Women are a “situationally disadvantaged” group in public places (Gardner 1995, p. 16; Valentine 1989). As such, they are disproportionately targeted for harassment by strangers. Such harassment is asymmetrical and pervasive, and it takes a gender-specific form (Gardner 1995). For this reason, scholars have investigated a range of behaviors that includes lewd comments, catcalls, whistling, gender-hostile remarks, staring, and groping. Sociological and legal research has documented that women experience feelings of discomfort and fear in relation to such unwanted, male-initiated behavior (Bowman 1993; Gardner 1995; Kissling 1991; Nielson 2000). However, researchers have not yet considered how the broader forms of everyday rude behavior that take place between strangers might impact women; activities such as pushing in line, tailgating, and cursing have been neglected. Although less obviously gendered, these experiences might also be frequent and have negative consequences. Moreover, a benchmarking of this wider universe could allow the specific, additional harms of intentional, communicative, and gendered forms of stranger harassment to be isolated.

In this vein, we seek to illuminate whether uncivil encounters between strangers can negatively impact women’s experience of public places and subsequently constrain their behavior. Using nationally representative survey data from the 2005 Everyday Life Incivility in Australia (ELIAS) project, we investigate the frequency with which negative, although non-criminal, encounters affect women, as well as how such experiences impact women’s subsequent emotional well-being and behavior. We demonstrate that uncivil behavior affects women more negatively than men and argue that public incivilities exact a gender-specific tax on women’s access to public places and related resources.

Gender, Street Harassment, and Public Incivility

The study of sexual harassment is central to research agendas examining how women are disadvantaged in everyday life. Although this activity can occur in the workplace, in organizational settings, or in the home, its most overt manifestation is unwanted attention in public space—an activity generally named “street harassment” in scholarship. Unlike men in public, women are presumed to be “open persons,” in Goffman’s (1963) sense, and hence available for unsolicited interaction. It has been established that, as with rape and domestic violence, there is gender-asymmetry in this context: men harass women in public frequently whereas women rarely harass men (Gardner 1995; Kearl 2014). These affronts can take any number of forms, ranging from threatening behaviors, sustained verbal abuse, and physical assault (Kissling 1991) to blatant violations of the norms of civil inattention and, consequently, to the tokenistic or refracted observance of women (Gardner 1995, p. 331).

Whatever the manifestation, harassment has been widely interpreted as a motivated and communicative act that reflects and reinforces broader structural relations of power. As Duneier (1999, p. 200–210) argues, men who harass women “use the privileged position [they] enjoy in the public sphere to influence what will happen on the street…knowing that even privileged women occupy vulnerable positions in public space.” Kissling (1991, p. 456) likens this activity to “terrorism” in that men “frighten, and through fear, dominate and control” women. Negative consequences are not temporally restricted to the immediate encounter. Harassment can amplify concerns about sexual violence as well as foster a more general fear of crime (Fairchild and Rudman 2008; Macmillan et al. 2000). Sociological research has shown that women may experience feelings of prolonged discomfort in response to such acts, with avoidance or defensive interactional strategies often the reluctant response (Cobbina et al. 2008; Lofland 1973; Macmillan et al. 2000). As a result, women’s capacity to enact and enjoy full citizenship within public places becomes diminished (Bowman 1993; Gardner 1995; Kissling 1991; Nielson 2000). This is doubly unfortunate given that even in the absence of such negative encounters, women’s ability to use and enjoy public space is often already compromised by a general sense of unease (Day 1999).

Clearly harassment in streets (Cobbina et al. 2008; Gardner 1995), leisure destinations (Grazian 2008), and other such public locations is a behavior that has significant collective and individual consequences; it is something that scholars must study. But what about the other forms of incivility experienced by women? What about actions difficult to describe as “terrorism” or even as “harassment”? One promising path has been to inquire into a broader range of gendered interactions. Specifically, researchers looking at everyday experiences defined as “gender prejudice” (Brinkman and Rickard 2009), as “prejudice-related” (Hyers et al. 2006) or as “sexism” (Ayers et al. 2009) have uncovered a raft of negative experiences such as being “checked out” by members of the other sex, putdowns based on stereotypes, exposure to sexist mass media materials, or being prejudicially ignored and excluded from group activities.

Falling short of conventional definitions of “harassment,” these experiences were shown to be both remarkably diverse and common, especially for women. Here we push this agenda even further. What about the full range of incivilities that women experience in their daily lives? What about low-intensity affronts such tailgating, the invasion of personal space, and pushing in line? Even where gender might make women a target for such acts by changing culturally defined, gendered situational opportunity structures (for example, through the perceived victim vulnerability of women and presumed offender impunity against retaliation), the everyday rude behaviors encountered by women need not take the form of prejudicial, intense communicative acts framing a sexed subject.

The scattered and diverse negative experiences experienced by women in public have rarely been studied. It is tempting to dub them “low-key,” “minor” or “trivial” in comparison to both full-blown street harassment and low-key forms of sexist interaction. However, these other events may well have their own negative consequences, such as constrained behavior, fear, and enforced changes to daily routines. It is worrying that research already shows notable gender disparities in well-being in domains of public life where we might predict women to encounter incivility from strangers. For example, female commuters are more stressed than are men (Roberts et al. 2011), and female leisure time is known to be precarious and compromised in its quality (Bittman and Wacjman 2000). Furthermore, research in organizational settings suggests that general incivility is connected in complex ways to gendered incivility and sexual harassment (Lim and Cortina 2005). These seem to reinforce each other, frequently experienced as a package of negative encounters. Looking to sexual harassment alone obscures this wider context.

In this way, when it comes to the core agendas of feminist scholarship, an investigation of the wider range of incivilities encountered by women is clearly important. It might help us generate analytic leverage in unpacking the focal issue of street harassment. With a comparison set of broader experiences, the additional penalties of intentional, sexually coded street harassment as a subset of antisocial situations might be isolated. Yet in contrast to research on public harassment or organizational culture, the existing knowledge base on the full range of “everyday incivilities” (Phillips and Smith 2003) experienced by women is sorely lacking. Whereas there is plenty of material available for a comprehensive literature review on street harassment and some on gender prejudice (our citations here are just a fragment), dedicated research asking women about their experiences of generic incivility is nearly invisible. A recent comprehensive national survey from Australia discovered that even after controlling for time spent in public places (routine activity), women were more likely to be targets of everyday incivility than were men, but less likely to be perpetrators (Smith et al. 2010). Yet this Australian study does not follow through on these observations and generally downplays the role of gender in interpreting the patterns and meanings of incivility. Other recent surveys also suggest that gender plays an important role in the experience of public incivility: for example, the New York based NGO Hollaback hosts a website that collects narratives from women who have been subject to “street harassment” broadly defined. Notably, a substantial proportion of the incidents logged at this website were not defined by the targets as sexual in nature, but simply as annoying (Hollaback 2014). An ABC News poll indicated that some disadvantage might come from the fact that women have different sensitivity levels than do men (Cohen and Langer 2006). For example, women were more likely than were men to be bothered by behaviors such as cursing, disrespect, and loud cell phone use. Although the picture we gain from such work is imprecise, it nevertheless hints that routine incivility is a real problem for women today.

Although research on the broad range of public incivilities perpetrated against women by strangers is thin on the ground, a vibrant tradition grounded in social psychology on workplace incivility signals that ours is a path worth exploring. Explicitly setting itself apart from the literature on workplace sexual harassment, much of this field examines the impact of experiencing or witnessing incivility on the job (Andersson and Pearson 1999; Cortina 2008). Actions such as shouting, eye rolling, and swearing are shown to be reasonably common and to have diverse negative impacts including reduced job satisfaction and productivity, as well as lower health satisfaction (Miner-Rubino and Cortina 2004). Men are more likely to be instigators than are women by a factor of two to one (Pearson et al. 2000) whereas women seem to be targets more often than men are (Schmidt 2010). Further, women are far more likely than are men to experience a wider package of negative encounters ranging from general incivility to sexual harassment (Lim and Cortina 2005). Like the research on “gender prejudice” we cited, this literature cumulatively suggests that encounters falling short of “harassment” may still have adverse impacts on women that are worthy of study.

The Present Study

With this thin research history as a background, we develop and empirically test four hypotheses. (a) Our first hypothesis is that women are more likely than are men to experience encounters with rude strangers, and this association will be independent of women’s other personal characteristics. Our additional hypotheses probe the consequences of such encounters in terms of emotions and behaviors. (b) Our second hypothesis is that women are more likely than are men to experience immediate fear in relation to encounters with rude strangers and that this effect will be amplified if the rude stranger is male. (c) Our third hypothesis is that encounters with rude strangers will have greater negative consequences for women’s sense of personal safety and well-being, compared to men’s. Specifically, we anticipate that women will be more likely to report emotional distress in response to the encounter. (d) Finally, we anticipate that women are more likely to modify their own behavior patterns in response to these negative experiences, with the intention of avoiding such encounters with rude strangers in the future.

Method

In order to pursue this agenda, we use data from the 2005 survey of Everyday Life Incivility in Australia (ELIAS), a cross-sectional study about commonplace encounters with rude strangers. The ELIAS survey used random sampling from the residential white pages, and the sample was stratified to ensure that it was nationally representative in terms of age, gender, and state of residence (Smith et al. 2010, pp. 16–20).

The ELIAS dataset is an apt choice for our research questions because the survey’s design allows us to establish benchmarks about the frequency and attributes of rude behavior. In particular, the survey captures detailed information about the event and the actors involved. These data include demographic attributes of the respondent as well as the rude stranger, details about how the interaction unfolded, situational and locational data, and the respondents’ emotional reactions to the event. Finally, we note that the ELIAS results are likely to be broadly representative of incivility in other multicultural, economically developed nations (Smith et al. 2010, p. 16); we therefore anticipate that our conclusions may be used to direct further research on gender and incivility in the United States and elsewhere.

As we reported, previously published research using this dataset had indicated noteworthy gender patterns. Smith et al. (2010) book found that the majority of respondents who reported uncivil encounters were women. Nevertheless, gender was a small and arguably residual component of their investigation. With other considerations taking center stage, we find different subsamples were used in the various analytic models that are presented (see pp. 113 and 123 for examples). This makes it impossible to draw coherent, unified conclusions about gender from their book, and, we suspect, provides a set of results that might underplay the role of gender in experiences of incivility. By contrast, we organize our analysis around gender and additionally use multiple imputation for missing data. As a result, we are able to run models on a consistent subsample and offer a comprehensive and systematic set of results.

Participants

Fully 1621 adults who were contacted agreed to participate in the ELIAS study, yielding a response rate of 30 %, which is fairly standard for telephone surveys of this nature (Hartmann et al. 2003). In terms of respondent demographics, 838 (51.7 %) were women and 783 (48.3 %) were men; the average respondent was approximately 45 years-old, and ages ranged from 18 to 85 across the sample; and 1459 (90 %) had lived in Australia for 20 or more years. Additionally, 406 (25 %) completed 10 years of schooling or less, 617 (38 %) completed high school or a trade certificate, and 586 (36 %) completed a Bachelor’s degree or higher. (Note that additional descriptive statistics may be found in Table 1.)

Table 1 Descriptive statistics and gender comparisons using imputed data
Full size table

Procedure

In order to administer the survey, trained telephone interviewers called the selected households after a letter introducing the study had been mailed to all potential study participants. Interviewers asked people about experiences of everyday incivility with a rude stranger within the prior month; this short time period ensures more accurate recall than is typically available in retrospective surveys. Rude strangers were defined as people who were not in a visible work role at the time of the encounter, and whom the respondent had never seen before. The survey captured detailed information on 508 recent rude stranger encounters, as well as demographic and attitudinal measures on the full sample of 1621 respondents.

Measures

Dependent Variables

We use four dependent variables to test our hypotheses. The first dependent variable consists of self-reported incidents of rude behavior: this binary variable indicates if the respondent experienced an encounter with a rude stranger during the previous month. Specifically, interviewers asked:

Now, can you think of an occasion within the past month when you came across a rude stranger? Can you recall such an event? We’re looking in particular at events that occurred in Australia, that involved another person you’d never seen or met before, just another member of the general public (rather than someone at work), came across in the course of simply going about your everyday life activities (rather than connected with your work).

If respondents had experienced more than one encounter with a rude stranger during the previous month, they were instructed to report details only about the most recent encounter. The location of these events excluded private venues such as home or work; respondents reported a wide variety of event locations, including (but not limited to) streets, highways, shops, and public transportation. In total, 508 (31 %) of 1621 respondents surveyed reported such an encounter. After taking a short summary of the incident, interviewers compiled event details through a layered survey protocol that lasted up to 45 min. This allows us to systematically analyze how women’s encounters differed from men’s. Survey interviewers took notes on the incident as described by the respondents. Examples of these female respondents’ descriptions are: “He stuck his middle finger out and said something abusive. I could not hear what it was, but clearly it was abusive.”; “My husband and two kids were out walking [with me] and a male just pushed past me on the pathway instead of waiting to pass or asking me to move aside; he nearly made me fall and just did not care.”; “We were sitting in a café and [women] were smoking nearby—I asked them to move and they refused.”; “I was in a supermarket and a lady pushed me out of the queue and got ahead of me.”; and “I was getting my mother-in-law out of the car and into a wheelchair in the parking bay, and I had the door open. The person tooted the horn as she tried to park in the next bay.”

Importantly, of the 508 respondents who reported an encounter, only four, or less than 1 %, noted that the rude stranger used sexual language. In other words, very few of these encounters were blatantly sexual or sexist in nature. This does not preclude them from being gendered in some other way that is indiscernible to respondents. However, we emphasize this point to highlight the routine, low-severity nature of these events and to reiterate that any gender patterns uncovered here are likely to be more pronounced in any higher severity and/or more explicitly gendered negative interactions.

Our second dependent variable is a dichotomous variable indicating if the respondent experienced fear at the time of the event. In an open-ended question, respondents were asked “What were the main emotions you remember feeling at that moment [during the encounter]?” Respondents who reported experiencing emotions such as fear, or very similar reactions such as worry and anxiety, were recorded as having experienced fear.

The remaining two dependent variables work as a pair. Our third dependent variable is a dichotomous variable indicating if the respondent reported a high number of coping behaviors in response to the encounter. Interviewers asked “People cope with their feelings about things that happen to them in different ways. In the time since the event, have you done any of these things as a way of dealing with your feelings about what happened?” The interviewer then listed an inventory of 11 possible coping behaviors, which were based on social psychological coping scales (Folkman et al. 1986). To prevent a gendered response bias, options that conformed to cultural norms for both genders were included; additionally, both active and passive responses were included. Possible responses included: (a) Tried to think about what happened in a different way, (b) Had a drink or took a pill, (c) Fantasized about what I should have said or done to the rude stranger, (d) Talked to someone about how I felt about what happened, (e) Tried to forget what happened by doing or thinking about something else, (f) Prayed to God for help, (g) Went out to get some exercise to make me feel better about what happened, (h) Yelled or hit something to let out my pent-up feelings about what happened, (i) Waited for my feelings about what happened to pass, (j) Tried to accept what happened, and (k) Thought about how to get revenge on the other person. The total number of reported coping behaviors per respondent ranged from 0 to 9, and the average number reported was three. Because it is difficult to equate different coping behaviors, we anticipate that an increase in one or two coping behaviors is less meaningful than having engaged in a relatively high number of coping behaviors. Thus, we used a binary outcome here, where a high number of coping behaviors is defined as greater than the average of three coping behaviors (coded 0 = 0–3; 1 = 4–9). The second variable of this final pair, our fourth dependent variable, is a dichotomous outcome measuring if the respondent reported one or more of three possible avoidance behaviors as a result of the encounter. Respondents were asked if they had been “avoiding the place where the event occurred,” “avoiding places like the one where the event occurred,” and/or “avoiding busy public places in general” since the time that the rude encounter had occurred (coded 0 if no avoidance behaviors were reported, and 1 if 1–3 avoidance behaviors were reported).

Independent and Control Variables

Our main independent variable is gender of the respondent, which was measured as a binary measure where response choices included female and male. We incorporate three sets of control variables in our models. The first set of controls consists of demographic characteristics of the respondent. Specifically, respondents were asked to report age, in years; education level, which had seven possible choices ranging from “did not complete high school” to “postgraduate degree” for highest level of education attained (collapsed into three categories for analyses); self-identified membership in a racial, ethnic, or religious minority group; length of residence in Australia in years; and residence in a major city (described to the respondent as 100,000 people or larger), a regional center (between 25,000 to 100,000 people), or a rural town (25,000 people or fewer).

In addition, we control for other relevant characteristics of the respondent. For cynical attitude about strangers, we constructed a binary variable that summed responses to three binary questions about trust (the summary score was coded 1 if the total summed to 3, and 0 if the total summed to 0–2): (a) “Do you think most people would try and take advantage of you if they got a chance, or would they try to be fair?” (coded 1 for “Would take advantage of you”); (b) “Would you say that most of the time people try to be helpful, or that they are mostly just looking out for themselves?” (1 = “Just look out for themselves”); and (c) “Generally speaking, would you say that people can be trusted or that you can’t be too careful in dealing with people?” (1 = “You usually can’t be too careful). Additionally, we control for the total number of crime victimizations that occurred in a public place during the previous year, as reported by the respondent, as well as if the respondent reports feeling unsafe when using public places in the daytime. For the latter variable, respondents were asked, “How safe or unsafe do you feel when out in public places by yourself during the day?,” selecting from five ordered responses ranging from “very safe” to “very unsafe” (“never out in public during they day” and “don’t know” were also options). We collapsed these responses to create a binary (0 or 1 coded) variable indicating that the respondent answered 1 (unsafe or very unsafe). We include the two attitudinal measures above and the crime victimization control to make sure that feelings of fear and/or any behavioral changes are related to the event of interest rather than a reflection of more general attitudes or previous crime victimization.

Finally, we control for time spent in public places, an index that captures the amount of time that a respondent reports spending in a variety of public places. Respondents were asked to report how often they typically spend time in 14 different public places, including (a) enclosed shopping malls, (b) outdoor shopping malls, (c) food courts, (d) supermarkets, (e) department stores, (f) large cinema complexes, (g) large casinos, (h) large food chains (e.g., McDonalds), (i) large video rental stores (e.g., Blockbusters), (j) events at large sporting grounds, (k) travel on a main roads/freeways/highways, (l) mass parking lots, (m) public transportation, and (n) planes. For each type of public place, respondents answered using a 7-point ordinal scale ranging from 0 (never) to 6 (almost every day), with a score of 98 as the potential maximum. In our data, the index ranges from 0 to 70, indicating that none of the respondents reported spending time in all 14 public places “almost every day.” The index is normally distributed, and we standardized it for ease of interpretation. By controlling for time spent in public, we ensure that we are not simply picking up on a propensity to experience rude behavior that is driven by having more opportunities for these encounters.

The second set of control variables measure characteristics of the rude stranger as perceived by the respondent. These include the stranger’s gender; and a binary measure indicating if the respondent perceived the rude stranger as 1 (very or quite “rough-looking”). The final set of control variables accounts for situational attributes, again as perceived by the respondent. The first of these is a binary variable indicating whether (coded 1) or not (0) the rude stranger’s behavior involved movement. We constructed this variable by drawing from an open-ended survey question in which respondents were asked to report exactly what the rude stranger did. Interviewers selected the closest match (or matches if more than one type of behavior was described) from a predetermined list of 25 rude but noncriminal behaviors, which fell into four categories: (a) movement, such as pushing in to the respondent or invading the respondent’s personal space; (b) bodily (mis)management, such as a rude gesture or spitting; (c) sounds, such as speaking loudly or screaming; and (d) language, such as sexual remarks or swearing. We included this control because we anticipated that a rude behavior involving movement would more often be perceived as threatening (rather than just offensive), and therefore needed to be accounted for in our models. For this reason, we also determined whether women and men experience other subtypes of rude behavior at different rates: we found no statistically significant gender differences in rude behaviors related to bodily mismanagement or sounds. We did find a gender difference in proportions of language-related rude incidents: 20 % of men experienced this type of encounter compared to 12 % of women. These results suggest that women are not more sensitive to particular types of rude behavior than are men. With the exception of language-related behaviors, women are not reporting substantially different types of rude behaviors than are men.

Additionally, we control for deliberate rude behavior, which indicates if the respondent perceived the rude stranger as deliberately meaning to offend or disadvantage him/her (0 or 1 coded); RS vehicle, which indicates if the rude stranger was either driving or a passenger in a vehicle (0 or 1 coded); night, which indicates if the event took place at night (0 or 1 coded); crowded location, which indicates if the respondent recalls that the location was somewhat or very crowded (0 or 1 coded); and respondent alone, which indicates if the respondent was traveling completely alone at the time of the event (0 or 1 coded). This third set of covariates enables us to control for event attributes that may make respondents more likely to experience fear or change their subsequent behavior patterns. For example, we would anticipate that rude encounters that take place at nighttime would cause more fear on average. These controls allow us to estimate the association between gender with our response variables more precisely.

Analytic Approach

Our analysis proceeds in two stages. First, we expand on Smith et al.’s (2010) findings by determining if the association between gender and uncivil encounters is maintained when we control for a host of demographic attributes for the respondent, using the full sample of 1621 respondents. Given the bivariate coding of this outcome variable, we employ a logistic regression model to estimate the associations with covariates. In stage two, we limit the sample to the 508 respondents who did experience rude behavior, and we employ a series of nearly identical logistic regression models to predict the likelihood of three key outcomes: experiencing fear at the time of the event, reporting a high number of coping behaviors after the event, and avoiding public places as a result of the encounter. In these three models, we control not only for individual characteristics of the respondent, but also for rude stranger characteristics and situational features of the incident. In sum, our analytic approach allows us to examine how respondents’ gender and rude strangers’ gender are associated with the frequency and impact of negative encounters between strangers in public places. Note that we used multiple imputation by chained equations to replace missing values in the data; for all variables used in our analyses, missing data ranged from .5–6.7 % of cases. We utilized Stata 12’s mi command series to generate five imputed datasets (Royston 2005). All analyses were conducted using the mi command series, which runs estimation commands on each imputed dataset separately, and then combines those estimates to produce the final output (StataCorp 2014).

Results

We first present descriptive results summarizing the characteristics of our samples. We then move to describe results from our four models. The first model assesses the relationship between respondents’ gender and the likelihood of experiencing rude behavior, using the full sample of 1621 respondents. The additional three models examine the relationships between gender and (a) experiencing fear during the encounter, (b) engaging in higher than average coping behaviors after the encounter, and (c) avoiding public places as a result of the event. As we noted, these last three models use the subsample of 508 respondents who did report an encounter with a rude behavior. We do not include data from the remaining respondents who did not report rude behavior because no information concerning rude stranger or situational characteristics was collected from those respondents. Taken together, the results of these models provide a robust understanding of how gender is related to the experience of rude behavior in public.

Descriptive Statistics

In Table 1, we present descriptive statistics for the full sample of 1621 respondents. We used I-tests (and chi-square tests) to examine gender differences in means (and proportions) in the first dependent variable and covariates. Descriptive results indicate that 34 % of women experienced rude behavior, compared to 29 % of men; this difference is statistically significant. Although women report that they typically spend slightly less time in public places than do men, the .10 standard deviation difference in means is not dramatic. Notably, women report fewer crime victimizations: whereas men reported .86 crime victimizations on average during the previous year, women reported .59 on average. Finally, whereas slightly fewer women than men report feeling safe in public during the day, the vast majority of women (96 %) and men (98 %) reported that they did feel safe in public during the daytime.

In Table 2, we conduct identical descriptive analyses but restrict the sample to the 508 respondents who report rude behavior. We first examine gender differences for the three remaining dependent variables. Of those who had an encounter with a rude stranger, significantly higher proportions of women experienced fear at the time of the event, engaged in a high number of coping behaviors, and reported avoidance behaviors after the encounter compared to men. Additionally, a number of other descriptive findings are noteworthy. In this subsample, men spend more time in public when compared to women (about .33 standard deviations more), and the difference is statistically significant. Again, we see that men report more crime victimizations than do women during the past year: men report an average of 1.32 victimizations, whereas women report .80. This difference is in line with previous research indicating that men are more likely to be victimized than are women (van Kesteren et al. 2001). We further note that significantly more male respondents experienced an encounter with a male rude stranger when compared to female respondents. Finally, women were less likely than were men to be traveling at night or alone when they encountered the rude stranger. Overall, descriptive results motivate a multiple regression analysis to determine if the associations remain when we incorporate key covariates.

Table 2 Descriptive statistics gender comparisons for respondents reporting rude behavior using imputed data
Full size table

Rude Behavior Victimization

Table 3 presents results from a bivariate model (Model 1), indicating that being female is associated with 1.27 times greater odds of reporting an experience of rude behavior. Model 2 results demonstrate a statistically significant association between gender and the odds of experiencing rude behavior after including covariates in the model. Specifically, being female is associated with 1.43 times greater odds of reporting an experience of rude behavior during the past month when we control for standard demographic attributes such as age, education, and race, as well as key characteristics including length of residence in Australia, city type, amount of time spent in public places, attitudes about strangers and public places, and crime victimization history. Additionally, we obtained predicted probabilities for Model 2 by varying gender and holding all other variables at the mean; these results indicate that women have about a one in three chance of reporting an experience of rude behavior, whereas men’s chance of experiencing rude behavior is about a one in four. This finding confirms the main thrust of our first hypothesis: women report experiences of rude behavior more frequently than do men, and this association with gender is robust to a host of controls that we might expect to affect this outcome.

Table 3 Odds ratios from logistic regression model predicting rude behavior victimization
Full size table

Notably, this association with gender is large in magnitude, even after we incorporate important covariates into the model. Being a racial, ethnic, or religious minority is associated with 48 % lower odds of experiencing rude behavior, net of other controls. This runs counter to our expectations because we anticipated that minorities would be subjected to rude behavior, including harassment, more often. As we might expect, time spent in public places has a significant, positive relationship with the odds of experiencing rude behavior: a one standard deviation increase in the amount of time spent in public is associated with 1.47 times higher odds of reporting rude behavior. Lastly, crime victimization history has a significant relationship with our dependent variable: one additional victimization during the previous year is associated with 1.13 times greater odds of reporting an experience with a rude stranger. This relationship suggests that some respondents, perhaps because of where they travel or with whom they associate, are more often subjected to rude behavior and crime; alternatively, people who have more often been targets of crime may be more sensitive to rude behavior. There are a few other associations that we note here. Having lived in Australia for less than 20 years decreases the odds of experiencing rude behavior by 36 %, compared to having lived there for 20 or more years, when we control for the other individual attributes noted above.

Rude Behavior and Fear

In Table 4, we first present the results of a bivariate model (Model 1) predicting respondents’ fear at the time of the encounter with a rude stranger. This model indicates that at baseline, being female is associated with 2.44 greater odds of experiencing fear at the time of the rude encounter. Model 2 shows that there is a statistically significant relationship between gender—of the respondent and the rude stranger—and respondents’ fear. In this way, Model 2 indicates that gender is a key factor in predicting fear in response to rude behavior victimization. Being female is associated with 3.61 times greater odds of experiencing fear, compared to being male, when we control for a rich array of control variables. These controls include respondents’ age, education, race, length of residence in Australia, city type, amount of time spent in public places, attitudes about strangers and public places, and crime victimization history. Additionally, we control for rude strangers’ gender, as well as whether the respondent perceived the stranger as “rough looking” and for situational attributes, including whether the behavior involved movement or vehicles, was deliberate, time of day, whether the location as crowded, and if the respondent was alone. The adjusted odds for experiencing fear are evidence of measurable gender differences in the consequences of negative encounters with strangers in public places and confirm our second hypothesis. We also note that respondents’ age has a small but significant positive association with the odds of reporting an encounter with a rude stranger; this may reflect a slightly greater sense of physical vulnerability among older adults.

Table 4 Odds ratios from logistic regression model predicting fear at time of event
Full size table

No other individual attributes, however, are important for predicting respondents’ fear. Characteristics of the rude stranger and the event itself, however, do play a role in fear because the gender of the rude stranger has a large and significant association with respondents’ fear. When we control for respondents’ gender and other key traits, as well as stranger and event attributes, we see that if the stranger is male, the odds that the respondent will experience fear are nearly four times higher than if the stranger is female. If the respondent characterized the stranger as “rough-looking,” the odds that the respondent experienced fear were 2.82 times as high. Two situational controls have significant relationships with respondents’ fear as well. If the rude behavior was perceived to be deliberate, then the odds that the respondent experienced fear increase markedly, having controlled for the aforementioned variables. Finally, if the stranger was driving or a passenger in a vehicle, odds of the respondent experiencing fear increase. We believe this reflects the increased risks of bodily harm eventuating from car accidents.

Using Model 2, we obtained predicted probabilities for experiencing fear by specific subgroups. Notably, our model indicates that about 10 in 100 women are expected to report fear during an encounter with a rude stranger, compared to 3 in 100 men. Additionally, we see that when the rude stranger is male, the respondent is more frequently expected to experience fear, whether the respondent is male or female. We also tested an identical model that incorporated an interaction between respondents’ gender and rude strangers’ gender, but the interaction term was not significant. This indicates that the increase in fear associated with encountering a male rude stranger is not more pronounced for women or men. In summary, results here indicate that respondents’ fear is primarily driven by respondents’ and rude strangers’ gender, as well as by a handful of event characteristics. Other respondent characteristics are largely irrelevant for this outcome, suggesting that, with the exception of gender, fear is not about some types of people being more sensitive than are others.

Rude Behavior, Coping, and Avoidance Behaviors

In Table 5, as well as in Table 6, we present results addressing our third and fourth hypotheses, that is, that women are more seriously affected by rude behavior and that they are more likely to adjust their use of public space in response to encounters with rude strangers. Our findings show that respondents’ gender is one of the few variables associated with respondents’ coping behavior patterns and avoidance of public places. We argue that the following results are the strongest evidence of gender differences in people’s experiences of public incivilities. Specifically, not only are women more likely to experience rude behavior and feel fear at the time of the event, but these events also have an impact on their emotional well-being and use of public space in ways that are measurably different from men’s.

Table 5 Odds ratios from logistic regression model predicting high coping behaviors
Full size table
Table 6 Odds ratios from logistic regression model predicting avoidance behavior
Full size table

Overall, odds ratios for the models predicting coping and avoidance behaviors (shown in Tables 5 and 6) show clearly not only that women are seriously impacted by rude behavior, but also that the consequences of facing rude behavior are quite different for women than for men. In Table 5, Model 2 demonstrates that women’s odds of reporting that they engaged in at least four coping behaviors as a result of an encounter with a rude stranger are 2.41 times greater than men’s—after controlling for respondents’ characteristics (including age, education, race, length of residence in Australia, city type, amount of time spent in public places, attitudes about strangers and public places, and crime victimization history), for rude strangers’ attributes (including gender and whether the respondent perceived the stranger as “rough looking”), and for situational characteristics (including whether the behavior involved movement or vehicles, was deliberate, time of day, whether the location as crowded, and if the respondent was alone). We additionally obtained predicted probabilities from Model 2 for reporting high coping behaviors for women and men, holding all other variables at the mean. For women, the predicted probability for reporting a high number of coping behaviors is .65, whereas for men it is .43. To the extent that coping behaviors are an indicator of a substantial emotional effect on the respondent, these results suggest that women face a greater negative impact on their emotional well-being after facing an encounter with a rude stranger.

We also find notable results for avoidance behaviors. The bivariate relationship between gender and avoidance behaviors is not significant. However, this relationship becomes statistically significant when we incorporate a full set of controls for respondents’ and rude strangers’ as well as key event characteristics. Model 2 (presented in Table 6) indicates that women’s odds of reporting at least one avoidance behavior as a result of the uncivil encounter are 2.37 times greater than men’s. Recall that possible avoidance behaviors included “avoiding the place where the event occurred,” “avoiding places like the one where the event occurred,” and/or “avoiding busy public places in general.” We argue that this measures a serious and sustained impact of rude behavior because the respondent found the event serious enough to change their behavior in an attempt to minimize the chance of having similar encounters in the future.

Additionally, we see that strangers’ gender and a handful of other variables are significant. When the stranger is male, the respondent has 2.56 times greater odds of reporting avoidance behaviors, after incorporating control variables. These gender patterns were affirmed when we obtained predicted probabilities from Model 2 (varying gender of the respondent and rude stranger, respectively, while holding all other variables at the mean). We find that for female respondents, the probability of reporting avoidance behaviors is .14, whereas for male respondents, this probability is .06. Moreover, results indicate that when a rude stranger is male, the probability of reporting avoidance behaviors is .13; by contrast, when a rude stranger is female, this probability is .05. These results hold true for both male and female respondents. We argue that this is evidence of a gender-specific pattern in encounters between strangers in public places. Finally, situational characteristics are significant here as well. If the rude behavior was perceived as being deliberate, or if the respondent was traveling alone, s/he had substantially greater odds of reporting avoidance behaviors.

It is important to note that the statistically insignificant bivariate relationship between gender and avoidance behaviors was unexpected, particularly because gender is significant in the full model for avoidance behaviors (shown in Table 6, Model 2). As such, we investigated these patterns further. To address concerns about multicollinearity in the full model (Model 2), we examined the variance inflation factor (VIF) for each independent variable. No value was greater than 2, which is well below the standard cut-off of 10 (Kennedy 1992). Additionally, we tested a series of models incorporating interactions between gender and other independent variables; however, no interactions were statistically significant. We suggest a few explanations for this pattern. First, there is substantial variability in both respondent and event characteristics, so it may only be possible to observe the relationship between being female and reporting avoidance behaviors after adjusting for important covariates. In other words, women may be more likely to avoid public places when they have encountered incivility that they perceive as particularly threatening. Additionally, because only 13 % of (66 of 508) respondents who encountered a rude stranger also reported avoidance behaviors, we have limited statistical power. This shortcoming underscores the need for further large-scale survey research on this topic.

Discussion

The main objective of our study was to determine how gender shapes negative encounters between strangers in public places beyond the more frequently studied and closely defined realm of sexual harassment. Our results demonstrate statistically significant, large associations between gender and the occurrence and consequences of such encounters. When a rude stranger is male, respondents of both genders are more likely to experience fear and report avoidance behaviors. However, women not only experience rude behavior more often than do men but also are far more likely to experience fear during the event. Perhaps most importantly, these encounters have a lasting impact on women’s emotional well-being, as evidenced by their coping and subsequent avoidance behaviors. We conclude that these public incivilities have a measurable and detrimental impact on women’s use of public places.

We suggest three explanations, or mechanisms, for the robust gender associations we described. First, women may objectively be subjected to rude behavior more frequently. This explanation is in line with Goffman’s theory that women are “open persons” in public places in that people feel more comfortable or entitled to engage women (1963, p. 128). This accounting is also broadly in line with the various studies and surveys in the field of organizational social psychology we cited earlier (e.g., Lim and Cortina 2005). A second explanation is that women may be more sensitive to certain types of behavior and/or are more likely to recall rude incidents. Such reasoning aligns with findings from research on domestic violence that indicates similar acts of violence affect women more seriously than they do men (Umberson et al. 1998), as well as those from organizational studies that point to gendered disparities in the labeling of identical actions (Montgomery et al. 2004). At a more macro level, this explanation also fits well with suggestions that the civilizing process has applied differentially across genders in Western modernity. Notably, since the Victorian era, women have been socialized and typecast as the upholders of modesty, propriety, and morality (see Davetian 2009, Chapter 6). Finally, perhaps men and women are subjected to rude behavior at similar rates, but women are targets of rude behaviors that are qualitatively different in some important way. A subtle shift in intensity or nature could mean that women would remember incidents more often. Lacking an omniscient view of what “really happened” (a task that would presumably require ethically problematic and pragmatically unrealistic, hidden video cameras in public settings; the real time interviewing of actors and spectators; ways to “objectively” code behaviors; etc.), our self-reported data does not allow us to discriminate among these three possible explanations.

Limitations and Future Research Directions

A limitation of our paper is that it shares the same disadvantages as every other non-experimental study that has examined experiences of incivility, crime, or sexism, that is, using a self-report method such as interview, diary, or survey questionnaires (and note that such studies form the backbone of our literature review). These shortcomings relate to respondents’ recall, to instrument calibration, to interviewer-effects, to self-censorship, to the tensions between researcher and folk definitions, and so on. Nonetheless, we believe our study and prior research has generated new knowledge that moves scholarship forward in a context of scant information. Moreover, the bottom line for even the most hardened skeptic must be that, regardless of the finer details of accuracy, memories of experienced realities are problematic for the women involved. As Brinkman and Rickard (2009, 472) put it, “…[T]he perceptions of the victims are often what are valuable. If a person believes they have been the victim ….they may face detrimental complications as a result.” And indeed, even when narrowly construed, our analysis clearly shows that the perceived reality of victimization is reliably associated with fear, coping activity, and constrained behavior. These findings are far from trivial.

Our results make clear that women are frequently disadvantaged when interacting with uncivil male strangers in public contexts. These findings stand in stark contrast with other more prominently studied life domains—in particular, work and family—where women have made large (if now stagnating and often threatened) status gains and where we have seen more opportunities for gender equality. We would like to emphasize two important implications of these results that ought to be examined further in future theoretical and empirical work. First, our findings suggest that in certain contexts, certain social attributes become more important. Here, anonymity among strangers makes it more likely that people will typify (Schutz 1973) each other and themselves by the most broadly salient and easily visible social categories. Our results hint that one implication of this information and complexity reduction is an amplified power asymmetry between men and women—something that cuts against the longstanding trend in urban studies to see the anonymity of cities and their visual logic as leveling forces (Benjamin 1983). Looking to environments where strangers can be typified along multiple or less-gendered lines (for example, if they are in wearing a uniform, gender might not be the “master status” that primes typification) might yield divergent or more complex results of considerable interest.

Second, our results showed that being female has measurable, gender-specific costs: in public, men enjoy a privileged position and on average are less likely to be negatively affected by rude behavior. By contrast, our findings on coping and avoidance behaviors indicate that women appear to be keenly aware of their subordinate and vulnerable status in these contexts, and rude encounters exact measurable costs in terms of subsequent emotional well-being and use of public places. Note that these implications hold whether women are more frequently subjected to rude behavior or if women are more sensitive to rude behavior, or both. The implication for research here is that more qualitative work is needed to understand women’s routine decision making and route planning, as well as the cognitive work that goes into making sense of unpleasant social encounters. We need to understand how this gender-specific tax is subjectively understood and, particularly, why it results in reduced civic participation for some women more than others. Survey research is hermeneutically thin, and we can only go so far in addressing such matters.

Practice Implications

So what is to be done? Our article has focused on identifying a problem; it has not been an effort in applied sociology. Yet we feel it is incumbent to discuss policy implications. A number of recent initiatives are relevant to our findings and could improve outcomes for women in the short-term. For example, campaigns for a more polite society, be they targeted at cities such as Portland and Beijing, or at specific institutions such as the Tulane and California Polytechnic college campuses (all four the subject of recent “civility initiatives”), might be placed higher on the list of feminist priorities. Our findings demonstrate that perceived subjective harms and reduced levels of citizenship eventuate for all types of women from rude and inconsiderate behavior. If such policies work, our study suggests they will improve the experiences of both women and men, but even more so for women. They have the further advantage of being gender-neutral in their branding and so may avoid backlash.

Policies that provide women with privileged or exclusive access to certain public places also come to mind. We think here, for example, of the “women’s nights” and the recent provision of special women-only transit buses, both in Bogota, Colombia (El Tiempo 2014). Intended to protect against sexual harassment, such policies have the collateral but unrecognized advantage of providing an opportunity to enjoy an altogether more civilized public space. Although our research shows that women can be uncivil to other women, our findings show that the most harmful and frequent patterns of rude stranger encounters involve men. Even if they fall short of guaranteeing risk-free civic participation, even if they somewhat unfairly tar all men with the same brush, such initiatives can draw attention to and perhaps briefly challenge what Gill Valentine (1989, p. 385) famously termed the “spatial expression of patriarchy.”

However, we concede that these types of civility campaigns are clearly limited in their capacity to address these problems in a lasting manner. Culturally, they run the risk of defining women as a vulnerable group in need of quarantine and protection. Transient gains in freedom might come at the cost of reinforcing entrenched ideology. Additionally, we argue that the amplified responses by women to rude behavior by men that we documented are not solely about incivility per se; rather, they reflect, darkly, the broader patterns of gender-based harassment and other forms of violence that reinforce female inferiority. To put it bluntly, women’s heightened sensitivity to rude behavior is almost certainly linked to fear of physical attacks or deliberate humiliation from male strangers. As we noted earlier, women are likely to be aware that a seemingly trivial act of incivility can escalate into a severe offense. If women are sensitive to their disadvantaged position in public, and suspect that men may be targeting them as a result of this disadvantage and perceived vulnerability, then incivility suddenly becomes much more menacing. Even general incivilities can become insulting or frightening if one suspects that she is being targeted because of her inferior position in the gender hierarchy—a supposition consistent with the documented “bundled” quality of general and sexual incivility as experienced by women (Lim and Cortina 2005). Thus, though civility campaigns may help women in the short-term, our findings underscore the far-reaching and complex consequences of gender inequality more broadly. They point towards the pressing need to simultaneously and comprehensively reduce gender-based harassment and other violence against women.

Conclusion

We began our paper by noting that relative to street (sexual) harassment, the impacts of broader forms of everyday incivility on women have been neglected by scholarship. Although our topic is novel, the findings from our research told a distressingly familiar story. When compared to men, women were more likely both to experience unwelcome encounters in public and to experience fear. In addition, these events had substantial lingering impacts, as indicated by coping and avoidance behaviors. We conclude that the entire spectrum of public antisocial behavior disproportionately taxes women. Our hope is that in recognizing this fact, both research and policy will be able to develop new approaches to ensure that women have truly equal access to public space.