Safety, Gender, and the Public Transport System in Santiago, Chile

Abstract

This research evaluated gender differences in the perception of safety in public transport in Santiago, Chile using quantitative and qualitative approaches. With data from the National Urban Citizen Security Survey 2019 (ENUSC), a gender comparison was made regarding the perception of safety in four scenarios: inside buses, inside the metro, at bus stops, and waiting for buses at night. Four ordinal logistic regression models were estimated to analyze how sociodemographic factors and variables associated with the perception of crime influence rider perceptions of safety in public transport. To complement the results, four focus groups were developed to obtain a deep understanding of the participants’ experiences with safety in the Santiago public transport system. We concluded that there is a high perception of insecurity in public transport for both men and women. In general, perceived insecurity inside buses, inside the metro, and waiting for public transport at night is greater among women, older people, and national citizens. Other influencing variables are the perception of insecurity regarding crime in general, the fear of being a victim of a crime, or negative situations that occur in the neighborhood, such as the presence of robberies, alcohol, and drug consumption. We proposed new variables such as fear of harassment, traffic accidents, discrimination, contagious diseases, and street protests among others. To carry out a precise public policy on this matter, a permanent scan on security issues in public transport should be developed, considering a complete set of variables. This result can be applied in Chile and all Latin American countries.

1. Introduction

An important goal of government policy is to ensure that public services effectively respond to the needs of all members of the community, including residents with disabilities, low-income residents, and members of disadvantaged minority social groups [1]. In 2015, the Member States of the United Nations approved and signed the 2030 Agenda for Sustainable Development to balance environmental, economic, and social sustainability, establishing the Sustainable Development Goals. One of the goals is to provide access to safe, affordable, accessible, and sustainable transport systems for all and improve road safety by expanding public transport and paying special attention to the needs of people in vulnerable situations, such as women, children, people with disabilities, and the elderly.

In this context, transport planners have made efforts to promote more inclusive public transport, proposing improvements in the accessibility and quality of service [2], in the dedicated infrastructure [3], as well as devices supporting traffic in the area of public transport such as P and R systems or bike sharing systems, which improve traffic in public transport points [4,5,6]. However, insecurity limits or conditions the transport options of riders [7,8,9,10,11]. Safety and security are one of the most significant concerns influencing public transport satisfaction [12], but their interpretations can vary between research disciplines. In broad terms, they refer to all those risks that can damage personal properties or individuals’ physical and mental health [13]. The feeling of insecurity and fear of crime are considered disadvantages of public transport [14,15,16,17] and are among the most important reasons why travelers decide not to use public transportation [18,19,20].

Different studies have shown that fear of crime and perceptions of safety are correlated with real levels of crime, surroundings, and characteristics of transport facilities [21]. Likewise, fear of crime has the potential to influence the behavior of travelers at each stage of the trip, from planning before the trip, through the journey, to the afterward evaluation [22,23,24,25].

In developing countries, the perception of security and fear of crime in public spaces have the potential to affect the behavior of travelers to the point of influencing the decision of avoiding traveling [26,27,28]. It has been reported that this behavior tends to be prevalent in women [13,29,30,31,32], and considering that a large percentage of women are captives of public transport [7], the fear of crime deserves the attention of researchers and policymakers.

Jaimurzina et al. [33] concluded that, in Latin America, it is necessary to adequately incorporate the gender perspective in the region’s policies, plans, and projects because transport plans are not engaged to adequately satisfy the needs of women; on the contrary, they are mainly oriented towards promoting the use of private vehicles, which favors the mobility of men.

In response, the objective of this article was to analyze the perception of insecurity in public transport in Santiago de Chile and to determine if gender has a moderating effect on the relationship between perceived safety in trains, buses, and bus stops, together with identifying what factors affect the perception of insecurity and what policies could be effective in closing existing gender gaps. For the above, we formulated and estimated different ordered logistic regression models, using data from the national urban citizen security survey (ENUSC), prepared by the Undersecretary for Crime Prevention of the Ministry of the Interior with technical support from the National Institute of Statistics (INE) of Chile. This quantitative data were complemented with a qualitative methodology, allowing the identification of new aspects of security that should be considered in future studies.

The remainder of this study is divided into four sections. Section 2 is a bibliographic review of the perception of safety in public transport and its main factors. Section 3 presents the methodological aspects of the ENUSC survey, the focus groups carried out, and the ordinal logistic regression models. We also provide a brief description of the public transport system of Santiago, Chile. Section 4 sets out the results of our estimated models. Finally, Section 5 presents our conclusions.

2. Bibliographic Review

Safety is not completely defined, which is why providing safety in public spaces is so difficult. It is a concept that can be subject to the interpretation of each individual and be understood as personal, public, human, individual, or even national [34]. A series of studies have examined the factors that make people more or less likely to feel insecure [12,24], among whom Delbosc and Currie [35] proposed four broad categories: psychological, situational, socioeconomic and neighborhood, as well as demographic.

The perception of insecurity and fear of crime on public transport also depends on the socio-demographic characteristics and individual experience [21,35,36], and the most used variables are age, sex, educational level, employment status, level of trust in justice institutions, levels of interpersonal trust, and knowledge of public security policies, among others [37]. Nevertheless, these variables are generally influenced by factors present in public spaces that are not controlled by travelers and are not always related to previous experiences of victimization.

Carr and Spring [38] identified several factors that contribute to the feeling of fear, highlighting the condition of vehicles, stations, and whereabouts; exposure to often scandalous and offensive groups; and the role of the media in exposing fear-generating incidents that occur both on the transportation system and in the community at large [39]. These factors create a self-reinforcing cycle that causes a reduction in the number of travelers and a reduction in real levels of safety.

Regarding demographic factors, studies show that the origin of insecurity is different according to the characteristics of groups; thus, security measures in the transport system that are effective for one demographic group may not be as effective for another [40]. Approaching this factor from a gender perspective, women have a greater tendency to be afraid, especially of sexual aggression; thus, they commonly use their strategies to evaluate and avoid possible threats [28,29,41]. Sexual minorities such as homosexual male adolescents should also be considered because they can be the target of aggression due to their dressing style [42].

How men and women commute is different when considering the purposes of the trip, time spent, means of transport, and the hours in which they travel [31,43]. Women travel more for purposes related to shopping, escorting children, and caring for the elderly, tending to travel in shorter periods and using mostly public transport or traveling by foot. They avoid rush hours, they travel usually when there is still sunlight, and they avoid large crowds. Trips are also more complex, solving in one journey several domestic necessities that require stopping at different places. In contrast, men travel more for work purposes, thus making more trips during rush hours and mainly by car [16,25,31,44].

Harassment in public spaces is mostly directed toward women, and the aggressor is usually a man [29]. Therefore, harassment is one of the most recurrent situations that influence the perception of safety in women while using public transport, forcing them to change their travel behavior [7,22,45]. Although lack of safety is common at night or in lonely places, several situations where harassment or abuse of women occur are in broad daylight and crowded spaces, such as public transport during peak hours [29,30,37].

Situations of harassment on public transport are also common in the presence of crowds [46], which is a relevant factor in the perceived lack of safety. Other situations that evoke fear in women are long waiting periods at bus stops or long walks when being alone [47]. These actions are perceived by men as part of their travel time, while for women it is more of an exposure time that they are forced to face. The regularity with which women experience street sexual harassment shows that their experience in the public space is plagued by a lack of autonomy. That is why these areas can be an enhancer of gender inequality, and since it influences public spaces, it can indirectly affect public transport [7,31,48].

The framework of security, crime, and harassment has led some cities to study and implement various measures to improve security in public transport, such as segregating spaces and services by gender [49,50,51]; public social marketing campaigns [52]; technological innovations such as mobile phone applications and computer applications to register drivers and passengers and to route vehicles [51].

In the case of Chile, in 2018, the Ministry of Transport and Telecommunications launched the first Gender Equity Policy in Transport, aiming to ensure that the gender approach is transversal to the implementation of policies that promote equity in aspects such as mobility, accessibility, safety, and efficiency in the transport system. The main characteristic of this new agenda is transversality, seeking to establish a sustainable gender approach and integrating this component as an essential part of the entire process of public transport policies, from the identification of problems to the design, implementation, and evaluation.

Allen, Cárdenas, Pereyra, and Sagaris [29] analyzed insecurity as one of the disadvantages that women face in public transport, preventing them to travel regularly. The insecurity that women perceive in public spaces can also be understood as a fear of situations that put their integrity and health, both physical and psychological, at risk. The research shows that 89% of women and 74% of males are harassed in public transport, and the aggressor is a man 83% of the time. Regarding the type of harassment experienced, 51% was verbal, 19% was physical, 6% was photographic, and 24% was considered severe harassment. The most relevant conclusions point to the issue of increasing the transit cost for low-income women who prefer to pay for taxi/Uber instead of being involved in risky situations; the consequences of the lack of safety for women in terms of their exclusion from public spaces; and the long-lasting effects of sexual harassment experiences because they are passed down through generations, as women teach their daughters how to deal with it or avoid public transport. This leads them to take actions that affect the way they travel, using strategies such as traveling accompanied, avoiding transportation through certain places, avoiding walking, avoiding trips that are too long, requesting to be picked up at a stop or station, and even avoiding travel [23,30].

Busco et al. [53] studied the relationship between transportation and social exclusion by observing the disadvantages faced by riders on the public bus transit system in Santiago, Chile. Through factor analysis, they distinguished disadvantages faced by riders of public transit buses in the Santiago metropolitan region, determining eight dimensions that together explained 58% of the data variability. Although poor service quality and accessibility are the issues commonly highlighted in evaluations of the public transit bus service in Chile, this study showed that security is the dimension users evaluated as the worst and presents a greater disadvantage. Harassment, although an element that is usually evaluated within security, conforms to a different dimension when analyzing the transportation system in Santiago de Chile. This dimension is particularly relevant for women, reducing their access to public bus travel in several conditions: when a bus is overcrowded, during the night, or crossing through specific urban zones defined as dangerous. More than 50% of riders were unable to use public transit buses at least once, mainly due to safety concerns among women, young people, and the elderly. This results in longer trips (women chose longer but safer routes), time disadvantages, more expensive means of travel (experiencing economic disadvantage), or the decision to stay home (avoiding the dangers of harassment), which results in social exclusion.

Finally, public space is not a safe place for girls and women, standing as an example of a lack of democracy, which is suffered by more than half of the population [54]. It is then important to consider the concept of social exclusion in transportation, which is defined as a transportation disadvantage or lack of basic mobility [14,16]. According to the Inter-American Development Bank, for the transport sector, this translates into the promotion of actions and projects that include a gender perspective in the different stages of development and having an adequate female representation in all steps in the value chain of the sector.

We concluded in this bibliographic review that the lack of security in the public transit system is experienced as a way of transit disadvantage, and consequently, it becomes an element that triggers social exclusion, specifically when gender is considered, resulting in our interest in deepening into this problem. Consequently, the research questions proposed in this study are: Are there any gender differences when analyzing insecurity perceptions in public transport in Santiago de Chile? and What are the factors that explain those differences?

3. Methodology

The data used for this research were obtained from the National Urban Survey of Citizen Security (ENUSC) 2019, which aimed to obtain information at the national and regional level on the insecurity and victimization of homes and people, as well as their reactions to criminal situations. The survey is carried out by the Undersecretary for Crime Prevention of the Ministry of the Interior and Public Security with technical support from INE.

ENUSC is a fieldwork face-to-face survey with a probabilistic, three-stage, and geographically stratified sample design by district and classification of the blocks according to the number of residences they contain. The first stage consists of the selection, with equal probability of communes or blocks, to then choose houses in the same way within the previous section. Finally, the Kish informant selection methodology assigns to each eligible person in the household the same probability of selection, thus preserving the probabilistic nature of the sample design applied in the previous selection stages. This approach is relatively easy to implement in the field, since the probability assignment mechanism does not require control in the distribution of predefined tables among the interviewers, and it is easy to supervise because it incorporates the ordered list of household members, always starting with the head of the family, followed by the rest of the members from oldest to youngest [55]. In this way, people over 15 years of age can be chosen.

The questionnaire structure consists of two parts, (i) registration of the household informant (RPH) and (ii) the central questionnaire. The first part (18 variables) seeks to register household socio-demographic characteristics, while the second part is made up of three units and is aimed exclusively at the respondent: (i) perception of insecurity and the reaction of people to crime; (ii) registers households’ victims of crimes and attempted crimes, as well as the characteristics of such events; and (iii) the evaluation of institutions related to the public security system. It also includes neighborhood context variables and the tenement occupied by the household surveyed.

The sample at the national level consists of 24,465 households (76,567 persons); however, in this research, we limited the analysis to the metropolitan region of Santiago, with a sample of 5785 households (19,088 persons). The sample size was designed to obtain estimates of the parameter of interest with theoretical absolute errors of a maximum 2.3%, together with theoretical relative errors of a maximum 8.5% for districts with more than 200,000 inhabitants. We focused on the questions about the perception of safety in public transport, where the respondents answered according to a Likert scale of 4 values (1 very insecure, 2 insecure, 3 secure, and 4 very secure) and their perception of safety in the following situations: inside buses, inside the metro, at bus stops, and waiting for buses at night.

To complement the results, a qualitative phase was developed that applied four focus groups held in May 2019 at Diego Portales University, aiming to obtain a deep understanding of the experiences, perceptions, opinions, feelings, and knowledge of the participants about safety in the Santiago public transport system. The sample consisted of a group of seven university students; six working women, mainly foreigners, who work as housekeepers in a high socioeconomic sector; five female university students; and five senior citizens. The conversations lasted between 70 and 90 min and consisted of a set of eight questions that addressed issues related to the disadvantages of public transportation, making it clear to the participants that, in the process, it was not necessary to generate agreements or a consensus; thus, everyone’s opinion, no matter how different, was welcome. Results were presented as a complement to the quantitative analysis.

The Santiago public transport system is managed by the Metropolitan Mobility Network (RED). RED connects all the public transport buses in the city (operated by six concessionaire companies) and the Santiago Metro physically and through the fare. The bus system covers around 6.2 million users in the city, in a geographical area of around 680 km² in urban areas. On a business day, around three million transactions are made on the system’s buses (the coverage of bus and metro services and population districts is presented in Figure 1).

According to the latest Origin and Destination Travel Survey 2012 [56], almost 18 million trips are made in Santiago during a typical business day, of which 61.5% correspond to motorized trips. A total of 26.1% of trips are made by car, and 25.9% are trips made by public transport (bus and metro). A total of 53.5% of public transport trips are made by women, while 56.3% of car trips are made by men. In the morning rush hour (07:30–08:30), 15.1% of total daily trips are made, and 58.3% of these trips are made by men.

Data analysis consisted of two parts. First, a gender comparison was made regarding the perception of safety in four scenarios: inside buses, inside the metro, at bus stops, and waiting for buses at night. Later, we estimated four ordinal logistic regression models to analyze how sociodemographic factors and variables associated with the perception of crime influence riders’ perceptions of safety in public transport and whether the effects of some of these factors differ for male and female passengers.

Ordinal logistic regression models correspond to an extension of the general linear model to ordinal categorical data, i.e., it allows the construction of models to predict a (dependent) response variable, in this case, categorical or ordinal, from one or several independent predictor variables (which can be continuous or categorical). The logistic regression model can be expressed as follows [57]:

$$\ln \left(Y_j^{\prime }\right)=\ln \left(\frac{{\pi }_j\left(x\right)}{1-{\pi }_j\left(x\right)}\right)={\alpha }_j+\left({\beta }_1{X}_1+{\beta }_2{X}_2+\dots +{\beta }_p{X}_p\right),$$

(1)

where ${\pi }_j\left(x\right)=\pi (Y\le j\left|x_1,x_2,\dots ,x_p)\right.$, which is the probability of being above or below category j given a set of predictors, $j=1,2,\dots ,J-1$; ${\alpha }_j$, are cut-off points between categories, and ${\beta }_1,{\beta }_2,\dots ,{\beta }_p$ are the logit coefficients for the respective independent variables.

4. Results

A summary of the statistical values of the main sociodemographic variables of the households interviewed in the metropolitan region of Santiago is presented in

Table 1. Descriptive summary of the sample.

Variable		%
Gender	Men	47.3%
	Women	52.7%
Age	15–20	25.6%
	21–30	15.9%
	30–40	14.6%
	41–50	12.0%
	51–60	12.7%
	>60	19.2%
Nationality	Chilean	89.3%
	Other	10.7%
Educational level	Incomplete school education	6.9%
	Basic school education	25.5%
	Complete school education	48.2%
	Higher education	17.2%
	Postgrads studies	2.2%
Activity	Study	10.7%
	Employed	58.4%
	Housework	9.6%
	Retired	11.9%
	Unemployed	5.4%
	Other	3.9%

.

Regarding the general results of perception of security, in 2019, 82.0% of people stated that they perceived an increase in crime in the country in the last twelve months, while 39.6% of people stated that they perceived an increase in crime in the neighborhood. A total of 12.1% of those surveyed declared having been the victim of a crime in the last twelve months, and 37.0% of people stated that they believed that they would be the victim of some crime within the next twelve months.

The main source of information regarding the opinion of the level of crime is the media with 53.3%, where the greatest influence on the construction of this perception is provided by newscasts and television programs in general. Other sources of information correspond to personal experience, information from other people, and family experiences with 17.7%, 15.3%, and 13.7%, respectively.

Figure 2 shows the distribution of the perception of safety on buses, on the metro, at bus stops, and waiting for buses at night. Using an aggregate comparison, we observed a greater degree of insecurity on buses (73.54%) followed by metro (50.32%). At bus stops, 50.51% of the sample feel insecure, while 10.65% perceive an extreme level of insecurity. When waiting for public transport at night was analyzed, there was a considerable increase in the perception of insecurity (82%). In addition, among those who use the bus system, 62.9% declared avoiding stops at night (within this group, 64.6% correspond to female users).

Figure 3 shows the percentage of respondents who feel insecure or very insecure in the four scenarios considered, differentiating by gender, and in all cases, the number of women who feel insecure is higher. A total of 81% of the women surveyed have a perception of insecurity waiting for public transport at night, while the perceived insecurity inside buses is 79% for women. In all scenarios, the gender difference was significant with 95% confidence.

Among those who use the bus system, 36% of the people surveyed stated that they had avoided taking empty buses because they felt that their safety was at risk. Within this group, 73.5% correspond to female users.

Delving into gender differences in the perception of safety in public transport, Figure 4 shows the female insecurity index, calculated as the number of women who feel insecure for every 100 men in the same situation. The biggest differences are inside the metro and buses; 188 women feel insecure inside the metro for every 100 insecure men, while 171 women feel unsafe on buses for every 100 insecure men.

Focus group participants pointed out a transversal concern to be a victim of crimes such as theft, assault, or other acts of violence influenced by drugs or alcohol. However, these are not the only fears that are associated with crime; the fear of being a victim of sexual harassment deserves a particular mention, mainly experienced by women in cases where physical contact is unavoidable due to large crowds within public transport.

In addition to the criminal aspect, the interviewees stated that they also perceive insecurity in the face of possible traffic accidents, which can be expected because of a usual irresponsible driving, which includes speeding, sudden maneuvers, or even a driver with obvious signs of fatigue. These imprudent behaviors are threatening because of possible collision, run overs, falls and accidents within the bus, and unforeseen arrests of the driver (and the resulting waste of time). These insecurities are especially relevant considering that this means of transport is also used by the elderly and disabled.

Table 2. Ordinal logistic regression models.

Variables	How Safe Do You Feel inside Buses?		How Safe Do You Feel inside the Metro?		How Safe Do You Feel at Bus Stops/Stations?		How Safe Do You Feel Waiting for Buses at Night?
	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR
Gender (female = 1)	−0.722 ***	0.485 ***	−0.796 ***	0.451 ***	−0.365 ***	0.694 ***	−0.982 ***	0.374 ***
	(0.062)	(0.0301)	(0.0618)	(0.0279)	(0.0582)	(0.0404)	(0.0655)	(0.0245)
Nationality (Chilean = 1)	−0.431 ***	0.650 ***	−0.497 ***	0.608 ***	−0.402 ***	0.669 ***	−0.381 ***	0.683 ***
	(0.108)	(0.070)	(0.108)	(0.0657)	(0.103)	(0.0688)	(0.108)	(0.0739)
Age (>30 years old)	−0.515 ***	0.597 ***	−0.213 *	0.808 *	−0.218 *	0.804 *
	(0.115)	(0.069)	(0.117)	(0.0944)	(0.113)	(0.0910)
Activity (employed = 1)	0.318 ***	1.374 ***	0.195 **	1.216 **	0.468 ***	1.596 ***	0.192 **	1.211 **
	(0.090)	(0.123)	(0.0951)	(0.116)	(0.0843)	(0.134)	(0.0955)	(0.116)
cut1	−3.284 ***	0.0375 ***	−3.699 ***	0.0247 ***	−2.828 ***	0.0592 ***	−2.159 ***	0.115 ***
	(0.161)	(0.00604)	(0.166)	(0.00412)	(0.154)	(0.00913)	(0.161)	(0.0186)
cut2	−0.294 *	0.745 *	−1.123 ***	0.325 ***	−0.132	0.876	0.980 ***	2.665 ***
	(0.151)	(0.113)	(0.155)	(0.0504)	(0.147)	(0.129)	(0.157)	(0.419)
cut3	3.382 ***	29.44 ***	2.109 ***	8.243 ***	3.617 ***	37.22 ***	4.230 ***	68.71 ***
	(0.206)	(6.077)	(0.165)	(1.363)	(0.192)	(7.134)	(0.229)	(15.71)
Observations	4393		4139		4568		4367
Log likelihood	−4165.7231		−4228.7339		−4550.1034		−3993.385
LR Chi2	179.10		429.49		79.15		247.61
R2	0.0210		0.0225		0.0086		0.0301

Note: OR: Odd ratios. Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the estimation of four ordinal logistic regression models, one for each scenario, where the response variable corresponds to the perception of security in the scenario (according to the Likert scale: 1 very insecure, 2 insecure, 3 secure, and 4 very secure), and the predictor variables correspond to sociodemographic dummy variables associated with gender (1 if female), nationality (1 if Chilean), age (1 if the respondent is over 30 years old), and activity status (1 if the respondent is employed).

Sensitivity tests showed that estimations, for all estimated models, were stable across specifications and robust to the inclusion of covariates. In all cases, the LR Chi2 test was significantly different from 0 (p < 0.001), i.e., the model with the sociodemographic variables provided a better fit than the null model without independent variables to predict the ordinal response variable. Even when the R2 value was low, most of the predictors were statistically significant; thus, conclusions can be drawn about the association between changes in predictor values and changes in response values [58].

Results confirmed a significant gender gap in the perception of safety. In all the estimated models, the female gender predictor was negative and significant at the 1% level; therefore, in all the scenarios, the probability that women feel more insecure compared to men is greater. It is important to note that the parameter for the metro safety model was higher than for the bus, as was the femininity index, which could be explained by the higher levels of overcrowding in the metro. Regarding safety within the metro, OR = 0.451 for the gender predictor indicated that the odds of being above a particular category of security perception decreased by a factor of 0.451 when the interviewee was a woman, when all the other predictors remained constant.

Concerning nationality, in the four estimated models, the predictor was negative and significant at the 1% level, indicating that Chilean users have a greater perception of insecurity compared to users of other nationalities. Foreign focus group participants explained these results when they compared crime insecurity in public transport in Santiago with their countries and cities of origin. Before that difference, they generally felt safer in Santiago. Nevertheless, they raised a different element of insecurity based on possible acts of discrimination and xenophobic attacks, which are usually expressed verbally but can become acts of violence.

In the case of the age variable, the predictor was also negative and significant at 10% for three of the four models. In the case of the safety perception model waiting for public transport at night, the parameter was not significant. Older adults focus group participants mentioned vulnerability to crime because of their reduced mobility, a condition that also leaves them prone to falls in the face of unforeseen situations regarding driving. Fear of accidents is not only associated with older adults but also with people with disabilities who often encounter difficulties such as access to preferential seats, the entry of wheelchairs, or the lack of cooperation on the part of drivers and passengers.

For the activity variable, it was only significant if the respondent was employed. In all models, OR was greater than 1, indicating that the odds of feeling secure or very secure increased when the interviewee was employed.

Going deeper into the analysis, Figure 5 shows the differences in the marginal effect between women and men. The marginal effect is calculated for each possible response on the Likert scale and represents the variation in the probability of choosing a given response for an average respondent considering that all other variables take the average value. It is possible to observe that women are more likely to respond that they feel very insecure (5%) or insecure (15%) on the metro compared to men. Inside the bus and waiting for public transport at night, this figure is 15%, and for waiting at a bus stop, the probability reaches 9%. These results are in agreement with Ouali, Graham, Barron, and Trompet [48], where a negative and significant gender gap was observed for safety statements as well as general satisfaction for metros and buses.

Focus groups helped to understand gender differences in this scenario. Although traveling at night is perceived as insecure by everyone, men are willing to use the public transport at night. After a party, they express that between sleeping back home and sleeping uncomfortably at a friend’s house, they are willing to take the risk of traveling at night for the increase in comfort. Women, on the other hand, have other choices; they sleep at a friend’s house or do not party at all, because late-night travel is out of the question. These conducts are taught from a young age and seem reasonable for them, and even though they are young adults that can make their own decisions, their parents usually prevent them from traveling alone at night.

Finally,

Table 3. Ordinal logistic regression models with perception of crime variables.

Variables	How Safe Do You Feel inside Buses?		How Safe Do You Feel inside the Metro?		How Safe Do You Feel at Bus Stops/Stations?		How Safe Do You Feel Waiting for Buses at Night?
	Coefficient	OR	Coefficient	OR	Coefficient	OR	Coefficient	OR
Gender (female = 1)	−0.438 ***	0.645 ***	−0.734 ***	0.480 ***	−0.0257	0.975	−0.682 ***	0.506 ***
	(0.0666)	(0.0430)	(0.0630)	(0.0302)	(0.0642)	(0.0626)	(0.0698)	(0.0353)
Nationality (Chilean = 1)	−0.287 **	0.750 **	−0.418 ***	0.658 ***	−0.291 ***	0.747 ***	−0.205 *	0.815 *
	(0.113)	(0.0848)	(0.109)	(0.0719)	(0.111)	(0.0827)	(0.113)	(0.0923)
Age (>30 years old)	−0.212 *	0.809 *					0.564 ***	1.758 ***
	(0.121)	(0.0982)					(0.128)	(0.224)
Worker (worker = 1)	0.193 **	1.213 **			0.285 ***	1.330 ***
	(0.0953)	(0.116)			(0.0917)	(0.122)
How much does crime currently affect your quality of life? (Ref: A lot)
Quite a bit	0.261 ***	1.298 ***	0.141 *	1.152 *	0.215 ***	1.240 ***	0.216 ***	1.241 ***
	(0.0792)	(0.103)	(0.0769)	(0.0886)	(0.0764)	(0.0947)	(0.0814)	(0.101)
Little	0.596 ***	1.815 ***	0.269 ***	1.309 ***	0.423 ***	1.527 ***	0.657 ***	1.929 ***
	(0.0852)	(0.155)	(0.0818)	(0.107)	(0.0823)	(0.126)	(0.0890)	(0.172)
Nothing	1.041 ***	2.831 ***	0.396 ***	1.485 ***	0.834 ***	2.302 ***	0.969 ***	2.635 ***
	(0.121)	(0.341)	(0.121)	(0.179)	(0.119)	(0.274)	(0.126)	(0.333)
Do you think you will be the victim of a crime in the next twelve months? (Yes = 1)	−0.351 ***	0.704 ***	−0.293 ***	0.746 ***	−0.370 ***	0.691 ***	−0.445 ***	0.641 ***
	(0.0672)	(0.0473)	(0.0640)	(0.0478)	(0.0652)	(0.0450)	(0.0698)	(0.0447)
Thinking about crime, would you say that during the last twelve months crime in your neighborhood: (Ref: Increased)
Remained the same	0.388 ***	1.475 ***	0.511 ***	1.667 ***	0.275 ***	1.316 ***	0.161 **	1.174 **
	(0.0721)	(0.106)	(0.0708)	(0.118)	(0.0700)	(0.0921)	(0.0749)	(0.0880)
Decreased	0.608 ***	1.837 ***	0.462 **	1.587 **	0.507 ***	1.660 ***	0.542 ***	1.719 ***
	(0.184)	(0.338)	(0.191)	(0.304)	(0.188)	(0.312)	(0.192)	(0.329)
During the last 12 months, how often would you say the following situations occurred in your neighborhood? Vandalism or damage to public or private property (Ref: Never)
Hardly ever	−0.0981	0.907			−0.141 *	0.869 *	−0.177 **	0.838 **
	(0.0830)	(0.0753)			(0.0808)	(0.0702)	(0.0866)	(0.0726)
Frequently	−0.291 ***	0.747 ***			−0.492 ***	0.612 ***	−0.391 ***	0.677 ***
	(0.0941)	(0.0703)			(0.0914)	(0.0559)	(0.0972)	(0.0657)
Always	−0.340 ***	0.712 ***			−0.679 ***	0.507 ***	−0.258 **	0.773 **
	(0.122)	(0.0868)			(0.118)	(0.0600)	(0.124)	(0.0959)
During the last 12 months, how often would you say the following situations occurred in your neighborhood? Alcohol/drug use on public roads (Ref: Never)
Hardly ever	−0.210 *	0.810 *	−0.179	0.836	−0.280 **	0.756 **	−0.325 **	0.723 **
	(0.120)	(0.0975)	(0.117)	(0.0982)	(0.117)	(0.0884)	(0.126)	(0.0913)
Frequently	−0.255 **	0.775 **	−0.328 ***	0.720 ***	−0.393 ***	0.675 ***	−0.386 ***	0.680 ***
	(0.115)	(0.0888)	(0.111)	(0.0801)	(0.111)	(0.0752)	(0.120)	(0.0817)
Always	−0.302 ***	0.739 ***	−0.431 ***	0.650 ***	−0.601 ***	0.548 ***	−0.554 ***	0.575 ***
	(0.117)	(0.0866)	(0.114)	(0.0740)	(0.114)	(0.0625)	(0.123)	(0.0709)
During the last 12 months, how often would you say the following situations occurred in your neighborhood? Robberies or assaults on public roads (Ref: Never)
Hardly ever	−0.0498	0.810 *	−0.0259	0.974	−0.277 ***	0.758 ***	−0.366 ***	0.693 ***
	(0.0870)	(0.0975)	(0.0842)	(0.0821)	(0.0845)	(0.0640)	(0.0917)	(0.0636)
Frequently	−0.348 ***	0.775 **	−0.103	0.902	−0.693 ***	0.500 ***	−0.582 ***	0.559 ***
	(0.0974)	(0.0888)	(0.0901)	(0.0813)	(0.0944)	(0.0472)	(0.102)	(0.0572)
Always	−0.384 ***	0.739 ***	−0.437 ***	0.646 ***	−0.996 ***	0.369 ***	−1.043 ***	0.352 ***
	(0.127)	(0.0866)	(0.119)	(0.0771)	(0.123)	(0.0455)	(0.131)	(0.0461)
During the last 12 months, for fear of being the victim of a crime (robbery, assault, battery, etc.), have you stopped doing any of the following activities? Take empty buses (Yes = 1)	−0.341 ***	0.711 ***			−0.217 ***	0.805 ***	−0.253 ***	0.776 ***
	(0.0728)	(0.0517)			(0.0698)	(0.0562)	(0.0748)	(0.0581)
How often do you avoid bus stops to prevent being a victim of a crime? (Ref: Always)
Only at nights	0.748 ***	2.112 ***			0.900 ***	2.461 ***	0.520 ***	1.682 ***
	(0.141)	(0.298)			(0.136)	(0.335)	(0.142)	(0.239)
Only in the days	0.779	2.179			0.754	2.126	1.446 ***	4.247 ***
	(0.484)	(1.055)			(0.465)	(0.989)	(0.511)	(2.172)
does not prevent it	1.205 ***	3.337 ***			1.680 ***	5.365 ***	1.671 ***	5.318 ***
	(0.153)	(0.510)			(0.148)	(0.796)	(0.157)	(0.835)
During the last twelve months, were you or any member of your household assaulted using violence, threats, or intimidation? (Yes = 1)	−0.169 *	0.844 *			−0.238 ***	0.789 ***	−0.149	0.862
	(0.0940)	(0.0793)			(0.0910)	(0.0718)	(0.0956)	(0.0823)
cut1	−2.544 ***	0.0785 ***	−3.721 ***	0.0242 ***	−2.673 ***	0.0690 ***	−1.925 ***	0.146 ***
	(0.254)	(0.0199)	(0.169)	(0.00410)	(0.214)	(0.0148)	(0.259)	(0.0378)
cut2	0.828 ***	2.288 ***	−1.056 ***	0.348 ***	0.555 ***	1.741 ***	1.924 ***	6.845 ***
	(0.251)	(0.574)	(0.157)	(0.0547)	(0.210)	(0.366)	(0.259)	(1.771)
cut3	4.756 ***	116.3 ***	2.292 ***	9.892 ***	4.712 ***	111.2 ***	5.487 ***	241.6 ***
	(0.289)	(33.56)	(0.168)	(1.663)	(0.244)	(27.11)	(0.309)	(74.77)
Observations	4393		4139		4568		4367
Log likelihood	−3821.8696		−4111.2863		−3996.1		−3477.689
LR Chi2	866.80		429.49		1187.15		1279.00
R2	0.1019		0.0496		0.1293		0.1553

Note: OR: Odd ratios. Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.

estimates the four previous models, adding from the same data base variables associated with the perception of crime. All estimated models were stable across specifications and the LR Chi2 test was significantly different from 0 (p < 0.001).

Again, the gender variable had a negative and significant effect in the four models. The model associated with the metro mode is the one that presented a parameter of greater magnitude in absolute value and with an OR = 0.480. The parameters of the other sociodemographic variables followed the same trend as the estimated models presented in Table 2.

In the four models, the probability of having a negative perception regarding security increased when the respondents believed that they would be the victim of a crime during the next 12 months or when a member of their household was the victim of a crime. The same effect was seen when respondents believed there was vandalism, robberies, and alcohol and drug use in their neighborhood. On the other hand, those who were not affected by crime in their daily life had a greater probability of answering that they were safe in the four proposed scenarios. The same trend was observed among those who believe that crime has decreased in their neighborhood.

5. Conclusions

This research evaluated gender differences in the perception of safety in public transport in Santiago, Chile using both a quantitative and a qualitative approach. Four ordinal logistic regression models were estimated to analyze how sociodemographic factors and variables associated with the perception of crime influence riders’ perceptions of safety in public transport. To complement the results, four focus groups were developed to obtain a deep understanding of the participants’ experiences with safety in the Santiago public transport system.

With data from the ENUSC 2019 survey carried out by the Undersecretary for Crime Prevention of the Ministry of the Interior and Public Security, a gender comparison was made regarding the perception of safety in four scenarios: inside buses, inside the metro, at bus stops, and waiting for buses at night. The estimated models provided information on different factors that influence perceived insecurity in public transportation and whereabouts. In general, perceived insecurity inside buses, inside the metro, and waiting for public transport at night is greater among women, older people, and national citizens. Other influencing variables are the perception of insecurity regarding crime in general, the fear of being a victim of crime, or negative situations that occur in the neighborhood, such as the presence of robberies, alcohol, and drug consumption.

In all the estimated models, the probability that women feel more insecure compared to men is greater. On the metro, women are more likely to respond that they feel very insecure (5%) or insecure (15%), while inside the bus and waiting for public transport at night, the probability increases 15% and for waiting at a bus stop, the probability reaches 9%.

The analysis carried out also allowed us to confirm that perceived insecurity also affects the use of public transport. A total of 36% of those surveyed stated that they had stopped using empty buses as a result of perceived insecurity, while 65% stated that they avoided bus stops at night. In both cases, more than 60% of those in this situation were female users. Although these results do not include harassment, they are showing a worrying reality for 50% of the population. This is ratified by focus groups where participants stated that the use of public transport at night was one of the scenarios that generates the greatest concern in terms of safety, especially for women who avoid traveling at night, choose longer but safer routes, or pay for more expensive but reliable modes of transport, such us Uber/taxi.

It is interesting to note that for the perception of insecurity in public transport, although it is affected by the previous experience of having been a victim of a crime, the ENUSC 2019 showed that only 12% of the people who declared perceiving insecurity had experienced a robbery, theft, or injury. If we consider the people who avoided using empty buses or bus stops at night, most of them had not been victims of a crime, which shows that previous experience as crime victims does not have a direct influence on the perception of insecurity and subsequent exclusion of using the transport system. A possible explanation is the influence of the media, the main source of information on the subject, widely surpassing sources such as their own experience or that of household members. Analyzing the data from other cities in Chile, although the general levels of insecurity may vary in magnitude, the gender difference is constant throughout the country, where some cities present more radical variations than others.

Although the Ministry of Transport has currently launched the first Gender Equity Policy in Transport, where one of the objectives is to promote equal levels of safety in the use of public transport, it is still necessary to reformulate the conceptual frame of insecurity, which in this survey is limited to fear of crime, excluding fears or insecurities related to harassment, traffic accidents, discrimination (genderphobia, xenophobia) expressed in words, gestures, and/or violence, contagious diseases, and street protests, among others. To carry out a precise public policy on this matter, a permanent scan of security issues in public transport should be developed, considering a complete set of variables, including the ones mentioned here and some of the factors mentioned in the literature that are not present in ENUSC, such as socioeconomic variables (income and neighborhood); psychological factors (traveling alone or accompanied; security element present in the transport mode, police presence); and travel variables (origin, destination, travel time). Unfortunately, the last two ENUSC (2020 and 2021) did not consider the variables studied by this research; therefore, policy makers do not have access to understanding how safety in the public transport system is evolving. Public opinion on the increase of insecurity in Chile and the relevance of avoiding this disadvantage in public transport are key for major public policies on social inclusion and gender equity. A deeper study is especially relevant nowadays given the increase of insecurity after the COVID pandemic, which is the current most relevant public issue in the country.

It is possible to conclude that there is a high perception of insecurity in public transport for both men and women, with women being the ones who declare feeling more insecure when using public transport. This lack of security in public spaces indicates a decrement of democratic standards that limit women’s freedom of movement and choices available, as well as an element of exclusion from social opportunities. Although a high level of insecurity perception is well known in Latin American public transport, the gender differences evidenced show a source of inequality that must be remedied by the public authorities of these countries.