Exploring Women’s Experiences of Gender-Based Violence and Other Threats to Safety on Public Transport in Bangladesh

Equal access to safe transport is increasingly conceptualised as a fundamental right for women, with demonstrated effects on health outcomes, social and economic mobility and societal participation. This study analysed qualitative and quantitative data to examine travel patterns and experiences among 200 women (aged between 18–64 years) using paid transport for work or educational purposes in Bangladesh. The results showed that the women faced multiple threats to their safety, including gender-based violence, harassment and crime, and traffic and non-traffic injuries, and that the use of paid transport was associated with high levels of anxiety and fear. Despite these circumstances, the women were captive travellers, forced to make transport choices based on price, availability and ease of travel rather than safety. Unable to choose safe transport, the women attempted to mitigate risks by changing their travel patterns and behaviour and by restricting their travel frequency. These findings are discussed within the context of women’s rights and mobility


Introduction
For many in the Global North, perspectives of the Global South follow a narrative of development in several domains, in particular the economy (its structure and performance), governance (democracy, legislation and implementation of fair and just systems) and society (moral and civic values and practices expressing the nature and degree of discrimination and hierarchies) (Carrington 2015). This is also typically the case with feminist research that has focused almost exclusively on issues facing English-speaking women in the Global North (Roberts and Connell 2016).
Women's equality was positioned at the centre of development in 2000 by the United Nations (UN) General Assembly Session 'Women 2000: Gender Equality, Development and Peace for the Twenty-first Century' (https://www.un.org/womenwatch/daw/followup/session/presskit/gasp.htm), building on the 1995 Beijing Platform for Action and representing a further reinforcement of instruments such as the 1979 Convention on the Elimination of all Forms of Discrimination Against Women (CEDAW). By now, it is widely acknowledged that gender issues are significant across all domains of development and that gender-based violence (GBV), in particular, both challenges and influences the success of development programs.
The domain of transportation of both passengers and freight is considered a critical area of development in the Global South. In recent years there have been serious attempts to place passenger transport (and mobility more generally) into a justice framework. Pereira, Schwanen and Banister (2017) analysed the insights and contributions to transport by different theories of justice, concluding that a combination of Rawls' egalitarianism and capability approaches, such as those of Sen (2009), pointed to the need to view accessibility as a human capability. From this, it follows that transportation systems need to respect the rights of individuals to have a minimum level of accessibility, 'prioritise disadvantaged groups, reduce inequalities of opportunities, and mitigate transport externalities' (Pereira, Schwanen and Banister 2017: 170). It is disappointing that the applicability of this transport justice approach to gender is only mentioned by Pereira, Schwanen and Banister (2017) in terms of gender as an interpersonal characteristic affecting accessibility, with gender-relevant issues such as access and equity seen as matters of service provision. More recently, the emerging area of mobility justice has taken a stronger critical focus that incorporates gender issues and power relations explicitly into transport systems analysis (Verlinghieri and Schwanen 2020). Although mobility justice research now encompasses the study of GBV in public transport in general (e.g., Lubitow, Abelson and Carpenter 2020), research such as this is still rare in the Global South. Our study aimed to address this gap.
The present research focuses on paid transport travel by women in Dhaka, Bangladesh, for education or work purposes. In low-and middle-income countries, paid transport involves private operators (e.g., private buses, informal motorcycle taxis and ride-sourcing) that offer transport services complementary to public infrastructure, which is often underdeveloped. Gender inequality in educational attainment and employment is documented worldwide but is more pronounced in the Global South (World Economic Forum 2020). Achieving gender parity in these areas is a social justice issue, with consequences for women individually and the societies they live in. At the individual and contextual level, education and employment have been linked to better health outcomes, increased economic and social mobility and widened political and civic participation (World Bank Group 2015). At the macro level, women's educational attainment and workforce participation are important drivers of economic growth, particularly in emerging economies (Hassan and Rafaz 2017;Kühn, Horne and Yoon 2017;Oztunc, Oo and Serin 2015).
Lack of access to safe transport and other constraints on women's mobility are increasingly being recognised as critical barriers to women's educational and workforce participation and broader freedom and autonomy (Clarke 2012;Priya Uteng and Turner 2019). This barrier is, in turn, reflective of gendered inequalities (income, legal and social norms around driving and travel) and exposure to a range of risks (GBV, as defined further below; robbery; and being injured in a traffic crash or fall). The notion of perceived rather than actual risk needs to be emphasised since it is an individual woman's perception of risk that will influence their travel decision. While statistics (if available) might contribute to that decision, it will also be based on personal experience, observation of the experiences of others, stories about the experiences of others, and the influence of opinion leaders. Research from both the Global North and South show that men and women's travel patterns differ. With fewer financial means and less access to private vehicles, women are more dependent on public transport systems and more likely to spend time on slower modes of transport, to combine trips and to spend a higher percentage of their journeys on foot (González et al. 2015;Uteng 2012). Research from developing countries indicates that investment in safe public transport has a positive impact on employment, particularly among the poorest, with emerging evidence suggesting that women benefit to a greater extent than men (Boisjoly, Moreno-Monroy and El-Geneidy 2017;Martinez et al. 2020;Seki and Yamada 2020).
Despite their reliance on public transport, women often report being afraid when using public transport and when moving in public spaces between public transport and their origin or destination. Studies from across the globe indicate that women universally face threats to their safety when travelling, with various acts of GBV named as key issues (Ceccato and Loukaitou-Sideris 2021;Ding, Loukaitou-Sideris and Agrawal 2020). GBV, defined as harmful acts that are directed at an individual based on their gender, is a broad concept that includes sexual, physical, mental and economic harm as well as threats of violence, coercion and manipulation both in public and private spaces (United Nations High Commissioner for Refugees [UNHCR] n.d.). During transport, GBV takes many forms, ranging from leering, stalking and indecent exposure to sexual assault, rape and murder (Ding, Loukaitou-Sideris and Agrawal 2020;González et al. 2015;Orozco-Fontalvo et al. 2019). Although GBV is present in public transport across the world, it is of particularly concerning prevalence and severity in South Asia and other countries of the Global South, where it is symptomatic of a broader societal subordination and devaluation of women (Bharucha and Khatri 2018;Islam MA and Amin 2016;Mitra-Sarkar and Partheeban 2011;Nahar, Van Reeuwijk and Reis 2013).

Study Setting: Bangladesh
As one of the world's most densely populated nations (163 million people at 1,116 people per square kilometre), Bangladesh has enjoyed a sustained period of economic growth, recording a reduction in population poverty from 43.8% in 1991 to 14.8% by 2016. In 2015, the country reached lower-middleincome status (The World Bank 2020). Rapid urbanisation and increased motorisation have followed on the heels of economic growth, and a lack of matching infrastructure investment and planning has resulted in severe pollution and traffic congestion in major cities across the country (Swapan et al. 2017;The World Bank 2020). Coupled with lax traffic law enforcement, high levels of unlicensed or unskilled drivers, and poor vehicle and road standards, the risk of traffic injury in both private and public transport is high by international standards (Barua and Tay 2010;Islam MR et al. 2020;Kamruzzaman, Haque and Washington 2014) and, on some indicators, is increasing (Biswas 2012;Hoque, Bin Alam and Habib 2003). Crash reporting is poor, with the World Health Organization (WHO) estimating that about 25,000 people are killed on the roads each year, 10 times the official figure reported by police (WHO 2018). For each fatality, there would be about 40 injuries, so about 1 million road users would experience traffic injury each year in Bangladesh.
Economic growth and expanding labour market opportunities have resulted in women increasingly seeking employment outside the home and entering professions previously dominated by men (Moon 2019;Rahman and Islam 2013). Despite these changes, Bangladesh remains, in many respects, a conservative and patriarchal society where cultural restrictions on women's autonomy and basic rights, early and forced marriages and domestic violence remain common (Banarjee 2020). The population is about 90% Muslim and 9% Hindu, and it is noted that patterns of GBV are similar to those in India and Pakistan, although Bangladesh has much higher rates than India or Pakistan on several indicators (https://evaw-global-database.unwomen.org/en/countries). Acid attacks and severe and fatal sexual assaults, some of which have occurred during transit on public transport, are extreme but not uncommon forms of violence directed at women in Bangladesh. In public places, women are often subjected to 'Eve teasing', a euphemism for sexual harassment that includes, but is not limited to, leering, stalking, catcalling and groping (Nahar, Van Reeuwijk and Reis 2013;Natarajan 2016;Talboys et al. 2017). Although sometimes trivialised as either an everyday nuisance or a harmless expression of men's romantic or sexual intentions (Ramasubramanian and Oliver 2003), Eve teasing is more aptly defined as 'an attitude, a mindset, a set of behaviours that is construed as an insult and an act of humiliation of the female sex' (Ghosh 2011: 100). As with other forms of violence towards women, Eve teasing operates to enforce and maintain male dominance by 'punishing' women for perceived lack of modesty (i.e., Eve as a temptress), increased participation in public life, or other gender norm transgressions (Naved et al. 2018;Zietz and Das 2018).

Theoretical Framework
Violence against women and girls remains a global phenomenon of significant concern. The safety of women across the world is a priority in the Sustainable Development Goals and never far from the news in the Global North and also the Global South. Specifically, violence, women, and transport in Bangladesh is a neglected area of research despite transport being a daily experience that women must endure if they are to access society. This research aimed to examine the use of transport to understand women's experiences of safety and security and what might be done to improve their access to safe transport.
While feminist theories were examined regarding their utility, intersectionality encompasses interacting dimensions of disadvantage, which means it enables contributions of different theoretical perspectives within the overall framework of intersectionality. Crenshaw, an African-American lawyer and civil rights advocate, first used the term intersectionality to examine and explain the oppression of Black American women, thus, connecting racism and identity politics (Crenshaw 1989). She argued that race, gender, class and other identity characteristics combined and overlapped, resulting in oppression. While rooted in feminism, the origins of intersectionality can be traced back to civil rights movements across the globe in the 1960s and 1970s (Collins 2015;Cooper 2016;Hancock 2016). However, most research on violence against women has been undertaken by feminist researchers from the Global North (Roberts and Connell 2016) to the neglect of violence against women in the Global South (Carrington 2015). Regardless, the problems that faced women at this time persist today in the Global North and the Global South, including GBV, rape and sexual assault, and patriarchal resistance to gender equality. Consequently, intersectionality has expanded over the past decades due to its utility in examining many intersecting categories of analysis, including ability/disability, socio-economic status, educational status and geographical location. Collins and Bilge (2020) also highlight the importance of using intersectionality as a reflexive method that can be used to unpack identity-based marginalisation. They reinforce that intersectionality is not just a theoretical lens, but they frame intersectionality as a form of critical praxis connected to social justice. Here, critique and intervention are purposefully paired to create positive change.
The utility of using intersectionality to explore and better understand the experiences of women in Bangladesh using transport is justified by the context. Bangladeshi society is patriarchal, where men are clearly privileged over women who are consequently oppressed, with consequences for how the different genders access and utilise the transport system. Notably, Lubitow, Abelson and Carpenter's (2020) discussion of their findings on gendered harassment and violence on public transport in Portland, Oregon, emphasised the intersectional nature of reported experiences; for them, race, sexuality, ability status, income and privilege all interacted with gender to shape their experiences. Intersectionality was, therefore, utilised in this study as an analytical tool that enabled both visible and invisible structures influencing the transport use of women to be considered. It allowed both power and oppression to be considered and to explain the experiences of women interviewed, with a view to identifying opportunities for intervention and change (Atewologun 2018).

Methodology
The aim of this research was to explore women's experiences of GBV while travelling to work or education on paid transport in Dhaka, Bangladesh, and the impact this experience had on access to education and employment. The research reported in this paper was drawn from data being collected as part of a larger project in Dhaka (Bangladesh), Hue (Vietnam), and Phnom Penh (Cambodia) that is still in progress and is comprised of two phases: a semi-structured questionnaire administered face-to-face and follow-up interviews with selected participants. The larger project has a main sample of women without disability and a supplementary sample of women with disability. The data used in this paper was restricted to the Bangladesh responses to the semi-structured questionnaire, not including women with disability; the analysis used both the free-text answers and the quantitative information, the latter of which was used to characterise the samples and complement the qualitative results.
To ensure that the questionnaire was sensitive to the contextual environments in Bangladesh, researchers in Australia worked closely with their counterparts in Bangladesh to capture the range of forms of paid transport available ('paid' transport was used to operationalise 'public' transport, given the wide range of modes available) and to ensure the translation of English words into local words did not change the meaning (this process was replicated in the other countries). Standard definitions such as violence, discrimination and assault were given in the questionnaire to ensure the same terminology was used across the three nations.
Before the commencement of data collection, co-investigator [name retracted] trained five female Bangladeshi postgraduate students in the delivery of the questionnaire, including gaining informed consent from participants, ensuring confidentiality and privacy and the safety of the participants. This research had ethical approval from the QUT Human Research Ethics Committee (Approval number: 1700000995).
Gender violence has long been recognised as one of the greatest challenges to peace and security in Pacific Island Countries (

Participants
Participants were 200 women who used transport either for work (n = 100) or for educational purposes (n = 100). They were mostly approached at bus stops in central Dhaka or at academic institutions, though the interviews were conducted elsewhere, such as cafes, parks, offices (for working women), academic institutions (for students) and, in some cases, at participants' houses. Note that the survey took place in 2019, prior to the emergence of COVID-19.
The sample was predominantly young adults, with 84.9% aged 18-34 years, 10.1% aged 35-44 years, 3.5% aged 45-54 years and 1.5% aged 55-64 years. On average, participants reported 3.94 adult members per household, ranging from 1-9. Annual household income (see Table 1) was reported by category (note that the number and percentage in each cell are identical as each sample had 100 participants). The Bangladesh Household and Income Expenditure Survey 2016 (Bangladesh Bureau of Statistics 2017) gives monthly household income figures for urban and rural areas. The average urban household income was converted from monthly to annual income and adjusted for the inflation rate to a 2019 equivalent (see https://www.worlddata.info/asia/bangladesh/inflation-rates.php), giving an average figure of 318,708 taka annually (about $3,771 USD at the time), around the middle of the second-lowest category in the survey. It can, therefore, be roughly estimated that around 82% of the education sample and around 90% of the work sample came from households with more than the average urban household income. However, to put this in perspective in terms of public versus private transport, the highest household income boundary (800,000 taka) equated to just under $9,500 USD annually, while the price of a new Toyota Yaris was about 3.35m taka, or about $39,600 USD (https://www.slideshare.net/mamunalo13/all-car-price-inbangladesh-2019). The women were asked a series of questions regarding the one or two most common trips they undertook for educational or work purposes. The average length of these trips was 7.24km (SD = 6.53), with an average price of 90 taka (around $1.10 USD). Around half of the sample (55.8%) described using one type of transport, while 42.2% reported using two modes of transport and 2.0% reported using three. Tempu, rickshaw and bus were the most frequent transport modes, used by 81.0%, 49.5% and 32.5%, respectively. Tempu and rickshaws are common and inexpensive modes of transport in Bangladesh. Rickshaws are pedal-driven with the cyclist at the front and typically a chair for one or two passengers behind the rider and open at the front; tempu (or tempo) are also known as autorickshaws, being a small, enclosed motorcycle with the rear wheel replaced with an attached small, enclosed cabin with two wheels, usually holding about six passengers. They are powered by diesel fuel, while the tricycle/CNG (noted later) is a similar vehicle powered by compressed natural gas (CNG). Most of the women (94.5%) travelled alone, and 98.5% reported that they walked as part of their journey.
Group differences between the work and education samples were explored. A Mann-Whitney U test found that the work sample was significantly older (mean rank = 135.74) than the education sample (mean rank = 63.90), U = 8524.0, z = 9.58, p < .001, and had a significantly higher household income (mean rank 113.86 compared to 87.14 for education), U = 6336.0, z = 3.38, p < .001. Independent sample t-tests found that the number of transport modes used during travel was significantly higher in the education sample (mean = 3.05, SD = 1.19) compared to the work sample (2.19, SD = 1.12), t(198) = 5.27, p < .001, but found no significant differences in travel length (p = .418) or in the price of the journey (p = .488) between the groups. Fisher's Exact Test found no significant associations between group membership and walking or not walking as part of the journey (p = .058) or travelling alone or with company (p = .621).

Data Preparation and Analysis
Analyses were weighted towards the qualitative component of the research. Descriptive statistics were calculated for the quantitative items and used to characterise the sample (as above) and to complement the qualitative data, and simple comparisons were made between the two samples to determine what overall differences there were, using parametric tests and non-parametric tests as appropriate. The qualitative data comprised the free-text responses to survey questions. All were translated by coinvestigator [name retracted], and a sample of the translations was back-translated by a bilingual researcher independent of the research team and compared with the original, with the comparisons showing very high correspondence. Thematic analysis was undertaken following the process outlined by Braun and Clarke (2006). Initial themes identified by co-investigator [name retracted] were checked and verified by chief investigator [name retracted] and reviewed by all other co-investigators. Quotes have been included to provide examples of the major themes.

Results
Analyses of both quantitative and qualitative data uncovered the nature and extent of the risks associated with participants' travel, the mechanisms underlying these risks, how they could be addressed, and the effects they had on participants' travel patterns, emotional wellbeing and quality of life.
Analysis of the open-ended questions found no substantive differences between the work and education samples, with participants in both groups describing similar realities and concerns around their travel. Statistical tests of group differences were calculated for the quantitative data, which, similarly, produced few significant results. The overall findings for the total sample are presented below. Each participant was given an identifier: W1-100 for the work sample and E1-100 for the education sample.

The Nature of Women's Experiences When Travelling
The initial quantitative analysis found that the women viewed the use of paid transport as an essential but highly unsafe activity. The entire sample (100%) indicated that access to transport affected their ability to study or work, while 3.0% reported that they 'always' felt safe while travelling, 11.5% 'very often', 53.5% 'sometimes', 21.5% 'rarely' and 10.5% 'never'. A Mann-Whitney U test found no significant differences in perceived safety between the education and work samples (U = 4308.5, z = -1.85, p = .064), indicating that these experiences were shared across all participants. The pervasive nature of participants' sense of fear was mirrored in quotes, such as the one below: In the context of our country, the everyday travel is not safe enough. Because crimes like road crashes, stealing, bag snatching, sexual harassment, physical assaults, stalking, and threats are common here. (E96) Participants commonly used words such as 'unsafe', 'worried', 'helpless', 'vulnerable', 'afraid' and 'scared' when describing their journeys. Some participants ascribed a persistent nature to their fears: 'I have a fear all the time' (W61) and 'due to collisions and men's abusive behaviour, I feel scared all the time' (W71).
Many also described a sense that the situation was deteriorating and that the Bangladeshi government was incapable of ensuring the security and safety of its citizens. This sentiment in itself contributed to the sense of dread and fear that was felt by many of the women: Broadly, women's concerns when travelling could be organised into three separate categories. When queried about what aspect of their travels and what particular part of their journey made them feel unsafe, GBV was cited as a concern by 49.0%, other forms of harassment and crime were mentioned by 54.0% of the sample, and the risk of traffic and non-traffic related injury was mentioned by 55.0% (45.5% specifically describing traffic injuries). An additional 11.0% stated that they were worried, without specifying the nature of their concerns (e.g., 'I am scared of the unsafe environment around us' E43).
When describing their fears of GBV, many women made reference to their own experiences while travelling. Although some did not elaborate on the nature of their abuse, others spoke of being subjected to groping, stalking, Eve teasing and other verbal abuse, for instance: 'victim of Eve teasing while walking, men's intentional physical touch' (W39), 'One man always stalks me that makes me uncomfortable' (W80), and 'I have to hear bad comments from the back. This will start even in front of my house' (E1). This abuse was encountered on the streets as well as from other male passengers on paid transport. The participants also spoke of the general risk of GBV that was associated with travel in Dhaka and how these instances were becoming more common: There is a group of Eve teasers who hang around on my way. Often, they make sexually abusive comments. In addition to that, the occurrence of rape, physical assaults have been increasing in recent time. Therefore, I feel this way. (E82) In addition to GBV, examples of other forms of harassment and crime were given, most commonly theft (bag snatching was mentioned by 48.0% of participants), followed by physical harassment and assault, which was mentioned by 18.0% of participants. Although most did not describe the nature of the physical harassment or assault they had encountered, those that did spoke of pushing and shoving from other pedestrians and passengers, particularly in crowded areas: 'it is difficult to walk in a crowded area. During that time, many people push intentionally' (W24). A further 10.0% of participants described a fear of crime, but without further specifying the criminal activity they were concerned about (e.g., by stating that they were fearful of walking on empty streets).
Traffic crashes were understood as the result of a combination of heavily congested roads, poor road conditions and driver behaviour. For instance, participants described how vehicles would drive on the wrong side of the road to avoid traffic jams or how, after spending hours in congested traffic, bus drivers would 'try to drive as fast as they can' (E31) to catch up on lost time. A general disregard for traffic rules and laws, including failure to stop at red lights, unlicensed and underaged driving (particularly among tempu drivers), a tendency among bus drivers to engage in 'competition with other bus drivers while driving a bus' (E1) and not always coming to a full stop at bus stops were also described and identified as contributing to crashes and other injuries. Damaged roads and the practice of leaving maintenance holes uncovered, particularly during the monsoon season, were identified as risk factors for traffic and nontraffic related injuries (e.g., tripping): There is an open manhole on my way. When my rickshaw needs to go that way, I feel anxiety.
In case of inattention of the rickshaw puller, there is a high risk of collision. (E70) Additionally, participants described how there was an absence of pedestrian crossings at busy roads and that designated pedestrian walkways (sidewalks or footpaths) were often occupied by street stalls, forcing pedestrians to walk on the road itself where they were exposed to vehicle traffic, creating additional safety risks for them.

The Underlying Mechanism of Crime and Gender-Based Violence
Quotes outlining the particular way in which different factors contributed to the risk of criminal activity, harassment and GBV were identified and analysed separately. While such quotes were relatively limited in the dataset, they nonetheless indicated that the area and type of transport, the time of the day and the number of people that were present were linked to the risk of incidents. Participants spoke of 'narrow lanes' as high-risk areas (narrow laneways are common in urban Bangladesh). These lanes were often empty, particularly after dark, and sometimes a gathering place for drug users. The fear of places where there were few or no other people present extended beyond these narrow lanes to include main roads (late at night) and buses: 'currently, there are many incidents that have been occurring in the empty bus' (W77); by 'empty', they would be referring to passengers, as buses often have conductors as well as drivers, and incidents involving conductors and drivers have been reported. Presumably, the presence of witnesses could act as a deterrent to criminal activity. Moreover, if attacks occurred, bystanders might offer assistance, provided they can be 'capable guardians', as noted by Valan (2020) in research on sexual harassment on public transport in India. One participant reported a perception along these lines: There are many passengers in the buses. Therefore, there is minimum risk of unacceptable incidents, like sexual assaults, bag snatching, physical assaults. In the case of an occurrence, it is possible to get help from others. (E88) As most participants (94.5%) travelled alone, they were particularly exposed to GBV and criminal activity when walking. Almost all participants (98.5%) walked as part of their journey, and among those that did, 4.1% reported that they 'always' felt safe when walking, 23.0% 'very often', 53.1% 'sometimes', 13.8% 'rarely' and 6.1% 'never'. No significant difference in perceived safety was detected between the education and work samples (U = 4717.0, z = -0.755, p = .451). However, while empty streets and transport vehicles posed a problem for the women, so could crowded areas. This duality is exemplified in the quote below: Stealing and GBV (mostly groping and Eve teasing or verbal abuse) on crowded buses were often mentioned: I feel helpless when the bus is overcrowded. I have this feeling due to stealing and experience of sexual harassment. (W69) Previous research has described the spatial and temporal dimensions associated with GBV in transit environments, linking groping to crowded buses and rape to empty bus stops, for instance (Ding, Loukaitou-Sideris and Agrawal 2020). While these dimensions were similarly noted here, a clear pattern linking certain types of crime to certain areas or modes of transport was not evident-likely due to the relative sparsity of quotes directly addressing these aspects.

Influences on Transport Choice
The transport mode chosen by participants on the day of the interview is displayed in Table 2, alongside the transport modes they avoided, those they felt were safest and those that they preferred. Buses were reported as available for their trips by 91.5% of participants, followed by rickshaws (75.5%), tempu (43.5%) and car services (20.5%), but the types chosen and preferred were quite different: rickshaws were most preferred (71.5%) and the second-most chosen on the day (50.5%), with this preference being consistent with other research on middle-income adults in Dhaka (Rahman 2019). Buses were not highly preferred (13.5%) but most often used on the day (67.5%), probably because of price and availability and the shorter distances and slower speeds of rickshaws. The relatively expensive car service was secondmost preferred (58.5%) but used by very few on the day (3%). Car service was considered safest (60%), just ahead of rickshaws (57%). Notably, the third-most used type was tempu (19%), but this type was preferred by no participants and regarded as safe by only one.
Despite the pervasive sense of fear associated with travel noted earlier, the women's choice of transport was, in open-ended questions, found to be mainly motivated by price (62.0%) and availability (36.0%), followed by other aspects relating to ease of travel such as convenience (27.0%), most direct route (21.0%) and quick connection (9.0%). This pattern is consistent with the patterns evident in Table 2. In the openended question responses, avoidance of certain transport modes was primarily motivated by safety, as mentioned by 81.5% of participants, followed by price (36.5%) and poor availability (17.5%). Factors relating to ease of travel (i.e., transport not being on time, being time-consuming or inconvenient) were only mentioned by 3.5% of the sample. Transport preference yielded different answers again, with comfort, as mentioned by 71.5% of participants, being the most commonly identified reason (e.g., 'I can travel with comfort' [E26]). Eight per cent of participants also spoke of preferring transport modes they felt were 'elegant', offered good service, were not suffocative, or enabled them to enjoy their surroundings as they travelled, presumably referring to car service and perhaps rickshaws. Just over half of the participants (57.0%) described safety as the reason for preferred transport mode. Availability was nominated by 28.5% of participants as a reason for preferring certain transport modes with aspects relating to ease of travel (time efficiency, direct route, convenience and mobility) being mentioned by 25.8%.  Overall, these results suggest the participants were captive travellers. In an identified unsafe travel environment, transport choice, a behaviour that is constrained by participants' circumstances, was primarily motivated by affordability and availability. Conversely, when the participants were asked about the transport mode that they preferred or that they would avoid, they were free to describe any mode of transport-when doing so, safety was featured as a key concern.

Precautions and Behavioural Adaptations
The notion that the participants were captive travellers was further supported by the answers they gave when asked what precautions they took to minimise the risks associated with their journeys during the night-time; only one woman indicated that she would modify her selection of transport: 'I would be extra careful to select my mode of transport and try not to be so late' (W27).
Unable to choose safer transport, the majority of the women instead described behavioural modifications, including modification to their travel patterns. Quantitative analysis showed that 3.0% of the sample were 'always' willing to travel after dark, 10.1% 'very often', 10.1% 'sometimes', 14.1% 'rarely', and 62.8% 'never'. Significant group differences were detected between the education (median = 5, interquartile range 1) and work samples (median = 5, interquartile range 2), U = 3856.5, z = -3.11, p = .002. In answers to the open-ended questions, the participants most commonly spoke of having someone accompany them if they travelled during the night-time (45.5%) ('I don't go outside without my family after dark' [E53]) or avoiding travel after dark (13.5%). Just over a fifth of the sample (21.0%) described arming themselves when travelling. Most commonly, they identified weapons and self-defence products such as knives, chilli powder and steel hair combs that were used, but ordinary everyday items were also described, including scissors, forks, safety pins, sharpened bobby pins, thick needles and perfume to be sprayed in the attacker's eyes. However, carrying arms offered no guarantees of safety: I keep a knife in my bag. Though it would be difficult to protect me with that knife but still I keep this for my safety. (W27) A further fifth of the sample (20.0%) indicated that they did not take any precautions during their nightime journeys. Most of these participants simply stated this fact: 'I do not take any precautions' (E8); however, a few indicated that they either could not afford to make any changes or that they were habituated to the circumstances they faced while travelling: 'no, I am habituated to travelling with fear' (W4).

Suggested Improvements
The type of incidents the women experienced and feared when travelling and the mechanisms through which they operated were reflected in the improvements the women suggested. Participants suggested strategies to decrease the number of passengers on buses, thus, reducing the opportunity for sexual harassment and theft. To this end, more buses, reduced fares on other paid transport to relieve pressure on buses and an enforceable limit on the number of passengers were suggested: Need to increase bus services. Every passenger should have a seat during the travels. That would reduce intentional physical touch, physical harassment. (W5) Over a third of the sample (35.0%) believed that comprehensive 'women-only' bus services were necessary (e.g., 'need to ensure the women bus service for all the routes and all the time' [E22]) to tackle the GBV women encountered on this transport. Chartered buses or other forms of transport for employees and policies giving female workers the right to leave before dark were suggested by 5.0% of the sample. To combat all forms of harassment and crime (including GBV), the participants most commonly described the need for better lighting (58.0%) and increased use of closed-circuit television (CCTV) (27.5%), as well as greater police and security guard presence (42.0%)-particularly at key areas and times such as in narrow lanes and on buses during peak time or night-time.
Twenty-eight per cent of the participants described a need for broader changes to policy, law enforcement and the judicial system. Tougher laws for street harassment and traffic violations were mentioned most often: 'I would also recommend for strong punishment for the offenders who are involved in various harassment on streets' (E4). This included punishment for Eve teasing, 'need to punish the Eve teasers' (E94), a behaviour that is punishable by law but rarely enforced in Bangladesh. Streamlined reporting of harassment and criminal activity was raised by three participants (1.5%), including the use of apps: 'there should be an app to ensure that a victim can call immediately during and after harassment' (W68). In contrast to the aforementioned policy changes, environmental modifications and legal sanctionsmeasures that are all reactive in nature-only 7.5% of participants described the need to tackle the underlying societal causes of crime and GBV, including the attitudes of men, poor economic conditions and unemployment.
Suggestions to reduce traffic crashes were given, including designated pedestrian walkways and crossings and improved road conditions ('it is important to repair the roads' [W15]), efforts to reduce the traffic jams and the instatement and enforcement of driver training and licence requirements: 'need to ensure that all the drivers are trained and adults' (E89) and 'need to ensure that drivers can drive a vehicle only with a legal drivers' licence' (E49).
Quantitative analysis showed that among participants who suggested improvements (96.7% of the sample), 92.3% stated that they would travel the same route they took on the day of the interview more often if these improvements were implemented (no significant association between education or employment group membership and increased travel was found: Fisher's Exact Test, p = .714). When asked why they would make these changes in an open-ended question, half the sample (50.0%) stated that their journeys would be less risky or that they would feel safer, more confident and less fearful during their travel: 'I would feel safe. There would not be any tension' (W13), 'I will be free from all fear' (W52). They also stated how this could affect a range of everyday activities: If these changes were implemented, I would travel confidently; there would be stability and peace in all activities, such as shopping, going to the market, walking, [and] travelling (W11). Some participants also indicated that greater safety would increase their willingness to travel or to simply go outside more ('I wouldn't have a problem going outside more often' [E92]), and others described how they would be able to change the way they travelled, including travelling alone or after dark ('I would feel safe and I would not need to bring my family members while I will travel after dark' [E1]). Without being prompted, five participants (2.5%) described how greater safety when travelling would affect their ability to work: As I do not feel safe while travelling, I try to avoid travel after dark and often keep my works pending for the next morning. If these changes were implemented, I would be able to go outside to complete work on time (W23).
Ten per cent of the women described, in a general sense, how increased safety would mean more freedom: 'there would not be a problem of going out as my wish' (E29), 'I wouldn't isolate myself' (W92), 'I would feel confident and wouldn't need to negotiate with my needs' (E54) and 'I don't go outside during my holidays due to hassles on the street. If these changes were implemented, I would go outside during my holidays'(W49).
The results presented here show the pervasive nature of GBV and other threats to personal safety that women encounter when using paid transport in Bangladesh, with an emphasis on the way in which it affects their emotional wellbeing and violates their right to freedom, dignity and quality of life.

Discussion
The present investigation examined women's experiences of safety when using paid transport in the city of Dhaka (Bangladesh) using intersectionality lenses. It was noted, when outlining the theoretical framework of this research, that the research conducted by Lubitow, Abelson and Carpenter (2020) found that race, sexuality, ability status, income and privilege interacted with gender to shape the experiences of GBV during travel by public transport in Portland, Oregon, which pointed to the value of an intersectional approach to our research in Bangladesh. It is of interest that the explanatory value of intersectionality is less clear in Dhaka, though there are several reasons why this might be the case. In terms of race, Bangladesh is relatively homogeneous, with 98% identifying as Bengali (https://www.worldatlas.com/articles/ethnic-groups-in-bangladesh.html). The primary religion is Islam, and there are regular news reports of religious discrimination and associated violence; however, none of the participants mentioned religion as an issue. Unfortunately, the survey did not ask for religious affiliation or for names (it is easy to recognise most names as either Muslim or Hindu). The frequent mention of Eve teasing has some religious overtones but is directed more at intra-religious, gender-related issues such as compliance with particular dress codes by women (Mazumder and Pokharel 2019). Once again, the religious nature of such harassment was not explicitly stated by participants.
Similarly, income was apparently not a factor, although it has been noted above that the participants came from relatively privileged households with a job or access to education and who can afford paid transport when commuting. In high-income countries, where private vehicle ownership is common, public transport is often seen as being necessary if people lack the income to afford a private vehicle. However, in low-and middle-income countries, public transport is often relatively inaccessible for low-income groups, while private car ownership (an 'elegant' form of transport) can be afforded by few. In addition, in Bangladesh, the proportion of women drivers is low, they are subject to harassment from male drivers, and in Dhaka it is difficult to park (https://www.thedailystar.net/shift/a-spike-in-women-drivers-in-bangladesh-1805248), all of which pushes higher-income women onto paid transport. As with income, there was also little difference between the experiences of women travelling for work or education, although the limitations that women placed on themselves (avoiding night-time travel, for example) may translate into the intersection of gender and occupation with a degree of self-restriction of travel: for instance, women who work deliberately avoid jobs where they have to travel at night or with too many mode changes. This would be worth exploring in future research.
The suggestion that intersectionality in terms of GBV and paid transport applies to avoidance behaviours more than to the experience of paid transport is consistent with the pattern of results for the avoidance of particular transport modes compared with the choice of mode on the day of the survey. The participants made strategic decisions to avoid certain transport modes, most often on the basis of safety, with safety itself having three dimensions: safety from GBV, safety from robbery and other forms of non-sexual harassment and crime and safety from traffic injury. By implication, the continued existence and use of these avoided modes demonstrate that men do not experience these threats to safety to the same extent (although, arguably, traffic injury risk would be the least differentiated by gender). It is plausible that there is an intersection between gender and safety from robbery, as women would usually be seen as easier targets if unaccompanied.
Once certain transport modes are ruled out for safety reasons, factors such as price and availability become important, and it would not be surprising to find the same pattern in males; that is, the intersection with gender is manifested in the modes avoided by women and not by men. At the same time, the participants still perceived a high level of risk associated with the modes they considered acceptable. This suggests that the participants are confronted by a lack of transport options that are as safe as they would like, that is, an ecological issue, which forces them to accept a level of risk-for three main kinds of risk-to travel to work or education. It is in this sense that they are captive travellers, forced by systemic, financial and social factors to expose themselves to GBV and other forms of risk.
The results made it clear that exposure to the risk of GBV while travelling is at least as great while walking to and from paid transport as it is on the transport itself. While the women are less 'captive' in these situations in the sense of not being confined to a vehicle, they are still limited to where they can feasibly walk and subject to factors outside their control, such as crowds. However, the participants also showed evidence of taking control of their safety to some extent, such as arming themselves for defence and managing their exposure to risk. Unfortunately, some of this taking control is also expressed by less travel, which can result in preventing women from meeting their emotional, social, health and economic needs. Mobility is essential to supporting quality of life because it fosters connections with the community and facilitates access to essential services and public spaces. This seems to be a pervasive problem in the Global South, with similar experiences of social exclusion also found in Asian countries such as Nepal (Neupane and Chesney-Lind 2014) and the Philippines (Ceccato and Loukaitou-Sideris 2021).
In general, the picture painted of women's travel by paid transport in Dhaka is one of restriction (because of the modes, places and times they must avoid) and fear of GBV, robbery and crime and traffic injury, contributing to considerably less amenity and access than men. In terms of mobility justice, women are clearly disadvantaged relative to men in terms of private vehicle use, paid public transport use and the range of paid transport modes available to them. The provision and operation of paid transport services are heavily biased towards meeting the mobility needs of men rather than women. Women avoid modes that men do not due to fear of GBV, bag snatching and traffic injury, which results in excluding themselves from work and education options. Importantly, it is plausible that this avoidance of transport modes can disadvantage women economically because they need to pay for transport with an acceptable level of safety and security (Infante-Vargas and Boyer 2021; Nguyen-Phuoc et al. 2021), while men do not necessarily change their travel behaviour due to this risk and may not incur the additional financial burden. Additionally, women could avoid safer transport modes such as buses and prioritise riskier transport options such as motorcycle taxis with a large likelihood of road crashes. These findings are consistent with previous research in South America (Orozco-Fontalvo et al. 2019).
It is clear that addressing this imbalance in mobility justice involves more than just providing more buses or even more women-only transport; the risks of GBV and robbery (and possibly traffic injuries) remain higher for women than for men regardless of the type of transport. If greater attention was focused on the characteristics of travel modes that women avoid, a more focused and effective set of strategies could be implemented. The participants themselves saw value in increasing surveillance of risky areas and addressing Eve teasing through legislation, and it is likely that these could be of benefit. An option not mentioned by participants is women police patrols such as those employed in India (Rabe-Hemp and Garcia 2020). However, such measures alone are unlikely to foster sustainable long-term changes in attitudes and behaviour because gender inequity is not restricted to mobility. The achievement of mobility justice for women in Bangladesh is interlinked with the establishment of gender equity across society and within relationships at all levels. This echoes the words of the Beijing Declaration of 1995 (UN 1995), which 'imagines a world where each woman and girl can exercise her freedoms and choices, and realize all her rights, such as to live free from violence, to go to school, to participate in decisions and to earn equal pay for equal work' (UN Women n.d.).
Correspondence: Dr Julie King, School of Public Health & Social Work, Faculty of Health, Queensland University of Technology, GPO Box 2434, Brisbane Queensland 4001, Australia. j.macknight-king@qut.edu.au