Predictors of Sustainable Student Mobility in a Suburban Setting

Abstract

Analyses of student mobility are typically conducted in an urban environment and are informed by socio-demographic or trip attributes. The prevailing focus is on individual modes of transport, different groups of commuters travelling to campus, students’ behavioural perceptions, and the totality of student trips. This paper starts with the identification of the determinants of student mobility that have received insufficient research attention. Utilising surveys, the study captures the mobility patterns of a sample of 1014 students and calculates their carbon footprint (CF; in kg/academic year) to assess whether the factors neglected in previous studies influence differences in the actual environmental load of student commuting. A regression analysis is employed to ascertain the significance of these factors as predictors of sustainable student mobility. This study exclusively focuses on the group of student commuters to campus and analyses the trips associated with compulsory activities at a suburban campus that is distant from the university centre and student facilities, which changes the mobility context in terms of commuting options. The under-researched factors identified in this research have not yet been quantified as CF. The findings confirm that only some of the factors neglected in previous research are statistically significant predictors of the local environmental load of student mobility. Specifically, variables such as student employment, frequency of class attendance, and propensity for ride-sharing could be utilised to forecast and regulate students’ mobility towards more sustainable patterns. However, all of the under-researched factors (including household size, region of origin (i.e., past experiences), residing at term-time accommodation while studying, and the availability of a family car) have an influence on the differences in CF magnitude in the studied campus.

1. Introduction

The transport sector of the European Union is responsible for the generation of approximately 25% of the total greenhouse gas emissions [1]. Notably, this statistic remains consistent among member countries that contribute significantly to the total emission levels, with transport being a primary contributor. The present volume of road transport results in 71% of the total EU transport-related greenhouse gas (GHG) emissions [2]. The EU has adopted a strategic approach to the transportation sector, with an objective to transform it in a manner that ensures enhanced resilience, sustainability, and the achievement of carbon neutrality. This transformation is being pursued in a manner that is simultaneously taking into account the economic and social aspects of the transport sector [3,4,5]. Such transformation should derive from a 90% reduction in greenhouse gas (GHG) emissions from transport [6], encouraging collective public transport and active mobility [7], whilst discouraging the use of motorised modes on fossil fuels [8,9], whose CO₂ emissions account for the bulk of total GHG emissions [10].

Higher education institutions (HEIs) attract large quantities of passenger flows to their location(s). The generated GHG emissions of students, teaching staff, administrative staff and visitors have a significant impact on the local environment. It is, therefore, important for the HEIs to recognise their potential to contribute to decarbonisation [11] and towards GHG emissions reduction [11,12,13,14,15,16,17] by supporting more sustainable mobility. Universities demonstrate readiness to reduce car-dependency and to simultaneously support active mobility modes [18], and therefore implement mobility management strategies that simultaneously upgrade campus accessibility and sustainability [19].

The sustainability of traffic flows generated by commuters to HEI’s location is subject to local emission reduction policies and strategies [20], especially in smaller towns, where the number of students often surpasses the size of the local population, which is the case in the studied campus settlement. Given the large number of students and employees at HEIs, the promotion and facilitation of sustainable mobility may have broader societal implications [21]. Consequently, research into the variables that influence student mobility should, therefore, help to better prioritise the necessary improvements in the transport sector and the associated investments. The sustainable planning of human mobility within a region is predicated upon a comprehensive understanding of the determinants of mobility. It is imperative that this understanding is then utilised to promote more sustainable choices, such as the transition from private vehicles to public transport (PT) [22]. In addition, sustainable mobility requires the adequate supply of the transport demand [23] as well as the understanding of its temporal distribution [24].

Mobility mode preferences and the impact of commuting choices on the environment pose a considerable challenge for society, whether it is in the form of a challenging university setting in congested urban surroundings or in a scarcely connected suburban or rural location [25,26,27]. The specific urban or suburban territorial context, in conjunction with the mobility options available, differentiates universities [28] and challenges their policies and objectives aimed at long-term sustainability. Research indicates that commuting has a significant impact on academic behaviour and achievement levels [29,30,31], hereby emphasising the importance of understanding the factors that influence mobility patterns, i.e., influence mode-choice decisions related to students’ compulsory activities.

There is considerable variance amongst students in their perception of the environmental impact of their mobility choices [24]. In order to effect a long-term transition towards more sustainable patterns of student mobility, it is necessary to consider not only mobility management and transport policies [32], but also housing and employment policies. The mobility patterns of students are indicative of future patterns [33,34,35] and have an influence on regional sustainability [36]. Therefore, students’ mobility patterns need to be examined independently from other commuters to campus [18] as they deviate from the general population [21,30,37,38,39,40] and necessitate comprehensive consideration in transport planning models [39]. Understanding the mobility patterns of students, which are predominantly influenced by routines and habits, and the factors that shape them, is a precondition for a transition towards smart and sustainable mobility over the long term [3]. Furthermore, policies need to recognise student preferences as part of shaping future demand [41,42].

Students’ daily time distribution is contingent on their academic schedules. Consequently, their commuting patterns exhibit less pronounced concentration during peak hours when compared to general population patterns. However, the flexibility exhibited by students depends upon the available route options, distance from transit hubs, and the existence of environmental and personal limitations [18,43]. The size of the city, the characteristics of the transport system, the campus location, and the mobility on campus form a local contextual setting that influences student mobility [44].

Irrespective of the number of factors that were considered as determinants of students’ mode choice preferences, previous studies demonstrated their interconnectedness. As indicated by some sources [23], key determinants of students’ mode choices are income, vehicle access time, trip duration, cost, and comfort. In other [45], social status, housing location, demography, and mode availability are considered relevant. Although the availability of mobility modes limits mode choice decisions [46], factors such as age, gender, and income [41], as well as the specific personal features, trip purpose, and infrastructural specifics [23], cannot be disregarded. Additionally, the perception of flexibility must be considered as well [47].

Some authors [48] suggest that when the utility function is considered, trip distance and duration are the most significant factors determining mode choice, and are frequently considered to be interchangeable [49]. The significance of trip distance and duration has been demonstrated in both urban and rural campus settings [25]. Trip duration is often found to be inversely proportional to the probability of selecting a mode of transport [20,41,50], and is considered to be more significant than cost [24]. However, in cases of suburban universities with commuting options to and from urban centres, personal characteristics, distance, and price interconnection were found to determine the mode selection [20]. There is a wide range of different factors pertaining to individual/household socio-demographic, built environment attributes, trip/service attributes, and student attitudes that could be considered [36] when modelling the influence of different factors on the frequency of mode usage in relation to an exhaustive set of mobility modes.

A positive correlation has been demonstrated between the vicinity of residence and university campus and the adoption of active mobility modes [24], provided that there is adequate cycling and walking infrastructure [18,51]. However, students tend to switch from walking and cycling to motorised modes if the alternative is reasonably priced, accessible, and app-based [34], which is why incentives are needed to support active-mode commuting [52]. The utilisation of smartphone applications for mobility purposes has the potential to contribute to more sustainable mobility [53]. However, it is anticipated that users of such applications are likely to be individuals who already frequently utilise shared and public mobility modes [54]. The integration of mobility options into a MaaS (Mobility as a Service) application that students would use necessitates real-time information and constant updates on traffic and timetables, time and cost optimisation options in trip planning, and discounts for more sustainable options, such as PT or bike-sharing [28].

A limited number of studies have been conducted on the GHG emissions of student mobility, with an even smaller number focusing exclusively on the student population. This stands in contrast to studies involving other groups that commute to campus locations [14,15,26,34,38,55], particularly those examining mobility related to academic schedules (thus related to compulsory, non-leisure time activities).

Unlike previous studies that mostly focused on universities located in urban surroundings, this research focuses on student mobility in relation to a suburban campus. This suggests that specific location-related factors may be influential, such as the absence of active and shared mobility alternatives, the cost of housing in close proximity, the necessity for term-time housing, the distance from the university centre and student services, and the paucity of employment opportunities in the local area. From a transport standpoint, accessibility of locations (in terms of spatial coverage and frequency) and mobility mode options are (more) constrained in suburban surroundings, thus warranting further investigation due to the presence of contextual differences and the potential ramifications on student mobility patterns. To the best of the author’s knowledge, the focus on suburban campuses in terms of sustainability of students’ mobility patterns, and to a more precise degree, on the factors that shape students’ mobility patterns related to commuting (compulsory activities) in the suburban context, has not been explored.

This research acknowledges solely the student population, excluding any other group of commuters to campus. It is evident that students constitute the majority of the total traffic flows to the campus location, and their mobility is previously proven to differ from that of the general population (i.e., teachers and administration staff employed at the campus) in terms of flexibility and mode choices. Whilst some authors recognise students’ leisure-time mobility [28,47,48,56], this research is focused on non-leisure time, compulsory activity-related travel. Class-related travel is a shared characteristic amongst students (it is scheduled), and this component of mobility patterns is not influenced by individual preferences, in contrast to involvement in various leisure-time activities. Furthermore, schedule-related mobility may be subject to mobility management actions on the part of the HEI. In contrast to the focus of many preceding studies on a single commute mode, this research acknowledges a range of mobility mode options, thereby offering a more comprehensive perspective on student mobility. Furthermore, the quantification of CF is contingent upon the academic year duration.

This paper hypothesises that factors that have received little academic attention in previous research on student mobility behaviour have a significant impact on the sustainability of student mobility in a suburban environment. In view of the complex nature of the hypothesis, the introduction of partial hypotheses is deemed necessary:

H1. 

The size of the CF of student commuting in relation to a suburban campus varies according to criteria related to factors that have been under-researched in previous studies.

H2. 

The under-researched factors influencing differences in CF of student mobility to and from campus are statistically significant predictors of the sustainability of students’ mobility.

The purpose of this paper is to identify the factors which have been under-researched in relation to the size of the carbon footprint (CF) of student mobility. The objective of this study is twofold. After establishing the size of the CF of student population on a suburban campus, it firstly seeks to reveal whether the factors neglected in previous studies influence differences in the local load of student commuting. Secondly, it aims to ascertain whether these factors are statistically significant, i.e., which factors act as predictors and could be used in shaping students’ mobility. The primary findings indicate that all of the factors of student mobility that have been under-researched influence differences in student mobility CF. However, it is important to note that only some of these factors are statistically significant predictors of sustainable mobility. Consequently, these factors should be further explored as part of a comprehensive set of determinants in researching student mobility.

Following the Introduction Section, the methodological approach is described in the Materials and Methods Section. The results of a two-step approach to exploring factors of student mobility that have received little attention, and to determining their relationships, are presented in the Results Section and discussed subsequently.

2. Materials and Methods

This study acknowledges the findings of prior research that has examined the factors influencing university-level students’ mobility. A range of approaches has been identified (single- or multiple-campus focus, involving one or more commuter groups at campuses; single- or multiple-mode context, identifying one or more types of influences, focusing on leisure-time or non-leisure time activities, etc.). This study contributes to the gap in research by narrowing the research focus exclusively to students and to compulsory activity-related commutes, and by focusing on a suburban campus, which is not often the case in similar research. Furthermore, this research avoids a single-mobility mode context and explores several potentially influential factors that have hitherto received insufficient recognition in previous research. The present study makes a contribution to the current body of knowledge by identifying the predictors of the size of students’ mobility environmental load among the under-researched determinants.

In accordance with the aforementioned points, the regression analysis in this research is preceded by desk research into factors that have received little academic attention. This is followed by an exploratory study of the factors that shape student mobility, conducted via an online survey. The survey of students is conducted following comprehensive research, and the research tool is developed based on existing tools. This research commenced with the identification of a gap in the existing academic literature, which supported the purpose and objective of the present paper. The research framework is presented in Figure 1.

Single university- or campus-based surveys are common in previous studies on student mobility [18,57,58,59,60,61]. Online surveys are frequently utilised in the collection of mobility-related data from the student population [11,14,15,20,33,34,47], and due to the advantages inherent in reaching a large number of potential respondents, online surveys were also employed in this research.

The survey was conducted among students enrolled at a suburban faculty in Croatia, located 20 kilometres distant from the main university campus and other student services and facilities in the urban centre, where the rest of the university is situated. A large number of students in pre-graduate and graduate studies, originating from the entire country, acted in favour of choosing the selected case to be studied.

The studied campus is situated within a small settlement, with a population of approx. 300 [62]. The number of students at this location is six times greater than the number of local residents. Consequently, the campus generates a significant demand for transportation within the surrounding area. The location has adequate road transport links, facilitating both public and private transport. The campus offers four parking areas for students, free of charge. The frequency of bus services is from half an hour to 45 min, and trips to the main university campus take from 45 min to over an hour, depending on urban bus connections. Although there is walking infrastructure in place, there is a lack of dedicated cycling infrastructure. It is evident that the area is not currently serviced by any bicycle-sharing, car-sharing or e-scooter systems. The range of available transport options is limited, and spatial constraints mean that the construction of infrastructure for alternative mobility modes is not a viable option. Nevertheless, transport demand management (TDM) offers a variety of potential actions to improve the sustainability of student mobility. A further constraint associated with campus location pertains to the absence of formal student housing in close proximity to the campus. It is evident that students have the option of either securing tourist accommodation (private apartments) in closer proximity to the campus, albeit at a significantly higher cost, or alternatively, opting for the university campus itself, which is comparatively more economical. Housing choices reflect mobility patterns in terms of trip duration, cost, and mode choice.

The survey was distributed through an online learning system, available only to the students of the studied campus, and accessible via a shared link for a period of three months during the summer semester. The collection of data was conducted anonymously, and the survey settings were configured to facilitate single entries per student. The objectives of the survey were communicated to students of all study years, all study programmes, and of different study status (full- and part-time), alongside the instructions for research participation. The survey had a substantial return rate (69.22%), which was reduced to 56.33% (n = 1014) after filtering the data and keeping only the valid surveys. The survey comprised nine sections of questions, which were grouped according to the sub-topic of research (respondents’ socio-demographic and household information; everyday mobility patterns and determinants; car-related behaviour; PT usage; cycling and bike-sharing; walking; car-sharing; e-scooter usage; and mobility-related apps). A number of questions were adopted from several previous studies on mobility (Special Eurobarometer 422a: Quality of transport [63]; Special Eurobarometer 495: Urban mobility and transport [64]; and Methodology and indicator calculation method for sustainable urban mobility survey [65]) with additional questions designed to obtain specific CF calculation-related data (personal vehicle type, fuel type, commute trip distances, carpooling and ride-sharing participation).

In contrast to the prevailing body of literature on the subject, some studies suggest that students’ mobility patterns are distinct yet homogeneous [42]. This suggests that a limited number of “expert” representatives could be adequate to ascertain the mobility preferences of the student body as a whole. However, in light of the findings of numerous preceding studies indicating that students’ mobility patterns are influenced by individual characteristics, social environments, available mobility options, the built environment, and service attributes, this research considers students’ mobility patterns to be heterogeneous. This assertion is further substantiated by the observation that previous studies frequently yield divergent conclusions regarding the factors determining mode choice or frequency, and whether the influence of identical factors is beneficial or detrimental in particular contexts. The sample in this study is therefore large, comprising 1014 students, which is more than 50% of the total targeted student population (attending classes at the campus locality), and includes their individual and contextual differences (

Table 1. Sample characteristics overview (n = 1014).

Attribute	Category	% Respondents
Gender	male	32
	female	68
Household size	1 or 2 people	14.69
	3 or 4 people	63.81
	5 and more	21.50
Age group	18–19	15.68
	20–24	76.33
	25–29	6.90
	30–39	0.69
	40+	0.39
Student status	full-time	92.60
	part-time	7.40
Mobility tool ownership	driving licence	76.04
	car	35.80
	PT pass	44.28
	bicycle	21.89
	(e-)scooter	1.97
Primary commute mode	car (driver)	36.19
	car (passenger)	3.55
	bicycle	1.28
	PT	30.97
	walking	14.40
	car and PT	6.02
	bicycle and PT	0.30
	bicycle and walking	0.10
	walking and PT	7.00
	taxi	0.20
Region of permanent residence	Adriatic Croatia	51.08
	City of Zagreb	17.36
	Northern Croatia	20.12
	Pannonian Croatia	11.44
Commute distance	<1 km	10.85
	1–5 km	23.08
	5–10 km	23.96
	10–20 km	21.01
	20–50 km	16.07
	>50 km	5.03
Commute duration	<15 min	24.75
	15–30 min	32.35
	30–60 min	31.56
	60–90 min	9.07
	>90 min	2.27

).

The average age of respondents in this study is 21 (mean = 21.37, SD = 2.95), with the majority of respondents falling within the 20–24 age range. Students aged 25 and above are not as represented. In general, the campus predominantly comprises younger individuals and a higher proportion of female students in terms of both age and gender distribution. The average household size of students is between three and four people (mean = 3.64, SD = 1.04), but a significant proportion of households are larger. One of the study’s limitations pertains to the under-representation of part-time students, whose mobility patterns differ from those of full-time students. This limitation has the potential to impact the findings concerning the size of CF for single trips undertaken by part-time students. However, the primary focus of this research is on compulsory activity-related mobility and the CF on an academic year basis. Notably, part-time students have no obligation to participate in classes at the studied university, in contrast to full-time students. The survey was made available to all students, under the same conditions, and participation was voluntary. While a greater proportion of part-time students would provide more insight into their mobility, the impact of full-time students is of higher significance. Furthermore, it is posited that the part-time students who participated in the study are considered valid representatives of the group.

The multiple-mode approach is evident from mobility tools ownership and primary commute mode choices. While car ownership is not as high as the possession of a driving licence, it does approximate the share of preferential car drivers in everyday commutes. The potential of PT is almost fully utilised, provided that all the mobility options involving PT usage are taken into consideration. Bicycle ownership is significantly higher than the vehicle’s modal share in commuting, while scooter ownership is minimal and not used for commuting purposes. The sample structure is almost evenly distributed among students originating from the university area (51.08%) and those from other regions. This is reflected in the proportion of students commuting to the campus from distances of 5 km and over, with the suburban location of the campus being a small, short-distance community. Conversely, distances exceeding 5 km are indicative of trips originating from the gravitational area, i.e., other cities in the region. The longest trip distances are travelled by car (approx. 16 km one-way), somewhat shorter by PT (13 km), much shorter by bicycle (approx. 4 km) and on foot (2.4 km). However, a limited number of students exhibit a willingness to travel more than 50 kilometres to attend classes, suggesting that even students residing within the region seek term-time accommodation during the academic year. The majority of respondents reported that they required between 30 and 60 min to travel to campus. On average, they spent approx. 41 min travelling by PT, 35 min by car, 24 by bicycle, and around 19 on foot.

As was outlined in the introductory section of this paper, there is a considerable body of research that focuses on either trip characteristics/service attributes (time/duration, distance, cost, etc.), or basic individual socio-demographics as variables of mobility mode choice decisions. The survey used in this research was extensive and detailed, but the paper focuses on neglected factors, i.e., variables found in only a small number of studies, and does not aim to provide a comprehensive list of mobility determinants. This study explores whether the factors that have received little research attention can lead to differentiation in environmental impact (CF size) among students, and if those factors are significant predictors in this context.

A significant number of studies on student mobility employ regression analysis [20,23,24,38,41,48,55], albeit with varying methodological approaches. Some form of regression modelling was used in approximately half of the studies examining the influential factors of student mobility patterns [36]. In order to examine whether the factors identified as under-researched are significant predictors of the sustainability of student mobility in the case of a suburban campus, this research also applies regression analysis to establish if those variables are relevant in the context of shaping students’ mobility.

3. Results

This section begins by demonstrating differences in student CF in relation to under-researched factors of influence. The importance of these factors is then analysed by regression, and their statistical significance in predicting student mobility CF is determined.

3.1. CF Quantification of Student Mobility in Relation to Under-Researched Factors

CF quantification (in kg of CO₂) is derived from the actual reported mobility patterns (number of trips, trip distance, primary mode choice used to commute to faculty, fuel type if primary mode used is a car, ride-sharing tendency, and the number of people the ride is shared with, among other), and the emissions are established for the period of academic year. In order to establish the environmental load of student mobility, CF is calculated based on the actual trip distances of individual commutes in relation to specific values of CO₂ emissions suggested for motorised [66,67] and active mobility modes [68,69]—in grams/km (

Table 2. Input data for CF estimation.

Mobility Mode	CO2 Emission (g/km) *	Average Distances (km)	Average Trip Frequency (d/w)
Car–fuel B7	171	15.8	4.7
Car–fuel E5	192
Car–EV	53
Car–LPG	160
PT–bus	105	13.1	4.5
Walking	56	2.4	4.6
Cycling	21	4.5	4.1

* CO₂ emission values are attributed according to relevant sources for motorised [65,66] and active mobility modes [67,68]. The upstream emission data is included for motorised modes (liquid fuels and electricity) and cycling according to referenced sources.

). It should be noted that CO₂ makes up 99% of the total direct GHG emissions [10].

An extensive review of the existing research reveals that factors that gained little academic attention, with regard to influencing differences in the environmental burden of student mobility, include household size [20,36]; student employment [20,54]; residing in term-time accommodation while studying [14]; region of origin (i.e., past experiences) [50,70,71]; household/family car at disposal [14,28]—unlike individual car ownership; and willingness to share rides (i.e., carpool) [38], and should, therefore, be reviewed in the terms of their actual significance in predicting students’ mobility (

Table 3. CF size (per person per academic year) relative to under-researched factors of influence (in kg of CO₂).

Variables	%	Mean	St. Dev.
Household size
1 person	3.35	743.39	1004.98
2 people	11.34	620.54	847.38
3 people	25.05	608.21	757.25
4 people	38.76	594.64	801.71
5+ people	21.50	697.51	1082.58
Employment status
Student/not employed	62.52	471.83	613.40
Student and employed	30.87	794.33	968.42
Student and unemployed	6.41	538.65	696.67
Regional distribution_NUTS-2
Study outside region of residence/term-time acc.	48.92	672.03	1000.46
Study in region of residence	51.08	585.84	724.43
Availability of a personal vehicle
Ownership	47.08	827.06	918.41
Household vehicle	51.75	490.02	252.00
No ownership/disposal	1.17	405.42	409.41
Primary mode choice 1—active modes 2
Public transport	30.97	387.01	435.97
Bicycle	1.28	23.16	19.05
Walking	14.40	42.26	91.90
Public transport combined with walking	7.00	315.47	374.61
Public transport combined with cycling	0.30	97.44	50.73
Walking combined with cycling	0.10	30.80	0.00
Active modes in total	54.04	149.36	162.04
Primary mode choice—non-active modes
Car (driver)	36.19	1060.11	1091.31
Car (passenger)	3.55	509.81	575.97
Public transport combined with a car	6.02	830.03	704.88
Other: taxi	0.20	271.44	267.80
Motorised modes in total	45.96	667.85	659.99
Class frequency (schedule)
1–2 times (part-time students)	8.62	309.46	449.96
3–5 times (full-time students)	91.38	572.76	740.01
Carpooling
Yes (always/often/sometimes)	68.66	459.90	475.78
No	31.34	956.51	1057.50

¹ primary mode choice is not part of the under-researched set of factors identified in this research, but it is considered relevant for the mobility pattern elaboration. ² active mobility modes include PT [14,59].

). A comprehensive analysis of students’ mobility should potentially include some of these attributes, in addition to level of service attributes (travel time or distance) or socio-demographic attributes, if confirmed as predictors.

In relation to the impact of individual factors on CF, the exploratory study suggests that the GHG emissions of students’ daily commutes to and from campus are influenced by a combination of socio-demographic and attitudinal attributes that have been identified as under-researched. As illustrated in Table 3, the factors identified as under-researched were confirmed to result in different levels of environmental load. In accordance with the focus on student compulsory activities, the frequency of class participation (attendance according to schedule) was also taken into consideration. Whilst the primary focus of this research is on the sustainability and environmental impact of student mobility, the table also demonstrates the impact of modal choice on the environment. Although not under-researched, the use of non-active or active forms of mobility is relevant to consider for the discussion of findings.

The mobility patterns of students from three- and four-person households are found to be the most sustainable, while influencing smaller CF (594.64 kg CO₂/ac. year at a four-person household) than students from the largest households and students living alone, whose environmental load levels are the highest (743.39 kg CO₂/ac. year). In contrast to students dealing solely with their student obligations (471.83 kg CO₂/ac. year), those who are simultaneously employed exhibit a substantially greater environmental impact (794.33 kg CO₂/ac. year), owing to their larger travel distances.

When comparing the students’ environmental load in relation to their region of origin, the difference is not as high when the data is observed in sum. However, when considered at the level of individual regions, there are substantial CF differences between certain NUTS-2 areas of Croatia and the region in which the university is located (585.84 kg CO₂/ac. year). For instance, students originating from Northern Croatia have the largest CF (824.68 kg CO₂/ac. year).

Approximately 36% of the respondents reported having their own vehicle (which is close to half of the students with a licence). However, around ¾ (76%) of individuals in possession of a driving licence may potentially utilise cars to commute to the campus. This results in a substantially higher environmental impact on the environment in case of car ownership (827.06 kg CO₂/ac. year) than in cases of using a family (or company) car (490.02 kg CO₂/ac. year). Approximately 31% of the students indicated that they do not engage in ride-sharing with other students, which could considerably lower their impact.

Fortunately, active and collective alternatives to cars, whose CF is much lower (149.36 kg CO₂/ac. year) in comparison to non-active mobility modes (667.85 kg CO₂/ac. year on average; max. 1060.11 kg CO₂/ac. year for car commutes), are the first choice for everyday commutes for more than half (54.04%) of the total sample. It is evident that certain active mobility modes, characterised by significantly lower CF, remain underutilised. PT is the primary choice among full-time students; however, their average CF is lower for each trip than that of part-time students in this instance. Nevertheless, the overall number of trips (which is the sum of class participation) results in a higher weekly and total CF per academic year (572.76 kg CO₂/ac. year in comparison to 309.46 kg CO₂/ac. year), although full-time students demonstrate more sustainable behaviour.

In order to ascertain the significance of the identified factors as predictors of the sustainability of student mobility, the factors were analysed using regression analysis in the following section.

3.2. Predictors of Sustainable Student Mobility

This research employs regression analysis to estimate the relationships between a dependent variable (the local environmental load of student mobility patterns/CF size) and several independent variables (factors identified in the exploratory part of this study). The objective is to establish if those variables are significant predictors of the sustainability of students’ mobility. As the indicators Durbin–Watson, VIF, and Tolerance are all within the permissible limits for each of the factors, it can be concluded that the conditions of independence of residuals and absence of multicollinearity among predictors are met. Therefore, the regression analysis is methodologically justified.

The results of the analysis (

Table 4. Regression analysis with household size as a predictor.

Variable	R2	F	β
Household size	0.000	0.036	−0.01

Legend: R²—coefficient of determination; F—F-ratio; β—beta coefficient.

) show that the number of household members is not a statistically significant predictor of students’ mobility CF (β = −0.01, p > 0.05). The household size variable did not prove to be a good predictor of CF because variations in household size did not account for changes in CF values.

In order to examine whether work status is a statistically significant predictor of students’ mobility environmental load, a hierarchical regression analysis was conducted (

Table 5. Regression analysis with employment status as a predictor.

Variable	∆R2	F	β
Block I
	0.032	33.89 *
Student			−0.18 *
Block II
	0.006	6.76 *
Student			−0.21 *
Student and unemployed			−0.09 *

Legend: ∆R²—change in the coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

). Indicator (dummy) variables were used as predictor variables: in the first block, a student (yes or no), and in the second block, simultaneously unemployed and a student (yes or no), while the CF value was used as a criterion variable. Given that the variable in question comprises three mutually exclusive categories (a student, simultaneously unemployed and a student, and simultaneously employed and a student), it was possible to include a maximum of two dummy variables within the model in order to avoid multicollinearity. The employed student category was omitted from the model. This category was instead used as a reference group against which the coefficients of the remaining dummy variables were interpreted.

The results of the analysis suggest a substantial correlation between specific work statuses and student mobility. In the first block of the model, incorporating a dummy variable to differentiate between students and the reference category (students who are employed), a statistically significant relationship was identified. Student status emerged as a substantial negative predictor (β = −0.18, p < 0.01). Therefore, students exhibit a lower CF compared to their employed counterparts. This finding suggests that employed students have a greater impact on the environment through mobility compared to other students. In the second block, which additionally includes a dummy variable that denotes persons who are both unemployed and students, statistical significance was also recorded, i.e., this status was shown to be a significant negative predictor (β = −0.09, p < 0.01). The negative beta coefficient obtained for the student and unemployed variable (β = −0.09) implies that students who are not employed have a significantly lower CF from mobility compared to students who are employed. The student/non-student variable remained negatively and significantly associated with mobility (β = −0.21, p < 0.01). This finding further confirms that employed students have a higher CF. As demonstrated in Table 5, the first block explained 3.2% of the variance of the criterion variable, and the second block explained an additional 0.6% of the variance.

According to the results in

Table 6. Regression analysis with the university campus region as a predictor.

Variable	R2	F	β
Region where the campus is located	0.001	1.21	−0.04

Legend: R²—coefficient of determination; F—F-ratio; β—beta coefficient.

, the region of origin (in terms of mobility habits) is not a statistically significant predictor of the students’ mobility environmental burden (β = −0.04, p > 0.05). Therefore, living in the same region where the university campus is situated does not necessarily imply more sustainable behaviour.

A hierarchical regression analysis was conducted in order to explore whether car ownership or access to a family car is a statistically significant predictor of the impact of student mobility (

Table 7. Regression analysis with car ownership as a predictor.

Variable	∆R2	F	β
Block I	0.042	34.22 *
Owning a car			0.21 *
Block II	0.000	0.00
Owning a car			0.21 *
Family car at disposal			0.01

Legend: ∆R²—change in the coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

). Indicator (dummy) variables were used as predictor variables: in the first block, car ownership (yes/no), and car availability (yes/no) in the second block, while the CF value was used again as the criterion variable. The reference variable was set to “not owning a car” (yes/no) and was excluded from the analysis to avoid multicollinearity.

The first block of the model was found to be statistically significant (F = 34.22, p < 0.01), with car ownership being a significant positive predictor of students’ mobility CF (β = 0.21, p < 0.01). This finding suggests that people who own a car have a higher CF compared to people who do not. The second block did not show statistical significance, meaning that having a household car available (β = 0.01, p > 0.05) is not a statistically significant predictor. The first model explained 4.2% of the variance, which is a small portion of the CF variability.

Although it is not one of the under-researched factors, having a driving licence is analysed (

Table 8. Regression analysis with having a driving licence as a predictor.

Variable	R2	F	β
Driving licence possession	0.027	28.002 *	0.16 *

Legend: R²—coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

), as it is considered a prerequisite for owning and using a car. The results show that the model is statistically significant. Therefore, having a driving licence is a significant positive predictor of students’ CF (β = 0.16, p < 0.01). The coefficient of determination (R² = 0.027) indicates that the predictor variable explains 2.7% of the variance in the criterion variable.

The analysis in

Table 9. Regression analysis with modal choice as a predictor.

Variable	∆R2	F	β
Block I	0.171	104.52 *
Car (driver)			0.42 *
Car (passenger)			0.04
Block II	0.005	3.01 *
Car (driver)			0.45 *
Car (passenger)			0.06
Bicycle			−0.04
PT			0.06
Block III	0.045	28.92 *
Car (driver)			0.59 *
Car (passenger)			0.11 *
Bicycle			−0.06
PT			0.20 *
PT combined with a car			0.24 *
PT combined with walking			0.09 *

Legend: ∆R²—change in the coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

uses mobility CF as the criterion variable and the primary commuting mode as the predictor. As each respondent could indicate only mobility mode, this variable was treated as multi-categorical and coded using dummy variables. The reference category was set to “walking” due to the low CF it results in.

In the first block, which includes motorised modes of transport as predictors, only travelling by own car has a significant positive impact on CF (β = 0.48, p < 0.01). The “car (passenger)” variable in this block was insignificant (β = 0.04, p > 0.05), indicating that driving a car (driver) contributes to a larger CF, whereas being a passenger has no significant impact. By adding more sustainable modes of transport to the second block, the beta coefficient for car driving remained significant and relatively stable (β = 0.45, p < 0.01), while being a car passenger was still not significant (β = 0.06, p > 0.05). Cycling and public transport were insignificant predictors of CF in this block (cycling β = −0.04; public transport β = 0.06; both p > 0.05). These results indicate that when considered independently of other modes of transport, sustainable modes of transport alone were not significantly associated with CF. In the third block, the model also included combined modes of transport, which resulted in changes to the significance of the individual predictors. Driving a personal car remained a significant and strong predictor (β = 0.59, p < 0.01). Riding as a car passenger (β = 0.11, p < 0.05) and using public transport (β = 0.20, p < 0.01) also became significant, while cycling remained insignificant (β = −0.01, p > 0.05). Additionally, the combination of public transport and car (β = 0.24, p < 0.01) and public transport and walking (β = 0.09, p < 0.05) were found to be significant CF predictors. The R² value in the first block indicates that the regression model explains 17.1% of the variance, which is the largest proportion in this study. Along with the R² values in the other two blocks (an additional 0.5% and 4.5%), transport mode choice explains the largest proportion of the CF data, which is not surprising.

The frequency of class attendance (

Table 10. Regression analysis with class frequency as a predictor.

Variable	R2	F	β
Class frequency	0.021	21.17 *	0.14 *

Legend: R²—coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

) and the practice of sharing rides with others (

Table 11. Regression analysis with carpooling inclination as a predictor.

Variable	R2	F	β
Ride-sharing	0.016	12.83 *	−0.13 *

Legend: R²—coefficient of determination; F—F-ratio; β—beta coefficient; *—statistically significant result.

) are statistically significant predictors of CF of student mobility.

The results in Table 10 confirm that class attendance (i.e., the frequency of trips to campus) is a statistically significant positive predictor (β = 0.14, p < 0.01) of the environmental load of students’ mobility. Unlike attendance frequency, carpooling is found to be a negative predictor of student mobility environmental load (Table 11).

Students who share their rides more often have a lower CF. Ride-sharing is also a significant negative predictor of mobility-related CF (β = −0.13, p < 0.01). The size of the coefficient of determination indicates the low explanatory potential of the independent variables in both cases, i.e., the model does not align well with the actual observed values of CF.

After analysing all the variables that were identified as under-researched in previous studies on student mobility, it can be concluded that although there are differences in the level of student environmental load based on these factors, not all of them are statistically significant predictors of CF when it comes to student mobility at a suburban campus. In addition, there are no variables with an R² value of less than 1% that would be dismissed from further analysis. The results are examined in the context of previous studies in the Discussion Section.

4. Discussion

Some studies indicated that students from households with a larger number of family members are more inclined to use PT, bicycles, or walking [36], whereas others [20,48,71] found that these students make less sustainable modal choices. The exploration of the influence of household size on mobility patterns and sustainable mobility, as presented in this study, resulted in the finding that the student population mobility in relation to a suburban campus demonstrates greater sustainability (lower CF) in households with two to four people, which is not the case in larger households. However, household size is not a statistically significant predictor of CF resulting from student mobility, meaning that the differences could be case-specific.

Employment could act as a strong determinant of choosing a car as the primary mode of transport for young people [32], since students that work full-time and study are more inclined to modes they perceive as faster and enabling shorter trip durations (cars), unlike PT and walking, in comparison to students that do not work or work part-time [36]. Full-time employment or part-time studying has been shown to increase the likelihood of choosing cars over PT [20]. Considering that, in the studied case, more than half of the employed students commute by car, around one in four use PT, and only 7% walk to and from campus; the findings of this research support previous findings and demonstrate that respondents who work and study are inclined to make less sustainable mobility choices, which result in a much larger environmental load. The analysis confirms that student employment status is a predictor of CF. Employed students have a significantly greater environmental impact through mobility than students or unemployed students.

Past experiences influence students’ mobility patterns [50,70] and act as predictors of modal choices [71], which is why the region of origin is important according to some previous studies. Every relocation, whether at the beginning of the study or between academic years, provides an opportunity to influence a change in students’ mobility patterns [72]. Exposure to “new mobility environments” and different mobility options could encourage students to commute more sustainably, by walking, cycling, or using PT, depending on their availability [71]. The limited availability of alternative modes of transport is one of the key campus limitations in this research. Therefore, the region of origin and past mobility habits cannot benefit the campus area.

On the other hand, living away from home encourages less sustainable choices, such as choosing to be car passengers rather than using PT, or using PT instead of walking [20]. It also influences the need to live in term-time accommodation [14], as was evident in the case study. This research found that students who live in the same region as their university have a smaller CF than those who live in term-time housing, regardless of trip distances. This finding is consistent with some previous studies [73,74] but differs from others [14]. In terms of habitual modal choice and sustainable mobility, the region of origin is not a statistically significant predictor of CF in this study. However, the higher CF among students from other regions is largely due to the campus location, with the only student accommodation provided at the university campus located 20 km away. Previous studies have confirmed that the distance from a residence to a campus negatively affects the choice of alternative (active and collective) modes [55,75,76]. However, distances up to 1 km increase the inclination to walk [20]. For commutes of more than 5 km, students are more inclined to use motorised means of transport [36]. This research revealed that students living within 5 km of the campus (34% of the respondents) mostly walk to campus (39%) or use PT (28%), with an additional 7% combining these two modes. Shorter distances are covered by active modes in an even higher share. However, there are extreme examples in every distance range category, with some students covering distances of less than 1 km by car. Researched campus would, therefore, benefit from student accommodation capacity at the micro-location in order to achieve shorter commutes and encourage more sustainable mobility [24,36].

Table 12. Discussion of findings in relation to prior research—an overview.

Variable	Findings of Prior Studies	This Study’s Findings
Household size	Students from larger households are more inclined to alternative modes of transport [36]. Students from larger households make less sustainable modal choices [20,48,70].	Students demonstrate greater sustainability in households with two to four people, not the case in larger, nor in smaller households 1.
Student employment	Employment could be a strong determinant of choosing a car as a primary mode [32]. Students who work are more inclined to cars [20,36].	Employed students make less sustainable mobility choices. More than half of employed students commute by car. Student employment is a predictor of CF.
Region of origin—past experience	Past experiences influence students’ mobility [50,69] and act as predictors of modal choices [70]. Every relocation has the potential to influence a change in students’ mobility [71].	Limited availability of alternative modes limits campus benefits from past mobility habits. The region of origin is not a predictor, but there are differences in CF relative to region of origin 1.
Region of origin—living in term-time accommodation	Living away from home encourages less sustainable choices [20]. The distance from a residence to a campus negatively affects the choice of alternative transport modes [55,74,75]. Distances up to 1 km increase walking [20] Inclination to motorised means rises on commutes of more than 5 km [36].	Students living at home have a smaller CF than those living in term-time housing, regardless of trip distances. Less sustainable transport choices are largely due to the campus location. Students living within 5 km of campus mostly walk or use PT; shorter distances are covered by active modes in a higher proportion.
Availability of cars	Availability of mobility modes limits mode choice decisions [46]. Choice of mobility mode is related to mode ownership, especially in the case of motorised vehicles [76].	Family car availability is not a statistically significant predictor of students’ CF, owning a car is. Unlike ownership, having a family car at disposal when commuting to campus results in a smaller CF 1.
Class frequency	Students’ mobility patterns are determined by their class schedules [18,39].	Class attendance has a statistically significant influence on CF, and full-time students show greater CF.
Carpooling	The majority of students do not share rides [18], or seldom do it [38].	Carpooling with other students is a strong negative predictor of CF. The majority of students who drive share rides (approx. 70%); more if employed (72%).

¹ case-specific results (no statistical significance).

provides an overview of the discussion of the findings of this study on student mobility in relation to prior studies, including the aforementioned factors and those discussed in the text following the table.

Possession of a driving licence is one of the significant predictors of mode choice, particularly in the case of car usage [47]. The proportion of students holding a driving licence in this study is higher than in some previous studies [24,33,36,47,56], which could influence differences in the findings. Nevertheless, driving licence possession has been confirmed to be a significant positive predictor of the environmental impact of student mobility. Students who obtain this mobility tool demonstrate a higher level of CF than those who do not have a licence.

The data on holding the licence is closely related to the availability of cars to students, and both factors have a significant impact on mode choice and usage [36], especially when it comes to greenhouse gas emissions [10]. The choice of mobility mode is often related to mode ownership, especially in the case of motorised vehicles [77]. In this study, fewer than half of the students with a driving licence use cars as their primary mode of transport to campus. This should be further discouraged, or students should be encouraged to share rides. Having a family car available is not a statistically significant predictor of students’ CF, although it does cause differences in environmental load. However, owning a car is a strong predictor, as in the previous studies.

Car ownership decreases the likelihood of commuting by PT instead of by car [20]. However, owning a PT pass (preferably discounted) makes it more likely that students will regularly use PT for their commutes [36]. In this study, PT is found to be underutilised relative to its potential. It is evident that alternative modes of mobility are also underrepresented. For instance, the proportion of students who own bicycles is significantly higher than the proportion who actually use them for commuting. The negative effect of lacking adequate infrastructure that would encourage active mobility [18,51,78] in commuting to campus could be the reason why fewer students at the campus in this study adopt walking or cycling. The results of this research show that in the model in which all motorised and non-motorised mobility options are included (along with intermodal options), modal options such as being a car passenger and using PT are predictors of students’ mobility CF, as well as intermodal options such as combinations of PT and cars or PT and walking. Driving a car is a significant and strong predictor in all scenarios, whether the transport system offers limited or multimodal options.

The probability of active mobility is reduced by suburban location [36]. Therefore, it is recommended that the studied campus start incorporating measures to support pedestrian and cyclist access, parallelly enhancing sustainability and liveability [51]. Such measures should include the removal of barriers to active commuting [79] and the promotion of the complementarity of PT and active mobility modes [18,80]. The sustainability of suburban campus students’ mobility would benefit from simultaneous improvements to PT (e.g., decreased trip duration and price [20,47]) and restrictions to car commuting (increased parking costs on campus).

Mobility patterns that reflect mobility due to class schedules could be subject to management actions by the HEI, in accordance with the finding that students’ mobility patterns are determined by their class schedules [18,39]. The findings of this research (Table 3) indicate that, although the individual CF per trip is lower for full-time students (attending classes three to five times a week), the frequency of commuting has a significant impact on the total commuting distance of full-time students, which exceeds that of part-time students (attending classes once or twice a week) on both a weekly and an annual basis. The findings of the regression analysis have demonstrated that class attendance exerts a statistically significant positive influence on CF, indicating that full-time students demonstrate higher levels of CF. It is, therefore, worth noting that in correspondence with the multiple weekly attendance requirement, students’ inclination to active mobility modes on shorter distances and the higher PT usage on larger distances increase [20], and should be encouraged and supported by the university and the built environment interventions.

The inclination to engage in carpooling with other students has been identified as a strong negative predictor of CF. Despite the fact that the proportion of students who do not share their rides (31%) is approximately equivalent to the proportion of employed students in this research, who, by assumption, have rather limited options for carpooling while balancing work and study obligations and often covering larger distances to get to class, the analysis demonstrates the opposite—that the majority (72%) of this subgroup engage in ride-sharing. However, it is evident that approximately 32% of the students who possess their own vehicle do not engage in carpooling. Furthermore, around 37% of the students who utilise their family vehicle exhibit a similar pattern of behaviour. Parking management measures (pricing, availability, and time limitation) are often used in encouraging carpooling [21,38]. This approach not only fosters modal shift [81] but also contributes to a reduction in GHG emissions and enhanced efficiency within transportation systems [20]. The implementation of the proposed measures in conjunction with PT improvements [50] has the potential to yield favourable outcomes in the context of the present study. It should be noted, however, that the majority of the students who drive opt to share rides (almost 70%), which makes the results different from the findings of previous studies that highlighted that the majority of students typically drive alone [18] or seldom share their rides [38].

5. Conclusions

In the context of planning for sustainable mobility patterns within any given environment, it is crucial to identify and comprehend the determinants, that is, the factors of transport-related behaviour. The influence on mode choice and subsequent shift towards a less environmentally burdensome commute is contingent on the availability of mobility options. The latter is inextricably linked to regional transport policies and the built environment. In light of the substantial number of commuters to university campuses, it is evident that HEIs have the opportunity to make a substantial contribution to local and regional sustainability through the implementation of mobility management measures that are oriented towards reducing GHG emissions. Establishing priorities for such actions and investments must commence with a comprehensive understanding of the mobility patterns of the largest group of campus-related commuters (students) and the factors that influence their mobility choices.

A comprehensive review of previous studies indicated that specific factors influencing student mobility have received minimal academic attention. The present study hypothesised that these factors are important predictors of the environmental load (CF) of students’ mobility in suburban surroundings.

CF calculations confirmed that there are differences in CF among students in relation to the following factors: household size, students’ employment status, region of permanent residence (in relation to term-time accommodation and mobility habits), car availability, class attendance, and carpooling tendency. However, the regression analysis revealed that household size, region of residence, and the availability of a family car did not significantly predict students’ environmental load. Nevertheless, the analysis demonstrated that other factors (identified as employment status, primary mobility mode choice (car (driver) in cases of lacking options, or car (driver and passenger), PT, and intermodal connections in cases of multiple options), class attendance frequency, and inclination to carpooling) are significant predictors of sustainability of student mobility in the context of the suburban campus location. These statistically significant predictors of the sustainability of student mobility should be included in future research to explore student mobility patterns more thoroughly.

The campus location setting, distant from the main university centre and student facilities, changes the mobility context related to students’ commuting options. The underlying issue in the case of the studied campus is the lack of alternative mobility options. This finding is also confirmed by the hierarchical results of the choice of modality. Remoteness from student facilities (lack of university housing at micro-location) also exerts a significant influence on the daily mobility of students. In order to facilitate more sustainable student mobility, HEIs (i.e., campus management) in question should consider the implementation of measures that encourage ride-sharing among students. Such measures may include introducing parking management measures, such as parking charges or prioritising car occupancy. Furthermore, it is essential that HEIs collaborate with PT operators to facilitate PT and active mobility intermodal connections. Simultaneous improvements to PT are also crucial, with a reduction in trip duration being achieved through the negotiation of direct lines to campus or the subsidisation of travel costs for students. Finally, restrictions on commuting by car must be implemented, including the imposition of increased parking costs on campus. Campus management should consider available (best practice) measures to encourage short-distance commuters to use active mobility modes. In the survey conducted for this research, students suggested the following incentives to motivate their modal shift: introducing a competition system or financial incentives for active commuters; installing simple infrastructure solutions (e.g., bicycle racks) that they could freely use; and the provision of education/workshops on the benefits of active and more sustainable mobility. The long-term sustainability of the campus would be significantly enhanced by the provision of student housing that is both appropriately sized and located in close proximity.

This research contributes to the existing body of knowledge by 1. focusing exclusively on students; 2. exploring mobility patterns related only to compulsory student activities; 3. case studying a suburban campus; 4. encompassing a variety of transport modes; 5. identifying and analysing factors that have received little academic attention; and 6. quantifying the sustainability of student mobility by CF. This paper makes a contribution to studying student mobility and the quantification of the environmental load that results from it at a local level. The identified factors had not previously been subject to quantification as the actual carbon footprint size, specifically in relation to a suburban campus location.

Shifting the focus away from urban campus-related mobility research, which is prevalent in the field, is both a limitation and a strength of this study. It limits the general applicability of the findings due to the influence of specific location factors on mobility decisions. However, the lack of research into suburban campus-related mobility provides an opportunity to contribute to filling the existing research gap. An important limitation of this study is the under-representation of part-time students, whose mobility patterns differ from those of full-time students due to more condensed schedules and fewer commutes to campus. However, focusing on compulsory attendance-related commutes minimises the limiting impact, as part-time students have no class attendance obligation.

In addition, applying the findings to urban campus locations and multi-location campuses is a promising direction for future research. Future research could also explore mobility in relation to suburban campuses by adding the factors presented in this paper to the sets of factors that have been more widely explored or could provide an interesting context, such as attitudinal or psychosocial factors. These factors were not the focus of this paper, but could provide interesting insights into managing student mobility in a way that is less environmentally burdensome. Furthermore, the potential indirect or interaction effects of variables with low R² (e.g., household size or region) present new research opportunities to delve deeper into these variables.