Pedestrian Decision-Making Behavior During Stair Evacuation: An Experiment Study on Stair Lane-Selection Preferences

Abstract

Improving the efficiency of stair evacuation plays a crucial role in emergency management, which may be shaped by pedestrians’ lane-selection behavior. However, most existing studies describe pedestrians’ lane-selection preferences during stair evacuation, while the mechanisms behind these preferences are not yet well understood. To solve this issue, a stair evacuation observation experiment and a questionnaire survey were carried out to investigate pedestrian stair lane-selection preferences. Based on 1793 pieces of experimental data and 397 questionnaires, it is found that (1) pedestrians in the middle lane are more inclined to proactively change lanes based on their personal preference when sufficient space is available. (2) The primary factors influencing pedestrians’ lane-selection preferences are perceived safety, shortest path, and behavioral habit. (3) As the distance to the wall increases, the preference for the wall-side lane gradually decreases. Notably, the rate of decline accelerates at first, then slows down as the wall becomes farther away. This study deeply deconstructs pedestrians’ stair lane-selection preferences which helps understand the interactions among pedestrians, between pedestrians and their surroundings. It offers a basis for the optimization of evacuation strategies, the design of emergency evacuation plans, and the calibration of evacuation simulation models.

1. Introduction

Multi-story buildings have become the dominant form of urban architecture and accommodate large numbers of occupants. Consequently, the emergency evacuation of such buildings, especially during fires or terrorist attacks, has emerged as a critical research topic [1]. For most building occupants, stairs and elevators serve as the primary evacuation routes [2]. In recent years, numerous studies have investigated the feasibility of elevator-assisted evacuation [3,4] as well as methods to enhance elevator efficiency during emergencies [5,6]. Nevertheless, elevator use in emergencies is constrained by the chimney effect [7], fire resistance limitations [8], waterproofing concerns, and power supply reliability. According to current international standards, stairs remain the only legally recognized evacuation route for high-rise buildings [9], and they are also the route most familiar to the public [10]. Therefore, improving the efficiency of stair evacuation has long been a focal point in evacuation research [11].

Stair evacuation fundamentally involves pedestrian flow, and understanding how pedestrians make decisions during evacuation is therefore essential for accurately capturing evacuation dynamics [12,13]. So, research on pedestrians’ decision-making behavior is important to improve the efficiency of evacuation. At every stage of stair evacuation, pedestrians must make appropriate decisions based on the circumstances they face, such as accelerating, decelerating or changing lanes, which affects the efficiency and safety of the evacuation [14]. Liu et al. emphasized that deeply understanding pedestrian dynamics on stairs is key for preventing accidents [15]. Relevant authorities may develop appropriate evacuation strategies informed by pedestrians’ decision-making preferences. Moreover, studying pedestrians’ lane-selection behavior during stair evacuation can also help guide architectural design [16]. Huo et al. used an extended lattice-gas model to study pedestrian flow at the floor–stair interface [17], and their findings suggest that the interface’s geometry and merging path layout should be optimized to reduce conflicts. In the field of computer simulation, the research on pedestrians’ lane-selection preferences provides empirical data for various pedestrian dynamics models and develops the realism of these models [18,19]. Therefore, studying pedestrians’ behaviors including pedestrians’ lane-selection preferences during evacuations is vital for preventing safety incidents and reducing casualties [20].

In recent years, many researchers have conducted experiments to investigate lane-selection behavior during stair evacuation. Some studies have analyzed pedestrians’ behavioral characteristics at stair corners and found that pedestrian densities are higher at inner corners [21,22]. Imanishi et al. also reported that crowd pressure increases on the inner side when pedestrians negotiate a corner [23]. Boltes et al. and Rui et al. validated that there is a clear tendency for pedestrians to walk on the inner side during cornering on stairways [24,25]. Dias et al. observed that during normal-speed walking, the two lanes formed before turning disappeared after the turn because of yielding and overtaking behaviors [26]. Beyond focusing on corner-preference behavior, other researchers have observed lane-changing and overtaking phenomena during stair movement [27,28,29]. In Ding Ning’s studies [30], he proposed the concept of pedestrians’ dynamic spatial demand on step treads. His experiment showed that, besides an inner-side walking preference, pedestrians maintain a certain distance from those ahead and change lanes when appropriate. Zhu Yu [31] further validated and extended the notion of dynamic spatial demand in his research. Zeng et al. [32] in the study indicated that it was also pointed out that when a stair contains multiple parallel lanes, pedestrians may choose either the inner or outer lanes, and may even overtake others. It is clear that pedestrians prefer to move toward the inner side during staircase evacuations [33,34,35].

However, most existing studies still remain at the level of investigating the manifestations of pedestrians’ lane-selection preferences, while the formation mechanism underlying pedestrians’ lane-selection preferences during stair evacuation remains unclear. Why do pedestrians exhibit these preferences, and what factors influence such preferences? Consequently, some controversies regarding pedestrians’ lane-selection preferences have emerged. For example, there is still major disagreement over whether pedestrians will proactively change lanes based on their own preferences when sufficient space is available. Some researchers argue that when there is enough space, pedestrians will proactively change lanes [1,36]. Other researchers argue that pedestrians tend to remain in their current lane and not change it [37]. A profound understanding of pedestrians’ lane-selection preferences can help elucidate the microscopic interactions among individuals as well as between individuals and their surroundings during stair evacuation [38,39]. To address these gaps, this experiment aims to answer the following core research questions:

(1)

What are the key factors influencing lane-selection preferences, and what is the hierarchy of their relative importance?

(2)

Do pedestrians in different initial lanes (handrail-side, middle, wall-side) exhibit distinct lane-changing behaviors when space allows? Are pedestrians in the middle lane more inclined to change lanes proactively?

(3)

How does the distance to the wall affect pedestrians’ preference for the wall-side lane, and what is the trend of this influence?

In this paper, the purpose is to explore the formation mechanism of pedestrians’ lane-selection preferences through a stair-evacuation observation experiment and a questionnaire survey—specifically, the factors influencing pedestrians’ lane-selection preferences during stair evacuation and the extent to which these factors affect such preferences. The rest of the paper is structured as follows: Section 2 describes the observation experiment and questionnaire survey methodology; Section 3 presents the data analysis; Section 4 discusses the results; and Section 5 draws conclusions based on the data and discussion.

2. Methodology

This study is mainly designed as a confirmatory experiment. So based on existing theories, literature reviews, and preliminary observations, we propose the following testable hypotheses to address the above questions:

(1)

Pedestrians’ lane-selection preferences are significantly influenced by three factors: perceived safety, shortest path, and behavioral habit.

(2)

When sufficient space is available, pedestrians will proactively change lanes based on their own preferences. If pedestrians are already in their preferred lanes, they will continue moving forward along the original lanes.

(3)

As the distance to the wall increases, the preference for the wall-side lane gradually decreases.

2.1. Research Methods

Current study methods on pedestrians’ stair evacuation behavior include computer simulations [40], virtual reality (VR) experiment [41], accident analyses [9], and evacuation drills [42]. Each approach has its strengths and limitations, and which method is suitable depends on the research goal [43]. Some studies studied the movement characteristics of pedestrians on stairs by the use of simulation [44]. However, the parameters in such models are manually specified, making it difficult to reproduce real-world crowd conditions; as a result, realistic pedestrian behavioral characteristics are difficult to extract. In addition, this study mainly examines psychological factors influencing pedestrians’ lane-selection preferences. Therefore, it is important to consider not only objective behavioral patterns but also pedestrians’ subjective attitudes. Based on these considerations, we designed a stair-observation evacuation and a survey questionnaire to explore the pedestrians’ lane-selection preferences.

2.2. Experiment Objectives

Before exploring pedestrian lane-selection preference, we need to explain the relationship between pedestrian lane-changing behavior and pedestrian lane-selection behavior. Lane-changing behavior constitutes a type of lane-selection behavior, which belongs to lane-selection behavior during stair movement. The observation experiment and the questionnaire survey support each other well. This paper evaluates the extent to which lane-selection preferences influence pedestrian behavior by analyzing the pedestrians’ lane-changing performance and their attitude to change lane. And this paper also explores the factors that influence pedestrians’ lane-selection preferences by analyzing the pedestrians’ lane-selection performance and their tendency to each lane.

2.3. Experiment Site

The experiment was conducted in the stairs of a laboratory building at a Chinese university. According to our survey, we find that the majority of stairs currently installed in buildings are three-lane scissor stairs. To ensure the generalizability of the experimental conclusions, we chose a three-lane scissors stair spanning the first to the third floor to conduct the observation experiment (as shown in Figure 1b). For clarity and to aid reader comprehension, we standardized the terminology for the two types of platforms connecting the floors: the platform connecting each floor to the stairs is referred to herein as the merging platform, while the platform between stair flights is referred to as the transfer platform (Figure 1a). The space involved in the experiment thus comprised three merging platforms (one for each floor), two transfer platforms, and four stair flight segments. The tread depth of each step was 0.28 m, and the riser height was 0.17 m, resulting in an incline angle of 28. The total tread width of the stair was 1.90 m. after subtracting the space occupied by handrails, the net width was 1.74 m. Considering the average shoulder width of Chinese adults (0.43 m) [45] and typical lateral body sway during walking, this width can adequately accommodate three pedestrians walking side by side. Between the lower merging platform and the transfer platform there were 15 steps, and between the transfer platform and the upper merging platform there were 12 steps. Each merging platform measured 3.80 m in length and 3.10 m in width; each transfer platform was 3.80 m long and 2.08 m wide, as illustrated in Figure 2.

2.4. Experimental Procedure

2.4.1. Stair-Evacuation Observation Experiment

The participants in this stair-evacuation observation experiment were 60 on-campus students from a university in China, aged between 18 and 22 years. Among these participants who volunteered to take part in the study through online registration, the ratio of males to females was 3:17. To eliminate visual interference due to varied clothing, all participants were required to wear identical outfits. Before the experiment, every participant was informed about the procedure and signed an informed-consent form.

The observation experiment was conducted in an uncontrolled scenario, allowing participants to behave freely: they descended from the third floor through the merging platform and exited at the first-floor exit as they normally would. To approximate behavioral dynamics under emergency evacuation conditions, a reward mechanism was used to motivate the participants’ engagement. During the interval between each trial, a 15 min break was arranged to allow participants to rest, and food and water were provided to help them recover. Each trial commenced with all participants positioned outside the third-floor safety exit, which had an effective width permitting four individuals to simultaneously pass. Upon a researcher’s signal, they began to evacuate and entered into the third-floor merging platform. A total of 10 repeated trials were conducted. The evacuation process was recorded by mounting GoPro11 action cameras at suitable positions on each platform to capture lane-selection behavior during stair evacuation. The experimental process is illustrated in Figure 3. Additionally, there was a safety officer stationed at each transfer platform and merging platform. To facilitate statistical data, the stairs used in this experiment comprised three parallel lanes, allowing three pedestrians to walk side by side. As in most real stairs, the handrails were located on the inner side and the wall on the outer side. For clarity, in this paper we refer to the lane adjacent to the handrail as the handrail-side lane, the lane adjacent to the wall as the wall-side lane, and the lane in the middle as the middle lane (Figure 4).

2.4.2. Questionnaire Survey

The objective data of pedestrians’ lane-changing behavior and lane-selection behavior during stair evacuation are collected by the observation experiment to explore the presentation of lane change and lane selection. In parallel, the questionnaire investigates lane-selection preferences from a subjective attitude standpoint, serving two functions: first, to cross-validate the experimental results, and second, to overcome the practical constraints of the observation experiment—namely, limitations in participant numbers and space condition. The effect of multi-lanes on pedestrians can be investigated.

For this questionnaire survey, we recruited a total of 397 volunteers to participate offline, including 297 males and 100 females. All students who participated in the observational experiment also participated in the questionnaire survey. The types of questions included in this questionnaire include seven-point Likert scale questions, five-point Likert scale questions, and ranking questions, which are shown in Appendix A.

In order to investigate pedestrians’ lane-selection preferences better, we designed a three-part set of questions. (1) Two seven-point Likert scale questions were designed to investigate pedestrians’ lane-changing tendency. Q4 assessed the lane-changing tendency of pedestrians in the middle lane under a three-lanes configuration. Q5 investigated pedestrians’ preference for changing lanes toward the wall-side lane or the handrail-side lane. This question is under the condition that pedestrians are situated in the middle lane with sufficient space for changing lanes. (2) Four seven-point Likert scale questions (Q6–Q9) were designed to investigate Pedestrians’ Lane Preference under different distances to the wall. It looks specifically at how the distance between a pedestrian’s lane and the wall influences their preference for the wall-side lane when the handrail-side lane is occupied. (3) Three five-point Likert scale questions (Q10–Q12) and three ranking questions (Q13–Q15) influencing factors of pedestrians’ lane selection preference were designed. Based on stair-evacuation studies by Huo et al. and Fang et al. [46,47], three key drivers of pedestrians’ lane selection during stair evacuation were identified: perceived safety, shortest path, and behavioral habit. So, these items mainly measured the influence of perceived safety, shortest-path considerations, and behavioral habits.

Before final deployment, each draft of the questionnaire was pilot-tested by seven researchers who completed it and provided feedback to ensure that all relevant information could be suitable. They evaluated readability, consistency of style and format, scale structure, and the clarity of item wording. To ensure data quality, one researcher was present during questionnaire completion to clarify any ambiguous items, thereby reducing the risk of misunderstanding and ensuring that responses accurately reflect respondents’ true lane preferences.

3. Data Analysis and Results

3.1. Stair-Evacuation Observation Experiment Results

Observational data points were extracted using a video-based trajectory tracking method. Due to crowding in the videos, some pedestrian trajectories were difficult to identify clearly. In addition, a small number of participants were delayed during the experiment for various reasons. These data were therefore excluded during the extraction process. Finally, a total of 1791 valid data points regarding pedestrians’ lane-selection behaviors were collected in this study, among which 753 data points related to pedestrians’ lane-changing behaviors were further filtered out. In analyzing the data, we took into account a structural peculiarity of the first-floor stair area: the stair exit is located directly in front of the first-floor landing (as shown in Figure 5a), which means that the movement direction of pedestrians on the step segment connected to that platform differs from that of other flight segments. To avoid bias resulting from this architectural asymmetry, we separated the data from that particular flight segment for analysis. So the stair region adjacent to the exit-connected platform is defined as Class A stair, whereas the remaining typical stair regions are denoted as Class B stair (as shown in Figure 5b).

3.1.1. Pedestrians’ Lane Selection Data

During stair evacuation, pedestrians’ lane-selection behavior in this study is categorized into two scenarios: (1) pedestrians choosing a lane when entering the stair flight from the transfer platform or the merging platform; (2) pedestrians already in the middle lane changing lanes or other lanes when space allows. The data for pedestrians selecting lanes upon entering the stair flight from the transfer and merging platforms are presented in Figure 6a. For Class A stair, the majority of pedestrians chose the handrail-side lane, with 245 individuals, accounting for approximately 42.10% of the total pedestrians on Class A stair, followed by the wall-side lane and the middle lane, with 189 and 148 individuals, respectively. For Class B stair, the handrail-side lane was also the most frequently selected, with 487 individuals, accounting for approximately 40.28% of the total pedestrians on Class B stair, followed by the outer wall-side lane and the middle lane, with 442 and 280 individuals, respectively.

The lane selection of pedestrians in the middle lane with changing lane is shown in Figure 6b. On Class A stair, seven pedestrians shifted to the handrail-side lane, representing 70% of lane-changing pedestrians. On Class B stair, 53 pedestrians moved to the handrail-side lane, accounting for approximately 74.64% of lane-changing pedestrians.

3.1.2. Pedestrians’ Lane-Changing Behavior Data

In this paper, we focus on pedestrians’ proactive lane-changing behavior during stair evacuation. Pedestrians who do not have conditions to change lanes are not considered in this analysis. The determination of available lane-changing space in the stair flight is based on the findings of Ding Ning [30,48], who reported that pedestrians on stairs, under the influence of gravity and step height, predominantly move forward or diagonally forward, with very limited backward or lateral movement on the same step. Accordingly, areas where lane-changing is obstructed are considered to include adjacent lanes on the same step and the preceding step, while pedestrians located behind an individual are not regarded as affecting that individual’s lane-selection decisions in this experiment. Therefore, in the observation experiment, a pedestrian is considered to have available space for lane changing if both the adjacent lanes on the same step and the preceding step are unoccupied. Otherwise, the pedestrian is considered unable to change lanes.

Based on the data presented in

Table 1. Statistics of pedestrian lane-changing behavior on the stairs.

Observational Area	The Wall-Side Lane		The Middle Lane		The Handrail Lane
	Lane Change	No Lane Change	Lane Change	No Lane Change	Lane Change	No Lane Change
Class A stair	7	122	10	77	7	88
Class B stair	13	130	71	38	15	185

and Figure 7, the proportion of pedestrians changing lanes in the wall-side lane and the handrail-side lane is below 10%. Even for pedestrians in the middle lane, significant lane-changing behavior is observed only on Class B stairs, with a proportion as high as 65.13%.

3.2. Questionnaire Survey Results

A total of 397 questions were collected in this survey. After excluding invalid or inconsistent responses, 334 valid questions were retained, resulting in an effective response rate of 84.13%. Among the valid respondents, 256 were male and 78 were female.

Before specifically analyzing the questionnaire survey data, we first conducted a correlation analysis across all variables (Figure 8). The results indicate that correlations between gender, age, educational attainment, and variables related to lane-selection behavior and lane-changing behavior are generally weak. Most correlation coefficients are small in absolute value and not statistically significant, suggesting that demographic characteristics do not primarily drive pedestrians’ lane-selection preferences during stairway evacuation. Instead, these findings imply that situational and psychological factors are more likely to influence evacuation decision-making.

Firstly, in terms of situational factors (Q6–Q9), when pedestrians occupy the middle lane and the handrail-side lane also is occupied by others, significant and progressively strong positive correlations were observed between preference for the wall-side lane and increasing distances to the wall. Additionally, in terms of psychological factors (Q10–Q12), behavioral habit demonstrated more stable and significant correlations with both lane-selection behavior and lane-changing behavior, exhibiting overall stronger correlations than factors related to shortest path and perceived safety. Finally, a significant negative correlation was observed between the ranking questions (Q10–Q12) and the corresponding impact scale (Q13–Q15). The participants’ subjective rankings of the influencing factors were consistent with their quantitative assessments, indicating good internal reliability and coherence in the questionnaire data.

In a word, according to the result of correlation analysis, both situational factors and physical factors affect pedestrians’ lane-selection preference.

3.2.1. Pedestrians’ Lane-Changing Tendency Data

Figure 9a shows pedestrians’ lane-changing willingness when pedestrians occupy the middle. Participants scoring above 4 accounted for 80.6%, and those scoring below 4 accounted for 11.7%. In particular, participants scoring 7 already accounted for 44.3%. Figure 9b shows pedestrians’ lane selection when pedestrians change lane in the middle lane. In total, 85.1% of participants tend to move toward the handrail-side lane, which score above 4, while 9.3% of participants tend to move toward the wall-side lane, which score below 4.

3.2.2. Data on Pedestrians’ Lane Preference Under Different Distances to the Wall

Figure 10a shows the effect of varying distances from the wall on pedestrians’ preference for the wall-side lane. Higher scores indicate a weaker preference for the wall-side lane. It is evident that when the distance is one lane to the wall, the proportion of participants most inclined to the wall-side lane is 34.13%. Figure 10b presents the trend in preference shifts for the wall-side lane, when the distance to the wall changes. Its data are from Figure 10a.

3.2.3. Data on Influencing Factors of Pedestrians’ Lane Selection Preference

Table 2. Scores of influencing factors.

	Factors	Shortest Path	Behavioral Habit	Perceived Safety
Score
1		6.59%	2.99%	4.97%
2		10.78%	5.99%	8.68%
3		23.35%	15.57%	17.37%
4		31.44%	35.93%	27.54%
5		27.84%	39.52%	41.62%
Average score		3.63	4.03	3.93

illustrates the extent to which the shortest distance, perceived safety and behavioral habit affect participants. After conversion, the score of behavioral habit is highest, 4.03 points.

To further investigate the relative influence of these three factors, we calculated final scores for the ranking item based on the situation where respondents ordered the options using Equation (1). These scores reflect the overall ranking of the factors according to participants’ preferences. Higher scores indicate that an option is ranked more prominently. The weighting of each option was determined by its position in the ranking. Three options were included: the option ranked first was assigned a weight of 3, the second a weight of 2, and the third a weight of 1.

$$\overline{X }=\sum (f\times w)/N$$

(1)

In this equation, $\overline{X}$ denotes the aggregated score of the option, $f$ represents the frequency of the corresponding ranking, $w$ is the weight assigned to that ranking position, and $N$ is the number of valid responses. The detailed results of the ranking question are presented in

Table 3. Detailed ranking of influencing factors.

Factor	Composite Score	First Place	Second Place	Third Place
Behavioral habit	2.06	134 (43.79%)	114 (37.25%)	58 (18.95%)
Perceived safety	1.85	119 (40.61%)	88 (30.03%)	86 (29.35%)
Shortest path	1.59	81 (28.03%)	79 (27.34%)	129 (44.64%)

. Table 3 also shows behavioral habit, perceived safety and shortest path-ranking scenarios.

4. Discussion

Participants in the observation experiment were aged between 18 and 22 years and participants in the survey questionnaire also included those under the age of 18 and those aged 22 and above beside the age of 18–22. Although the age range of participants in the observation experiment and the questionnaire survey differs, they all belong to the youth, both whose mental and physical condition are similar. The effect of age differences can be considered negligible in this paper.

4.1. Pedestrians’ Lane-Choice Preference

In the stair-evacuation observation experiment, both for Class A and Class B stair, pedestrians entering the stair flight from the platform (Figure 6a) most frequently selected the handrail-side lane, followed by the wall-side lane, and least frequently the middle lane. During movement within the stair flight (Figure 6b), pedestrians in the middle lane who had sufficient space to change lanes still showed a stronger tendency to shift toward the handrail-side lane rather than toward the wall-side lane, with 70% choosing the handrail-side lane in Class A stair and 74.64% in Class B stair. From the perspective of subjective attitudes, pedestrians in the middle lane also indicated a greater willingness to shift toward the handrail-side lane compared with the wall-side lane. Based on both the observation experiment and the questionnaire survey results, it is clearly shown that pedestrians most prefer the handrail-side lane, then the wall-side lane, and the middle lane is the least favored.

Moreover, in the stair-evacuation observation experiment, the average probability of pedestrians choosing the handrail-side lane was 41.20%, which is closely aligned with the 41.80% reported in the previous study. This observation experiment was conducted under high pedestrian density and this situation happened frequently that some pedestrians were unable to change the current lane because the handrail-side lane was occupied. Therefore, the actual preference for the handrail-side lane is likely even stronger than what was observed. The agreement between the results of the stair evacuation observations, the questionnaire survey, and previous studies further supports the validity of this observation experiment.

Based on the characteristics of Pedestrians’ Lane Preference, the crowd density varies across different lanes during stair evacuation, with the handrail-side lane having the highest density compared to the other two lanes. To further improving efficiency of evacuation, when formulating evacuation strategies to guide pedestrians, the handrail-side lane should be designated as an express lane to allow faster-moving pedestrians with fewer belongings to move more quickly; the wall-side lane as a safety lane for slower-moving pedestrians with safety needs; and the middle lane as a lane-changing lane to meet the lane-changing demands of faster-moving pedestrians. This evacuation strategy not only aligns with pedestrians’ lane-selection preferences but also avoids congestion and improves evacuation efficiency.

4.2. Pedestrian Lane-Changing Behavior

As shown by the stair-evacuation observation experiment (Figure 7), pedestrians exhibited consistent lane-changing behavior in both Class A stair and Class B stair when occupying the wall-side lane or the handrail-side lane. Most pedestrians chose to continue along their current lane rather than change lanes. But the lane-changing behavior was different in different lanes when pedestrians occupied the middle lane. In Class A stair, the majority of pedestrians chose to continue along their current lane. And Class B stair, the majority of pedestrians proactive changed their current. It is considered that this situation results from stair structure. As shown by Figure 5, the difference in Class A stair and Class B stair is that there is an exit in front of the Class A stair. The effect of stair structure will be discussed. Considering that the structure of Class A stair varies from the site of exit and the structure of Class B stair can remain relatively stable, the section mainly focuses on Class B stair to discuss the pedestrians’ lane-changing behavior.

From the perspective of pedestrians’ subjective attitudes (Figure 9a), pedestrians present strong tendency to change lanes. Specifically, 44.3% of pedestrians opted for the highest possible score (7 points). This result is consistent with the observation experiment result. Accordingly, it is concluded that pedestrians in the middle lane are more inclined to change lanes based on their own preferences when enough space is available. In contrast, when they occupy either the wall-side or handrail-side lane, they generally continue along their current lane. From another perspective, this lane-changing performance further indicates pedestrians’ preference for the handrail and wall lanes.

In general, pedestrians’ preference for proactive lane-changing does not affect the overall crowd evacuation efficiency. However, when the distance between consecutive pedestrians is very close, sudden lane-changing by the front pedestrian will force the rear pedestrian to stop abruptly, thereby reducing the overall evacuation efficiency. To address this, sudden lane-changing by front pedestrians should be strictly restricted during evacuation guidance to avoid disrupting the movement of pedestrians behind.

4.3. Influencing Factors and Their Relative Importance

As shown in Section 4.1, pedestrians clearly prioritize the handrail-side lane, then the wall-side lane, with the middle lane being the least favored. Moreover, when they are in the middle lane with enough space, they often change lanes to match their preference. Drawing on previous studies, we consider that pedestrians’ behavior is influenced by three key factors: perceived safety, shortest-path, and behavioral habit. According to the questionnaire results (Table 2), mean scores for all three factors exceed 3 points on a five-point Likert scale, which indicates that they each have a meaningful influence on pedestrians’ lane-selection preferences.

4.3.1. Perceived Safety

During stair evacuation, pedestrians often experience substantial movement pressure due to high crowd density. Under such conditions, pedestrians tend to seek physical stability by identifying support points to maintain balance, which leads them to move toward the wall-side and handrail-side lanes. At the same time, fear of potential trampling accidents may prompt them to move toward regions with lower local density. The most salient manifestation of safety perception influencing lane-selection preference is seen in the choice between the wall-side and middle lanes: when the handrail-side lane is occupied, pedestrians always prefer the wall-side lane (as illustrated in Figure 11). Although the shortest-path consideration continues to exert some influence, perceived safety becomes the dominant factor shaping pedestrians’ lane-selection preferences under these conditions (as shown in Figure 10). This explains why, in Section 3.1.2 for the Class B stair, more pedestrians chose the wall-side lane than the middle lane.

4.3.2. Behavioral Habits

The notion of habitual behavior discussed in this section refers only to a part of the lane-selection preferences investigated in this study—specifically, the pedestrians’ conditioned responses. This section analyses pedestrians’ lane-selection behavior in the Class A stair to illustrate the influence of habitual behavior (Figure 12). For the Class A stair, where the exit is located directly in front of the stair, the shortest-path factor exerts an equal influence across all lanes. Both the wall-side lane and the handrail-side lane provide physical support that helps pedestrians maintain balance. Hence, when external conditions are disregarded, habitual behavior becomes the primary determinant of lane selection. The participants in this experiment were students from a Chinese university, who generally exhibit a right-side walking habit. Consequently, in the Class A stair, the number of pedestrians selecting the handrail-side lane was bigger than those selecting the wall-side lane.

4.3.3. Shortest Path

Section 4.2 states that pedestrians occupying the middle lane exhibited different lane-changing behavior in the Class A and Class B stair, primarily due to the influence of the shortest-path factor resulting from structural differences between the two stair classes. In the Class A stair (Figure 12), the exit is located directly ahead of pedestrians. Whichever lane they occupy, the directional attraction of the shortest path is oriented toward the front. In contrast, in the Class B stair (Figure 13), pedestrians’ target is the stair on the next floor, and the attraction of the shortest path is directed diagonally forward. Consequently, in the Class B stair, most pedestrians in the middle lane proactively change lanes, whereas in the Class A stair, most pedestrians continue along their original lane.

Therefore, from the perspective of individual pedestrians, we conclude that lane-selection preferences during stair evacuation are primarily shaped by three factors: perceived safety, behavioral habits, and the shortest path. Furthermore, based on the ranking results presented in Table 3, the relative influence of these factors on pedestrians’ lane-selection preferences can be clearly ordered from strongest to weakest as follows: behavioral habits, perceived safety, and the shortest path. Meanwhile, when setting parameters for pedestrian stair evacuation simulations, the weights of the influences of perceived safety, shortest path, and behavioral habit on pedestrians’ decision-making follow the order: behavioral habit > perceived safety > shortest path.

4.4. The Influence of Stair Width on Pedestrians’ Lane Selection Preference

This section discusses the influence of wall distance on pedestrian lane-selection preferences, where the preference of lane-selection preference specifically refers to the inclination toward the wall-side lane. According to the Section 4.1 discussion, pedestrians most prefer the handrail-side lane, then the wall-side lane, and the middle lane is the least favored. However, these results were obtained under a three-lane stair structure. When pedestrians are two, three, or even four lanes away from the wall, do they still prefer the wall-side lane?

Based on the questionnaire data (Figure 10a), the proportion of pedestrians scoring below four points shows a declining trend as the distance to the wall increases. To accurately examine how preferences change with wall distance, we plotted the mean scores at different wall distances (Figure 10b). The analysis shows that preference for the wall-side lane declines as the distance to the wall increases. Interestingly, the rate of decline first accelerates and then gradually levels off.

This shift in preference to the wall-side lane is primarily driven by a desire for perceived security and the pursuit of the shortest path. During stair evacuation, pedestrians select the wall-side lane based on two factors: (1) the wall provides pedestrians with a point of support, helping to stabilize their bodies. (2) Pedestrians may also choose less crowded lanes due to fear of crowd crush events. However, as the distance to the wall increase, pedestrians must expend more energy, walk a greater distance, and spend more time evacuating to reach the wall-side lane. Consequently, their preference to select the wall-side lane may decrease, and the rate of decline first accelerates and then gradually levels off. It can thus be concluded that the influence of stair width on Pedestrians’ Lane Preference is limited. When the stair width reaches a certain threshold, fewer and fewer pedestrians choose the outer lane, and most pedestrians still cluster in the lanes adjacent to the handrail. This leads to a significant reduction in space utilization and results in space waste. Therefore, in the design of stair lanes, a wider stair width is not necessarily better; instead, the design should be strictly based on the maximum number of occupants in the building.

5. Conclusions

This paper investigates the factors driving pedestrians’ lane-selection preferences and examines the extent to which these preferences influence pedestrians’ behavior. To do so, both a stair-evacuation observation experiment and a questionnaire survey were conducted. The following conclusions are drawn:

• When sufficient space is available, pedestrians in the middle lane are more inclined to proactively change lanes according to their preferences. In contrast, those occupying the wall-side or handrail-side lane are more likely to stay in their current lane and continue forward.
• Pedestrians’ lane-selection preferences during stair evacuation are primarily influenced by perceived safety, the shortest path, and behavioral habits. The relative influence of these factors on pedestrians’ lane-selection preferences from strongest to weakest are as follows: behavioral habits, perceived safety, and the shortest path.
• The effect of the distance to the wall on pedestrians’ lane-selection preference is varied. As the distance to the wall increases, the preference for the wall-side lane gradually decreases. Notably, the rate of decline accelerates at first, then slows down as the wall becomes farther away.

This study provides an in-depth analysis of the factors influencing lane-selection preferences and the extent to which these preferences affect pedestrians’ behaviors. Understanding both aspects contributes to a better comprehension of interactions among pedestrians and between pedestrians and the physical environment during stair evacuation.

However, there are still some limitations existing in the paper. (1) The sample was homogeneous: all participants were young adults, while older adults, pregnant women, children, and individuals with disabilities were not represented. (2) Insufficient consideration has been given to the impact of external conditions on the experiment, such as smoke, light, noise, temperature and so on, which may affect pedestrians’ lane-selection preferences. (3) The range of stair types considered in this observation experiment was limited. Stairs wider than three lanes, spiral staircases, and bisected or split parallel staircases were not included in the study. (4) The above conclusions are derived from the analysis of data collected through stair-evacuation observation experiments and questionnaire surveys. However, questionnaire surveys often suffer from recall bias, subjectivity, and behavioral validity issues, especially in emergency evacuation scenarios. When pedestrians are in different specific scenarios, their actual behaviors may deviate from their stated behaviors due to the influence of various external factors.

Therefore, to further investigate pedestrians’ stair lane-selection preferences, future research should not only conduct controlled experiment across different scenarios and populations but also improve in the following three aspects: (1) applying video-based systematic trajectory tracking to capture pedestrians’ movement trajectories; (2) utilizing kinematic variables, such as walking speed, local density, and interaction forces, to conduct quantitative analysis of pedestrians’ lane-selection preferences; and (3) implementing statistical modeling to correlate environmental variables and psychological factors with observed behaviors.