The transportation system of Dhaka Metropolitan City (DMC) is predominantly road-based and covers a total area of about 1530 square kilometers [14] . Every day, approximately 1.9 million passengers travel by bus [15] , yet the operators and government fail to offer an average service to the commuters. Many researchers investigated the quality of service provided by public buses in Dhaka city, and unfortunately, all of them discovered that the public bus service in Dhaka city is tragic and inconvenient to use. Due to long waiting times, delays on plying, long boarding times, overloading, discomfort, and a long walking distance from the residence or workplace to bus stoppages, current bus services are inefficient, unproductive, and unsafe [16] [17] [18] . Apart from all that, a female commuter has to face some specific gender-based harassment on public transport. Groping, stalking, Eve-teasing, intentional physical touch by fellow male passengers and conductor, and other verbal or gestural abuse are examples of gender-based harassment that a woman encounters on a daily basis [19] . However, the public bus service quality in Dhaka city is predominantly low due to the lack of proper Crime Prevention through Environmental Design (CPTED) facilities, lack of safety, shortage of buses, overcrowding, misbehavior of the bus drivers and conductors, lack of cleanliness, etc.

Willingness to pay (WTP) refers to the maximum amount of money a customer or consumer is willing to spend on a product or service [20] . In general, it is a person’s desire and ability to pay for something in order to protect, improve, or form something good [21] . If measured accurately, WTP can become a crucial factor for finding or deciding the ideal price or value to sell a product or service to the consumer. WTP was previously widely used by environmental economists to value non-traded and non-property-rights goods such as air, forests, water, and wildlife populations [22] . At present, with the increasing population and number of consumers, WTP is used by researchers from a variety of fields or majors for the purposes of policy planning [23] . In order to standardize the country’s transportation infrastructure, transportation engineers and policy makers from different areas have been exploring consumers’ willingness to pay for different service attributes of transport. [24] decided to investigate commuters’ WTP for improving the service quality of intermediate public transportation (IPT) modes by developing a Multinomial Logit Model (MNL) and collecting data through a stated preference survey. They discovered that IPT modes with high service quality, taking into account commuters’ safety and comfort, are crucial for the commuters’ willingness to pay. A Multiple Regression Model was developed by Nyamaliza [21] to estimate the WTP for improved public transport in Dodoma City, Tanzania. The study explored various factors affecting commuters’ willingness-to-pay for improved public transport, such as; travel distance in kilometers, travel time in minutes, monthly income, type of transport used, employment status, respondents’ level of education, marital status, age, and gender. Several researchers have explored that household income, occupation, age, and price bid are statistically significant factors that influence commuters’ WTP [23] [25] [26] . Chen et al. [27] conducted a study in China, collecting data via a stated preference survey and analyzing the data using an ordered logistic model. The findings disclosed the effect of age on WTP and how, keeping those findings in mind, transport as well as the tourist industry can flourish. They found that the older respondents were 1.6 - 2.4 times more likely to spend an extra penny to reduce waiting time for transport than the younger respondents. By adopting the Contingent Valuation Method (CVM), Pujiati et al. [28] provide a list of service attributes based on which commuters are more willing to pay, and these are ticket service, rate suitability, conditions, speed and length of waiting for the bus, and the placement and conditions of good stops. Jenatabadi et al. [1] came up with a new theoretical approach for measuring WTP accurately. A combination of Fuzzy Analytical Hierarchy Process (F-APH) and the Taguchi method were introduced to assess de-biased WTP and acquire a better understanding of the factors that stimulate public transport users to be willing to pay for the fare.

When measured correctly, willingness-to-pay (WTP) can help an operator understand the utility function of his product or service or how much satisfaction this service can provide to consumers. From time to time, various researchers have established various frameworks of WTP measurement methods. The effectiveness of services has long been assessed through “customer satisfaction surveys” in which participants express their perceptions of the services under consideration, typically by evaluating them on a scale of one to five [29] . But this kind of survey has many biases, such as a lengthy questionnaire, no interpretation of the respondents’ expectations, and erroneous results. To eliminate this bias while collecting data [3] [27] [29] , and used an alternative and more convenient way to capture customer expectations, or importance, through the Stated Preferences (SP) surveys, which actually allow for an indirect capture of the service attributes that are important to customers. One significant benefit of these types of experiments, according to [29] , is that hypothetical services can be evaluated by both current and potential users, who can express their opinions even if they do not use the service. As a result, the analyst can both capture potential users’ opinions and the service aspects that influence the users’ decision to use or not use public transportation [29] .

The state reference survey has two categories: direct surveys and indirect surveys. In the direct customer survey, consumers’ preferences are the ultimate concern, i.e., how they value a service and what amount they will be willing to pay to consume better services are directly analyzed to assess WTP. It is stated in [30] that 76% of the surveyed firms use a direct approach due to its obvious advantages. A direct approach has three major advantages over any other measurement methods of WTP, and these are: it is simple and very easy to implement, especially in the case of data collection and analysis; it is also very budget-friendly and less time-consuming [31] .

A comprehensive questionnaire survey was conducted with commuters of a public bus at major city Dhaka in Bangladesh. The survey was conducted during working days and holidays in March-April 2022 when users of the public bus services commute to different destinations. A total of 12 service attributes were asked to the respondents for evaluating their willingness-to-pay by considering the quality level and overall satisfaction level and from these 12, 8 service attributes are used to predict the possible improvement of service quality to mitigate the women commuters harassment, safety and security of public bus. To better understand the users.

To determine the sample size of the popularly we have adopted the following equation

Sample Size = 2500 × N × (1.96)²/{25 × (N − 1) + 2500 × (1.96)²} (1)

N: Total Population of the desired study area (N = 24,950,381 person)

Confidence coefficient: Z-score = (1.96) for 95% confidence level

The desired sample size is 385 person, which is below the entire 400 respondents who participated in this survey. The respondents are current users and customers of the general public bus companies based within the city, which is ensured by the respective questions.

A stated preference survey was conducted to estimate the commuters’ willingness-to-pay for the improvement of the existing bus service quality in Dhaka city. The study was carried out in some major points of Dhaka city and got an overall of 400 effective respondents who use public bus as their main mode of transport. Keeping the objectives in mind, the stated preference questionnaire was divided into five sections and included a wide range of questions (socio-demographics, attitudinal, and WTP) in the form of ratings and multiple choices.

A total of 400 public bus commuters in Dhaka city were selected as the subjects of the study. Table 2 shows the profile of the respondents’ sociodemographic information as well as highlights whether or not they are satisfied with the present bus service quality. Out of 400 respondents, 54.2% were male and 45.8% were female. It indicates that the frequency of male and female respondents in this study is almost equal. Most of the respondents in this study belong to the category of 30 - 40 years of age, and the percentage for that is almost 43.75. The demographic results show that most of the respondents are higher educated, i.e., out of 400 respondents, 227 are graduated and 119 are postgraduate. A total of 203 respondents, or 50.75%, are service holders by profession, and 110, or 27.5%, are businessmen. A major group of the respondents do not belong to the higher income group. Only about 12% of respondents to this survey earn between 40,000 and 50,000 BDT (Bangladeshi taka), with only 1.5 percent earning more than 50,000 taka. Respondents’ allocation for monthly travel expenditure is also very poor, i.e., only 6.75% of respondents spent more than 30% of their monthly income on travel. Moreover, a majority of respondents (40.5%) travelled 10 - 15 km per day, whereas the number of people who travelled 15 - 20 km was 119. Later, respondents’ satisfaction with the current public bus service in Dhaka city was measured. Table 2 shows that almost 333 out of 400 respondents said that they are not satisfied with the quality of bus service.

In general, Figure 1 shows that 94% of the respondents think that women commuters are harassed while travelling on public buses. Previously, numerous researchers presented the actual scenarios of bus conditions for female passengers. In short, at present, public bus service quality is a kind of dystopia for female passengers. Most of the respondents of this study, as shown in Figure 2 think that it’s the lack of service, planning, and policies that’s the ultimate reason for female passengers’ insecurity on public buses. The passengers already had a negative view towards public buses, and according to Figure 3, (89%) of them considered that overcrowding on public buses intensifies the present poor condition. From Figure 4, it is clearly visible that while answering what they think,

the female commuters are mostly harassed, about 63% answered that female commuters are mostly harassed while boarding and alighting the buses. On the other hand, to 21% of respondents, it’s when the contractors tickets inside the bus. However, almost 8% respondents respectively, considered that female passengers are mostly harassed while they are coming to the bus stand and while they are waiting for a bus at the bus stop. Figure 5 shows, however, that nearly 76% of commuters agreed that female passengers are harassed inside buses and 24% of them respond that women are mostly harassed at bus stop. Moreover, Figure 6 gives us a clear picture of how the majority (93%) of total respondents think public buses in Dhaka city have no implement of CPTED (crime prevention through environmental design). Well-planned and well-designed transport can mitigate the number of crimes, but in Bangladesh, transport infrastructure is so poor that it results in the most harassment. Figure 7 represents how 67% of respondents believed it’s the male co-passengers by whom female passengers are mostly harassed. On the other hand, 33% of respondents viewed female passengers as the most harassed by bus staff. According to Figure 8, 80% of total respondents said women commuters feel unsafe and insecure inside buses, while

14% and 8% said women feel insecure at bus stops and on the street, respectively. Lastly (shown in Figure 9), when answering, a majority (47%) of the respondents agreed that female passengers are mostly unsafe between 9.00 and 10.00 pm. Following this, 42% responded that female passengers feel unsafe between 10.00 and 11.00 pm.

Respondents were asked to reveal their expected service quality level for the

twelve most influential attributes they want to experience in the future while traveling by public bus in Dhaka city. Table 3 shows the percentages of the respondents’ expected service quality level based on the twelve attributes. The most expected “high” level of service is in boarding and alighting facilities, as respondents prefer a “high” level of service in boarding and alighting facilities. The second highly preferred attribute is the behavior of bus staff. The commuters preferred a “high” level of professionalism from the bus staff. The respondents expected the same high level of service from the two attributes, namely women commuters’ harassment and the fitness of the bus. On average, 5.5% and 11.5% of respondents expected “same as now” and “medium” level of service for the formal attribute. On the other hand, for the latter attribute, there were respectively 0.5% and 16.5% of responses for “same as now” and “medium” level of service.

Furthermore, the least three “high” preferred services are, in order, travel cost, comfort, and overcrowding mitigation. The respondents felt that an improved bus service can be assured if the authorities at least ensure a “medium” level of service quality for these three attributes, whereas a great concern should be on the mostly high-level preferred service attributes for providing better service quality.

Public transport services in Bangladesh, especially in its capital Dhaka, need to be improved. Apart from implying better policies and observation, a better service quality that ensures women’s safety and comfort is needed to improve the present state. As shown in Figure 10, the majority of respondents, i.e., almost 85% of total respondents, revealed their willingness to pay for the improvement of public bus service quality in Dhaka city, while 15% of respondents didn’t agree to pay more. A binary logistic model was run to find out the most influential service features the respondents would like to pay for. Table 4 presents the result of the binary logistic model.

In comparison to the null model, the model was statistically significant x² (8, N = 400) = 65.010, p < 0.001, indicating that the model has great explanatory power in addressing the most influential service facilities for which the respondents

are willing to pay. The model correctly identified 89.5 percent of the instances and explained between 15% (Cox & Snell R square) and 26.6% (Nagelkerke R square) of the variation in the dependent variable. The Hosmer and Lemeshow test result is non-significant (0.152) or p > 0.05, which in fact shows that the model fits the data well.

Table 4 shows that all the coefficients of improved service quality features are negative except for two features, named “CCTV camera and GPS tracker inside buses” and “fare collected inside bus using ticketing machine”. For “CCTV camera and GPS tracker inside buses,” almost 48.5%, of total male respondents, said “yes,” while the number of females with “yes” responses was 43.5%. Males and females, admitted that fares should be collected inside the bus using a ticketing machine in order to reduce women’s harassment inside the bus. This actually means that these two service features have a greater impact on the respondents’ willingness to pay for better service quality, or in other words, the respondents’ like to see these types of improved features in public buses more than the rest of the six attributes. The p-value for these service features is all non-significant, suggesting a non-zero correlation between the dependent and independent variables. Nevertheless, when asked about the preference, most of the respondents said “yes” to these improved service facilities and agreed to use this type of public bus in the future. However, the six improved service facilities after the two most influential service features are respectively described as follows. “No standing passengers are allowed”, with 48% and 44.25% of male and female passengers who said “yes” to this improved feature. Provide a special security service bus for female commuters during off-peak hours, to which 53.5% of males and 45.75% of females said “yes”, with no female and only 3 male respondents saying “no.” The next important feature is “Setup possible nearest bus stop”, and the number of male and female respondents who made “yes” responses to this feature was respectively 48.75% and 44%. Furthermore, the rest of the three most influential service features are “Separate seat arrangement for male and female”, which received 48.5% and 43.5% positive responses from male and female respondents; “Provide special security for female passengers on bus”, which received 49% and 44% positive responses from male and female respondents; and finally “A separate door for male and female passengers”, which received 54% and 45.75% positive responses from male and female respondents.

The amounts respondents were willing to pay were ordinal in nature. As a result, an ordinal logistic regression model was developed to determine which factors were influencing the number of respondents who agreed to let go in exchange for better service quality. The regression model offers a satisfactory level of significance and a high explanatory power of the predicators over the baseline intercept-only model. Table 5 presents the result of the ordinal logistic regression model.

The result shows that respondents’ gender and age range have nothing to do with the amount they are willing to pay, whereas the respondents’ WTP amount

OLR results: −2 Log-likelihood = 435.942. Chi-Square test (with 13 degrees of freedom) = 130.947; the corresponding p-value is p < 0.0001. Goodness of fit test (Deviance = 0.580 and Pearson = 0.000). Nagelkerke R² = 0.310.

is highly affected by their monthly income. The negative coefficient of all income groups prior to the reference category (i.e., the highest income group) suggests that all of those income groups are less likely to pay WTP > 4.00 BDT than the highest income group. The coefficient of the lowest income group is (−5.718), indicating their unwillingness to pay more compared to the second or third income group. The more a respondent’s income raises the more likelihood he or she has of paying WTP > 2.50 BDT. However, the respondents’ likelihood of paying WTP > 4.00 BDT or WTP 2.50 BDT isn’t influenced by their highest education level. The positive coefficient (1.034) of education level (3) suggests the graduate respondents are more likely to pay more than the reference group, which is the highest educational group. The negative coefficient (−0.353) of the second education group indicates they are less likely to pay more than the reference category. On the other hand, the positive coefficient (0.127) of the lowest education group means that they are more likely to pay more than the reference category. Overall, respondents’ highest education does not indicate that they are willing to pay a premium for higher service quality. Previous studies like [21] [32] , and [33] found some socio-demographic features of the respondents that have a great impact on the respondents’ WTP for the targeted service or products. [21] attempted to value WTP for improved public transport in Dodoma, Tanzania, and found that the commuters’ education level, marital status, age, monthly income, and employment status were all responsible for affecting the commuters’ willingness to pay. However, compared with the results of the [21] and [37] , the present study shows that it’s the commuters’ monthly income that is affecting their WTP for an improved service quality for public buses in Dhaka city.