Exploring Melbourne Metro Train Passengers’ Pre-Boarding Behaviors and Perceptions

Abstract

The focus on sustainable transportation has increased interest in promoting sustainable modes of transport, such as rail. Understanding train passengers’ behaviors and perceptions is essential to enhance their travel experience and increase train ridership. Pre-boarding behaviors and perceptions are crucial in shaping the overall train travel experience. However, there are limited studies that have developed a systematic framework for investigating train passengers’ pre-boarding behaviors and perceptions. This paper examines the train passenger’s pre-boarding behaviors and perceptions about the station and platform. The study adopts a mixed-methods approach by developing a pre-boarding decision framework and combining it with questionnaire surveys to explore passengers’ behaviors and perceptions on the platform before boarding. A total of 429 valid responses from Melbourne metro train users were used for analysis. Descriptive statistics and correlation techniques were applied to identify patterns and relationships. The findings reveal common pre-boarding behaviors and perceptions. Furthermore, the study uncovers factors influencing these behaviors and perceptions, such as passenger demographics, travel patterns, and specific trip characteristics. For example, carrying large items and travel frequency significantly impact passengers’ travel experience in the pre-boarding phase. Waiting time, group travel, carrying small items, gender, and age group also significantly impact some pre-boarding behavior variables. Travel time, on the other hand, makes no significant impact on any of the pre-boarding variables that we examined. This research provides valuable insights for rail service operators and policymakers to enhance the pre-boarding experience, optimize station design, and improve passenger satisfaction.

1. Introduction

The increasing concern for sustainable transportation has encouraged researchers and policymakers to explore ways to promote environmental-friendly modes of travel. In this context, rail transportation has gained significant attention as a sustainable alternative to other modes. Understanding the behaviors and perceptions of train passengers is crucial for enhancing their travel experience and hence promoting train ridership.

The rapid expansion in highly populated cities has presented many challenges in public transportation. Melbourne, a major Australian city, also faces similar challenges. The infrastructure of the Melbourne Metro network is vast, consisting of 998 km of track, 15 lines, and a total of 222 stations. The network operates with over 220 trains, carrying approximately 450,000 commuters on a daily basis [1]. Understanding how passengers behave as group norms or why individuals behave the way they do may lead us to a better solution to the station and platform design and service planning. However, the most recent study that we found about customers’ perceptions of Metro Train services in Melbourne was published more than a decade ago [2]. The level of services that the train operator aims to provide to the public could directly or indirectly influence every train user’s satisfaction [3,4]. On the other hand, the behavior of each passenger could also positively or negatively influence service performance.

One of the key motivations behind our research is the recognition that passengers’ behavior on the platform is critical for the safe and efficient operation of the train services. Existing studies have demonstrated that passenger behavior during boarding and alighting significantly impacts train dwell time [5,6,7,8]. By understanding the specific pre-boarding behaviors and decision-making processes of passengers, we aim to address the challenges in station operation. Another key consideration is the observation of overcrowding at various locations along the platform. The existing literature shows that uneven passenger distribution along the platform can cause irregular and extended dwell times [9]. The passenger cluster can also increase the risk of falling. By investigating passengers’ pre-boarding behaviors, perceptions, and decision-making processes, our research aims to contribute to the development of strategies and interventions that enhance both the efficiency and safety of train operations.

Several studies have examined passengers’ behaviors or perceptions, each with a different focus [10,11,12,13]. Most of these studies have investigated user-induced delays. However, a systematic framework for understanding the relationship between passengers’ perceptions and their pre-boarding behavior is lacking in the literature. This study aims to fill that gap by systematically exploring the pre-boarding behaviors and perceptions of train passengers.

The relationships between passengers’ behaviors/perceptions and various traveler/trip characteristics are also examined. By identifying and examining the factors that may influence passengers’ behaviors on the platform, one can gain valuable insights into passenger behaviors and attitudes before they board the train. This knowledge is essential for designing interventions and strategies that can improve the passenger experience and hence promote the sustainable mode of public transport.

2. Review of Relevant Literature

This section provides an overview of the relevant studies, especially focusing on pedestrian flow modeling, queue simulations, passenger crowding and distribution on the platform, and an approach to studying pre-boarding behaviors. By exploring these topics, we aim to identify key findings and knowledge gaps that contribute to the motivation of this study.

The examination of passenger behavior at an aggregated level has led to extensive discussions in the literature, particularly regarding pedestrian flow at platforms. Researchers have recognized the importance of understanding and managing pedestrian movements in order to ensure efficient and safe operations at train stations and platforms [14,15]. One of the earliest publications about the level of service (LOS) for pedestrian flow can be traced back to the 1970s. Fruin [16] introduced the concept of pedestrian level of service based on space occupied, flow rate, and speed of the pedestrians. Additionally, research has identified differences in walking speeds between senior individuals and younger adults [17]. Mathematical models for pedestrian movement have also been developed. It is often assumed that an individual’s decision is not just random but with certain uniformities, which can be determined by utility maximization [18].

The literature on queue simulations has provided valuable insights into passenger flow dynamics and on-platform behaviors. Many studies have taken the First-In-First-Out (FIFO) approach to model passenger movements on the platform. For example, Random-In-First-Out (RIFO) approach was used and compared with the FIFO approach in [19], which shows that the difference in passengers’ waiting time between the two approaches increases as the congestion level increase.

Passenger crowding is a common issue encountered by train passengers. It is often observed that during peak hours, crowding can result in increasing boarding and alighting time and extended dwell time. According to Station Planning Standards and Guidelines [20], a crowded situation is reached when density exceeds 2 passengers/m² along the walking area or above 5 passengers/m² in the waiting area.

It is also noticed that the crowding levels differ at different waiting areas along the platform. Both field studies and literature confirm that passengers waiting on platforms tend to position themselves near platform entrances, stairs, or escalators. For example, a field observation primary set up to study the dwell time at some Dutch stations revealed that passengers were not evenly distributed along the platform [21]. A similar phenomenon was found and studied in the UK. According to the video recording of nine trains at Oxford Station, the distribution of passengers along the platform was mapped. It is found the most concentrated area on the platform is within 16 m of the main entrance, where about 60% of the passenger board the train from that concentrated zone [22].

Various approaches have been employed in studying pre-boarding behaviors in the existing literature. The Social Force Model introduced in the 1990s by Dirk Helbing and Peter Molnár [23] has become a fundamental knowledge base for simulating and understanding human behavior. For instance, based on the adaption of the social force model, passenger behaviors are simulated at either the station level [24] or the platform level [25]. Apart from the social force model, Cellular automata models were also used to replicate individual behavior [26,27,28].

New approaches have also been adopted in recent years. An innovative approach to modeling the passenger flow and dwell time used an Origin-Destination (OD) matrix approach with fuzzy logic intelligence [29]. The Artificial Neural Network approach is used to forecast passenger flow [30]. Furthermore, agent-based modeling has emerged as a promising technique, as highlighted in a literature review paper [31]. Daamen [32] used the simulation software SimPed to model the pedestrains’ behaviors. An agent-based simulation approach was used to model passengers’ behavior on the platform. Moreover, a recent study introduced an integrated model combining two sub-models with different boarding strategies to accurately simulate uneven passenger distribution on the platform [9]. These advancements in modeling techniques have greatly facilitated investigations into passenger behavior.

Upon reviewing the relevant literature, it is found that the existing studies have primarily focused on modeling passenger behavior. The underlying motivation that drives different platform behavior is not well understood. There is a noticeable gap in research that explores passenger perceptions and the potential factors that influence their behavior. Furthermore, the literature lacks a systematic framework for understanding the relationship between passengers’ perceptions and their pre-boarding behavior.

3. Methodology

The research framework for pre-boarding assessment is proposed at the beginning of this section. It is then followed by a discussion of data collection and data analysis methods.

3.1. Pre-Boarding Assessment Framework

A conceptual framework was designed to identify and evaluate the factors influencing passengers’ decision-making processes before they board the train. Relevant survey questions in the pre-boarding phase include passengers’ general perception of the station and platform, their experience with delays, their route choice to the platform and waiting areas, their waiting and boarding behavior, and their opinions about the boarding queue, as well as the perceptions of passenger distribution on the platform.

The aspects that form the decision-making process outlined in the conceptual framework were synthesized according to the authors’ personal experience as train passengers, knowledge gained from field observations, and insights derived from relevant studies in the literature. Passengers’ familiarity with the station and platform shapes their general perception and affects their behavior [12,33]. Delays have been studied extensively in the literature, including user-induced delays [34]. Route and activity location choice is discussed as an essential component when studying pedestrian traffic flow on the platform [32,35]. Queuing behavior [19], boarding behavior [24,36], crowd behavior [37,38], and the uneven passenger distribution on the platform [21,22] are also studied in the existing literature.

The framework focuses on the pre-boarding phase, starting from the moment that passengers arrive at the station and ending when they step into the train (i.e., finish boarding). It is designed in a way that the main actions and decisions are represented in a flowchart, where relevant survey questions are placed closer to the actions or decisions. The survey questions were organized into two groups. Group one focused on the general perceptions of the station and the overall experience with delays, while group two focused on specific behavior choices and preferences.

Table 1. Linking survey questions to intended insights and implications.

Survey Questions	Intended Insights	Implications
Q1.1 Select the most suitable option—How familiar are you with the station layout?	Insights into passengers’ comfort and ease in navigating the station environment	Providing better navigation system and direction sign at station
Q1.2 Select the most suitable option—Have you experienced unexpected delays or extra waiting time on the platform?	Understanding the general experience of delays	Improving service frequency and schedule planning
Q1.3 Select the most suitable option—How often do you have to wait extra time on the platform due to the platform is too crowded?	Insights into the frequency of delay caused by overcrowded platform	Shaping the strategy to manage crowd on the platform and better managing dwell time.
Q2.1 Agree or disagree—I prefer the shortest path if there are multiple routes to the platform.	Understanding the trade-off in choosing the route to the platform (prioritize time/walking distance)	Designing clear signage and providing wayfinding information
Q2.2 Agree or disagree—I prefer the less crowded path to the platform.	Understanding the trade-off in choosing the route to the platform (prioritize comfort/safety)	Provide clearer and wider pathway
Q2.3 Agree or disagree—I notice that passengers are not evenly distributed along the platform.	Perception of uneven passenger distribution on the platform	Improving platform layout for efficient use of available space
Q2.4 Agree or disagree—I don’t mind waiting for the next train if the current one is too crowded.	Assessing passengers’ willingness to wait for the next train under the scenario of crowded carriage	Help determine the need to increase service frequency or carriage capacity
Q2.5 Agree or disagree—Regardless the crowd level, I just want to get onboard as soon as possible.	Understanding passengers’ boarding behavior (prefer boarding sooner than waiting)	Guide boarding process and manage passenger flow
Q2.6 Yes or no—Do you have a preferred waiting area on the platform?	Identifying the chance of having a preferred waiting area	Signage pointing to various waiting areas
Q2.7 Agree or disagree—I would rather walk longer distances for a less crowded waiting area.	Understanding the preferred choices in the selection of waiting area	Improving and optimizing waiting area design
Q2.8 Agree or disagree—I’d like to sit down and wait for the next train.	Exploring preferences for sitting down while waiting	Increasing or decreasing seating on the platform
Q2.9 Agree or disagree—I am often confused as where to line up for boarding.	Assessing passengers’ level of confusion when it comes to queuing and boarding	Boarding queue indication/floor marking

Note: Agreement scores are rated on a scale of 1–5, where 1 represents “strongly disagree” and 5 represents “strongly agree”.

provides a summary showing the linkage between the survey question, the intended insights, and its implication on station and platform planning.

This type of framework can be a useful tool for ensuring that all necessary considerations are taken into account. By mapping out the process in a flowchart, one can clearly see the steps involved and identify potential areas of concern or opportunities for improvement. Additionally, by including survey questions at relevant points in the flowchart, decision-makers can gather feedback and data to inform their choices and better understand the impact of their decisions. The proposed conceptual framework to assess passengers’ pre-boarding behavior is presented in Figure 1.

The framework starts by assessing passengers’ familiarity with the station layout through survey question Q1.1, “How familiar are you with the station layout?”. Introducing this question aims to provide insights into passengers’ level of comfort and ease in navigating the station environment, which could potentially influence their behavior and decision-making in the following process.

The framework includes questions (Q1.2, Q1.3) to understand passengers’ perceptions and experiences with delays. Specifically, survey question Q1.2 asks, “Have you experienced unexpected delays or extra waiting time on the platform?”. The purpose of this question is to gain insight into passengers’ overall experience with delays. The term “unexpected delays” subjectively reflects the actual time beyond the train schedule. The term “extra waiting time” is used to capture passengers’ perception of extended waiting periods. Survey question Q1.3 asks, “How often do you have to wait extra time on the platform due to the platform is too crowded?”. This question aims to gain further insights into the cause of the delay, which will help us identify and address issues related to overcrowding on the platform. By asking the passengers to report delays, we can understand how common these experiences are. These insights can help inform potential strategies to improve passenger experience and address issues related to overcrowding and delays.

Studies have shown that individuals consider multiple factors when making route choices, including travel distance, travel time, and comfort [39,40,41,42,43]. Survey questions Q2.1 (Agree or disagree—“I prefer the shortest path if there are multiple routes to the platform”) and Q2.2 (Agree or disagree—“I prefer the less crowded path to the platform”) examine the trade-off between time and comfort in choosing the route to the platform. The term “shortest path” that we used in Q2.1 is in relation to walking distance, while the “less-crowded path” in Q2.2 implies quicker or more comfortable walking. Having these two questions aims to understand how passengers perceive and prioritize different factors when navigating to the platform, which can inform the design of a wider pathway and better wayfinding system.

Once passengers arrive at the platform, the survey question (Q2.3) focuses on understanding passengers’ awareness of the uneven passenger distribution on the platform. Survey question Q2.3 (Agree or disagree—“I notice that passengers are not evenly distributed along the platform”) can be useful for understanding the perception of passenger flow and common behavior. It helps identify the need for improving platform layout to ensure efficient use of available space.

Passengers’ likely behavior in crowded situations is examined by survey question Q2.4 (Agree or disagree—“I don’t mind waiting for the next train if the current one is too crowded”). This question aims to understand passengers’ willingness to wait for the next train if the current one is too crowded. By knowing their willingness, transportation providers can determine if they need to increase the frequency of trains during peak hours or adjust the capacity of the train cars to meet the demand. Another likely behavior is to start boarding. Survey question Q2.5 (Agree or disagree—“Regardless the crowd level, I just want to get onboard as soon as possible”) aims to reveal passengers’ general boarding priorities, which provides complementary information to the specific crowded scenario set in Q2.4. By knowing whether passengers prioritize getting onboard as soon as possible over considering the crowd level, transportation providers can better design boarding processes and manage passenger flow.

The design of waiting areas plays a crucial role in platform configuration. Whether passengers have a preferred waiting area is assessed through survey question Q2.6 (“Do you have a preferred waiting area on the platform?”). Understanding whether passengers have a preferred waiting area can help improve station design and layout to better meet the needs and preferences of passengers. Additionally, this information can also help inform communication strategies, such as providing clear signs or directions for passengers to various waiting areas. The framework also explores passengers’ path choice to the preferred waiting area. Survey question Q2.7 (Agree or disagree—“I would rather walk longer distances for a less crowded waiting area”) aims to gain insights into passengers’ preferences and the trade-off between walking distance and the level of crowding in the waiting area. These preferences can be influenced by various factors such as the passenger’s age, gender, and travel purpose. Understanding these preferences can help train operators optimize waiting areas and improve passengers’ waiting experience, potentially leading to increased satisfaction and ridership.

Survey Question Q2.8 (Agree or disagree—“I’d like to sit down and wait for the next train”) aims to explore passengers’ preference for sitting down and waiting for the next train. This question provides insights into passengers’ desire for seating options near the waiting area. Understanding this preference can help identify the need for providing adequate seating facilities. However, it is worth noting that the decision to increase or decrease seat offering is also constraint by the physical infrastructure (available space) and the level of crowding. In the context of the Melbourne metro system, public seating is available on the platform for most stations, even for one of the busiest stations such as Melbourne Central or Richmond station.

Once the train arrives, passengers will get ready for boarding. Their next decision is whether to form a queue for boarding. From the field study, we observed that some passengers are confused about where to line up for boarding. The framework uses one question (Q2.9) to assess passengers’ level of confusion when it comes to the queue location. Survey question Q2.9 (Agree or disagree—“I am often confused as where to line up for boarding”) aims to assess passengers’ level of confusion when it comes to lining up for boarding. Understanding whether passengers clearly understand where to queue for boarding can help identify areas for improvement. If issues are identified from this question, effective signage or communication should be introduced to make the process smoother and less confusing for passengers. Clear and well-placed floor markings can help passengers queue in an organized and efficient manner, reducing confusion and frustration during boarding. This pre-boarding phase finishes when boarding is complete.

3.2. Passenger Survey

This study utilized an online survey, approved by RMIT University’s Human Research Ethics Committee (Reference: 2021-23822-13326), to collect relevant information for achieving the research objectives. The survey was conducted in collaboration with the market research firm Qualtrics. The target audience is Melbourne Metro train users who are aged 18+ and have experience with train travel. Their travel experience is not limited to any specific metro lines. Having comprehensive coverage ensures a representative sample of Melbourne metro train passengers. After a soft launch and some modifications, the survey was conducted at full scale in March 2021. With two outliers removed from 431 returned forms, a total of 429 valid samples were used for analysis. To verify the representativeness of the data, the distributions of age and gender groups were compared with the 2021 Census data. The result of Chi-square tests shows no statistically significant difference between the two data sources with regard to the gender distribution and age group distribution.

3.3. Data Analysis

Various statistical techniques were used to analyze and comprehend the survey data. Descriptive statistics were used to gain an understanding of passengers’ behaviors and perceptions. To assess the behavioral differences, the Chi-square test and Spearman correlation test are used. In total, twelve behavior/perception variables were included in the analysis. Those variables are derived from the survey questions that are labeled in the conceptual framework. The differences in behavior by gender were assessed using the Chi-square test, as gender is a nominal variable. If the Chi-square test identifies a significant relationship, Cramer’s V is then reported to show the strength of association. It is noticed that the cutoff values for interpretation of Cramer’s V vary in the literature. Taking the general guidance [44], practice [45], and the context of this study into consideration, we take Cramer’s V value of less than 0.05 as a negligible association; a weak association has Cramer’s V between 0.05–0.2; a moderate association has Cramer’s V between 0.2 to 0.3; a strong association has Cramer’s V greater than 0.3.

Apart from the Gender variable, there are seven other traveler/trip characteristic variables, which include Age group, Travel Frequency, Average Travel Time, Average Waiting Time, Group Travel, Carry Small Item, and Carry Large Item. Please find the descriptive statistics about passenger groups with different characteristics presented in Appendix A. Those variables were derived from survey questions that provided answers with a natural ranking (except for Gender); therefore, the Spearman correlation test is used to reveal the associations between those variables and pre-boarding survey variables. Spearman’s rank correlation coefficient (rho) is used to assess the strength of correlation. According to commonly used guidelines for interpreting the magnitude of the coefficient (Field, 2013) and the nature of this study, we take a weak correlation that has an absolute rho value range from 0.0–0.3; a moderate correlation has an absolute rho value range from 0.3–0.7; and a strong correlation has an absolute rho value range from 0.7–1.0. We have further specified the weak correlation band into “very weak” (rho < 0.1), “weak” (0.1 <= rho < 0.25), and “weak to moderate” (0.25 <= rho < 0.3).

4. Results

This section presents the findings from the proposed pre-boarding behavior assessment framework. Through descriptive analysis of twelve behavior/perception variables, insights into the passengers’ pre-boarding behaviors are gained. Furthermore, the main influences that contribute to passengers’ behaviors and perceptions are also examined and dissed in the following subsections.

4.1. Pre-Boarding Behaviors and Perceptions

A summary frequency table of passengers’ general experiences at the station and platform is presented in

Table 2. Frequency table of station/platform experiences.

Variable	Survey Questions	Category	Frequency	(%)
Station Familiarity	Q1.1 How familiar are you with the station layout?	Not at all familiar	6	1.4
		Slightly familiar	42	9.8
		Somewhat familiar	71	16.6
		Moderately familiar	124	28.9
		Very familiar	186	43.4
Delay_ Frequency	Q1.2 Have you experienced unexpected delays or extra waiting time on the platform?	Never	24	5.6
		Sometimes	270	62.9
		Often	135	31.5
Delay_ Crowded Platform	Q1.3 How often do you have to wait extra time on the platform due to the platform is too crowded?	Never	54	12.6
		Rarely	144	33.6
		Sometimes	161	37.5
		Often	56	13.1
		Always	14	3.3
Preferred Waiting Area	Q2.6 Do you have a preferred waiting area on the platform?	No	119	27.7
		Yes	310	72.3

.

The first relevant question in the pre-boarding phase is to assess passengers’ familiarity with the station layout. The results of the survey indicate that the majority of train users have a good level of familiarity with the station layout. A negligible small percentage (1.4%) of respondents reported being “not at all familiar” with the layout, and only 9.8% indicated “slightly familiar”. In contrast, a significant 88.9% of train users rated themselves as “somewhat familiar”, “moderately familiar”, or “very familiar” with the station layout.

Using the pre-boarding framework, we also examined train users’ experience about crowds and delays. The results of survey question Q1.2 suggest that delays are a common experience among train users. Only 5.6% of the participants reported never experiencing unexpected delays or extra waiting time on the platform. The majority of respondents (94.4%) do have some experience with unexpected delays or extra waits, with 62.9% reported sometimes and 31.5% reported often. The result from Q1.3 shows that 87.4% of participants reported experiencing extended waiting times due to overcrowded platforms, while only 12.6% reported never experiencing that.

Experienced train users may use the knowledge to make more rational decisions, such as choosing the shorter or less crowded route to the platform, and walking to the preferred location to wait and board the train. The result of survey question Q2.6 suggests that most passengers have a preferred waiting area on the platform. In particular, 72.3% of participants answered yes, while 27.7% answered no.

To gain a deeper understanding of train users’ behaviors at the station or platform, we have also analyzed the relevant survey questions of given statements related to passengers’ choices of actions and observations on the platform. The descriptive statistical results, including the average agreement score and the standard deviation, are presented in

Table 3. Descriptive statistics of statement questions in the pre-boarding phase.

Variable	Survey Statements	Mean	SD
Choice_ Shortest Path	Q2.1—I prefer the shortest path if there are multiple routes to the platform.	4.09	0.768
Choice_ Less-crowded Path	Q2.2—I prefer the less crowded path to the platform.	4.10	0.761
Platform Passenger Distribution	Q2.3—I notice that passengers are not evenly distributed along the platform.	3.86	0.837
Choice_ Wait Next Train	Q2.4—I don’t mind waiting for the next train if the current one is too crowded.	3.25	1.187
Choice_ Boarding ASAP	Q2.5—Regardless the crowd level, I just want to get onboard as soon as possible.	3.40	1.114
Choice_ Walk Longer	Q2.7—I would rather walk longer distances for a less crowded waiting area.	3.79	0.975
Choice_ Sit for Waiting	Q2.8—Agree or disagree—I’d like to sit down and wait for the next train.	3.59	1.012
Boarding Confusion	Q2.9—I am often confused as where to line up for boarding.	2.79	1.146

.

The result shows that both survey questions, Q2.1 and Q2.2, have higher mean values (above 4 out of 5) and smaller standard deviations compared to the other statement questions. This suggests that the responses are more in agreement with the statements, with less variability in the answers. Survey question Q2.1 shows that the majority of respondents agree with the statement, “I prefer the shortest path if there are multiple routes to the platform”. Specifically, 82.1% of respondents either agree (51.3%) or strongly agree (30.8%) with the statement, while only 3.2% either disagree (3%) or strongly disagree (0.2%). The remaining 14.7% of respondents were neutral towards the statement. Survey question Q2.2 yielded similar results, with a total of 82.3% of respondents indicating agreement with the statement “I prefer the less crowded path to the platform”, and only 3% indicating disagreement.

Passengers’ general perception of pedestrian flow on the platform is assessed by survey question Q2.3. The average agreement score for the statement “I notice that passengers are not evenly distributed along the platform” is 3.86 (SD = 0.837). According to the survey results, 69% of participants noticed uneven passenger distribution along the platform, indicating a potential issue with overcrowding or congestion in certain areas. Only 5.6% of respondents disagreed with the statement, while 25.4% remained neutral. These results show that uneven passenger distribution on the platform is commonly recognized among passengers.

Survey questions Q2.4 and Q2.5 were used to assess passengers’ likely behavior and strategy when it comes to boarding decisions. With regard to the statement, “I don’t mind waiting for the next train if the current one is too crowded”, about 47% of the participants don’t mind waiting, while slightly less than 30% do mind to some extent. This statement has a relatively lower average agreement score (M = 3.25, SD = 1.187) compared to other statements. On the other hand, the statement “Regardless the crowd level, I just want to get onboard as soon as possible” was agreed by the majority (52%), while 11% disagreed with it. Comparing the results of these two statements, it is possible to interpret that passengers may prioritize boarding the train earlier over waiting for the next train if the current one is crowded.

Passengers may behave differently in response to crowds, with some choosing to strategically take different routes for less crowded areas. The average agreement score for the statement “I would rather walk a longer distance for a less crowded waiting area” is 3.79 (SD = 0.975). The results of survey question Q2.7 revealed that 67.8% of respondents agreed with the statement, while only 10.5% of respondents disagreed. 21.7% neither agree nor disagree. This suggests that the majority are willing to prioritize personal space over convenience.

Survey question Q2.8 sought to explore differences in passengers’ sitting and standing behavior on the platform. The average agreement score for the statement “I’d like to sit down and wait for the next train” is 3.59 (SD = 1.012), which is in between neutral (score of 3) and agree (score of 4). The results revealed that more than half of the participants (56.9%) agreed with the statement. On the other hand, 15.2% of respondents disagreed with this statement. These results indicate that over half of the participants prefer sitting while waiting on the platform.

Passengers’ perceptions of the boarding queue may have an impact on their boarding behavior. Survey question Q2.9, with the statement “I am often confused as to where to line up for boarding”, aims to reveal the opinions about the boarding queue. The average agreement score for this question is the lowest in the set (M = 2.79, SD = 1.146), which is in between disagree (score of 2) and neutral (score of 3). The result also shows that only 28% agreed with the statement. These results suggest that, overall, passengers do not feel confused about where to line up for boarding.

4.2. Pre-Boarding Behavior Differences by Traveller and Trip Characteristic Variables

The following analysis aims to explore the ways in which traveler and trip variables may impact train passengers’ behaviors and perceptions during the pre-boarding phase and how these insights can be applied to improve the overall passenger experience. There are eight traveler and trip characteristics variables. Two variables are related to demographics (including Gender and Age group), and six relates to travel pattern and trip characteristics (including Travel Frequency, Average Travel Time, Average Waiting Time, Group Travel, Carry Small Item, and Carry Large Item).

4.2.1. Gender-Based Differences Revealed in Pre-Boarding Variables

The associations between gender and all the relevant survey variables in the pre-boarding phase are examined. The results found gender shows no difference in the perception of uneven passenger distribution on the platform, no difference in the choice of walking longer distances for a less crowded waiting area, no difference in sitting down for waiting, no difference in the strategy of boarding (wait for next train or board ASAP). The results suggest that gender may play a role in the preference for the shortest path when there are multiple routes to the platform. On the contrary, there is no association between gender and the preference for the less-crowded path.

On average, female participants reported a higher station familiarity score. The result shows that females are more likely to experience and report delays “sometimes” or “often” compared to males. Females were more likely to report experiencing delays due to a crowded platform “sometimes” compared to males, while males were more likely to report experiencing delays due to a crowded platform “often” compared to females.

Overall, there are in total five significant relationships between gender and relevant survey variables in the pre-boarding phase, including the association to “Station Familiarity” (χ² (4, N = 429) = 10.097, Cramer’s V = 0.153), “Delay_Frequency” (χ² (2, N = 429) = 6.179, Cramer’s V = 0.120), “Delay_Crowded Platform”(χ² (4, N = 429) = 10.998, Cramer’s V = 0.160), “Boarding Confusion” (χ² (4, N = 429) = 12.864, Cramer’s V = 0.173) and “Choice_Shortest Path” (χ² (4, N = 429) = 17.320, Cramer’s V = 0.201). With regards to the strength of association, four out of five are weak associations (“Station Familiarity”, “Delay_Frequency”, “Delay_ Crowded Platform”, and “Boarding Confusion”). One association (between gender and “Choice_Shortest Path”) is moderate.

4.2.2. Pre-Boarding Behavior Differences by Age Group

Age can be another important factor that affects train passengers’ behavior, perception, and preference. Different age groups may have different needs, expectations, and priorities when it comes to training travel. The Spearman correlation test shows that age group has a statistically significant relationship with three survey variables in the pre-boarding phase, including the association to “Delay_Frequency” (rho = −0.172, p < 0.001), “Delay_Crowded Platform” (rho = −0.181, p < 0.001), and “Boarding Confusion” (rho = −0.109, p = 0.024). All those relationships are negative and relatively weak.

The result indicates that as the age group increases, the frequency of experiencing and reporting unexpected delays or extra waiting time on the platform tends to decrease slightly. One possible reason is that senior passengers may have more experience with public transportation and have learned how to better plan their trips to avoid delays. Additionally, senior individuals may have more flexible schedules and are able to avoid peak travel times, which may be more prone to delays. It is also found that as the age group increases, participants are less likely to report being confused about where to line up for boarding.

4.2.3. Pre-Boarding Behavior Differences by Travel Frequency

According to the frequency of train travel, passengers are re-classed into regular users and non-regular users. It is possible that regular and non-regular train users may have different opinions regarding various aspects of the train travel experience. The Spearman correlation test shows that travel frequency has a statistically significant relationship with seven pre-boarding survey variables, including the association with “Station Familiarity” (rho = 0.178, p < 0.001), “Preferred Waiting Area” (rho = 0.318, p < 0.001), “Delay_Frequency” (rho = 0.268, p < 0.001), “Delay_Crowded Platform” (rho = 0.298, p < 0.001), “Choice_Shortest Path” (rho = 0.116, p = 0.017), “Choice_Wait Next Train”, and “Boarding Confusion” (rho = 0.244, p < 0.001).

Among those significant associations, four are considered weak associations, two are considered weak to moderate associations, and one is considered moderate associations. The strongest association is between preferred waiting areas and travel frequency. This correlation suggests that as people’s travel frequency increases, they are more likely to have a preferred waiting area at the station. This could be due to a number of factors. For example, individuals who travel frequently may become more familiar with the station and develop preferences for certain waiting areas based on factors like comfort, convenience, or proximity to their destination. Additionally, those who travel more frequently may have a greater need for a designated waiting area as they frequently need to spend time waiting at the station. According to the previous finding that approximately sixty percent of the users are regular travellers; thus, providing amenities and services that cater to their preferences may help improve their experience and satisfaction with the services.

These findings suggest that there are differences in opinions and behaviors between regular and non-regular train users during the pre-boarding phase. Based on this, transport agencies and train operators should aim to comprehend and address these distinctions to ensure a positive experience for all passengers.

4.2.4. Pre-Boarding Behavior Differences by Travel Time

How travel time impacts the passengers’ pre-boarding behaviors and perceptions is also studied. The Spearman correlation test results show that the associations between pre-boarding survey variables and travel time are generally very weak. There is a mix of positive and negative correlations between the relevant survey variables and travel time. For example, travel time has very weak positive correlations with survey variables including “Station Familiarity”, “Delay_Frequency”, “Delay_Crowded Platform”, “Choice_Shortest Path”, “Choice_Less-crowded Path”, “Passenger Distribution_Platform”. On the other hand, travel time has very weak negative correlations with survey variables including “Preferred Waiting Area”, “Choice_Walk Longer”, “Choice_Sit for Waiting”, “Choice_Wait Next Train”, “Choice_Boarding ASAP”, “Boarding Confusion”.

Overall, it appears that while certain pre-boarding factors may be related to travel time, all those associations are not particularly strong enough or consistent. As a matter of fact, none of the relationships are statistically significant at the p < 0.05 level. Therefore, it can be concluded that passengers’ pre-boarding behaviors are not significantly affected by travel time.

4.2.5. Pre-Boarding Behavior Differences by Waiting Time

The duration of waiting time on the platform can also affect a passenger’s opinion and behavior. The Spearman correlation test shows that waiting time has a statistically significant relationship with five pre-boarding survey variables, including the association to “Station Familiarity” (rho = −0.142, p = 0.003), “Delay_Frequency” (rho = 0.130, p = 0.007), “Delay_Crowded Platform” (rho = 0.212, p < 0.001), “Choice_Boarding ASAP” (rho = −0.115, p = 0.017), and “Boarding Confusion” (rho = 0.125, p = 0.010). The strength of the associations for those five pairs is relatively weak. Among those relationships, only two pairs are negative (association with “Station Familiarity” and “Choice_Boarding ASAP”, and the rest are all positive.

The result indicates that as station familiarity increases, waiting time tends to decrease. This result makes good sense as passengers who are less familiar with the station may arrive earlier than necessary. On the contrary, passengers who are more familiar with the station normally are more confident about what to expect, thus they may arrive at the station closer to the train arrival time. The negative association with “Choice_Boarding ASAP” also makes sense. Passengers who prioritise boarding ASAP are more proactive in finding a spot on the train, resulting in them getting on board earlier and potentially avoiding long waiting times.

4.2.6. Pre-Boarding Behavior Differences by Group Travel

It is possible that passengers who travel in a group may have different opinions compared to passengers who travel alone. The presence of others in a group can impact a passenger’s comfort and convenience, as well as the general behavior. Spearman correlation test shows that group travel (i.e., the frequency of travel with friends or family) has a statistically significant relationship with four pre-boarding survey variables, including the association to “Delay_Frequency” (rho = 0.098, p = 0.043), “Delay_Crowded Platform” (rho = 0.222, p < 0.001), “Choice_Wait Next Train” (rho = 0.195, p < 0.001), and “Boarding Confusion” (rho = 0.148, p = 0.002). All these associations are positive.

The result shows that passengers who more frequently travel in groups are slightly more likely to experience and report delays or extra waiting time on the platform. The group travelers are more likely to choose to wait for the next train if the current one is too crowded. This result makes sense as group travelers desire to stay together. Hence, they are more likely to wait for the next train to avoid being split up in a crowded train carriage. It is also found that passengers who travel with friends or family may be more likely to experience confusion during the boarding process. It is possible that coordinating with others, especially if there are multiple people in the group, could add an extra layer of complexity and confusion to the boarding process.

4.2.7. Pre-Boarding Behavior Differences by Carry Small Item

Carrying a backpack can impact a passenger’s comfort and convenience. Passengers who frequently carry backpacks on board may behave differently compared to passengers who do not. The Spearman correlation test shows that carrying small items (i.e., the frequency of carrying small items onboard such as backpacks or bags of similar size) has a statistically significant relationship with four pre-boarding survey variables, including the association to “Preferred Waiting Area”(rho = 0.108, p = 0.025), “Delay_Frequency” (rho = 0.231, p < 0.001), “Delay_Crowded Platform” (rho = 0.240, p < 0.001), and “Boarding Confusion” (rho = 0.127, p = 0.009). All these associations are positive.

The result shows that passengers who frequently carry small items such as backpacks are more likely to have a preferred waiting area on the platform. They are more likely to experience and report delays. This may be because passengers who carry small items are more likely to use public transportation regularly and are, therefore, more likely to encounter delays over time. Additionally, passengers carrying small items may be more likely to have valuable or important items that they want to keep secure, leading to increased stress or anxiety in crowded or delayed situations. The result suggests that passengers who frequently carry small items may be more aware of the uneven distribution of passengers on the platform. This result makes sense as these passengers are more likely to pay attention to the passenger distribution on the platform because they are more likely to be impacted (slow down) by other passengers when carrying backpacks.

4.2.8. Pre-Boarding Behavior Differences by Carry Large Item

Passengers who frequently take large items onboard may also behave differently compared to passengers who do not. The Spearman correlation test shows that carry large item (i.e., the frequency of carrying large items onboard such as bicycle, luggage, stroller/pram, shopping trolley, etc.) has a statistically significant relationship with seven pre-boarding survey variables, including the association to “Station Familiarity” (rho = −0.123, p = 0.011), “Preferred Waiting Area” (rho = 0.174, p < 0.001), “Delay_Frequency” (rho = 0.273, p < 0.001), “Delay_Crowded Platform” (rho = 0.318, p < 0.001), “Choice_Shortest Path” (rho = 0.124, p = 0.010), “Choice_Wait Next Train” (rho = 0.237, p < 0.001), and “Boarding Confusion” (rho = 0.246, p < 0.001).

All these associations are positive except the one between “Carry Larry Item” and “Station Familiarity”. Out of the seven correlations found, five are considered weak associations; one is considered weak to moderate, while one is considered moderate associations. Moderate associations exist between “Carry Larry Item” vs. “ Delay_Crowded Platform.

Comparing to the significant associations between survey variables and Carry Small Item, Carry Large Item has four similar associations, including “Preferred Waiting Area”, “Delay_Frequency”, “Delay_Crowded Platform”, and “Boarding Confusion”. However, the strength associations between Carry Large Item and those four survey variables are stronger than the associations between Carry Small Item and those five survey variables.

Apart from those four significant associations, there are three more survey variables (“Station Familiarity”, “Choice_Shortest Path”, and “Choice_Wait Next Train”) that are significantly associated with Carry Large Item. The result suggests that passengers who are less familiar with the station are more likely to carry large items such as bicycles. This is an interesting finding. There could be a few potential reasons why a weak negative correlation is found between carrying large items and station familiarity. One possibility is that those who carry large items on board may not be regular train users and, as a result, are less familiar with the station layout. Another possibility is that those who are familiar with the station layout are well aware of the difficulties of carrying large items on board and try to avoid them. Passengers who frequently carry large items are slightly more likely to prioritize taking the shortest path to the platform. One possible explanation is that choosing the shortest path helps minimize the amount of time and effort required to transport the large items to the platform. Passengers who frequently carry large items are more likely to choose to wait for the next train, given the current one is too crowded. This makes good sense as carrying large items can be very challenging, especially under crowded conditions, and therefore, passengers may prefer to wait for the next train to increase the chance of a more comfortable and efficient journey.

4.2.9. Overall Assessment

The summary table incorporating the results of the Spearman correlation test between all the pre-boarding survey variables and all the traveler and trip characteristic variables is shown in

Table 4. Significant associations between pre-boarding survey variables and traveller/trip variables.

	Gender	Age Group	Travel Frequency	Travel Time	Waiting Time	Group Travel	Carry Small Item	Carry Large Item
Q1.1 Station Familiarity	−0.088	−0.010	0.178	0.039	−0.142	−0.074	0.066	−0.123
Q1.2 Delay_Frequency	0.118	−0.172	0.268	0.017	0.130	0.098	0.231	0.273
Q1.3 Delay_Crowded Platform	0.095	−0.181	0.298	0.074	0.212	0.222	0.240	0.318
Q2.1 Choice_Shortest Path	−0.106	0.019	0.116	0.032	0.022	0.093	0.029	0.124
Q2.2 Choice_Less-crowded Path	−0.026	0.034	0.059	0.027	0.037	0.018	0.095	0.002
Q2.3 Platform Passenger Distribution	0.051	−0.022	0.091	0.051	0.073	0.031	0.127	0.059
Q2.4 Choice_Wait Next Train	0.116	0.027	0.244	−0.068	0.002	0.195	0.060	0.237
Q2.5 Choice_Boarding ASAP	−0.011	0.005	−0.026	−0.074	−0.115	−0.048	−0.044	−0.024
Q2.6 PreferredWaiting Area	0.019	−0.084	0.318	−0.014	−0.029	0.061	0.108	0.174
Q2.7 Choice_Walk Longer	0.032	0.015	0.009	−0.013	−0.040	0.075	0.011	0.010
Q2.8 Choice_Sitting for Waiting	−0.040	0.014	0.045	−0.028	0.069	0.042	0.069	0.050
Q2.9 Boarding Confusion	0.117	−0.109	0.130	−0.044	0.125	0.148	0.034	0.246

Color Legend: .

. For easier comparison and pattern identification, we introduced a color scheme on the correlation coefficients, where the color gradient runs from red (most negative) to white (around zero) to blue (most positive).

By examining the color pattern in the table, it is found that the Travel Time column appears to have lighter colors compared to the other columns. The result shows that the correlations between Travel Time and all the relevant survey variables are very weak and insignificant. This suggests that travel time does have a significant impact on passenger perceptions or behaviors at the pre-boarding phase. Additionally, other factors such as trip purpose, mode of transportation, and passenger demographics may have a stronger influence on passenger perceptions and behaviors than travel time alone. It’s important to note that correlation does not necessarily imply causation, so even if there are no significant correlations between travel time and the survey variables, travel time may still be an important factor in shaping passenger experiences and behaviors.

Additionally, it was observed that the shading in three specific rows (Q2.2, Q2.7, Q2.8) appears to be noticeably lighter compared to other rows. The fact is that the correlations between three survey variables (“Choice_Less-crowded Path”, “Choice_Walk Longer”, “Choice_Sit for Waiting”) and all the traveler/trip variables are very weak and insignificant. This result suggests that passenger’s choice of less-crowded path is not affected by their gender, age, travel frequency, or specific trip characteristics. It is also realized that passengers prioritize avoiding crowds and choose to walk longer distances regardless of their personal characteristics or trip details. Also, passengers’ preference for sitting down and waiting for the next train is s not influenced by personal or trip factors. It is also found that the shading in two additional rows (Q2.3, Q2.5) is also generally lighter. The result shows that the survey variable “Distribution_Platform” is only significantly associated with Carry Small Item. Survey variable “Choice_Boarding ASAP” is only significantly associated with “Waiting time”. Both correlations are weak, though.

5. Key Findings and Discussion

There are some general patterns identified through the assessment. The analysis indicates a high level of station familiarity among Melbourne Metro train users. Over 88% of participants reported being at least “somewhat familiar” with the station layout. This finding aligns with similar studies conducted in Melbourne, where 80% of respondents expressed familiarity with the train station and ease of movement [46]. The consistent finding offers credibility to the survey results and indicates that participants have reliable knowledge about the pre-boarding phase.

Regarding the overall perception of delays, it is evident that passengers commonly experience delays. A significant number of passengers, 87.4%, reported experiencing delays caused by overcrowded platforms. These findings highlight that passengers have significant concerns regarding the punctuality of the services. However, the Victoria Department of Transport Annual Report presents contrasting information, showing high service punctuality scores of 92.1 for the year 2019-2020 and 95.2 for the year 2020–2021 [47]. It is worth noticing that these scores measure the percentage of nominal 30-min trips being completed within 35 min on normal weekdays within school terms, whereas our results reflect the perceptions of rail users. These contrasting perspectives highlight the need to address the issues of delay. Efforts should be made to improve service frequency and schedule planning, as well as to shape strategies for managing crowds on the platform and effectively managing dwell time.

Our result shows a significant percentage of respondents were willing to walk longer distances for less crowded waiting areas. A previous study revealed that passengers consider the walking distance when selecting their boarding car. The study found that 69.7% of passengers choose a specific boarding car to minimize the walking distance to the exit at the destination station. Additionally, 13.5% of passengers choose a specific boarding car to minimize the walking distance from the entrance at the origin station [48]. Another study found that the congestion level of the platform tended to decrease as the distance from the platform entrance increased [9]. These findings suggest that designing clear signage, providing wayfinding information, and ensuring clearer and wider pathways are important design considerations to enhance passenger experience and facilitate efficient movement.

The study found that over two thirds of participants (69%) noticed uneven passenger distribution along the platform, indicating a potential issue with overcrowding or congestion in certain areas. A few existing studies show similar findings of uneven passenger distribution along the platform [21,22]. These results identified the issues related to overcrowding and congestion in certain areas on the platform, which further emphasized the importance of improving the platform layout and optimizing the use of available space.

The assessment found that more than half of the participants preferred to sit and wait for the next train, indicating a preference for comfort and relaxation while waiting. This result highlighted the need for seating facilities. However, if it is not feasible due to limited space or other constraints, alternative strategies such as increasing service frequency and implementing crowd control measures should be considered. The result also shows that most passengers have a preferred waiting area on the platform. This information can be used in shaping the communication strategies, such as providing clear signs or directions for passengers to various waiting areas.

When it comes to boarding decisions, passengers tend to prioritize boarding the train earlier over waiting for the next one if the current one is crowded. There is a similar finding in the literature which shows that morning commuters are slightly less likely to wait for a later train [49]. Regarding the boarding process, the finding that 28% of participants reported feeling some degree of confusion about where to line up for boarding highlights the need for improved guidance and signage. This insight motivates us to propose the implementation of floor markings and clear signages to facilitate the boarding process.

It is common to observe differences in behaviors and opinions among various demographic groups. However, the specific differences are not discussed in the existing literature. Our correlation tests revealed that behaviors and perceptions could also vary depending on the nature of the trip. The novel findings from the correlation tests are presented in the following paragraphs.

It is found that passengers’ familiarity with station layout is positively correlated to travel frequency, but negatively correlated to waiting time and carrying large items. Passengers having a preferred waiting area are positively associated with travel frequency, carrying small items, and carrying large items. Passengers experiencing delays are negatively correlated to age group but positively correlated to travel frequency, waiting time, group travel, carrying small items, and carrying large items. Senior passengers are less likely to experience and report delays. They are also less likely to feel confused about where to line up for boarding. Passengers perceiving confusion about boarding lines are positively correlated to gender, travel frequency, waiting time, group travel, and carrying large items.

Passengers’ choosing a less-crowded path or a longer path for a less crowded waiting area are common behaviors that are not affected by any personal or trip characteristics. Passengers’ choosing the shortest path to the platform is negatively correlated to gender but positively correlated to travel frequency and carrying large items.

Passengers’ choosing to sit while waiting for the next train is not affected by any traveler/trip characteristic variables. Uneven passenger distribution on the platform is perceived differently by passengers frequently carrying small items.

The differences in passengers’ boarding behaviors are also identified. Passengers choosing to board the train as soon as possible is negatively correlated to waiting time. Passengers choosing to wait for the next train (given the current one is too crowded) is positively correlated to travel frequency, group travel, and carrying large items.

6. Conclusions

In conclusion, this study provided a systematic assessment of the behaviors and perceptions of train passengers in the pre-boarding phase using the proposed pre-boarding framework. This framework can provide valuable insights into passengers’ pre-boarding behaviors, preferences, and decision-making processes, which can inform the design and management of station environments, wayfinding systems, and boarding processes. The survey data collected from this study was used to test the framework, and the main findings were presented.

The analysis shows some general patterns that indicate most participants were familiar with the station layout and that delays are commonly experienced. Passengers tend to prioritize boarding the train earlier over waiting for the next one if the current one is crowded, and most do not feel confused about where to line up for boarding. The correlation tests revealed that behaviors and perceptions might vary depending on the nature of the trip and personal characteristics. Carrying large items and travel frequency significantly impact passengers’ travel experience in the pre-boarding phase. Waiting time, group travel, carrying small items, gender, and age group also significantly impact some pre-boarding behavior variables. Travel time, on the other hand, makes no significant impact on any of the pre-boarding variables that we examined.

The proposed framework provides a systematic approach to investigating passengers’ pre-boarding behaviors and perceptions. Adopting this framework enables a comprehensive understanding of train passengers’ pre-boarding experiences. There is an opportunity for this framework to be adapted to other public transport fields, including high-speed trains, light trams, and bus rapid transit (BRT) systems. Additionally, by adjusting the context and slightly altering the focus of the survey questions, this framework can also be applied to other industries, such as the aviation sector.

The findings from this study could be useful for transportation planners and operators to better understand passenger behaviors and perceptions in the pre-boarding phase. By identifying factors that influence passengers’ choices and behaviors, transport agencies and service operators can develop strategies to improve their service accordingly. This may increase passengers’ satisfaction and subsequently encourage the adoption of train travel over less sustainable alternatives such as private vehicles.

The data from this study was limited to Melbourne, Australia. In future, a similar survey can be conducted in different Australian states or other countries to gain insight into geographical bias. The passenger survey conducted in this study aimed to capture passengers’ general behaviors and perceptions. It did not specifically focus on any specific time periods. However, we acknowledge the importance of considering peak periods. Future research could consider conducting surveys during peak periods or including a question to identify the trip period; hence a deeper understanding of passengers’ behaviors and perceptions during the high-demand periods can be obtained.

Another limitation of our study is the absence of a suburb-specific analysis of passenger behavior and perception. Our survey data collection did not include information about the specific suburbs from which the respondents originated. The survey we conducted has a primary focus on capturing pre-boarding behaviors and perceptions at a city level rather than analyzing the data at a suburb level. As a result, we were unable to explore potential variations in behavior and perception across different suburban areas, including affluent and non-affluent suburbs. To address this limitation, future studies should consider incorporating additional survey questions related to other socioeconomic factors such as income level, education level, or specific suburb information to better understand the variation of behaviors and perceptions.

It is also worth noting that assessing human behavior can be very challenging. While this proposed framework and survey responses provide valuable insights into passengers’ perspectives, there may be instances where the reported perceptions do not fully reflect the reality of their actions. To mitigate this bias, future research could consider incorporating additional methods such as extensive field observations or video analysis. By combining self-reported data with direct observation techniques, a more objective and comprehensive understanding of passengers’ behavior can be achieved.