Comparative Analysis of Public Transportation Comfort in Addis Ababa: Objective and Subjective Performance Metrics

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

The rapid economic growth and continuous advancements in automobile production result in an increasing overall number of vehicles in society. While this development has undoubtedly improved comfortability in both work and daily life, it simultaneously exacerbates issues such as traffic congestion, safety, and environmental pollution. To address these challenges, prioritizing public transport by enhancing its service quality is a crucial strategy [1, 2]. This can be achieved by performing continuous performance evaluations. These measures help to attract more passengers to opt for public transportation, thereby alleviating traffic congestion and minimizing environmental and safety impacts. Among various mass transit service quality attributes, comfort is the commuters’ primary concern when making mode choice decisions, and it is a growing concern after the introduction of private transportation [3]. Additionally, other studies, such as the one by Obsie, Woldeamanuel, and Woldetensae [4], also highlight the significance of comfort in influencing user satisfaction in Addis Ababa.

While many recent urban mobility studies, such as Obsie, Woldeamanuel, and Woldetensae [4] and Girma et al. [3], have revealed that comfort has a significant impact on commuters’ mode choice decisions, in Addis Ababa, a multitude of studies have been conducted to evaluate other aspects of mass transport services. For instance, studies on mass transport affordability [5], reliability [6], and an analysis of stakeholders’ opinions on accessibility [7] have been undertaken. Moreover, mode-specific mass transit performance evaluations have been conducted. For example, Debebe and Asteray [8] conducted an assessment of the current and future performance of the Addis Ababa light railway transit service in 2022 using mathematical modeling. Additionally, Berhan, Beshah, and Kitaw [9] conducted a study in 2013 to assess the operational and financial performance of the Anbesa City Bus Service Enterprise (ACBE).

Moreover, Addis Ababa is experiencing rapid population growth, and according to the Addis Ababa Transport Bureau report, the city’s private vehicular ownership has been increasing greatly, even though the city is heavily reliant on public transportation. However, the introduction of private transportation could raise the concept of comfort [10], and such studies are not abundant in developing countries like Addis Ababa.

Therefore, this study is aimed at conducting a mode-wide comfort performance evaluation to enhance public transport comfort because it is pivotal for service improvement and encouraging modal shifts. Furthermore, in a competitive environment, mode-wide performance comparison using key performance indicators (KPIs) to specify the strengths and weaknesses of the respective city bus services has made an essential contribution to government authorities and private companies.

2. Literature Review

The extensive research on mass transit comfort segregates into two core domains. The first focuses on vehicle performance and operational factors profoundly impacting passenger comfort—factors like vibrations, acceleration, jerk magnitude, and vehicle noise [11–13]. Conversely, the second domain centers on defining comfort aboard, embracing clean, ergonomic seating, optimal temperature control, minimal crowding, and maintaining low levels of noise and odors [14]. The indicator most frequently used for evaluating comfort during the journey is linked to the degree of crowding on transit service. Many scholars have studied and proved the importance of passenger load in determining transit comfort. For instance, Vovsha et al. [15] conducted a survey and showed that when a passenger has less than a 40% probability of getting a seat, he or she feels uncomfortable. In addition, other studies also showed that in-vehicle passenger loading, in-vehicle travel time, sound, and vibrations are the main factors that affect vehicle ride comfort [16, 17]. Moreover, another study performed in Harbin City, China [18], also revealed that in-vehicle time considerably affects passenger comfort perception, as does the passenger load factor. Another study also revealed that factors such as bus crowding level and travel time were identified as major priorities which affect passengers’ comfort perceptions [19].

The evaluation of transit service comfort performance involves both objective (quantifiable indicators) and subjective (user viewpoints) approaches [20]. However, sole reliance on such measures has its drawbacks, including the subjective nature of users’ evaluations, the inability to account for the perceptions of nonusers, and potential significant statistical errors due to inadequate respondent sampling or user judgment heterogeneity. Furthermore, the objective nature has limited flexibility, lacks context, and may not capture the full range of factors contributing to a positive transit experience [21]. In the present time, when evaluating transport performances, mode-wide performance evaluation and comparison are commendable rather than evaluating one specific mode of transport to enhance service quality [19].

For instance, a mode-wide comfort performance study was conducted in Dhaka, the capital of Bangladesh, in 2018 [22], and it revealed the comparison of comfort levels across various transport modes, contributing to a better understanding of user perceptions of both passengers and companies. However, the available studies conducted in Addis Ababa have primarily focused on specific modes, considering only the subjective nature of performance. Furthermore, although mass transit comfort significantly influences passenger satisfaction and mode choice decisions, there is a lack of studies in this area in Addis Ababa. Therefore, this study is aimed at addressing this gap by conducting a comparative mass transport comfort performance study in Addis Ababa, utilizing both objective and subjective performance indicators across five modes of transport by incorporating four variables (passenger load, vehicular noise, vibration, and in-vehicle time) and integrating the findings from both objective and subjective analyses. The results of these surveys can assist bus operators and authorities in designing more appealing measures to enhance bus comfort levels. Moreover, this study could help the commuters during mode choice decisions by showing which mode of mass transport is comfortable in respective KPIs.

In this study, traditional variables such as temperature inside the vehicle and cleanliness, commonly used to assess comfort in public transport, have been excluded. This decision is rooted in several factors, including climatic considerations, the specific context of Addis Ababa, and the study’s objectives. Addis Ababa’s high altitude, found 2400 m above mean sea level, typically maintains a moderate climate year-round, minimizing significant temperature variations inside vehicles. Thus, temperature fluctuations may not be a primary discomfort factor for passengers, as observed in cities with more extreme climates. Moreover, while cleanliness is essential to the overall passenger experience, it is solely subjective and unable to collect the data objectively. Therefore, focusing on variables like passenger loading, travel time, vibration, and sound aligns with the local context and the overarching goal of enhancing transportation services.

3. Methodology

3.1. Description of Study Area

Addis Ababa, located at coordinates 9°0${}^{\prime }$19.4436${}^{″}$ N and 38°45${}^{\prime }$48.9996${}^{″}$ E, is the capital of Ethiopia and ranks as the third major diplomatic city globally, after New York City and Geneva. Projections estimate a population increase of six million by 2030, covering 11 subcities (Figure 1).

[figure(s) omitted; refer to PDF]

The city offers both private and public transport options, constituting more than 56% of Ethiopia’s total number of vehicles, according to the Addis Ababa Transport Bureau report. Among public transport alternatives, minibus taxis accommodate 12–16 passengers, midi buses have a capacity of up to 37 passengers (22–27 seated and up to 10 standing), ACBE and Sheger Mass Transport Enterprise (SMTE) accommodate 70–90 passengers, and public service transport buses (mainly for government employees) offer 60 seats and 20 standing spaces for passengers. Additionally, a light rail transit system operates in Addis Ababa. However, in this study, the abovementioned five modes of mass transport are included, except for the Addis Ababa Light Rail, as it is not experienced by many commuters due to its limited alignment and unique characteristics. A specific route, from Megenagna to Piassa via Kebena, was chosen to reflect the operational diversity of modes and typical traffic conditions within Addis Ababa (Figure 2).

[figure(s) omitted; refer to PDF]

3.2. Sample Size and Sampling Technique

The study population comprises passengers primarily utilizing public transport. Ensuring the sample size strikes a balance between statistical significance and ethical considerations is crucial. The Cochran formula is employed to determine the minimum sample size, considering factors such as a population exceeding 100,000, aiming for a 95% confidence interval, a 0.5 (50%) degree of variability, and a ±5% precision. This yields a calculated sample size of 384. However, for robust evaluations and comparisons among different modes of transport while minimizing uncertainties and biases, it is deemed essential to include an equal number of samples for each mode. Therefore, the study incorporates 400 participants for each respective mode, totaling 2000 participants, utilizing random sampling to ensure each unit within the population has an equal chance of selection, thereby promoting unbiased and more precise results.

3.3. Data Collection

In this study, measurements of mass transport in-vehicle travel time, noise, vibration, and passenger loading, along with commuters’ perceptions regarding these comfort indicators, were conducted for five modes of public transport. The data collection took place from March 5, 2023, to May 23, 2023.

3.3.1. Collection of Objective Data

The objective data collection involved in-vehicle passenger counting and travel time measurement for the selected five modes of public transport along the designated road segment, which were conducted during peak hours (morning 7:30–9:30 am, evening hours 5:00–7:00 pm) and off-peak hours (10:30–11:30 am), as illustrated in Figure 2, for a total of 50 trips for each mode. In-vehicle sound and vibration were collected by using smartphones equipped with GPS devices and validated using Extech instruments, specifically a sound level meter described in the work of Murphy and Franks [23], conducted during those specified periods and road segments. Collecting in-vehicle sound levels and vibrations using an Android app and validating with instruments is a common practice in contemporary research. For instance, Rahman et al. [22] used an Android app to collect objective data when evaluating the comfort levels of different modes of transport in Dhaka City, Bangladesh.

3.3.2. Collection of Subjective Data

Commuters’ comfort perceptions regarding in-vehicle travel time, noise, vibration, and passenger loading for five modes of public transport were gathered through questionnaires using a 5-point Likert scale (Table 1). This data collection method is common and has been used in various studies, such as one performed in Harbin City, China [18], to collect the comfort perceptions of passengers. Another study conducted in Addis Ababa by Deyas, Woldeamanuel, and Erena [24] specifically focused on minibus taxi service satisfaction, utilizing a 5-point Likert scale questionnaire. For this study, the surveys were distributed onboard using a self-administered survey technique. In practice, interviewers clarified the questionnaire to groups of users, and users independently completed it to maximize the number of interviews during the extensive journey duration. A total of 2500 questionnaires were distributed on the selected buses of the chosen line. Out of these, 400 were returned fully completed, contributing to a total of 2000 questionnaires across the five modes of transport. Regarding the respondent rate, taxi users had a 94.1% response rate (425 distributed and 400 returned).

Table 1

Comfort perception level and scoring criteria.

Perception level	Extremely uncomfortable	Very uncomfortable	Slightly uncomfortable	Comfortable	Very comfortable
Scoring criteria	1	2	3	4	5

This high response rate can be attributed to taxis accommodating fewer passengers, making management easier. However, the ACBE had the lowest response rate at 88% (455 distributed and 400 returned) because this bus accommodates around 130 passengers during peak hours, making it difficult to manage. In general, this survey achieved an overall 80% response rate. The notably high response rates can be attributed to the interview administration method.

3.4. Data Analysis

3.4.1. Objective Data Analysis

Frequency analysis, statistical analysis, and signal processing stand out as essential techniques for delving into data on transport noise and vibration. In this study, the choice for analyzing transport noise and vibration aligns with the recommendation put forth in the FTA Transit Noise and Vibration Impact Assessment by Khan and Burdzik in 2023 [25], favoring a statistical analysis approach. Additionally, travel time analysis is also treated by statistical analysis methods. Turning attention to the analysis of passenger loading in the different modes, the methodology adheres to Exhibit 27.5 of HCM 2000. Two prevalent approaches exist for calculating in-vehicle passenger level of service (LOS): the first one is passengers per seat (P/seat), primarily applied when a vehicle is designed with a focus on seating, and the second is standing passengers per area, utilized when a vehicle accommodates more standees than seated passengers [14]. Consequently, this study opts for the analysis of P/seat LOS, given that the city transit system is predominantly designed for seated passengers. Exhibit 27.5 of HCM 2000 (Table 2) outlines the threshold values for passenger load LOS in terms of P/seat or passengers per square meter [26].

Table 2

Passenger loading level of service for bus.

LOS	m2/P	P/seat	Comments
A	> 1.20	0.00–0.50	No passenger needs to sit next to another
B	0.08–1.20	0.51–0.75	Passengers can choose where to sit
C	0.60–0.79	0.76–1.00	All passengers can sit
D	0.5–0.59	1.01–1.25	Comfortable loading for standees
E	0.40–0.49	1.26–1.50	Maximum schedule load
F	< 0.40	> 1.50	Crush load

Source: Highway capacity manual 2000.

3.4.2. Subjective Data Analysis

Prior to selecting any statistical methods, checking the nature of the data is mandatory. Therefore, the data underwent parametric statistical tests to assess assumptions such as normal data distribution, independent observations, equal variance across groups, and random sampling. Methods like the Shapiro–Wilk test or Kolmogorov–Smirnov test, along with skewness and kurtosis values, were used to evaluate data normality and distribution. A skewness between −1 and 1 suggests near symmetry, and a kurtosis of 3 denotes normal distribution. For normality tests, a $p$ value above 0.05 is considered ideal. The results of the normality test show that both skewness and kurtosis fall within acceptable ranges, indicating conformity to a normal distribution (Table 3). With all necessary assumptions met, a one-way analysis of variance (ANOVA) is chosen to compare means across the five transportation modes.

Table 3

Normality test result.

			Kolmogorov–Smirnova		Shapiro–Wilk
	Skewness	Kurtosis	Statistic	$\mathit{p}$ value	Statistic	$\mathit{p}$ value
Taxi	−0.127	2.74	0.075	0.056	0.986	0.061
Midi bus	0.196	2.96	0.065	0.063	0.982	0.053
PSETSE	0.180	2.98	0.053	0.077	0.991	0.070
ACBE	0.204	2.84	0.068	0.088	0.989	0.074
SMTE	0.043	2.89	0.051	0.074	0.990	0.079

3.4.3. Study Participant Summary

The study participant information is summarized in Table 4. This table encapsulates data pertaining to gender, income level, frequency of travel, and age distribution among the 2000 participants across the five modes of public transport, facilitating a holistic understanding of the demographics and travel behaviors involved in the study.

Table 4

General characteristics of the respondent.

General information		Frequency
		Taxi	Midi bus	Public service bus	Anbesa city bus	Sheger bus
Gender	Male	207	243	202	282	273
	Female	193	137	198	118	127
Monthly income (ETB) $1\text{ETB}=\text{\$}0.0167$	Below 3000	53	126	32	78	65
	3000–5000	72	159	132	178	156
	5000–10,000	83	78	150	108	132
	10,000–20,000	142	34	67	34	42
	Above 20,000	95	3	19	2	5
Frequency of travelling per week	Once a week	17	9	11	6	13
	2 days	26	19	32	13	19
	3 days	56	72	66	43	34
	4 days	84	98	94	79	87
	5 days	150	123	197	165	132
	6 days	56	45	—	50	63
	7 days	11	34	—	44	52
Age	Below 18	33	25	43	23	31
	18–24	67	55	83	68	76
	25–34	126	66	39	43	34
	35–44	89	112	63	79	77
	45–54	51	97	141	121	111
	55 and above	34	45	31	66	71
	Total	400	400	400	400	400

4. Results and Discussions

4.1. Objective Performance Result

Table 5 presents the objective comfort performance across five public transport modes in Addis Ababa with respect to KPIs. Taxis accommodate an average of 1.167 P/seat during both peak and off-peak periods, resulting in a consistently good LOS, likely due to their ability to accommodate 12–16 passengers overall. This relatively small capacity ensures that all seats are typically occupied, maintaining a consistent passenger loading rate.

Table 5

Peak and off-peak period performance results.

Modes	Parameters	Peak period	Off-peak period
Taxi	In-vehicle passenger loading (passenger/seat)	1.167, LOS D	1.167, LOS D
	Travel time (min/km)	2.6	1.67
	Vibration (vibrometer rating)	3.5	4.1
	In-vehicle sound level (dB)	72.3	73.53
Midi bus	In-vehicle passenger loading (passenger/seat)	1.537, LOS F	1.403, LOS E
	Travel time (min/km)	3.8	3.57
	Vibration (vibrometer rating)	3.9	4.4
	In-vehicle sound level (dB)	77.47	79.20
PSETSE	In-vehicle passenger loading (passenger/seat)	1.542, LOS F	1.403, LOS E
	Travel time (min/km)	3.5	3.01
	Vibration (vibrometer rating)	2.9	3.03
	In-vehicle sound level (dB)	59.67	60.57
Anbesa bus	In-vehicle passenger loading (passenger/seat)	2.113, LOS F	1.548, LOS F
	Travel time (min/km)	4.2	3.34
	Vibration (vibrometer rating)	3.3	3.34
	In-vehicle sound level (dB)	72.53	72.63
Sheger bus	In-vehicle passenger loading (passenger/seat)	1.9, LOS F	1.31, LOS D
	Travel time (min/km)	4.2	3.32
	Vibration (vibrometer rating)	3.2	3.31
	In-vehicle sound level (dB)	72.53	72.63

Midi buses and public service employee transport service (PSETSE) have similar passenger loading experiences, with averages of 1.537 and 1.542 P/seat during peak periods and 1.403 P/seat during off-peak periods. Both buses represent the second most comfortable means of transport in Addis Ababa in terms of passenger loading. ACBEs have the highest passenger loading during peak (2.113) and off-peak periods (1.548), followed by Sheger buses with 1.9 P/seat during peak and 1.31 P/seat during off-peak periods.

Regarding travel time, significant differences exist between peak and off-peak travel times among the modes of transport. Taxis take 2.6 min during peak periods and 1.67 min during off-peak periods to cover 1 km, while midi buses take 3.8 and 3.57 min, PSETSE takes 3.5 and 3.01 min, and Anbesa and Sheger buses take 4.2 and 3.34 min, respectively, during peak and off-peak periods (Table 5). The results showed that passenger crowdedness and travel time, except for the small bus, which is a taxi, exhibit significant variation during peak and off-peak periods. This might be due to the high number of trip makers during peak periods such as morning and evening, resulting in vehicles being overloaded to accommodate the travel demand. Larger buses, with both standing and seating passengers, have the flexibility to accommodate standing passengers even when available standing areas are fairly occupied. However, smaller mass transit modes like minibus taxis, with fixed numbers of seats and unable to accommodate standing passengers, are obligated to provide service for a fixed number of seats. Regarding travel time, larger buses show no significant variation, which may be due to their larger passenger load. Other studies have indicated that passenger crowdedness increases travel time [27] because it requires extended time for loading and unloading passengers.

In-vehicle sound and vibration also vary throughout the day and across modes of public transport. Larger buses generally exhibit lower vibration levels during both peak and off-peak periods. For example, PSETSE records vibration levels of 2.9 and 3.03 during peak and off-peak periods, followed by Sheger bus and ACBE at 3.2 and 3.3 and 3.31 and 3.34, respectively. Midi buses exhibit the highest vibration levels at 3.9 and 4.4 during peak and off-peak periods. Similarly, midi buses also produce the highest sound levels measured in decibels, recording 77.47 and 79.20 dB during peak and off-peak periods, followed by taxis at 72.3 and 73.53 dB. In contrast, PSETSE exhibits the lowest sound levels at 59.67 and 60.57 dB during peak and off-peak periods (Table 5). From the results, taxis and midi buses show significant variations in vehicle vibration and sound during off-peak and peak time measurements. This may be due to the increase in vehicle speed during off-peak times, leading to higher levels of vibration and noise experienced by passengers. The rise in vibration and sound primarily stems from factors such as road surface irregularities, engine vibrations, and aerodynamic effects [28, 29].

Furthermore, at higher speeds, vehicles encounter more bumps, potholes, and imperfections on the road surface, resulting in greater vibrations transmitted through the vehicle’s suspension system to the passengers. Additionally, engine vibrations and aerodynamic forces become more pronounced at higher speeds, further contributing to the overall vibration levels experienced inside the vehicle [30, 31]. However, for larger buses like PSETSE, ACBEs, and Sheger buses, there is not a significant difference in peak and off-peak vibration and sound. This may be attributed to their suspension systems and vehicle designs, as studies have revealed that these factors help mitigate vibrations even at higher speeds [32].

4.2. Subjective Performance Result

To comprehensively capture commuters’ perceptions regarding these variables and validate the objective performance results, subjective data collected during peak periods are analyzed as follows. Table 6 presents the average comfort variable values as well as the overall comfort values of different transport modes. The results reveal varying comfort scores across transportation modes. The minibus taxi obtained the highest overall comfort score of 10.13, followed closely by the PSETSE bus with a score of 9.62. Meanwhile, the midi bus scored 7.81, and the ACBE scored 8.84. To statistically test whether comfort level changes with a change in mode, a single-factor ANOVA test is conducted (Table 7). Here, the null hypothesis is that the mean variance will remain the same between various modes of transport for a particular variable.

Table 6

Overall comfort performance perception across five modes of transport.

Modes	Predictor variables
	Sound (1–5)	Vibration (1–5)	Travel time (1–5)	Passenger loading (1–5)	Overall comfort (1–20)
Minibus taxi	2.61	2.23	2.25	2.92	10.13
Midi bus	2.13	1.94	1.89	1.89	7.81
PSETSE	2.88	2.65	2.24	2.278	9.62
Anbesa bus	2.66	2.39	2.04	1.69	8.84
Sheger bus	2.79	2.44	2.11	1.89	9.45

Table 7

Analysis of variance (ANOVA) test result.

	Sum of squares	df	Mean square	$\mathit{F}$	$\mathit{p}$ value
Sound
Between groups	134.877	4	33.719	30.399	0.000
Within groups	2212.903	1995	1.109
Total	2347.780	1999
Vibration
Between groups	113.075	4	28.269	21.062	0.000
Within groups	2674.907	1995	1.342
Total	2787.982	1999
Travel time
Between groups	36.552	4	9.138	8.038	0.000
Within groups	2265.794	1995	1.137
Total	2302.346	1999
In-vehicle passenger loading
Between groups	390.392	4	97.598	101.32	0.000
Within groups	1921.608	1995	0.963
Total	2312.000	1999

The one-way ANOVA table (Table 7) indicates a significant difference in the comfort performance among the groups, with a $p$ value below 0.05. However, the result does not specify the particular groups that differ from each other. To determine the specific differences, a post hoc multiple comparison analysis is conducted, and the results are presented in Tables 8, 9, 10, and 11.

Table 8

In-vehicle passenger loading post hoc analysis.

Mode of transport	$\mathit{N}$	Subset for $\mathbf{a}\mathbf{l}\mathbf{p}\mathbf{h}\mathbf{a}=\mathbf{0.05}$
		1	2	3	4
Anbesa city bus	400	1.69
Midi bus	400	1.8475	1.8475
Sheger bus	400		1.89
PSETSE	400			2.275
Minibus taxi	400				2.92

Note: Means for groups in homogeneous subsets are displayed. Uses harmonic mean sample size = 400.000.

Table 9

Travel time post hoc comparison.

Mode of transport	$\mathit{N}$	Subset for $\mathbf{a}\mathbf{l}\mathbf{p}\mathbf{h}\mathbf{a}=\mathbf{0.05}$
		1	2	3
Midi bus	400	1.88
Anbesa city bus	400	2.04	2.04
Sheger bus	400		2.11	2.11
PSETSE	400			2.24
Minibus taxi	400			2.25

Note: Means for groups in homogeneous subsets are displayed. Uses harmonic mean sample size = 400.000.

Table 10

In-vehicle vibration post hoc comparison.

Mode of transport	$\mathit{N}$	Subset for $\mathbf{a}\mathbf{l}\mathbf{p}\mathbf{h}\mathbf{a}=\mathbf{0.05}$
		1	2	3	4
Midi bus	400	1.94
Minibus taxi	400		2.22
Anbesa city bus	400		2.38	2.38
Sheger bus	400			2.44
PSETSE	400				2.65

Note: Means for groups in homogeneous subsets are displayed. Uses harmonic mean sample size = 400.000.

Table 11

In-vehicle sound level post hoc comparison.

Mode of transport	$\mathit{N}$	Subset for $\mathbf{a}\mathbf{l}\mathbf{p}\mathbf{h}\mathbf{a}=\mathbf{0.05}$
		1	2	3	4
Midi bus	400	2.13
Minibus taxi	400		2.602
Anbesa city bus	400		2.66	2.66
Sheger bus	400			2.79	2.79
PSETSE bus	400				2.88

Note: Means for groups in homogeneous subsets are displayed. Uses harmonic mean sample size = 400.000.

Table 8 presents the perception of in-vehicle passenger loading comfort across five modes of transport. The ACBE received the lowest score of 1.69, followed by the midi bus at 1.85, without significant performance differences as revealed by the post hoc test. In contrast, the minibus taxi and PSETSE bus had the highest scores of 2.92 and 2.28, respectively. This indicates that passengers perceive the ACBE as uncomfortable regarding passenger crowdedness, while the taxi is perceived as a comfortable mode of transport regarding passenger loading.

This result aligns with the objective analysis presented in the above section, which revealed that the ACBE has an average of 2.113 P/seat, potentially reducing passenger comfort perception. A study done by Vovsha et al. [15] further supports this, stating that when passengers have a lesser chance of getting a seat, they feel uncomfortable.

Passenger feedback on travel time (Table 9) shows that the midi bus had the lowest score (1.88), indicating a very uncomfortable experience. Followed by, the ACBE and Sheger city bus had scores of 2.04 and 2.11, respectively, without significant variation. On the other hand, minibus taxis had higher scores of 2.25, suggesting the highest comfort perception by commuters regarding travel time compared to other modes. Regarding in-vehicle vibration, the findings indicate that the midi bus displayed the lowest vibration comfort perception level at 1.94, while PSETSE showed the highest at 2.65 (Table 10). This suggests that passengers perceived the midi bus as very uncomfortable regarding vibration, followed by the minibus taxi, while the PSETSE had a comfortable performance. The post hoc multiple comparison test for in-vehicle sound level (Table 11) revealed that the midi bus exhibited the lowest perception at 2.13, followed by the minibus taxi scoring 2.6, while PSETSE had the highest comfort perception at 2.88.

Combining both objective and subjective analysis methods has the advantage of addressing the limitations of each method and helps to capture the full spectrum of phenomena. To summarize the results from both peak period objective and subjective evaluations, correlation analysis was performed and presented in Table 12. The results showed that there is a correlation value of 0.6 and 0.77, indicating a moderate positive correlation regarding in-vehicle vibration and travel time objective and subjective results, respectively. Moreover, the highest correlation is observed in passenger loading and in-vehicle sound, with values of 0.9 and 1, respectively. On average, there is an 81.75% correlation found between the two performance metrics. The correlation results provide a benchmark for evaluating the effectiveness of interventions aimed at improving passenger experience. By monitoring changes in both objective and subjective metrics over time, transit authorities can assess the impact of implemented improvements and adjust strategies accordingly.

Table 12

Correlation between objective and subjective results.

	Vibration
	Subjective performance	Objective performance
Subjective performance	1
Objective performance	0.6	1
	Sound
	Subjective performance	Objective performance
Subjective performance	1
Objective performance	1	1
	Travel time
	Subjective performance	Objective performance
Subjective performance	1
Objective performance	0.77	1
	Passenger loading
	Subjective performance	Objective performance
Subjective performance	1
Objective performance	0.9	1

5. Conclusion

Nowadays, the comfort of mass transportation plays a pivotal role in commuters’ mode choice decisions. Therefore, to attract commuters and remain competitive in the market, available modes must offer comfortable services. Achieving this necessitates a comprehensive evaluation and comparison of comfort performance across different modes. However, such studies are often lacking in developing countries like Ethiopia. Hence, this study is aimed at evaluating the comfort performance of five public transport modes found in Addis Ababa: taxi, midi bus, PSETSE, ACBE, and Sheger bus by utilizing crowding, noise, vibration, and travel time as predictor variables. While many global studies on transport comfort tend to focus solely on objective measures such as passenger density, a few, exemplified by the work of Mohd Mahudin, Cox, and Griffiths [33], argue that objective treatments alone cannot fully capture the passenger experience. They suggest the incorporation of passenger perceptions into the evaluation process [21]. In Addis Ababa, however, many mass transit studies have a subjective nature. Therefore, this study adopts a comprehensive approach by integrating both objective metrics and subjective evaluations and finally establishes correlations between the two. Regarding subjective analysis, this study utilizes statistical methods such as ANOVA and post hoc multiple comparison tests, in addition to objective analysis performed through statistical and LOS analyses.

The results highlight variations in comfort levels across five transport modes concerning passenger loading, travel time, vibration, and sound. Minibus taxis emerge as the most comfortable, offering 2.6 min/km during peak hours and 1.67 min during off-peak periods, with an average of 1.167 P/seat. Following closely is the PSETSE, which exhibits the lowest noise production at 59.67 and 60.57 dB during peak and off-peak periods, respectively, along with the lowest vibration ratings of 2.9 and 3.03 during peak and off-peak periods, respectively. In contrast, midi buses emerge as the least comfortable mode, attributed to elevated noise levels of 77.47 and 79.20 dB during peak and off-peak periods, respectively, and an average vibration rating of 3.9 and 4.4 during peak and off-peak periods, respectively. Despite ranking third in travel time efficiency at 3.8 and 3.57 min/km during peak and off-peak periods, respectively, among the studied modes, they exhibit notable discomfort.

Regarding the Sheger city bus and ACBE, they are ranked third and fourth in comfort performance, respectively, displaying good performance in vibration and sound levels (3.2 and 3.31 for the Sheger bus and 3.3 and 3.34 for the Anbesa bus during peak and off-peak periods, respectively) and consistent sound levels of 72.53 and 72.63 dB during peak and off-peak periods, respectively. However, these two modes have the least performance in terms of in-vehicle passenger loading and travel time. ACBE has 2.113 and 1.548 P/seat and takes 4.2 and 3.34 min to cover 1 km during peak and off-peak periods, respectively, putting it at the lowest performer in these two parameters. On the other hand, the Sheger bus accommodates 1.9 and 1.31 P/seat and takes 4.2 and 3.32 min to cover 1 km during peak and off-peak periods, respectively. Regarding the subjective responses, the minibus taxi obtained the highest overall comfort score, followed closely by the PSETSE bus. Meanwhile, the midi bus scored the least, followed by the ACBE and the Sheger bus, respectively. In general, this objective result and subjective responses demonstrate a substantial correlation, averaging 81.75%, with values ranging from 60% to 100%.

6. Policy Implication and Future Research Direction

• Passenger loading or crowdedness is classified as a key determinant affecting commuter satisfaction and mode choice considerations. However, larger buses like the ACBE, Sheger bus, and midi bus accommodate a higher P/seat and are classified as crowded vehicle alternatives. Therefore, to increase their chances of being chosen by commuters and to enhance passenger satisfaction, these modes should aim to reduce passenger crowdedness. One way to alleviate this problem is for the respective companies to increase the frequency of trips. This could be achieved by implementing scheduling, which is currently lacking.

• Travel time is another important parameter that significantly affects passenger satisfaction, as identified by many studies such as Shen et al. [18]. In this case, larger buses like the ACBE and the Sheger bus often have longer travel times, so efforts should be made to reduce it. One approach could be to decrease passenger crowdedness, which in turn would reduce the time taken for loading and unloading processes. A study conducted by Shao et al. [27] in Shanghai, China, states that passenger crowdedness significantly increases travel time. Therefore, reducing the number of passengers on board could decrease travel time by minimizing loading and unloading times. This would enhance passenger satisfaction and significantly impact mode choice decisions.

• Based on the results, minibuses, taxis, and midi buses generate higher levels of sound and vibration, leading to these specific modes being ranked as uncomfortable forms of mass transport in Addis Ababa based on this parameter. Therefore, the company should address this issue to remain competitive in this business and attract more passengers. This could be achieved by conducting ongoing maintenance, as regular maintenance can improve their performance by reducing noise and vibration generation.

In conclusion, the findings from this study provide valuable insights into the comfort dynamics within Addis Ababa’s transport system. By identifying the comfort performance of various modes and highlighting the correlation between objective and subjective evaluations, this research lays the groundwork for potential enhancements in service quality. Future research could expand on these findings by exploring additional modes and variables, contributing further to the establishment of mass transport comfort standards for the city.

Ultimately, improving comfort attributes can play a vital role in attracting commuters and ensuring a positive experience for passengers, thereby enhancing the overall efficiency and effectiveness of the transportation network in Addis Ababa.

Disclosure

The research is conducted as a partial fulfillment of the degree of Master of Science in Civil Engineering (Road and Transport Engineering) at Addis Ababa University.

Funding

The authors received no specific funding for this work.

Acknowledgments

The authors extend their sincere gratitude to the Addis Ababa Road & Transport Bureau (AARTB) for their generous provision of data and unwavering collaboration. We also thank the Department of Civil Engineering at Addis Ababa University (AAU) for supplying the essential materials and resources that were crucial to achieving the objectives of this study. Furthermore, we are deeply appreciative of the insightful comments and suggestions provided by the editor and reviewers, which significantly enhanced the quality of this paper. We acknowledge the use of Grammarly and ChatGPT for writing assistance, particularly in the editing and language refinement of the Introduction and Data Analysis sections of the manuscript.

References

[1] Q. Bao, Y. Shen, L. Creemers, B. Kochan, T. Bellemans, D. Janssens, G. Wets, "Investigating the minimum size of study area for an activity-based travel demand forecasting model," Mathematical Problems in Engineering, vol. 2015 no. 1,DOI: 10.1155/2015/162632, 2015.

[2] H. Zhao, L. L. Guo, X. Y. Zeng, "Evaluation of bus vibration comfort based on passenger crowdsourcing mode," Mathematical Problems in Engineering, vol. 2016,DOI: 10.1155/2016/2132454, 2016.

[3] M. Girma, B. Woldetensae, G. Nuriye, T. Lika, "Importance performance analysis for evaluating city bus service quality in Addis Ababa, Ethiopia," Urban, Planning and Transport Research, vol. 10 no. 1, pp. 451-465, DOI: 10.1080/21650020.2022.2099968, 2022.

[4] A. Obsie, M. Woldeamanuel, B. Woldetensae, "Service quality of Addis Ababa light rail transit: passengers’ views and perspectives," Urban Rail Transit, vol. 6 no. 4,DOI: 10.1007/s40864-020-00135-2, 2020.

[5] M. Mohammed, S. Senadheera, "Public transport affordability evaluation for Addis Ababa City," Sustainability, vol. 14 no. 11,DOI: 10.3390/su14116883, 2022.

[6] F. A. Gidebo, J. Szpytko, "Reliability assessment of the transport system, Addis Ababa case study," Journal of Konbin, vol. 49 no. 4, pp. 27-36, DOI: 10.2478/jok-2019-0073, 2019.

[7] A. Aklilu, T. Necha, "Analysis of the spatial accessibility of Addis Ababa’s light rail transit: the case of East–West Corridor," Urban Rail Transit, vol. 4 no. 1, pp. 35-48, DOI: 10.1007/s40864-018-0076-6, 2018.

[8] G. Debebe, B. Asteray, "Assessment on the current and future performance of Addis Ababa light rail transit service using mathematical modeling," Advances in Civil Engineering, vol. 2022 no. 1,DOI: 10.1155/2022/5097929, 2022.

[9] E. Berhan, B. Beshah, D. Kitaw, "Performance analysis on public bus transport of the city of Addis Ababa," International Journal of Computer Information Systems and Industrial Management Applications, vol. 5, 2013.

[10] J. Chen, S. Li, "Mode choice model for public transport with categorized latent variables," Mathematical Problems in Engineering, vol. 2017 no. 1,DOI: 10.1155/2017/7861945, 2017.

[11] J. C. Castellanos, F. Fruett, "Embedded system to evaluate the passenger comfort in public transportation based on dynamical vehicle behavior with user’s feedback," Measurement, vol. 47 no. 1, pp. 442-451, DOI: 10.1016/j.measurement.2013.08.068, 2014.

[12] D. Sekulić, V. Dedović, S. Rusov, S. Šalinić, A. Obradović, "Analysis of vibration effects on the comfort of intercity bus users by oscillatory model with ten degrees of freedom," Applied Mathematical Modelling, vol. 37 no. 18–19, pp. 8629-8644, DOI: 10.1016/j.apm.2013.03.060, 2013.

[13] K. Zhang, K. Zhou, F. Zhang, "Evaluating bus transit performance of Chinese cities: developing an overall bus comfort model," Transportation Research Part A: Policy and Practice, vol. 69, pp. 105-112, DOI: 10.1016/j.tra.2014.08.020, 2014.

[14] L. Eboli, G. Mazzulla, "Performance indicators for an objective measure of public transport service quality," European Transport-Trasporti Europei, vol. 51, 2012.

[15] P. Vovsha, M. G. S. Oliveira, M. St, N. W. Suite, W. Davidson, "Statistical analysis of transit user preferences including in-vehicle crowding and service reliability," Transportation Research Board 93rd Annual Meeting. January 12-16, .

[16] C.-Y. Lin, L.-J. Chen, Y.-Y. Chen, W.-C. Lee, A comfort measuring system for public transportation systems using participatory phone sensing, 2010.

[17] A. Tirachini, D. A. Hensher, J. M. Rose, "Crowding in public transport systems: effects on users, operation and implications for the estimation of demand," Transportation Research Part A Policy and Practice, vol. 53, pp. 36-52, DOI: 10.1016/j.tra.2013.06.005, 2013.

[18] X. Shen, S. Feng, Z. Li, B. Hu, "Analysis of bus passenger comfort perception based on passenger load factor and in-vehicle time," Springerplus, vol. 5 no. 1,DOI: 10.1186/s40064-016-1694-7, 2016.

[19] A. Páez, K. Whalen, "Enjoyment of commute: a comparison of different transportation modes," Transportation Research Part A: Policy and Practice, vol. 44 no. 7,DOI: 10.1016/j.tra.2010.04.003, 2010.

[20] R. Narayanamoorthy, V. K. Goel, V. H. Saran, "Vibration measurement-a tool for evaluating activity comfort in trains," . 2009. http://www.researchgate.net/publication/266214233

[21] Z. Li, D. A. Hensher, "Crowding in public transport: a review of objective and subjective measures," Journal of Public Transportation, vol. 16 no. 2, pp. 107-134, DOI: 10.5038/2375-0901.16.2.6, 2013.

[22] M. Rahman, K. A. Haque, M. Asif, H. Anik, M. Hossain, "Comparing level of comfort for various transportation modes in Dhaka City durability of concrete in marine environment view project," . 2018. https://www.researchgate.net/publication/330350109

[23] W. J. Murphy, J. R. Franks, "Revisiting the NIOSH criteria for a recommended standard: occupational noise exposure," The Journal of the Acoustical Society of America, vol. 111,DOI: 10.1121/1.4778162, 2002.

[24] G. T. Deyas, M. Woldeamanuel, S. H. Erena, "Customers’ satisfaction towards Addis Ababa City’s minibus taxi service," Heliyon, vol. 9 no. 11,DOI: 10.2139/ssrn.4177735, 2022.

[25] D. Khan, R. Burdzik, "Measurement and analysis of transport noise and vibration: a review of techniques, case studies, and future directions," Measurement: Journal of the International Measurement Confederation, vol. 220, article 113354,DOI: 10.1016/j.measurement.2023.113354, 2023.

[26] National Research Council (US), Highway Capacity Manual, 2000.

[27] M. Shao, C. Xie, T. Li, L. Sun, "Influence of in-vehicle crowding on passenger travel time value: insights from bus transit in Shanghai, China," International Journal of Transportation Science and Technology, vol. 11 no. 4, pp. 665-677, DOI: 10.1016/j.ijtst.2021.09.001, 2022.

[28] V. Kumar, "Modelling and simulation of a passenger car for comfort evaluation," International Journal for Research in Applied Science and Engineering Technology, vol. 6 no. 4, pp. 4013-4018, DOI: 10.22214/ijraset.2018.4662, 2018.

[29] U. Kırbaş, M. Karaşahin, "Investigation of ride comfort limits on urban asphalt concrete pavements," International Journal of Pavement Engineering, vol. 19 no. 10, pp. 949-955, DOI: 10.1080/10298436.2016.1224413, 2018.

[30] Y. S. Park, J. H. Jeon, Y. J. Kang, "ISO 8608-based pavement roughness classification with artificial neural networks using suspension vibration measurements," INTER-NOISE and NOISE-CON Congress and Conference Proceedings, vol. 261 no. 1, 2020.

[31] M. Agostinacchio, D. Ciampa, S. Olita, "The vibrations induced by surface irregularities in road pavements - a Matlab® approach," European Transport Research Review, vol. 6 no. 3, pp. 267-275, DOI: 10.1007/s12544-013-0127-8, 2014.

[32] X. Bai, L. Lu, C. Zhang, W. Geng, "Research on height adjustment characteristics of heavy vehicle active air suspension based on fuzzy control," World Electric Vehicle Journal, vol. 14 no. 8,DOI: 10.3390/wevj14080210, 2023.

[33] N. D. Mohd Mahudin, T. Cox, A. Griffiths, "Measuring rail passenger crowding: scale development and psychometric properties," Transportation Research Part F: Traffic Psychology and Behaviour, vol. 15 no. 1,DOI: 10.1016/j.trf.2011.11.006, 2012.