4%
	Transit pass	39.5%
	Bicycle	47.7%
Use of modes in the past month	Private vehicle	33.3%
	Public transit	76.8%
	Bicycle	10.7%
	Walking	27.4%
Note: The sample size of SMTO was 15,226 responses

To help examine the influence of land use attributes on the location choice decisions of post-secondary students, two additional datasets were used to supplement the SMTO data. The first is the Enhanced Points of Interest (EPOI) dataset, which includes the coordinates and classification of each point of interest in Canada (DMTI Spatial Inc., 2016). Examples of classifications include accommodation and food services, retail trade, and educational services. The second dataset was a shapefile that segmented the study area into mutually exclusive land use categories. The categories used in the dataset include residential, commercial, and parks and recreational (DMTI Spatial Inc., 2014). In addition to land use information, data from the 2016 Canadian Census was also incorporated into the analysis to examine the influence of zonal socio-demographic attributes on location choice decisions.

Level-of-service (LOS) information was also incorporated into the analysis to help determine the available alternatives for the location choice decisions made by the SMTO respondents. Because traffic analysis zones (TAZs) were used as the basis for this analysis, transit travel times were imputed using the Tool for Incorporating Level of Service attributes (TILOS) tool described in Hasnine et al. (2017). TILOS utilizes a Google application programming interface (API)-based framework to obtain the in-vehicle travel time, access time, egress time, total travel time, walking distance, and the number of transfers required to travel from a given origin to a given destination. For this analysis, TILOS was used to impute the travel time from the centroid of a given TAZ to the centroid of every other TAZ in the study area. Due to cost constraints, LOS information was only obtained for three periods – 6 AM to 10 AM, 10 AM to 3 PM, and 3 PM to 7 PM. These periods are consistent with those used by the TTC when planning their service.

Location Choice Model

Location choice models are often used to analyze the determinants of the decision of where to live, where to establish a firm, and where to participate in an activity. Studies of activity location choice typically use the MNL model and tend to include inter-zonal impedance measures, zonal size measures, and land use information (Jianwei Yang et al., 2009; Kim & Lee, 2017). This study uses the MNL model to examine the determinants of the TAZ chosen by post-secondary students engaging in discretionary activities. The formulation of the MNL model is based on the concept of random utility maximization, which assumes that each individual $i$ obtains a certain level of utility by choosing alternative $j$ $\left({U}_{ij}\right)$ (Train, 2009). This utility is comprised of a deterministic component $\left({V}_{ij}\right)$ and a random component $\left({\epsilon }_{j}\right)$. The probability that individual $i$ chooses alternative $j$ is given by (Ben-Akiva & Lerman, 1985):

$${P}_{i}\left(j\right)=P\left({U}_{j}>{U}_{k}\right)=P\left({V}_{ij}+{\epsilon }_{j}>{V}_{ik}+{\epsilon }_{k}\right)=P\left({\epsilon }_{k}-{\epsilon }_{j}<{V}_{ij}-{V}_{ik}\right) \forall k\ne j$$

The deterministic component of utility can be modelled as a function of observable attributes corresponding to the decision-maker and/or the alternatives $\left({{X}}_{{i}{j}}\right)$ and a set of parameters $\left(\beta \right)$:

$${V}_{ij}={{\beta }}^{{\prime }}{{X}}_{{i}{j}}$$

However, because TAZs are unlabelled alternatives, the utility function for a given alternative cannot include an alternative-specific constant. In the MNL formulation, the random components of utility are assumed to follow the type I extreme value distribution and to be independent and identically distributed (Train, 2009). Consequently, the probability of individual $i$ choosing TAZ j is given by:

$${P}_{i}\left(j\right)=\frac{\text{e}\text{x}\text{p}\left(\mu {V}_{ij}\right)}{{\sum }_{k\in {C}_{i}}\text{exp}\left(\mu {V}_{ik}\right)}$$

Where:${C}_{i}$ is the choice set for individual $i$

$\mu$ is the scale parameter

The location choice model presented in this study was estimated using the maximum likelihood estimation module written for the GAUSS programming language (Aptech Systems Inc., 2012).

Choice Set Formation

When developing a discrete choice mode, a key consideration is the process through which the choice set of each decision-maker will be determined. In the context of location choice models, where there are potentially hundreds or thousands of alternatives, a common practice is to estimate the model using a subset of the universal choice set (Lee & Waddell, 2010). In the literature, choice set formation tends to take one of three approaches: 1) the random selection of alternatives from the universal choice set, 2) the consideration of all plausible alternatives, and 3) the application of deterministic rules to identify a set of feasible alternatives (Rashidi et al., 2012). This study utilizes the third approach by applying the concept of the time-space prism proposed by Hägerstrand (1970). This approach has been applied by prior studies into location choice decisions for "flexible" activities that occur between two fixed locations (Xiang Chen & Kwan, 2012).

In this study, the choice set formation process began by classifying the locations visited by each SMTO respondent in their travel diary. Activities were classified as non-discretionary (i.e., school, work, home) or discretionary based on the reported purpose. The next step of the process was to examine the attributes of the trips made to reach the discretionary activity. Based on the available LOS information, activities accessed using public transit on a weekday between 6 AM and 7 PM were retained in the analysis. For each of these activities, the set of feasible alternative locations was determined based on the available time budget; this value was defined as the difference between the arrival time at the location of the next mandatory activity and the departure time from the location of the previous mandatory activity, minus the duration of the activity in question. Consider the activity schedule shown in Fig. 2: for activity 3 (denoted as A3), the time budget is determined by subtracting TD2 and D3 from TA6. For each observation in the dataset, a TAZ $k$ was considered to be a feasible alternative if it satisfies the following constraint:

$$TT\left(P, k\right)+TT\left(k, N\right)\le TB$$

Where:$TT(a,b)$ is the time needed to travel from zone $a$ to zone $b$ using public transit

$P$ is the TAZ where the previous mandatory activity took place

$N$ is the TAZ where the next mandatory activity will take place

$TB$ denotes the available time budget

$TT(a,b)$ was determined based on the LOS information obtained from TILOS.

Accessibility Measures

Since the seminal work on the topic was published by Hansen (1959), the concept of accessibility has taken various forms. Despite the different approaches to defining and measuring accessibility, two common themes exist in the literature: 1) accessibility arises from the interaction between land use and transportation systems, and 2) accessibility is related to the ability to reach goods, opportunities, or services (Albacete et al., 2017; Geurs & van Wee, 2004). Measures of accessibility tend to be grouped into one of three categories: count-based, gravity-based, and utility-based measures. Count-based measures are determined based on the number of opportunities reached within a given time or cost threshold (Dong et al., 2006). Although these measures are easy to interpret, this approach has two key shortcomings. First, the values of count-based measures will be influenced by the choice of the threshold value, with opportunities just outside of the threshold being omitted (Cascetta et al., 2013). Second, this approach assumes that opportunities are equally attractive (Curtis & Scheurer, 2010; El-Geneidy et al., 2016).

To help address these shortcomings, gravity-based accessibility measures were developed. While these measures account for the impedance associated with reaching these opportunities, they inherently assume that the impedance is the same for all individuals (Geurs & van Wee, 2004). Moreover, both count- and gravity-based measures can potentially overestimate accessibility due to their tendency to overlook the context in which trips are made (Hasnine et al., 2019). This issue can be addressed through the use of utility-based accessibility measures, which have the potential to provide a more context-sensitive measure of accessibility by accounting for the perceptions of travellers and the context in which trips are made.

As part of this study, utility-based and count-based measures of accessibility are calculated on a TAZ-by-TAZ basis. Because the time-space prism concept cannot be applied to determine the choice set, travel from one zone to another was considered feasible if it could be reached within 105 minutes (the maximum travel time captured in SMTO). A count-based measure of accessibility was determined for each TAZbased on the number of TAZs that can be reached by public transit within the 105-minute threshold. To calculate a utility-based measure of accessibility for each TAZ, the location choice model was applied to the set of zones that were included in the set of feasible alternatives for TAZ $k$ $\left({F}_{k}\right)$. The utility-based measure of accessibility for students in $TAZ k$ (denoted as $A\left(k\right)$) is then calculated based on the logsum value of the location choice model (Páez et al., 2012):

$$A\left(k\right)=\frac{1}{\mu }\text{ln}\left(\sum _{z\in {F}_{k}}\text{exp}\left(\mu {V}_{z}\right)\right)$$

Empirical Model

As part of developing the location choice model, variables concerning level-of-service, land use, and zonal attributes were tested. In addition, interactions between these attributes and socio-demographic information were tested to examine whether the latter mediates the extent to which the former impact location choice decisions. The decision of whether to retain a variable in the final specification of the model was based on the sign and significance of the corresponding parameter. However, a small number of statistically insignificant parameters were retained in the model due to the insights they offer into location choice decisions. As part of this study, two models are estimated – one with the interaction terms and one without; the latter was used to derive the utility-based accessibility measures. Variables with the expected sign that were found to be statistically significant in the model without interaction terms were retained in the model with interaction terms. Overall, the results offer insights into the determinants of location choice decisions and how the impacts of said determinants can differ based on socio-demographic attributes. As shown in Table 2, the adjusted ${\rho }^{2}$ value of the final model was 0.10.

Table 2

Final model specification
	Final Model (w/o interaction terms)		Final Model
Variable	Estimate	t-stat.	Estimate	t-stat.
Level-of-Service Attributes
Number of transfers – mean effect	-0.623	-8.624	-0.584	-6.102
Number of transfers * Primary campus is located outside downtown Toronto [0/1]	-	-	-0.141	-1.003
In-vehicle travel time per km – mean effect	-0.017	-1.643	-0.013	-1.002
In-vehicle travel time per km * Person has a driver's license and access to a private vehicle [0/1]	-	-	-0.016	-0.670
In-vehicle travel time per km * Household income over $120,000 [0/1]	-	-	-0.047	-0.979
Access time – mean effect	-0.046	-1.937	-0.048	-1.986
Egress time * TAZ is outside the central business district, and the trip purpose is shopping [0/1]	-	-	-0.032	-1.480
Points of Interest and Interactions
ln(no. of manufacturing POIs per km²) – mean effect	-0.518	-6.373	-0.502	-7.385
ln(no. of education POIs per km²) – mean effect	0.123	1.987	-0.012	-0.187
ln(no. of education POIs per km²) * Person is a full-time graduate student [0/1]	-	-	0.200	2.309
ln(no. of food and accommodation POIs per km²) – mean effect	0.564	5.867	0.701	6.745
ln(no. of food and accommodation POIs per km²) * Person is female [0/1]	-	-	-0.157	-2.576
ln(no. of food and accommodation POIs per km²) * Household owns at least one vehicle [0/1]	-	-	-0.106	-1.770
ln(no. of retail trade POIs per km²) – mean effect	0.388	3.536	0.538	5.220
ln(no. of retail trade POIs per km²) * Person is over the age of 22 [0/1]	-	-	-0.126	-2.002
Zonal Size Measures
TAZ area (km²) – mean effect	0.539	5.815	0.566	6.446
% of zone classified as residential – mean effect	-1.076	-5.346	-0.895	-4.591
% of zone classified as commercial – mean effect	0.659	1.385	0.149	0.287
Goodness-of-fit measures
Adjusted${\rho }^{2}$	0.06		0.10
Number of observations	539		539

The LOS attributes have a negative impact on the probability that a given TAZ is chosen, which is consistent with the findings of Eom et al. (2010). Notably, the number of transfers required to reach a zone reduces the probability that it is chosen. Additionally, the number of transfers appears to have a greater impact on the location choice decisions of students studying at campuses outside of downtown Toronto, all else being equal. Similarly, longer egress times reduced the likelihood of choosing TAZ outside the central business district (CBD) for shopping trips. This could stem from areas outside the CBD having land use patterns that are less conducive to walking. This is consistent with prior studies, which have found that out-of-vehicle travel time is perceived more negatively than in-vehicle travel time (Wardman, 2004). Besides, the negative impact of in-vehicle travel time per kilometer may imply that students prefer zones accessed by transit routes operating at relatively high speeds (i.e., those operating in a dedicated right-of-way).

Moreover, land use attributes were also found to influence location choice decisions for post-secondary students. Interestingly, the results suggest that the impacts of the concentration of certain POIs on location choice decisions can differ based on socio-demographic attributes. For example, the concentration of food and accommodation POIs was less impactful for students from households that owned at least one vehicle and female students. Similarly, the concentration of retail trade POIs appears to be less impactful among students over 22. Additionally, zonal size measures were also found to influence location choice decisions, consistent with the findings of Garikapati et al. (2016). In particular, zones predominantly comprised of residential areas were less likely to be chosen, and zones with higher proportions of commercial areas were more likely to be chosen.

Comparison of Accessibility Measures

A comparison of count- and utility-based accessibility measures is presented in Fig. 3. The two sets of measures are compared in qualitative terms because the differences in how they are derived preclude them from being compared in quantitative terms. To facilitate this comparison, the range of values was divided into 11 equally spaced categories, and the value for each zone was assigned to the corresponding category.

Two important trends emerge from comparing the count- and utility-based accessibility measures. First, accessibility by public transit for university students is highest in the downtown core and the TAZs near rapid transit stations. This is expected given the relative proximity of TAZs in the downtown core to rapid transit stations and the relative speeds at which vehicles on the rapid transit lines travel. Second, count-based measures appear to paint a more optimistic picture of accessibility by transit. The differences in the relative values of the two sets of accessibility measures land use are particularly prominent in the northeast and north-west portions of Toronto, where land use patterns tend to be more auto-centric and relatively few rapid transit stations are present.

The relatively optimistic values produced by the count-based approach to deriving accessibility measures likely stem from these values being determined based on the number of zones that can be reached by transit within a set travel time threshold. While these values are relatively easy to interpret, this approach can overlook the impacts of various factors on the perceived attractiveness of a given TAZ. For example, the out-of-vehicle travel time and the number of transfers associated with reaching a destination are perceived more negatively than in-vehicle travel time (Wardman, 2004). Besides, count-based measures do not account for the impacts of land use attributes on the relative attractiveness of a given TAZ. Consequently, the ability of the utility-based measures to capture the impact of these factors on accessibility is likely why these measures produce less optimistic values. However, an important shortcoming of utility-based measures is that they are more difficult to interpret and compare across spatial and temporal dimensions.

Implications for Future Work

The findings of this study offer further insights into an aspect of the activity-travel of post-secondary students that has received relatively little attention in the literature – their activity location choice decisions. The location choice models presented in this study highlight the impacts of different components of transit travel time and land use attributes on the location choice decisions of university students using transit to participate in discretionary activities in Toronto. Additionally, the results suggest that the influence of level-of-service and land use attributes can differ based on socio-demographic attributes. Given their relatively unique travel needs, the flexibility of their schedules, and the differences in their travel behaviour compared to the general population, activity-based models would likely benefit from explicitly considering the behaviour of post-secondary students. However, attempts to do so should not treat these individuals as a homogenous group, given the differences in the impacts of various attributes on location choice decisions that have been identified in this study.

The comparison of accessibility measures also demonstrates the potential shortcomings of count-based measures of accessibility. In particular, they can neglect the impacts of contextual factors on the attractiveness of opportunities, such as the impacts of the various components of transit travel time and land use attributes. The former could be addressed by using a composite measure that accounts for the relative disutility of out-of-vehicle travel time and transfers compared to in-vehicle travel time. Similarly, count-based measures could be determined based on a generalized cost threshold, which would allow for the impacts of both travel time and cost on the attractiveness of an opportunity to be captured. In the context of public transit, the use of a generalized cost could be particularly useful in jurisdictions that lack the integration of fares across different transit providers or areas where transfers between different transit agencies are common. This would allow for the benefits of count-based measures to be maintained while helping to ensure that they can more explicitly account for the impacts of contextual factors.

This paper presented the results of an investigation into the factors influencing the location choice decisions of university students when using public transit to participate in discretionary activities in Toronto. While there is a growing body of literature on the travel behaviour of post-secondary students, a limited number of studies have examined the activity component of the activity-travel behaviour of these individuals. Additionally, relatively few studies have explored the accessibility by transit among post-secondary students. Using data from a web-based survey administered to students attending the four universities in Toronto, this study estimates a location choice model for discretionary activities to help shed light on the determinants of the activity location choices of these individuals. The model is then used to derive utility-based measures of accessibility by transit for discretionary activities, which are compared to count-based measures on a qualitative basis.

The results highlight the influence of different components of transit trips (in-vehicle travel time, access time, and the number of transfers) on the location choice decisions of the respondents. Moreover, the findings suggest that the impacts of level-of-service and land use attributes on location choice decisions can vary based on socio-demographic attributes such as age, household vehicle ownership, and student status. Additionally, the comparison of count- and utility-based approaches to quantifying accessibility suggest that the former has the potential to produce relatively optimistic values. This could stem from the former being unable to account for the impacts of contextual factors on the attractiveness of each opportunity.

While the results of this study offer insights into the location choices of post-secondary students, there are two key limitations. First, the sample did not include students attending the colleges in Toronto, whose needs and travel behaviour differ from those of university students. Second, the results of the comparison of the count- and utility-based accessibility measures were based on a subjective categorization of two sets of values, meaning that the results could be influenced by the approach taken to categorize the values. Similarly, the discretization of the study area into TAZs could also influence the specification of the location choice model. There are several ways in which future work can build on the current studies. First, future studies should consider trips made outside of the 6 AM to 7 PM period considered in this study. This could offer further insights into the location choice decisions that are made during a period when transit services operate relatively infrequently. Second, this study treated the level-of service attributes as fixed. However, there is typically a feedback loop between the location choice, mode choice, and traffic assignment components of activity-based travel demand models (Castiglione et al., 2015). Consequently, future studies could capture this feedback in their choice set formulation procedure. Similarly, future studies could also take a probabilistic approach to determine the choice set of each individual, which would allowing contextual factors to influence the time budget for a given activity.

ACKNOWLEDGEMENTS

The study was partially funded by an NSERC Discovery grant and a Percy Edward Hart professorship awarded to the second author.

AUTHOR CONTRIBUTIONS

Conceptualization: P. Loa, K.N. Habib; Methodology: P. Loa, K.N. Habib; Formal analysis: P. Loa, Writing – original draft preparation: P. Loa, K.N. Habib; Writing – review and editing: K.N. Habib; Supervision: K.N. Habib.

DATA AVAILABILITY STATEMENT

The data collected through the StudentMoveTO survey are available from the project website: https://www.studentmoveto.ca/resources-2/2019-survey/. However, the authors are solely responsible for analyzing and interpreting the results presented in this paper.

STATEMENTS and DECLARATIONS

The authors do not have any conflicts of interest to declare.

Aghaabbasi, M., Shekari, Z. A., Shah, M. Z., Olakunle, O., Armaghani, D. J., & Moeinaddini, M. (2020). Predicting the use frequency of ride-sourcing by off-campus university students through random forest and Bayesian network techniques. Transportation Research Part A: Policy and Practice, 136, 262–281. https://doi.org/10.1016/j.tra.2020.04.013
Akar, G., Fischer, N., & Namgung, M. (2013). Bicycling Choice and Gender Case Study: The Ohio State University. International Journal of Sustainable Transportation, 7(5), 347–365. https://doi.org/10.1080/15568318.2012.673694
Akar, G., Flynn, C., & Namgung, M. (2012). Travel Choices and Links to Transportation Demand Management: Case Study at Ohio State University. Transportation Research Record: Journal of the Transportation Research Board, 2319(1), 77–85. https://doi.org/10.3141/2319-09
Albacete, X., Olaru, D., Paül, V., & Biermann, S. (2017). Measuring the Accessibility of Public Transport: A Critical Comparison Between Methods in Helsinki. Applied Spatial Analysis and Policy, 10(2), 161–188. https://doi.org/10.1007/s12061-015-9177-8
Aptech Systems Inc. (2012). Maximum Likelihood Estimation 5.0 for GAUSS Mathematical and Statistical System. Aptech Systems Inc.
Ben-Akiva, M. E., & Lerman, S. R. (1985). Discrete Choice Analysis: Theory and Application to Travel Demand. The MIT Press.
Cascetta, E., Cartenì, A., & Montanino, M. (2013). A New Measure of Accessibility based on Perceived Opportunities. Procedia - Social and Behavioral Sciences, 87, 117–132. https://doi.org/10.1016/j.sbspro.2013.10.598
Castiglione, J., Bradley, M., & Gliebe, J. (2015). Activity-Based Travel Demand Models: A Primer. https://doi.org/10.17226/22357
Chen, C., & Mokhtarian, P. L. (2006). Tradeoffs between time allocations to maintenance activities/travel and discretionary activities/travel. Transportation, 33(3), 223–240. https://doi.org/10.1007/s11116-005-2307-4
Chen, Xiang, & Kwan, M.-P. (2012). Choice set formation with multiple flexible activities under space–time constraints. International Journal of Geographical Information Science, 26(5), 941–961. https://doi.org/10.1080/13658816.2011.624520
Chen, Xueming. (2012). Statistical and activity-based modeling of university student travel behavior. Transportation Planning and Technology, 35(5), 591–610. https://doi.org/10.1080/03081060.2012.701818
Cooper, H., Okamura, L., & Gurka, V. (1992). Social activity and subjective well-being. Personality and Individual Differences, 13(5), 573–583. https://doi.org/10.1016/0191-8869(92)90198-X
Coutts, S., Aird, B., Mitra, R., & Siemiatycki, M. (2018). Does commute influence post-secondary Students' social capital? A study of campus participation at four universities in Toronto, Canada. Journal of Transport Geography, 70, 172–181. https://doi.org/10.1016/j.jtrangeo.2018.06.006
Curtis, C., & Scheurer, J. (2010). Planning for sustainable accessibility: Developing tools to aid discussion and decision-making. Progress in Planning, 74(2), 53–106. https://doi.org/10.1016/j.progress.2010.05.001
Daisy, N. S., Hafezi, M. H., Liu, L., & Millward, H. (2018). Understanding and Modeling the Activity-Travel Behavior of University Commuters at a Large Canadian University. Journal of Urban Planning and Development, 144(2), 04018006. https://doi.org/10.1061/(asce)up.1943-5444.0000442
Danaf, M., Abou-Zeid, M., & Kaysi, I. (2014). Modeling travel choices of students at a private, urban university: Insights and policy implications. Case Studies on Transport Policy, 2(3), 142–152. https://doi.org/10.1016/j.cstp.2014.08.006
Data Management Group. (2018). TTS 2016 City of Toronto: Summary by Ward. http://dmg.utoronto.ca/pdf/tts/2016/2016TTS_Summaries_Toronto_Wards.pdf
De Vos, J., Schwanen, T., Van Acker, V., & Witlox, F. (2013). Travel and Subjective Well-Being: A Focus on Findings, Methods and Future Research Needs. Transport Reviews, 33(4), 421–442. https://doi.org/10.1080/01441647.2013.815665
Delmelle, E. M., & Delmelle, E. C. (2012). Exploring spatio-temporal commuting patterns in a university environment. Transport Policy, 21, 1–9. https://doi.org/10.1016/j.tranpol.2011.12.007
Devi, M. K., Fitria, L. M., Roychansyah, M. S., & Herwangi, Y. (2019). Making the connection between accessibility and travel behaviour in university setting (Case study: Babarsari Area, Yogyakarta). IOP Conference Series: Earth and Environmental Science, 340(1), 012030. https://doi.org/10.1088/1755-1315/340/1/012030
Dharmowijoyo, D. B. E., Susilo, Y. O., & Karlström, A. (2018). On complexity and variability of individuals' discretionary activities. Transportation, 45(1), 177–204. https://doi.org/10.1007/s11116-016-9731-5
Diana, M. (2008). Making the "primary utility of travel" concept operational: A measurement model for the assessment of the intrinsic utility of reported trips. Transportation Research Part A: Policy and Practice, 42(3), 455–474. https://doi.org/10.1016/j.tra.2007.12.005
DMTI Spatial Inc. (2014). CanMap RouteLogistics User Manual.
DMTI Spatial Inc. (2016). CanMap Content Suite Data Dictionary.
Dong, X., Ben-Akiva, M. E., Bowman, J. L., & Walker, J. L. (2006). Moving from trip-based to activity-based measures of accessibility. Transportation Research Part A: Policy and Practice, 40(2), 163–180. https://doi.org/10.1016/j.tra.2005.05.002
El-Geneidy, A., Levinson, D., Diab, E., Boisjoly, G., Verbich, D., & Loong, C. (2016). The cost of equity: Assessing transit accessibility and social disparity using total travel cost. Transportation Research Part A: Policy and Practice, 91, 302–316. https://doi.org/10.1016/j.tra.2016.07.003
Eom, J. K., Stone, J. R., & Ghosh, S. K. (2009). Daily Activity Patterns of University Students. Journal of Urban Planning and Development, 135(4), 141–149. https://doi.org/10.1061/(ASCE)UP.1943-5444.0000015
Eom, J. K., Stone, J. R., & Kang, K. (2010). Empirical Case Study of Spatial–Temporal Student Activity Population. Transportation Research Record: Journal of the Transportation Research Board, 2157(1), 11–21. https://doi.org/10.3141/2157-02
Garikapati, V. M., You, D., Pendyala, R. M., Patel, T., Kottommannil, J., & Sussman, A. (2016). Design, Development, and Implementation of a University Travel Demand Modeling Framework. Transportation Research Record: Journal of the Transportation Research Board, 2563(1), 105–113. https://doi.org/10.3141/2563-15
Geurs, K. T., & van Wee, B. (2004). Accessibility evaluation of land use and transport strategies: review and research directions. Journal of Transport Geography, 12(2), 127–140. https://doi.org/10.1016/j.jtrangeo.2003.10.005
Hägerstrand, T. (1970). What About People in Regional Science? Papers in Regional Science, 24, 6–21.
Hansen, W. G. (1959). How Accessibility Shapes Land Use. Journal of the American Institute of Planners, 25(2), 73–76. https://doi.org/10.1080/01944365908978307
Hasnine, M. S., Graovac, A., Camargo, F., & Habib, K. N. (2019). A random utility maximization (RUM) based measure of accessibility to transit: Accurate capturing of the first-mile issue in urban transit. Journal of Transport Geography, 74, 313–320. https://doi.org/10.1016/j.jtrangeo.2018.12.007
Hasnine, M. S., Kamel, I., & Habib, K. N. (2017). Using Google Map to Impute Transportation Level-of-Service Attributes - Application in Mode and Departure Time Choice Modelling. Presented at the 11th International Conference on Transport Survey Methods.
Hossain, S., Loa, P., Ong, F., & Habib, K. N. (2022). The determinants of commute mode usage frequency of post-secondary students in the Greater Toronto and Hamilton Area. Transportation Research Part A: Policy and Practice, 166, 164–185. https://doi.org/10.1016/j.tra.2022.10.010
Jianwei Yang, Sun, D., Yanping Du, & Yunyun Zhao. (2009). The effect of sampling of alternatives on MNL models: An empirical analysis in the context of shopping-destination choice models. 2009 2nd IEEE International Conference on Computer Science and Information Technology, 75–80. https://doi.org/10.1109/ICCSIT.2009.5234758
Kamruzzaman, M., Hine, J., Gunay, B., & Blair, N. (2011). Using GIS to visualise and evaluate student travel behaviour. Journal of Transport Geography, 19(1), 13–32. https://doi.org/10.1016/j.jtrangeo.2009.09.004
Khattak, A., Wang, X., Son, S., & Agnello, P. (2011). Travel by University Students in Virginia: Is This Travel Different from Travel by the General Population? Transportation Research Record: Journal of the Transportation Research Board, 2255, 137–145. https://doi.org/10.3141/2255-15
Kim, J., & Lee, S. (2017). Comparative analysis of traveler destination choice models by method of sampling alternatives. Transportation Planning and Technology, 40(4), 465–478. https://doi.org/10.1080/03081060.2017.1300242
Lee, B. H. Y., & Waddell, P. (2010). Residential mobility and location choice: a nested logit model with sampling of alternatives. Transportation, 37(4), 587–601. https://doi.org/10.1007/s11116-010-9270-4
Limanond, T., Butsingkorn, T., & Chermkhunthod, C. (2011). Travel behavior of university students who live on campus: A case study of a rural university in Asia. Transport Policy, 18(1), 163–171. https://doi.org/10.1016/j.tranpol.2010.07.006
Mitra, R., & Nash, S. (2019). Can the built environment explain gender gap in cycling? An exploration of university students' travel behavior in Toronto, Canada. International Journal of Sustainable Transportation, 13(2), 138–147. https://doi.org/10.1080/15568318.2018.1449919
Nahal, T., & Mitra, R. (2018). Facilitators and barriers to winter cycling: Case study of a downtown university in Toronto, Canada. Journal of Transport & Health, 10, 262–271. https://doi.org/10.1016/j.jth.2018.05.012
Nash, S., & Mitra, R. (2019). University students' transportation patterns, and the role of neighbourhood types and attitudes. Journal of Transport Geography, 76, 200–211. https://doi.org/10.1016/j.jtrangeo.2019.03.013
Nguyen-Phuoc, D. Q., Amoh-Gyimah, R., Tran, A. T. P., & Phan, C. T. (2018). Mode choice among university students to school in Danang, Vietnam. Travel Behaviour and Society, 13, 1–10. https://doi.org/10.1016/j.tbs.2018.05.003
Ortúzar, J. de D., & Willumsen, L. G. (2011). Modelling Transprt (4th ed.). John Wiley & Sons.
Ory, D. T., & Mokhtarian, P. L. (2005). When is getting there half the fun? Modeling the liking for travel. Transportation Research Part A: Policy and Practice, 39(2–3), 97–123. https://doi.org/10.1016/j.tra.2004.09.006
Páez, A., Scott, D. M., & Morency, C. (2012). Measuring accessibility: positive and normative implementations of various accessibility indicators. Journal of Transport Geography, 25, 141–153. https://doi.org/10.1016/j.jtrangeo.2012.03.016
Polzin, S. E., Chu, X., & Godfrey, J. (2014). The impact of millennials' travel behavior on future personal vehicle travel. Energy Strategy Reviews, 5, 59–65. https://doi.org/10.1016/j.esr.2014.10.003
Rashidi, T. H., Auld, J., & Mohammadian, A. (Kouros). (2012). A behavioral housing search model: Two-stage hazard-based and multinomial logit approach to choice-set formation and location selection. Transportation Research Part A: Policy and Practice, 46(7), 1097–1107. https://doi.org/10.1016/j.tra.2012.01.007
Rodrı́guez, D. A., & Joo, J. (2004). The relationship between non-motorized mode choice and the local physical environment. Transportation Research Part D: Transport and Environment, 9(2), 151–173. https://doi.org/10.1016/j.trd.2003.11.001
Schwanen, T., Banister, D., & Anable, J. (2012). Rethinking habits and their role in behaviour change: the case of low-carbon mobility. Journal of Transport Geography, 24, 522–532. https://doi.org/10.1016/j.jtrangeo.2012.06.003
StudentMoveTO. (2018). Summary Findings 2015. http://www.studentmoveto.ca/resources-2/2015summary/
Taylor, R., & Mitra, R. (2021). Commute satisfaction and its relationship to post-secondary students' campus participation and success. Transportation Research Part D: Transport and Environment, 96, 102890. https://doi.org/10.1016/j.trd.2021.102890
Train, K. E. (2009). Discrete Choice Methods with Simulation (2nd ed.). Cambridge University Press. https://doi.org/10.1017/CBO9780511805271
Wang, X., Khattak, A. J., & Son, S. (2012). What can be Learned from Analyzing University Student Travel Demand? Transportation Research Record: Journal of the Transportation Research Board, 2322, 129–137. https://doi.org/10.3141/2322-14
Wardman, M. (2004). Public transport values of time. Transport Policy, 11(4), 363–377. https://doi.org/10.1016/j.tranpol.2004.05.001
Whalen, K. E., Páez, A., & Carrasco, J. A. (2013). Mode choice of university students commuting to schooland the role of active travel. Journal of Transport Geography, 31, 132–142. https://doi.org/10.1016/j.jtrangeo.2013.06.008
Zhan, G., Yan, X., Zhu, S., & Wang, Y. (2016). Using hierarchical tree-based regression model to examine university student travel frequency and mode choice patterns in China. Transport Policy, 45, 55–65. https://doi.org/10.1016/j.tranpol.2015.09.006
Zhou, J. (2012). Sustainable commute in a car-dominant city: Factors affecting alternative mode choices among university students. Transportation Research Part A: Policy and Practice, 46(7), 1013–1029. https://doi.org/10.1016/j.tra.2012.04.001

No competing interests reported.

Download PDF

Journal Publication

published 24 Feb, 2024

Read the published version in Transportation →

Editorial decision: Major revision
23 Oct, 2023
Reviews received at journal
29 Sep, 2023
Reviewers agreed at journal
25 Jul, 2023
Reviewers invited by journal
22 Jul, 2023
Editor assigned by journal
07 Jul, 2023
Submission checks completed at journal
27 Jun, 2023
First submitted to journal
26 Jun, 2023

You are reading this latest preprint version

How does Public Transit Serve Post-Secondary Students in Toronto? A Utility- based Analysis of Accessibility by Transit for Discretionary Trips

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Literature Review

Data for Empirical Investigation

Methodology

Location Choice Model

Choice Set Formation

Accessibility Measures

Results and Discussion

Empirical Model

Comparison of Accessibility Measures

Implications for Future Work

Conclusion

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1