Risk en route: Women’s mobility and exposure to gender-based violence in public spaces in Curitiba,Brazil

Abstract

Chinese

Sexual violence against women in public spaces remains a persistent and troubling societal issue. The daily rhythms of people’s routine activities offer important insights into when, where, and who may be most exposed to gender-based violence. This study proposes an exploratory spatio-temporal approach to investigate women’s potential exposure to gender-based violence in public spaces in Curitiba, Brazil, adopting an intersectional perspective that considers gender, education level, income, and age. Using sample data from Curitiba’s Origin-Destination survey, we analyze the associations between the spatio-temporal patterns of the ambient population and those of gender-based crimes. Our results reveal that the correlation between population flows and gender-based crime incidents varies throughout the day, with stronger associations during daytime hours, especially morning and afternoon, when movements typically relate to commuting between home, work, and other destinations. Findings show that young women, those from low-income groups, women with secondary education, and those relying primarily on walking or public transportation tend to be more exposed to gender-based violence throughout the day, especially during peak hours. High-incidence crime areas are more closely linked to women’s destinations than to their places of residence and tend to align with public transport hubs. We conclude with a discussion on the implications of these findings for public policy and research agendas.

Keywords

gender mobility gender-based violence intersectional analysis public space space–time approach

Introduction

Gender, mobility, and violence are interconnected aspects that significantly impact women’s experiences in urban environments. A central factor in these experiences is the ability to navigate the city safely while carrying out routine activities. In the Global South, this mobility is constrained by fear of violence in public spaces, especially in crowded transport environments (buses, trains, and trams), where crimes such as harassment, assault, and rape disproportionately target women and gender-nonconforming individuals (Ceccato and Uittenbogaard, 2014; Loukaitou-Sideris and Ceccato, 2022).

Despite growing scholarly attention to women’s perception of fear in urban environments, gender-based violence in public space remains one of the most persistent and underexamined threats to women’s urban mobility (Balbontín and Arredondo, 2015; Ceccato and Paz, 2017). As Sagaris et al. (2024) argue, the prevailing emphasis on perceived safety rather than lived experiences not only obscures the structural nature of violence but also reinforces narratives that place the burden of self-protection on victims. Bridging this gap requires empirical analyses capable of identifying and quantifying when and where such violence occurs, who is most exposed to it, and how patterns of mobility intersect with these risks.

This study aims to contribute to bridging this gap that limits our understanding of women’s vulnerability to Gender-Based Violence in Public Spaces (GBVPS), compromising the design of effective, evidence-based policies for safer and more equitable cities. Adopting an exploratory analytical approach, we examine the spatio-temporal dynamics of women’s potential exposure to GBVPS, focusing on when and where risks are heightened, and which groups face greater vulnerability due to their social characteristics and mobility patterns. The research is guided by the following questions:

Q1: What are women’s mobility spatio-temporal patterns and how do they relate to security strategies against sexual crimes?

Q2: What is the relationship between GBVPS and the ambient population?

Q3: What are the characteristics of women at risk to GBVPS?

Q4: What are the spatio-temporal patterns of GBVPS?

The contributions of this study are twofold. First, it adopts a quantitative approach that integrates mobility data with records of gender-based crimes in public space, employing a spatially and temporally sensitive approach to explore their associations. We focus on women’s mobility patterns throughout different times of day and night, portraying their potential exposure to GBVPS across both time and space. This analysis builds on the routine activity theory in criminology, emphasizing how daily movement patterns shape vulnerability to victimization.

Second, the study brings much-needed attention to a Global South context, where women’s experiences are compounded by broader challenges such as limited connectivity, unaffordable transit options, and restricted access to employment and essential services (Balbontín and Arredondo, 2015; Loukaitou-Sideris, 2016). These challenges are further intensified by intersecting forms of marginalization that underscore the importance of applying an intersectional lens (Collins, 2022) when examining mobility and vulnerability in urban contexts.

To do so, we analyze mobility data from the Origin–Destination (OD) survey and combine them with GBVPS reports for Curitiba, the largest capital in southern Brazil. In addition to the availability of mobility data and the city’s renowned efficient transport system, Curitiba presents a core–periphery socio-spatial structure and the unequal nature typical of Latin American cities, making it an interesting case study. As such, it allows us to generate insights that resonate beyond the local context and contribute to broader Global South debates.

Background

Research across various geographical contexts identifies safety as a central barrier to women’s daily mobility (Loukaitou-Sideris, 2016; Sánchez de Madariaga, 2013; Vera-Gray and Kelly, 2020). Concerns about harassment and violence in public spaces routinely influence how women navigate the city, often resulting in restricted mobility and limited access to urban opportunities (Pain, 2001; Vera-Gray and Kelly, 2020; Zhang et al., 2022). The literature also highlights the disproportionate impact of fear on specific intersectional groups, such as low-income mothers and women with disabilities, often resulting in “immobility” (Whitley and Prince, 2005).

As an act of resistance, women incorporate “safety work,” a set of self-protection strategies embedded in everyday mobility decisions (Vera-Gray and Kelly, 2020). These include choosing certain types of clothing, traveling with an escort, or avoiding specific areas or public transportation altogether (Pain, 1991; Valentine, 1989).

The role of fear in shaping mobility and immobility within gendered power structures is undeniable. This is especially evident in highly patriarchal societies, where fear functions as a social mechanism that reinforces cultural norms designed to discourage girls and women from moving freely, limiting both the places they can access and the times they travel (Valentine, 1989). A range of studies has examined how specific environmental features contribute to women’s fear in public spaces (Kern, 2021), often drawing on frameworks from criminology (Fowler, 1987). While these approaches offer valuable insights into how the built environment influences feelings of safety, they often emphasize women’s heightened perception of risk compared to men (Johansson and Haandrikman, 2023).

Loukaitou-Sideris and Ceccato (2022) demonstrated that women’s perception of fear in public transport systems is higher in countries with higher crime rates. The pervasive fear of women related to violence in Latin American cities is linked to the fact that 60% of women have experienced harassment during their commutes (Boros, 2014). Yet, there are few studies linking crime rates to women’s fear (see Fox et al., 2009) or examining the associations between women’s spatio-temporal mobility patterns and crime rates (see Ceccato and Uittenbogaard, 2014). This focus on perceived rather than actual risk can obscure the structural and material conditions that expose women to real dangers in public space.

Such studies are limited by the nature of crime data, which lacks individual-level information about victims. The inability to directly link crimes to victims is a broader issue in criminological research (Browning et al., 2021), which presents further challenges in cases of gender-based crimes, a category that is underreported.

In criminology, crime rates are used to quantify crime as a ratio of the population. However, spatially disaggregating crime rates to investigate their geographic distribution requires determining which population is most pertinent to the specific crime under investigation (Haleem et al., 2021; Malleson and Andresen, 2016). Thus, the ambient population, comprising both residents and non-residents present in a given area at a specific time (Hipp et al., 2019; Malleson and Andresen, 2016), is often used in crime analysis as the relevant population denominator for certain types of crime.

According to routine activity theory (Cohen and Felson, 1979), certain types of crimes occur when a motivated offender, a suitable victim, and a lack of capable guardians converge in the same location and time. This theory recognizes the relevance of people’s routine behaviors to understand crime, suggesting that people’s mobility and crime spatio-temporal patterns are associated (Wilcox and Cullen, 2018).

In the context of GBVPS, the population at risk consists of females who are exposed to such crimes while engaging in routine activities. These include commuting to and from places of work, leisure, or study, as well as performing everyday tasks in public spaces. The population present in these spaces comprises not only victims (women, girls and gender-nonconforming persons) but also offenders (typically men) and potential guardians, defined as anyone who might prevent a crime by intervening or witnessing.

Understanding women’s mobility patterns is essential to identifying their exposure to risk. Low-income women in the Global South rely more on walking and public transportation as their main means of mobility, with greater dependence than men (Muhoza et al., 2021; Roos et al., 2022), a pattern also observed in Curitiba (Araujo et al., 2025). These women tend to spend extended periods in public spaces, increasing their exposure to potential offenders.

In Latin America, spatial and social inequalities further intensify women’s vulnerability (Falú, 2009). Factors such as income, race, and age intersect to shape not only how women move through the city but also how safe they feel. Gonzalez et al. (2020) demonstrate that fear of violence significantly restricts women’s mobility in Rio de Janeiro, Buenos Aires, and Lima, corroborated by evidence for Bogotá, México City, and Santiago (Balbontín and Arredondo, 2015). Harassment on public transport is widespread, and young, low-income, and working or student women are the most affected due to their daily reliance on transit and limited mobility alternatives (Balbontín and Arredondo, 2015).

In Brazil, low-income women living in peripheral urban areas often face long commutes, sometimes exceeding two hours daily (Pereira et al., 2020). Poorly located bus stops, overcrowded transport, and unsafe walking conditions significantly contribute to their sense of vulnerability. Fear is especially acute when waiting for public transport or walking home afterward (ITDP, 2018).

As Balbontín and Arredondo (2015) highlight, gender-based violence on public transport is widespread and socially tolerated in many Latin American cities, reflecting deeper structural patterns of inequality. Despite its prevalence, it is often overlooked in public agendas, and the lack of data hampers effective policy development to ensure safe and equal mobility for women.

Curitiba: A Global South metropolis

Curitiba has approximately 2 million inhabitants (IBGE, 2022), ranking highly amongst Brazilian capitals in terms of Human Development Index (PNUD, 2021). Despite being recognized as the world’s smartest city by the World Smart City Awards in 2023, Curitiba exhibits the typical inequalities of Latin American cities. Its macro-segregation pattern, like other Brazilian capitals, is characterized by a concentration of higher-income groups in the center, while lower-income groups mainly reside in the underserviced outskirts (Bittencourt and Faria, 2021).

This socio-spatial pattern is reinforced by the city’s public transportation system. Curitiba was a world pioneer in developing the Bus Rapid Transit (BRT) model, with dedicated lanes and an efficient boarding process to improve urban mobility. Yet it struggled to adapt to the city’s evolving demographic and geographic demands, with issues of accessibility and speed that disproportionately affect peripheral communities (Turbay et al., 2024). Curitiba’s BRT, like most transport systems, was designed around the typical male home-to-work commute, overlooking the diverse mobility needs of women and other users (Lindau et al., 2010).

Curitiba is located in one of the regions with the lowest rates of violence against women in Brazil (Moroskoski et al., 2022). However, it still records significant levels of GBVPS. In 2022 alone, Curitiba had 335 cases of sexual harassment in public environments (Aguiar, 2023) with almost 100 on public transport settings (Foggiato, 2023).

In response, the municipality introduced measures to reduce harassment on public transportation (Foggiato, 2023), including two enacted laws. State Law No. 19,582/2018 grants women, elderly individuals, and people with disabilities the right to request disembarkation from public transport at locations they consider safer, particularly between 22:00 and 5:00, in order to reduce their vulnerability during nighttime travel. Municipal Law No. 15,883/2021, framed as a campaign against sexual harassment on public transport, functions primarily as a self-reporting initiative. In 2014, a proposal to segregate women on public transport (in exclusive vehicles) was rejected in Curitiba. Critics argue that such measures fail to address the root issue, men’s inappropriate behavior, and risks implying that women who do not opt for women-only transportation are willing to tolerate harassment (ANTP, 2019).

Exploring spatio-temporal patterns of gender-based violence in public spaces

This study adopts an exploratory spatial and temporal analysis to examine women’s potential exposure to GBVPS. It focuses on identifying when and where risks are higher, as well as which social groups face greater vulnerability based on their mobility patterns and demographic characteristics. The following sections outline the data sources and the analytical approach employed in the study.

Data

In the absence of a comprehensive dataset on Curitiba’s ambient population, the sample in the OD survey was adopted as a proxy for this population. The OD survey was conducted between 2015 and 2017 (IPPUC, 2017), developed to capture the regular travel patterns of Curitiba’s population and assess its public transport system demand. Since individual mobility tends to follow consistent spatial and temporal rhythms, mobility patterns can be extrapolated (Ceccato and Uittenbogaard, 2014), making this dataset suitable to explore ambient population potential exposure to crime.

The dataset comprises travel diaries from a sample of individuals and includes individual-level information such as gender, age, income, trip purpose, departure and arrival times, modes of transport, and locations aggregated across Curitiba’s 137 OD zones. It represents the most complete and open-source mobility dataset currently available for the city, offering granular insights into individual travel behavior.

We selected all trips made by men and women aged 18 years or older within Curitiba, encompassing all modes of transport and travel purposes. Individuals within this subdataset were then categorized into five income brackets (A—highest to E—lowest) based on their reported gross income. Respondents with no information on income, age, or educational level were excluded from the analysis.

The final sample included 7069 individuals (51.64% women and 48.46% men) and 35,587 trips. The dataset overrepresented Groups A and D and underrepresented B when compared to ABEP’s (2017) population income distribution for Curitiba’s Metropolitan Region. Since the survey was not designed to ensure representativeness across different income and educational levels or age groups, nor did it incorporate key social markers such as race, the sample does not accurately reflect Curitiba’s overall demographic composition. Consequently, the results presented here pertain specifically to the surveyed sample and cannot be generalized to the broader population. The representativeness of relevant subgroups within the sample was considered when analyzing how crime affects particular subgroups of the sample in the results.

Figure 1 presents the distribution of residences and work destinations of income groups. Group A tends toward central residences and destinations, while other groups are more dispersed but still show notable concentration of destinations in the city center.

Figure 1.

Curitiba’s OD zones, BRT infrastructure, location of gender-based crimes (a) and distribution of income groups in residence areas and work/study destinations (b) Labels A–E indicate the various income groups analyzed, A being highest and E being lowest.

This study utilizes records from the Municipal Guard of Curitiba from 2009 to 2023 (PMC, 2023), representing a broader temporal scale than the mobility data from the OD survey. We selected 2638 reported GBVPS, including rape, attempted rape, violent indecent assault, obscene and lewd acts, sexual harassment, and offensive harassment, as offenses that predominantly target women (Uggen and Blackstone, 2004). Official crime statistics significantly underrepresented the actual occurrence of such offenses due to underreporting. Although the dataset includes the location and timing of these incidents, as well as the type of place they occurred (e.g. public streets, parks, bus terminals), it lacks information on victims’ characteristics. Figure 1 also displays the locations of reported GBVPS in Curitiba, revealing a higher concentration of cases in the city center and along the BRT infrastructure.

Methodology and analytical strategy

A series of exploratory methods was used to examine the associations between spatial and temporal patterns of GBVPS and the presence of an ambient population, addressing each research question.

Q1: On the variation of women’s mobility patterns and security strategies against gender-based crimes

A descriptive overview of the travel behavior of the sampled population was employed, including the distribution of transport modes by gender, the temporal distribution of trips by purpose across hours of the day, and the age profiles of female travelers in each income group. We conducted a descriptive analysis disaggregating the number of women’s trips by mode of transport and by hour, which were plotted alongside the frequency of gender-based crimes in public spaces by hour to explore security strategies against gender-based crimes, such as changes in modal choices.

Q2: Association between GBVPS and ambient population

Two complementary analytical strategies were employed to address Q2. First, a descriptive analysis was conducted combining travel patterns of women and men with the frequency of GBVPS across different times of the day. This was followed by the calculation of the ambient population (detailed below), and a statistical analysis using Spearman’s rank correlation coefficient (rho) to assess the strength and direction of the association between the number of crime incidents and the size of the ambient population. Spearman’s rho was chosen to account for the non-linear patterns, outliers, and overdispersion typical of crime data, as well as to test for a monotonic relationship between variables, following Ratcliffe (2002). Together, these approaches provide an overview and a statistically robust examination of whether higher levels of ambient population correspond to increased occurrences of gender-based crimes.

The ambient population was used as the denominator in the calculation of GBVPS rates, for which the sample in the OD survey was adopted as a proxy. The ambient population for each hour and OD zone was calculated based on individuals’ presence in a given zone (i) at a specific time (x).

p_{a m b} = O_{(i, x)} + D_{(i, x)} + S_{(i, x)}

(1)

where:

p_{a m b} = ambient population

\begin{matrix} O_{(i, x)} = individuals whose trip o r i g i n a t e s \\ in zone i at time x \end{matrix}

\begin{matrix} D_{(i, x)} = individuals whose trip a r r i v e s \\ in zone i at time x \end{matrix}

\begin{matrix} S_{(i, x)} = individuals who arrived earlier in zone \\ i and whose stay includes time x \end{matrix}

Individuals located at their place of residence were excluded from the ambient population counts unless they were departing or arriving at home, which we considered as a presence in public space. To aggregate by hourly time bins, individuals were assigned to the relevant hour intervals based on their trip timings. For instance, a person leaving home at 8:30 was counted in the origin zone during the 8:00–8:59 time bin; if they arrived at their destination at 9:10, they were counted there during the 9:00–9:59 bin. If they stayed at the destination through subsequent hours, they were included in those corresponding hourly bins until they left.

Two versions of the GBVPS rate were calculated for each OD zone and hour. The first used the total ambient population as the denominator, which encompasses potential victims, offenders, and guardians (equation (2)). The second considered only women in the ambient population, a proxy for potential victimization risk (equation (3)).

R_{(i, x)} = \frac{C_{(i, x)}}{P_{a m b (i, x)}}

(2)

where:

R_{(i, x)} = GBVPS rate in zone i at time x

\begin{matrix} P_{a m b (i, x)} = total ambient population at \\ zone i at time x \end{matrix}

\begin{matrix} C_{(i, x)} = number of GBVPS in zone i \\ at time x \end{matrix}

and

R_{(i, x)} = \frac{C_{(i, x)}}{P_{t f (i, x)}}

(3)

where:

R_{(i, x)} = GBVPS rate in zone i at time x

\begin{matrix} P_{t f (i, x)} = women in the ambient population at \\ zone i at time x \end{matrix}

\begin{matrix} C_{(i, x)} = number of GBVPS in zone i \\ at time x \end{matrix}

Q3: Characteristics of women at risk to GBVPS

Spearman’s rho was used to investigate associations between GBVPS rates and the sociodemographic and mobility characteristics of women in the ambient population in different spatial and temporal contexts. Variables such as income, educational level and age were examined alongside mobility patterns and transport mode. The proportion of women in the ambient population with these sociodemographic characteristics and mobility patterns was considered as a proxy for potential victimization risk, reflecting how individual profiles and daily movement patterns may influence vulnerability.

Q4: Spatio-temporal patterns of GBVPS

To capture spatial and contextual variations, GBVPS data were disaggregated by both time of occurrence and type of public space, including green areas, bus terminals, public streets, and glass-tube stops, which are a distinctive feature of Curitiba’s public transport system.

To identify statistically significant spatial clusters of GBVPS, Getis-Ord Gi* Statistics (Getis and Ord, 1992) was applied. This spatial technique detects local concentrations of high or low values by comparing the crime rate of a feature and its surrounding features to the overall distribution within the study area. A feature is only flagged as a significant hotspot if both the location itself and its neighbors exhibit similarly high values. The resulting Z-scores and p-values indicate the intensity and statistical significance of clustering: positive Z-scores signal hotspots, while negative ones denote cold spots. The Getis-Ord GI* statistics are given by:

G i^{*} = \frac{\sum_{j = 1}^{n} w_{i, j} w_{j} - \bar{X} \sum_{j = 1}^{n} w_{i, j}}{S \sqrt{\frac{[n \sum_{j = 1}^{n} w_{i, j}^{2} - {(\sum_{j = 1}^{n} w_{i, j})}^{2}]}{n - 1}}}

(4)

where:

\bar{X} = \frac{\sum_{j = 1}^{n} w_{i, j}}{n}

and

S = \sqrt{\frac{\sum_{j = 1}^{n} x_{j}^{2}}{n} -} \bar{(X})^{2}

Notations:

x_j is the attribute value for feature j,

w_i,j is the spatial weight between feature i and j, and

n is equal to the total number of features.

Results

Women’s mobility patterns and security strategies against gender-based crimes (Q1)

The results in Figure 2 show that, although the car is the most frequently used mode of transport overall, the proportion of women’s travel trips as passengers is nearly twice that of men’s. Bus use and walking are also more common among women, while men are more frequent users of bicycles and motorcycles, as reported by other studies.

Figure 2.

Income groups analysis (modal choice, trip motives and age distribution). Labels A–E indicate the various income groups analyzed, A being highest and E being lowest.

Women’s mobility patterns within the sample vary significantly by income and age group. Higher-income women (Group A) tend to be older, travel mostly by car, and commute for work, aligning with typical work hours (7:00–8:00 and 17:00-18:00). In contrast, lower-income women (Group E) are younger, walk more, and travel mainly for education, especially in the morning and early afternoon. Only 3% of women in Group E travel for work, compared to about 60% in other groups. Groups B–D show typical work patterns. Group B presents a similar pattern of Group A, with women traveling mostly by car. In groups C and D, bus usage and walking trips dominate among women.

Figure 3 underscores a critical temporal alignment between women’s mobility and the incidence of GBVPS. Figure 3(a) shows that over 70% of reported incidents occur between 6:00 and 18:00, aligning with peak hours of women’s mobility as captured in the OD survey. With the exception of rape, of which 46.3% of cases occurred during daylight, all the other types of GBVPS predominantly took place in daytime hours (e.g. sexual harassment 73.4% and lewd acts 69.4%). Rape also represents the smallest share of total incidents, accounting for only 2.05% of all GBVPS cases.

Figure 3.

(a) Frequency of men’s and women’s travel and frequency of gender-based crimes per hour and(b) women’s transport modal choice per hour and per income group.

The most prominent peaks, at 7:00, 12:00, and especially 17:00, correspond to heightened frequencies of crimes, suggesting greater exposure during periods of intense urban movement. It also shows a reduction in nighttime trips, which aligns with typical travel behavior across the general population. Women’s trips decrease more significantly than men’s after 10 pm, which may reflect self-protection strategies driven by the fear of crime and victimization (Vera-Gray and Kelly, 2020), known to impose self-restrictions on women’s nocturnal mobility (Zhang et al., 2022).

Relationship between GBVPS and the population in transit (Q2)

Table 1 presents Spearman’s correlation coefficients for different time periods aggregated by OD zones, examining the relationship between the number of reported GBVPS and the ambient population. The correlation is strongest and statistically significant during the morning hours (6:00–12:00), indicating that higher concentrations of people in transit are associated with increased GBVPS, especially during peak commuting hours (6:00–8:00). This suggests that GBVPS tend to occur in areas and hours of high mobility, reinforcing the idea that routine activities and ambient population play a critical role in shaping women’s exposure to risk. In contrast, during theevening (19:00–23:00) and early morning (24:00–5:00), this relationship weakens considerably. This result aligns with Ceccato and Paz’s (2017) study for São Paulo.

Table 1.

Correlation between ambient population and gender-based crimes.

Time period(hours)	Spearman’srho	p-Value
24:00–5:00	0.081	0.014
6:00–8:00	0.452	<0.001
9:00–10:00	0.369	<0.001
11:00–13:00	0.406	<0.001
14:00–16:00	0.320	<0.001
16:00–18:00	0.299	<0.001
19:00–23:00	0.230	<0.001

Figure 3(b) shows disaggregated travel patterns of women in the sample by income group, revealing differences in mobility behaviors and transport modes. After 18:00, when the statistical correlation weakens, there is a notable decline in the overall number of trips, a natural consequence of decreased activities at nighttime and lower availability of transport services. For the lowest-income group (Group E), this shift is accompanied by a marked decrease in the proportion of walking trips by women in comparison to those by men. This is compensated for by an increase in the proportion of trips made by bus (7.6%) and as car passengers (18.6%) in comparison to the 16:00–18:00 period. This group of women relies most heavily on walking and public transport during daytime hours and such a change in model choice might suggest they are seeking safer travel alternatives during nighttime hours despite the additional cost, a behavior consistent with women’s strategies to safely navigate urban environments (Pain, 1991; Valentine, 1989; Vera-Gray and Kelly, 2020).

Characteristics of women at risk (Q3)

Figure 4 presents the statistically significant Spearman correlations (p < 0.05) between hourly gender-based crime rates with only women in ambient population as denominator (equation (3)) and the characteristics of these women from the sample, disaggregated by transport mode, age group, education level, and income group. These correlations serve as indicators of which mobility profiles may be more exposed to GBVPS.

Figure 4.

Spearman correlations between gender-based crime rates and the proportion of women in the ambient population, by (a) transport mode, (b) age group, (c) educational level, and (d) income group.

Moderate positive correlations are predominant, as shown Figure 4(a). The strongest positive correlations are observed for women traveling by bus and walking through the day. Between 19:00 and 21:00, the highest correlations are found among women traveling by bus, suggesting that women with that mobility profile may be more exposed to victimization. Conversely, car and motorcycle users show weaker and less consistent correlations. These corroborate other studies’ findings that women who rely on public or active modes of transport often perceive a greater risk of gender-based violence, largely due to increased time spent in public spaces and waiting areas (ITDP, 2018; Loukaitou-Sideris, 2014).

Figure 4(b) shows that woman in the 18–30 year old age bracket exhibit the strongest and most consistent positive correlations with gender-based crime rates across nearly all hours of the day. In evening hours (19:00 to 21:00) the correlation is stronger. Women aged 30–40 show moderate correlations, which weaken notably after 40, especially among those 60+, where few remain significant. This is consistent with literature indicating that young adult women are disproportionately targeted in public spaces (Fileborn and Hardley, 2023).

Figure 4(c) and (d) show that women with lower levels of education, particularly those with only elementary or high school schooling, alongside those in lower income groups (d) and (e), exhibit stronger correlations with gender-based crime rates, especially during early morning hours (5:00–6:00) and in the evening (18:00–21:00). These time periods are consistent with structural constraints experienced by low-income women, who tend to leave peripheral residential areas early and endure extended workdays, or even multiple shifts, to sustain household livelihoods (Tacoli, 2012). Their reliance on walking and public transportation further increases their exposure during these hours.

Women with no formal education exhibit fewer significant correlations with gender-based crime rates. Similarly, women from higher-income groups (a and b) and with higher education levels also display weaker and often non-significant correlations. The latter can be explained by their greater reliance on private transportation, which reduces their exposure to public space and, consequently, to potential risks.

Spatio-temporal patterns of GBVPS (Q4)

Figure 5(a) illustrates the hourly distribution of reported GBVPS across different types of public spaces based on the location data from crime dataset. Public streets are by far the most frequent settings for these incidents, with intensity peaking between 12:00 and 18:00, aligning with periods of high urban activity. Green areas and bus terminals also show increased occurrences during daylight hours, though to a lesser extent. Glass-tube bus stops display a pattern of crimes concentrated during commuting peaks (6:00–8:00 and 17:00–19:00). This temporal distribution points to heightened vulnerability in everyday public spaces used for transit and circulation.

Figure 5.

(a) Temporal and spatial distribution of gender-based crimes by type of public space and time of day and (b) hotspots and coldspots of GBVPS in Curitiba.

Figure 5(b) presents the spatial distribution of gender-based crimes using Getis-Ord Gi* to identify statistically significant hotspots and coldspots. Hotspots (in red) are concentrated in central and well-connected areas of the city—zones characterized by a high concentration of jobs, commerce, and services. These are not areas where most low-income women or those with lower education reside (shown in Figure 1), but rather destinations including work, study, and essential services. In contrast, central areas are often residential zones for higher-income women, who, in turn, experience less exposure to gender-based crimes due to greater access to private transport.

Discussion

This study generated five main findings, which shed light to key aspects GBVPS of spatio-temporal patterns in Curitiba, and bring up important reflection on the roles of witnesses and on the potential exposure of specific women’s profiles to these types of crime.

Contrary to common assumptions, gender-based crimes in public spaces and transport settings are predominantly concentrated during daytime hours. One frequent concern among women is the “last mile” (the walk home from the bus stop), which has been partially addressed through policies introducing on-demand bus stops—a valuable measure that acknowledges women’s safety concerns and aligns with a whole-journey approach to improve travel experiences. Our findings suggest that, as a standalone intervention, this policy is insufficient to enhance women’s safety in public transport in Curitiba. A more comprehensive set of strategies is required to address the multiple layers of vulnerability that women face throughout the urban mobility network.

Our results show that gender-based crimes tend to happen at very busy times and places in daylight. It is known that crowded transport settings can be conducive to violence, and that women report feeling safer when traveling during off-peak hours (Gonzalez et al., 2020). The high incidence of such crimes in broad daylight and in areas with a high number of potential guardians corroborates the idea that the presence of others does not necessarily contribute to women’s safety. Cook and Reynald (2016) suggest that men are more likely to intervene when they perceive clear physical aggression, while women tend to be more empathetic toward victims but hesitate to intervene when alone. This is relevant given that the most common offense in our dataset—obscene or lewd acts, which often do not involve physical contact—is frequently under-recognized as a crime in patriarchal societies. Hence, such acts might not be perceived as serious enough to prompt intervention from potential guardians. Research also shows that individual responsibility to report or intervene tends to diminish in group settings (Cook and Reynald, 2016) and that individuals on the move are less likely to act as guardians (Ceccato and Uittenbogaard, 2014).

Hence, the times and places where women supposedly should feel safer are also the ones where they are most at risk, pointing to the need for a wider reflection on what the Brazilian society considers as acceptable behavior. We argue that there is a need for awareness campaigns on what constitutes a gender-based crime as well as on guidance to the population on how to react when witnessing GBVPS, including what is generally considered as minor forms of violence against women. More attention needs to be given to the role of society as guardians and preventers.

Our analysis shows that intersectionality significantly influences women’s exposure to gender-based crimes in public spaces. Young women, and those from lower-income groups or with lower levels of education, are more exposed to GBVPS in Curitiba. This appeared as a significant association in our results despite the underrepresentation of low-income groups in the OD sample. Although socio-economic vulnerability is known to increase exposure to GBVPS, our analysis adds nuance to this understanding. Contrary to the assumption that women from lower-income backgrounds are primarily at risk near their homes, which are often located in high-crime and deprived areas, our findings indicate that the highest risks occur during peak transport hours and in central urban areas that are not this group’s areas of residence. This pattern holds true even for crimes occurring after dark. The areas of greatest risk are consistently central zones and transport hubs, suggesting that these women are more at risk where they travel for work or study than where they live. Their reliance on public transportation and their need and determination to pursue employment or education appear to be key factors contributing to their higher exposure. This trend is particularly pronounced among women in early adulthood (ages 18–30), a group known to be more susceptible to unwanted male attention.

Finally, our results demonstrate that women who are potentially more exposed to gender-based violence in public spaces tend to employ strategies of resistance and self-protection. They suggest that women often avoid using public transport at night, either by reducing travel during those hours or, when possible, choosing private cars, a well-documented self-protection strategy (Pain, 1991; Valentine, 1989; Vera-Gray and Kelly, 2020). This pattern was especially prominent among low-income women despite their limited access to private vehicles and economic constraints.

Conclusions

The spatio-temporal analysis of GBVPS in Curitiba reveals nuances of the phenomenon, offering insights that challenge common assumptions often used as the foundation for policy design and intervention strategies. The findings that gender-based crime is more prevalent during daylight hours and in busy central urban areas contrasts with much of the focus of existing literature on the geography of women’s fear and of public safety policies, which typically emphasize nighttime risks. Our results highlight the need for further research on the spatial and temporal dynamics of mobility and GBVPS in order to better understand when and where protective interventions are most required.

Another key takeaway from our study is the urgent need to deepen our understanding and response to a pervasive form of crime that appears deeply ingrained in Brazilian society. It is crucial to reflect on and further investigate the extent to which crime witnesses, who could act as guardians for victims, may inadvertently become enablers by silently acquiescing to patriarchal norms. Beyond strengthening policing and implementing more robust safety measures to protect women, there must be greater public awareness about what constitutes a gender-based crime, accompanied by clear, actionable guidelines on how to respond when witnessing such incidents. Traditionally, society has placed the burden of preventing GBVPS on women themselves, expecting them to remain vigilant, protect themselves, and officially report offenses, while managing their daily routines and responsibilities. The literature rightly criticizes these patriarchal approaches for failing to adequately address the male perpetrators or the broader societal structures that enable them. Without integrating multiple layers of evidence, combining quantitative and qualitative insights to fully understand the where, when, and who are the victims of these crimes, our capacity to effectively confront and promote policies in order to reduce gender-based violence will remain limited.

In the midst of a global “data revolution” and the explosion of data availability that has fueled advances in artificial intelligence, our study remains exploratory due to persistent gender data gaps. These gaps leave us in a restricted position, lacking sufficient data to conduct more robust and comprehensive analyses. Nevertheless, our results reveal important associations which, when considered alongside existing literature, allows us to challenge long-standing narratives based on common assumptions and to highlight critical areas for further research and policy intervention. While this study might not capture the entire picture, it emphasizes that the spatio-temporal dynamics of gender-based crime represent a vital story that must be told using solid evidence, for which more and better disaggregated data on gender and other social markers are needed.

Footnotes

ORCID iDs

Agnes Silva de Araujo

Geisa Tamara Bugs

Joana Barros

Jaqueline Massucheto

Phâmela Alves

Flávia da Fonseca Feitosa

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

ABEP (2017) Critério Brasil - Critério de classificação econômica. Available at: https://abep.org/criterio-brasil/ (accessed 24 July 2025).

Aguiar

(2023) Curitiba registra mais de 300 casos de importunação sexual no ano de 2022. Available at: https://www.portalcomunicare.com.br/curitiba-registra-mais-de-300-casos-de-importunacao-sexual-no-ano-de-2022/ (accessed 8 October 2024).

ANTP (2019) O que as experiências com o vagão rosa nos ensinaram até agora. Associação Nacional de Transportes Públicos) Revista dos Transportes Públicos. Available at: http://files.antp.org.br/2019/4/10/rtp151-3.pdf (accessed 23 July 2025).

Araujo

ASD

Massucheto

Bugs

, et al. (2025) Unveiling inequalities: The intersection of gender and income in accessibility in Curitiba, Brazil. Journal of Transport Geography 123: 104136.

Balbontín

Arredondo

(2015) Violencia de género en el transporte público: Una regulación pendiente. Naciones Unidas.

Bittencourt

Faria

JRVD

(2021) Distribuição de investimentos públicos, infraestrutura urbana e desigualdade socioespacial em Curitiba. urbe. Revista Brasileira de Gestão Urbana 13: e20190300.

Boros

(2014) POLL: Latin American cities have most dangerous transport for women, NYC Best. Reuters, 29 October. Available at: https://www.reuters.com/article/markets/commodities/poll-latin-american-cities-have-most-dangerous-transport-for-women-nyc-best-idUSL6N0S32MQ/ (accessed 25 July 2025).

Browning

Pinchak

Calder

(2021) Human mobility and crime: Theoretical approaches and novel data collection strategies. Annual Review of Criminology 4(1): 99–123.

Ceccato

Paz

(2017) Crime in São Paulo’s metro system: Sexual crimes against women. Crime Prevention and Community Safety 19(3-4): 211–226.

10.

Ceccato

Uittenbogaard

(2014) Space–time dynamics of crime in transport nodes. Annals of the Association of American Geographers 104(1): 131–150.

11.

Cohen

Felson

(1979) Social change and crime rate trends: A routine activity approach. American Sociological Review 44(4): 588–608.

12.

Collins

(2022) Bem Mais Que Ideias: A Interseccionalidade Como Teoria Social Crítica. Boitempo.

13.

Cook

Reynald

(2016) Guardianship against sexual offenses: Exploring the role of gender in intervention. International Criminal Justice Review 26(2): 98–114.

14.

Falú

(2009) Mujeres En La Ciudad: De Violencias y Derechos. Ediciones SUR.

15.

Fileborn

Hardley

(2023) ‘I’d be in my school uniform’: The informal curriculum of street harassment. Gender and Education 35(4): 330–347.

16.

Foggiato

(2023) Sugestões à Prefeitura: Proposta de combate à importunação sexual em Curitiba. Available at: https://www.curitiba.pr.leg.br/informacao/noticias/sugestoes-a-prefeitura-proposto-o-combate-a-importunacao-sexual-em-curitiba-1 (accessed 8 October 2024).

17.

Fowler

(1987) Street Management and city design. Social Forces 66(2): 365–389.

18.

Fox

Nobles

Piquero

(2009) Gender, crime victimization and fear of crime. Security Journal 22(1): 24–39.

19.

Getis

Ord

(1992) The analysis of spatial association by use of distance statistics. Geographical Analysis 24(3): 189–206.

20.

Gonzalez

Machado

Bianchi Alves

, et al. (2020) Why Does She Move?: A Study of Women's Mobility in Latin American Cities. World Bank. Available at: http://hdl.handle.net/10986/33466

21.

Haleem

Do Lee

Ellison

, et al. (2021) The ‘exposed’ population, violent crime in public space and the night-time economy in Manchester, UK. European Journal on Criminal Policy and Research 27(3): 335–352.

22.

Hipp

Bates

Lichman

, et al. (2019) Using social media to measure temporal ambient population: Does it help explain local crime rates? Justice Quarterly 36(4): 718–748.

23.

IBGE (2022) Censo demográfico 2022. Instituto Brasileiro de Geografia e Estatística.

24.

IPPUC (2017) Consolidação de dados de oferta, demanda, sistema viário e zoneamento – Relatório 5 pesquisa origem-destino domiciliar. Available at: https://ippuc.org.br/visualizar.php?doc=https://admsite2013.ippuc.org.br/arquivos/documentos/D536/D536_002_BR.pdf (accessed 11 September 2024).

25.

ITDP (2018) O acesso de mulheres e crianças à cidade. Available at: http://itdpbrasil.org.br/wp-content/uploads/2018/01/ITDP-Brasil-_-O-Acesso-de-Mulheres-e-Criancas-a-Cidade-_-ABR-2018.pdf (accessed 14 February 2022).

26.

Johansson

Haandrikman

(2023) Gendered fear of crime in the urban context: A comparative multilevel study of women’s and men’s fear of crime. Journal of Urban Affairs 45(7): 1238–1264.

27.

Kern

(2021) Cidade Feminista: A Luta Pelo Espaço Em Um Mundo Desenhado Por Homens. Oficina Raquel.

28.

Lindau

Hidalgo

Facchini

(2010) Bus Rapid Transit in Curitiba, Brazil. Transportation Research Record: Journal of the Transportation Research Board 2193(1): 17–27.

29.

Loukaitou-Sideris

(2014) Fear and safety in transit environments from the women’s perspective. Security Journal 27: 242–256.

30.

Loukaitou-Sideris

(2016) A gendered view of mobility and transport: Next steps and future directions. Town Planning Review 87(5): 547–565.

31.

Loukaitou-Sideris

Ceccato

(2022) Sexual harassment on transit: A global, comparative examination. Security Journal 35(1): 175–204.

32.

Malleson

Andresen

(2016) Exploring the impact of ambient population measures on London crime hotspots. Journal of Crime and Justice 46: 52–63.

33.

Moroskoski

Brito

FAM

Oliveira

(2022) Time trend and spatial distribution of the cases of lethal violence against women in Brazil. Revista Latino-Americana de Enfermagem 30: e3609.

34.

Muhoza

Wikman

Diaz Chavez

(2021) Mainstreaming gender in urban public transport lessons from Nairobi, Kampala and Dar es Salaam. Stockholm Environment Institute.

35.

Pain

(1991) Space, sexual violence and social control: Integrating geographical and feminist analyses of women’s fear of crime. Progress in Human Geography 15(4): 415–431.

36.

Pain

(2001) Gender, race, age and fear in the city. Urban Studies 38(5–6): 899–913.

37.

Pereira

RHM

Braga

CKV

Serra

, et al. (2020) Desigualdades socioespaciais de acesso a oportunidades nas cidades brasileiras - 2019. Instituto de Pesquisa Econômica Aplicada (IPEA.

38.

PMC (2023) SiGes guarda: Consulta de bases. Available at: https://www.curitiba.pr.gov.br/dadosabertos/busca/?termo=guarda (accessed 1 October 2024).

39.

PNUD (2021) Atlas do Desenvolvimento Humano no Brasil. Programa das Nações Unidas para o Desenvolvimento. Available at: https://atlasbrasil.org.br/ (accessed 24 July 2025).

40.

Ratcliffe

(2002) Aoristic signatures and the spatio-temporal analysis of high volume crime patterns. Journal of Quantitative Criminology 18(1): 23–43.

41.

Roos

Sprei

Holmberg

(2022) Traits and transports: The effects of personality on the choice of urban transport modes. Applied Sciences 12(3): 1467.

42.

Sagaris

Baker

Woodcock

(2024) Gender, transport, and health: Emerging trends and gaps in global research. Advances in Transport Policy and Planning 13: 379–434.

43.

Sánchez de Madariaga

(2013) Mobility of care: Introducing new concepts in urban transport. In: Sánchez de Madariaga

Roberts

(eds) Fair Shared Cities: The Impact of Gender Planning in Europe, 1st edn. Routledge, pp. 33–48.

44.

Tacoli

(2012) Urbanization, Gender and Urban Poverty: Paid Work and Unpaid Carework in the City. Human Settlements Group, International Institute for Environment and Development.

45.

Turbay

ALB

Pereira

RHM

Firmino

(2024) The equity implications of TOD in Curitiba. Case Studies on Transport Policy 16: 101211.

46.

Uggen

Blackstone

(2004) Sexual harassment as a gendered expression of power. American Sociological Review 69(1): 64–92.

47.

Valentine

(1989) The geography of women’s fear. The Royal Geographical Society 21(4): 385–390.

48.

Vera-Gray

Kelly

(2020) Contested gendered space: Public sexual harassment and women’s safety work. In: Ceccato

Nalla

(eds) Crime and Fear in Public Places: Towards Safe, Inclusive and Sustainable Cities, 1st edn. Routledge, pp.217–231.

49.

Whitley

Prince

(2005) Fear of crime, mobility and mental health in inner-city London, UK. Social Science & Medicine 61(8): 1678–1688.

50.

Wilcox

Cullen

(2018) Situational opportunity theories of crime. Annual Review of Criminology 1(1): 123–148.

51.

Zhang

Zhao

Tong

(2022) Constructing women’s immobility: Fear of violence and women’s constricted nocturnal travel behaviour. Travel Behaviour and Society 26: 178–192.