Factors affecting college students’ attitudes towards carpooling

Abstract

As a sustainable mode of travel, carpooling is an effective solution to environmental and energy problems, but it has not been fully utilized. The research on carpooling attitudes, especially for college students, still needs to be improved. In response to this problem, this paper aims to investigate the effect of personal attributes, travel attributes, attitude attributes and other attributes of students on carpooling attitudes in China. To collect the data, an online questionnaire survey was implemented at Shenzhen University, and 514 students participated. Based on the samples, the multinomial logit model is used to explore the contribution of all variables. The model results showed that personal attributes such as sex and age had no significant effect on the attitude of college students towards carpooling. The safety of carpooling and its high cost make college students have a neutral attitude towards it. The most concerning factor for college students in carpooling is comfort. These findings can provide valuable suggestions for measures to be taken in response to different attributes affecting students’ attitudes and are particularly important for the university and government to make optimal decisions to motivate students to select carpooling.

1. Introduction

Currently, many countries have started to promote green transportation in the face of environmental problems. Carpooling, as an eco-friendly transportation mode, can reduce the number of automobiles needed by travellers, promote the reduction of energy consumption and gas emissions and decrease parking demands [1,2]. According to a recent report by the international energy agency, carpooling has been considered to have the greatest potential for saving fuel among all measures [3]. In Beijing, it is estimated that the direct energy savings from carpooling are approximately 26,600 tons of standard coal per year and the annual emission reductions of CO₂ and NO_x are approximately 46,200 tons and 253.7 tons [4]. Although carpooling has so many advantages and has been practised to some extent around the world, it is still not fully used and promoted.

Some studies show that the percentage of college students selecting carpooling is still relatively low. Khattak et al. [5] conducted a quantitative analysis of the travel patterns of college students at four major universities in Virginia and found the proportion of college students selecting carpooling ranges from 11% to 15%, which is significantly lower than the proportion of the total population. Akar et al. [6] reported the same findings that only a few college students try carpooling. The above results also showed that the travel behaviour of college students is significantly different from that of other populations, predominantly during off-peak periods. Balsas [7] studied how college students are encouraged to shift from cars to other travel modes, noting that they are in a privileged position to be able to spread sustainability and help reshape social transportation patterns. Thus, the college student is a significant potential group for carpooling.

In previous studies, college students’ attitudes towards carpooling have received little attention. Zhou [8] conducted a study at the University of California, Los Angeles, and found that students carpooled at a higher rate than college employees. Tezcan [9] analysed the potential for carpooling among undergraduate students and found that females and frequent campus travellers were more likely to select carpooling. Moreover, as trips get longer and costlier, students are more likely to choose to carpool to travel. The factors affecting the attitudes of college students are different and unclear. Investigating the factors that influence their attitudes towards carpooling can help suggest some measures to promote it. Thus, it is necessary to conduct a study that can reveal the contribution of attributes of the college student group in selecting carpooling.

2. Literature review

The literature of all groups has been reviewed and summarized to find the potential contributing factors and make the questionnaire. For example, Bulteau et al. [10] explored the possible determinants of the use of carpooling for commuting among a comprehensive set of socio-demographic, socio-economic, interpersonal and contextual variables. Mitropoulos et al. [11] described carpooling in terms of online platforms, user factors and barriers affecting carpooling services. Based on the existing literature, the factors affecting carpooling attitudes are summarized in the following categories: socio-demographic characteristics, travel characteristics and subjective attitudes and perceptions.

2.1 Socio-demographic characteristics

Socio-demographic characteristics, such as sex, age, income, occupation and education level, are essential factors that influence carpooling attitudes. For instance, Koppelman et al. [12] studied the factors influencing travel mode choice, and the results suggested that sex and the number of cars owned in the household played important roles in the propensity of carpooling. Specifically, women and individuals in households with fewer cars were more likely to carpool, whereas people with variable work schedules, trip chaining and higher incomes were less likely to carpool. Dias et al. [13] found that carpooling users tended to be young, well-educated, higher income, employed and residing in higher-density neighborhoods. Socio-demographic characteristics are all personal attributes of the travellers and are necessary to be considered in the questionnaire.

2.2 Travel characteristics

Travel characteristics also significantly affect carpooling, including travel mode, purpose, cost, travel distance, travel time and so on. For example, many studies have found that cost savings are among the most important factors influencing people to carpool [14–16]. Travel purpose and travel time also affect the attitudes of residents towards carpooling. O’Fallon et al. [17] found that people tend to choose cars for commuting if they need to transport children to school during their commuting. And Habibian and Kermanshah [18] indicated that residents are more likely to choose public transport to avoid morning or evening peak congestion.

2.3 Subjective attitudes and perceptions

Subjective attitudes and perceptions, including people’s attitudes towards carpooling, the environment and different contextual models, are essential determinants of carpooling. Amey [19] demonstrated the importance of people’s attitudes in carpooling decisions and found their attitudes towards carpooling to be a significant barrier to increasing carpool participation. Attitudes determine the level of trust among persons, and a positive attitude promotes better carpooling development. For example, Correia et al. [20] studied the importance of attitudes in influencing the propensity to carpool, and the results showed that attitudes play an important role in carpooling decisions.

2.4 Summary

College students are different in the following dimensions than the general population, particularly employees. First, there is a continuous movement of college students throughout the day and irregular schedules. They have some control over their course schedule and commute time, which can often allow them to avoid the two peak hours typically faced by most employees. For example, Khattak et al. [5] found that college students’ single peak commute was around midday, while employees’ peak hours were around 7:00 am and 5:00 pm. Second, most college students are unmarried and/or have no children, and therefore do not have any daycare responsibilities to constrain their carpooling options. For example, Park et al. [21] examined the factors affecting individual carpooling decisions and role preferences, and the results showed that carpooling interest and choice were related to personal attitudes. The young people and graduate students were more willing to participate in carpooling. Finally, carpooling is more attractive to college students as a younger and busier population group with relatively low incomes. Ayaz et al. [22] analysed the factors of travellers’ attitudes towards different carpooling scenarios. The results showed that most of the surveyed respondents are oriented towards using carpooling services for their educational (university/school) purposes, meaning carpooling will be more attractive for people like students. Another study by Shaheen and Martin [23] in Beijing also showed that those interested in carpooling tended to be younger people and college students. The university has a unique group of young people and a more active group of individuals. Balsas [7] reported that the unique educational environment of college campuses is a privileged place to communicate sustainability and help reshape society’s transportation patterns.

Based on reviewing the literature and considering the difference between college students and the general population, this paper comprehensively selects personal attributes, travel attributes, attitude attributes and other attributes to study the factors that affect the attitude of college students towards carpooling. To investigate the factors contributing to the attitude towards carpooling in the university, a questionnaire survey of students was conducted at Shenzhen University as an example, and a multinomial logit model is used to explore the relationships between independent and dependent variables.

3. Data

Shenzhen University has 41,164 students, including 28,425 undergraduates, 11,887 graduate students, 561 doctoral students and 291 international students. In this paper, 514 questionnaires were collected through the online questionnaire, among which 463 were valid, with an efficiency rate of 90%. The questionnaire includes four parts: personal attributes, travel attributes, attitude attributes and other attributes. The first part includes personal information such as sex, age, educational background, college major, monthly living expenses and whether they are single. Based on the feedback from a pilot survey of a few students, it was found that the average monthly living expense is about 2000 CNY, and so this is used to evaluate the consumption level of a student. The second part includes the daily travel mode, travel purpose, travel time, travel distance, etc. In addition, the route from Shenzhen University to Shenzhen North Station is designed to observe the impact of the combined cost of time and money on carpooling attitudes. The third part is an intuitive survey of carpooling attitudes, including attitudes of college students towards the effect of carpooling on traffic congestion and psychological attitudes towards carpooling with the opposite sex. The fourth part investigates the carpooling information platform and the way to obtain carpooling information, the reasons why college students are reluctant to carpool, the most concerning factors about carpooling and suggestions for improvement.

The survey results are shown in Table 1. It can be seen that 52% of the respondents were male, indicating that the ratios of male and female are relatively close. The difference between the number of undergraduates and the number of graduate students in the questionnaire is also tiny, while the difference between the number of science subjects and the number of arts subjects is relatively large because Shenzhen University has a high proportion of science students. In addition, 62.2% of the students’ monthly living expenses are less than 2000 CNY. Another interesting finding is that most of the respondents are single. Most respondents' daily travel mode is public transportation, and more than half of students’ daily travel purpose is for leisure entertainment. Among the respondents, 66.1% of the student dormitories are less than 1000 metres from the nearest bus stop or subway station and more than 60% prefer to travel during off-peak periods. The daily travel habits of students are still very different from the general public. As shown in Fig. 1, more than half of the respondents were neutral about carpooling, and only 10% were opposed to carpooling. Attitudes towards carpooling suggest that it is a familiar concept or that college students believe that it is possible but just not yet widely used among college students. Many respondents said that the current market in carpooling is not formal enough, security is not guaranteed and more than one carpool will increase travel time. Among the respondents, the top two concerning factors of carpooling are travel costs and security. The suggestion that a reputable and safe carpooling information platform needs to be built is put forward [11, 24].

4. Methodology

This paper uses the multinomial logit model to explore the relationships between independent and dependent variables [25–27]. The independent variables contain personal, travel, attitude and other attributes. The multicollinearity of the independent variables has been checked and the results show no multicollinearity among the independent variables [28–30]. The dependent variable is the attitude of college students towards carpooling, which is classified into three discrete categories: support, against and neutral. The probability that student n has an attitude of i towards carpooling is written as:

where P_n(i) is the probability that the attitude of the nth student for carpooling is i and U_ni is a function to determine the utility function of the corresponding attitude. To estimate this probability, the linear function of U_ni is expressed as:

where β_i is a vector of estimable coefficients for carpooling attitude i, x_ni is a vector of explanatory variables and ε_ni is a random vector to explain unobserved influences on carpooling attitude.

If the random variable ε_ni obeys the general assumption that it is independently and identically distributed (IID) and belongs to the extreme-value distribution, then the multinomial logit model can be expressed as:

where all variables are defined as before, the coefficient β_i is estimated using the maximum likelihood method and the coefficient of the reference category is considered to be 0. In this study, neutral was used as the reference category.

Due to the nonlinear nature of the multinomial logit model, the estimated coefficients of the independent variables do not reflect their direct effects on the dependent variable. The odds ratio parameter is used to assess the effect in this study. If the odds ratio value is equal to 1, it means that the factor does not influence the choice of carpooling attitude; if the odds ratio value is greater than 1, it means that the influence of the factor on the carpooling attitude is negative; if the odds ratio value is less than 1, it means that the influence of the factor on the carpooling attitude is positive.

Before the modelling, variables are firstly selected based on the Akaike information criterion (AIC) which measures the goodness of fit of a statistical model. AIC is based on the concept of entropy and provides a criterion for weighing the complexity of the estimated model and the goodness of the fitted data. The basic formula is defined as:

where K is the number of unknown parameters in the model and L is the likelihood function. Among a set of available models, the smallest AIC is usually selected as the best model.

The filtering of variables is divided into two main steps. The first step is to screen out variables significantly different from college students’ carpooling attitudes. Suppose X = {x₁, x₂,..., x_s} is the set of significant variables and S = {s₁, s₂,..., s_m} is the set of insignificant variables. Then the set X is taken as the independent variable, carpooling attitudes are taken as the dependent variable to carry out multinomial logit modelling and the obtained model is model_x.

In the second step, select any n (1 ≤ n ≤ m) elements from the set S for combination and |$N\,(N=\sum _{n=1}^mC_m^n)$| combinations can be obtained, denoted as T = {T₁, T₂,..., T_N}. The multinomial logit modelling is performed that the set V = {T₁∪X, T₂∪X,..., T_N∪X} elements are taken as independent variables, and carpooling attitudes are taken as dependent variables. The obtained set of models is M = {model₁, model₂,..., model_N}. The model with the smallest AIC value is selected from the set M and compared with the AIC value of model_x. Finally, the model with the smallest AIC value is selected as optimal.

5. Results and discussions

The results of statistically significant variables in the multinomial logit model are shown in Table 2, including coefficient estimates, standard deviations, p-values and odds ratios of the independent variables. To compensate for the reduction in efficiency, we use a more generous confidence level of 90% instead of the traditional 95% [31, 32].

5.1 Travel attributes

From Table 1, it can be seen that the majority of students select public transportation as a daily travel mode. One of the reasons is that the transportation around Shenzhen University is convenient, and the subway and bus lines can directly reach the campus entrance. As shown in Table 2, the coefficient estimates for public transportation are negative, and the odds ratio is exp(−1.060) = 0.347 < 1. The results mean college students whose daily travel mode is non-public transportation are more likely to oppose carpooling compared to college students who choose public transportation. The finding is consistent with previous research. Javid et al. [33] obtained the same results and found that college students whose daily travel mode is non-public transportation care about personal safety, comfort and privacy.

In terms of travel purposes, the estimated coefficient for leisure entertainment and part-time work are negative. The odds ratio of part-time work is exp(−0.926) = 0.396, and the odds ratio of leisure entertainment is exp(−1.766) = 0.171. The results show that college students who travel for part-time work and leisure entertainment hold a neutral attitude towards carpooling compared with those who travel for other purposes. The result may be because the current cost of carpooling is higher than the subway and bus prices in the market. As a group without income, travel costs are very important to college students. In addition, when the purpose of travel is part-time work and leisure entertainment, carpooling may not provide them with flexible and available options. Thus, college students may prefer to use other options instead of carpooling.

From the result of parameter calibration, the odds ratio of carpooling willing to travel during holidays is exp(0.707) = 2.029 > 1, which indicates that college students who are willing to choose to carpool during holidays are inclined to support carpooling. The reasons may be divided into the following two points. First, there are a lot of tourists travelling on holidays, which is the peak travel period. College students choosing to carpool can save the cost of subway or bus transfer time and avoid traffic congestion. Second, the transportation costs generated by holiday travel are also a considerable expense. Considering the travel costs, some college students choose to share the carpooling costs with others, which can save travel costs.

It is found that college students with carpooling experiences tend to support carpooling. This is consistent with some previous findings. Cao and Mokhtarian [34] pointed out that an individual’s experience greatly affects their consideration of travel-related strategies. The reason may be that carpooling reduces the pollution from car driving and the probability of car congestion. Also, participants believe carpooling is comfortable, safe and low cost [35]. Carpooling experiences help carpoolers decide if carpooling is a good travel choice, and the result shows that people with carpooling experience are indeed inclined to support it.

5.2 Attitude attributes

The estimated results show in Table 2 that the attitude that accepts carpooling with the opposite sex obviously influences carpooling. The odds ratio is 0.246, which means that college students who do not accept carpooling with the opposite sex are more likely to oppose carpooling. One possible reason for opposing carpooling with the opposite sex may be that carpooling with the opposite sex will squeeze the space in the car, and even some members of the opposite sex have a peculiar smell, which brings a bad carpooling experience. Another reason may be due to people’s distrust of others. It is consistent with earlier findings that the main barriers to carpooling are distrust towards sharing trips with strangers [36]. On the contrary, Ayaz et al. [22] found that females, whether single or married, are more inclined to ride-share with males, and the same applies to married males with females.

5.3 Other attributes

The results show that the odds ratio for the reason of carpooling is not formal enough and safety is not guaranteed is 0.478, which indicates that college students with higher safety requirements have a neutral attitude towards carpooling. The odds ratio for the reason that carpooling is still more expensive and not as cheap as bus and subway is 0.379, indicating that college students who are more concerned about the cost of carpooling have a neutral attitude towards carpooling. The result may be because some carpooling vehicles are on the road without business certificates. Many college students have little legal awareness, and illegal carpooling services will bring significant security risks. This is consistent with earlier findings that travelling with other passengers may reduce the feeling of safety for potential carpooling users [37]. In addition, Ashraf Javid and Al-Khayyat [38] identified three crucial factors affecting college students in carpooling through factor analysis, including safe and relaxed travelling, friendly and cheaper travelling and carpooling personal norms. Delhomme and Gheorghiu [39] and Ciari et al. [40] also mentioned that carpooling costs are an essential factor affecting college students’ carpooling. The results in the present paper are consistent with previous studies.

The results show that travel time is more important than low-carbon and environmental protection for college students choosing carpooling services. They show that schedule reliability is also a very important factor, the biggest concern of carpooling participants is the possibility of missing travel time [35]. Such results are consistent with previous studies [16, 41]. An important condition for carpooling with others is that the carpoolers have a common destination or the destinations they want to reach is in the same direction, and the travel time should be similar to meet the requirements of carpooling. Therefore, students with higher requirements on travel time are more likely to object to carpooling.

In terms of the online information platform for carpooling, the results show that college students living in cities with online carpooling information platforms are more likely to support carpooling. Although Shenzhen has many carpooling websites, there is no platform that serves college students specifically for carpooling. According to the findings of Liakopoulou et al. [42], the main characteristic of such a carpooling platform is the official provider of a user-friendly platform organized by the university, ensuring their security and the cost calculation. In addition, the results show that college students are more likely to support carpooling if a specific carpooling information platform for students is available. If a city (e.g. Shenzhen) has an independent, safe and reliable carpooling information platform for students, the potential for carpooling among college students would be fully tapped.

5.4 Policy recommendations

The findings of this study suggest that a specific carpooling platform should be established to serve college students. The following functions should be embedded into the platform. Firstly, the platform should be connected with third-party carpooling enterprises (e.g. Didi), and the legitimacy of the enterprises should be audited by strict rules. Meanwhile, a real-name certification system should be provided to drivers and passengers by the platform. Secondly, the user could sort and filter information. For example, passengers can select the information that meets their carpooling requirements, such as carpool time and location, sex and age of the carpooling partner and the number of carpoolers. Finally, users could evaluate drivers and share their experiences. Passengers can evaluate and score carpool participants and drivers at the end of each carpool, and the results can be shared on the platform. And personal reputation points can be established based on the evaluation and scoring. To promote the platform, the following measures should be taken by different roles.

5.4.1 Governments

First, the government should work with universities to build a formal and safe carpool information platform specifically for college students and publicize reliable information to college students through the internet and other communication media. Second, the government also needs to formulate relevant laws and regulations to clarify the rights and obligations of college students and drivers, as well as the division of responsibilities in case of accidents and disputes to protect the legitimate rights and interests of both parties. Finally, the government should formulate relevant preferential measures to guarantee the legality of carpooling and encourage college students to carpool. For example, free parking permits or financial subsidies can be issued to carpooling vehicles. In addition, a carpool discount strategy can be given based on the number of carpoolers. The same trip should be provided to no more than four passengers at the same time, so the maximum number of passengers in the car is five, including the driver. If you carpool with one person you get 10% off, if you carpool with two people you get 15% off and if you carpool with three people you get 20% off.

5.4.2 Universities

Universities should collaborate with the local government to establish a carpooling information platform for college students. The carpooling opinions of college students should be collected by the schools, and the relevant carpooling information should be updated regularly. In addition, regular lectures should be held to popularize knowledge about carpooling for college students and encourage them to carpool through formal channels.

5.4.3 College students

As carpooling practitioners, college students should improve their cognitive level of carpooling, learn relevant legal knowledge and increase their legal literacy. They also should learn to defend their legitimate rights and interests with the law when encountering problems in carpooling.

5.5 Limitations

Several limitations should be pointed out. First, the collected sample size of carpooling is limited. Second, we did not measure objective carpooling use, so the analyses were based solely on self-reported measures, known to be sensitive to social desirability [43]. More variables can be added for further research to build a more comprehensive model. Third, the data collected in this study were based on Chinese students, which may be less applicable to foreign schools or the general population. Due to cultural differences, students in different countries have many different characteristics [44]. For example, Chinese college students are generally boarding, and their families provide their living expenses. In contrast, college students in most Western countries tend to rent houses and rely on part-time work to cover their living expenses.

6. Conclusions

This study has investigated the factors influencing the attitude of college students towards carpooling using a multinomial logit model. An online questionnaire survey of 463 college students was conducted at Shenzhen University in China. The influencing factors include personal, travel, attitude and other attributes. The main conclusions can be summarized as follows:

1. Individual attributes such as sex and age have no significant impacts on the attitude of college students towards carpooling.

2. In terms of travel properties, college students whose daily travel mode is non-public transportation are inclined to oppose carpooling, while college students whose travel purpose is for part-time work and leisure entertainment have a neutral attitude towards carpooling. College students willing to travel by carpooling during holidays are more likely to support carpooling, and carpooling experience has a significant and positive effect on supporting carpooling.

3. College students who could not accept carpooling with the opposite sex are more likely to oppose carpooling.

4. The most important concern for college students to choose to carpool is travel time. Carpooling safety and cost may be the key reasons why carpooling has not become a popular travel mode among college students. Most importantly, a safe and reliable carpooling information platform specifically for students will make college students more likely to support carpooling.

These results are of great significance to promoting the carpooling market and formulating campus transportation policy in the future. Although this study focuses on a specific population from a large university setting, the results can still provide general policy making and be applied to other university campuses.

Acknowledgement

This work was supported by the National Key R&D Program of China (Grant No. 2020YFB2103503).

References