SEXUAL HARASSMENT IN PUBLIC SPACES: ANALYSIS OF NEW SURVEY DATA FROM PESHAWAR, PAKISTAN • Pakistan Gender and Social Inclusion Platform • Poverty and Equity Global Practice NOVEMBER 2020 This Note analyses sexual harassment in public spaces as reported by women in Peshawar city in the Khyber Pakhtunkhwa province of Pakistan and is part of the World Bank's multimethod Women in the Workforce study. The data have been collected as part of the World Bank’s Peshawar Urban Household Survey (PUHS). Earlier qualitative research by the World Bank’s Pakistan Gender Plat- form team revealed that sexual harassment is pervasive, resulting in a strong impact on women’s agency, ability to engage in society, and decision making (https://openknowledge.worldbank.org/ handle/10986/30197). It also limits their access to education and employment opportunities. Quanti- tative data on sexual harassment in public spaces are limited, and this Note is a first step toward mea- surement of this phenomenon. About 32 percent of respondents said that they had experienced at least one form of sexual harassment. The analysis shows that the threat of sexual harassment and per- ceptions around its prevalence and impact guide mobility and other day-to-day decisions for women and girls (potentially deterring participation in education and the labor force and curtailing access to health services). Most women and men are not aware of Pakistan’s laws against sexual harassment. Introduction how sexual harassment is associated with other as- pects of women’s lives in Peshawar. Where relevant, we supplement survey analysis with findings from This Note shares findings from a multipurpose qualitative research on sexual harassment con- household survey that also provides estimates ducted in Peshawar by the Pakistan Gender and of sexual harassment in Peshawar city, Pakistan. Social Inclusion Platform, in collaboration with local The data presented here are part of the Peshawar consulting firms, in 2018–2019. In the rest of this Urban Household Survey (PUHS), a multipurpose Note, we use the term “our survey” to refer to sexual household survey conducted by the Pakistan Gen- harassment, gender, and other relevant modules der and Social Inclusion Platform and the Poverty from the PUHS. and Equity Global Practice of the World Bank. Fol- lowing appropriate informed consent procedures, Sexual harassment can be defined as “any unwel- fieldwork took place between February and April come sexual advance, request for sexual favor, 2018. The sample was drawn from 2,357 house- verbal or physical conduct or gesture of a sexual holds covering a total of 21,506 individuals. Data nature, or any other behaviour of a sexual nature were collected separately from 5,870 working-age that might reasonably be expected or be per- men and 5,629 working-age women 15 to 64 years ceived to cause offence or humiliation to another.”1 of age (see Appendix 1 for more details). In this UN Women notes that, like other forms of gender-​ Note, we present reported prevalence of sexual based violence, sexual harassment of women in harassment in public spaces, measurement chal- public spaces is a human rights violation based lenges and methods to overcome harassment, and on gender discrimination and rooted in unequal 1 UN Women, “Prohibition of Discrimination, Harassment, Including Sexual Harassment, and Abuse of Authority,” 2013, https://tinyurl.com/y98agw3z. 1 power relations and a male-dominated culture.2 The sexual harassment module was embedded This is certainly the case in much of South Asia, in a broader module on gender, additional con- including Pakistan, where sexual harassment of sent was taken for the sexual harassment section, women in public spaces arises from an underly- and enumerators were specially trained to ask ing patriarchal structure that defines the role of questions on sexual harassment, including how women to be subordinate to men such that men to switch to less sensitive sections if privacy was are in control of women’s life choices in public and compromised. To reach all women, each house- private spheres.3 hold was visited multiple times, contributing to the rigor of the data collected. Quantitative data—whether national or regional— measuring sexual harassment are scant. As with Sexual harassment must be measured rigor- other forms of gender-based violence (GBV), col- ously because of its pervasiveness, its impact lecting data on sexual harassment is challenging on women, and the very lack of rigorous, quan- with respect to defining “abuse”: selecting an ap- titative data about this type of gender-based vi- propriate sample, training enumerators effectively, olence. Recent multicountry research shows that a and ensuring an ethical implementation of data majority of women and girls across the world have collection. Even the most rigorous quantitative es- faced sexual harassment in public spaces. Estimates timate of the prevalence of sexual harassment is of sexual harassment are as high as 86 percent of likely to be an underestimate, as women are likely to women in Thailand and in Brazil, according to a be reluctant to report such abuse for fear of shame, recent poll conducted across the globe.5 Women blame, and other consequences detrimental to and girls in South Asia also face high risks of sexual them.4 While data collection and measurement of harassment. Such harassment can be experienced other common forms of GBV, such as intimate part- on the street, on public transport,6 in offices, and ner violence and child marriage, have nonetheless almost anywhere that women and girls are out in been surveyed frequently and across the world, public spaces. Box 1 highlights some recent esti- sexual harassment has received less attention. This mates from across the region. Note starts to fill this gap. Studies also suggest that the experience and To contribute to addressing this data gap, and fear of sexual harassment can negatively affect to ensure that the most accurate data were col- women’s lives, including their labor force par- lected in an ethical manner, all women in the ticipation7 and access to education. Women who sample households were interviewed in rele- do venture outside the home try to avoid harass- vant modules of the PUHS. A key aspect of the ment by adopting an appearance that cleaves to PUHS that distinguishes it from other, more typical, the patriarchal notion of how women should dress community-based multipurpose surveys is that all and behave, especially in public, such as covering modules that sought information on women’s own part or all of their body. Those who experience ha- outcomes were fielded to all women over the age rassment are frequently unlikely to report such in- of 15 in sample households. This included labor cidences to police or to families for fear of further force participation, maternal and child health, gen- restrictions on their movement.8 Survivors of sexual der norms and inequality, and sexual harassment. harassment may also face mental trauma.9 2 UN Women, Towards an End to Sexual Harassment: The Urgency and Nature of Change in the Era of #MeToo. (New York: UN Women, 2018). 3 Jennifer L. Solotaroff and Rohini Prabha Pande, Violence against Women and Girls: Lessons from South Asia (Washington, DC: World Bank, 2014). 4 M. Ellsberg, L. Heise, and World Health Organization, “Researching Violence against Women: A Practical Guide for Researchers and Activists” (Geneva: World Health Organization: 2005), https://tinyurl.com/yxs7vfvx. 5 Actionaid, “Three in Four Women Experience Harassment and Violence in UK and Global Cities,” 2016, https://www.actionaid.org.uk/latest-news/three-in-four-women-experience-harassment-and-violence. 6 Huma Daha, Assessing Female Mobility in Lahore: A Step towards Sustainability (Washington, DC: World Bank Group, 2014). 7 T. Chakraborty, A. Mukherjee, S. R. Rachapalli, and S. Saha,” Stigma of Sexual Violence and Women’s Decision to Work,” World Development 103 (2018): 226–38. 8 Bilal Ahmed, Farhan Yousaf Navid, and Umm-e-Rubab Asif. “Combating Street Harassment: A Challenge for Pakistan,” Women & Criminal Justice, doi: 10.1080/08974454.2019.1644697. 9 Muhammad Nasir and Aisha Shoukat, A Study to Investigate the Physical and Psychological Effects of Sexual Harassment on Working Women in Kallur Kot, International Journal of Educational, Science and Research 3, no. 3 (August 2013), 21–34. 2 BOX 1: ESTIMATES OF SEXUAL HARASSMENT IN PUBLIC SPACES IN SOUTH ASIA ƒ In Afghanistan, 90 percent of 376 women surveyed in seven provinces in 2015 reported having been harassed in public spaces, 87 percent in workplaces, and 91 percent in educational institutions; the survey was conducted in educational institutions, workplaces, and public spaces, and among survivors.10 ƒ In Bangladesh, in a national random survey carried out by the Bangladesh Bureau of Statistics in 2015, supported by the United Nations Population Fund, almost one-third of women mentioned the workplace as the most likely location for sexual violence.11 ƒ In Delhi, India, nine out of ten women in a representative household survey reported having experienced at least one form of sexual violence in public spaces, with comments, sexual jokes, whistling, leering, or obscene gestures being the most common (88 percent). More than half (53 percent) said their breasts or buttocks had been touched or groped, while 32 percent reported having been stalked.12 ƒ In Nepal, an online survey garnered responses from 1,000 women across the country, of which 98 percent of surveyed women reported having been harassed in public spaces, and 71 percent reported being harassed in transportation.13 ƒ In Pakistan, in an online survey administered to female university and college students in the cities of Islamabad, Lahore, and Rawalpindi, only 2.8 percent of respondents surveyed about experiences of nonverbal, verbal, or physical sexual ha- rassment said they never experienced any of them.14 Underlying Context: The Situation Pashtun cultural and religious norms strongly influence the situation of women in Peshawar of Women in Peshawar City city. A key aspect of these norms is that the notion of family honor is strongly tied to women’s move- The study site, Peshawar city, lies in Peshawar ment and bodies, and household decision makers district, the largest district by population in the restrict women to home or to nearby areas to pro- province of Khyber Pakhtunkhwa (KP) in Paki- tect them against sexual harassment and maintain stan. Peshawar is the sixth most populous city15 family honor.18 Men are supposed to control and in Pakistan and the provincial capital of KP with an safeguard women, in particular women’s sexual- estimated population of 1.97 million.16 Ethnically, ity. Any woman leaving her house without purdah the population is mainly Pashtun with a significant (or covering herself) is considered to be risking representation of Afghan refugees. Peshawar, and her sexual purity and thereby potentially violate KP more broadly, have suffered from violent conflict the honor of the household. Male members of a and acts of terror for many years.17 household go to great lengths to avoid such a sit- uation via their control of female household mem- bers. Men and women both are socialized in these 10 Women and Children Legal Research Foundation, Research on Sexual Harassment against Women in Public Places, Workplace and Educational Institutions of Afghanistan. (Kabul: WCLRF, 2015), https://tinyurl.com/y9wg9ba9. 11 Bangladesh Bureau of Statistics, 2015 Report on Bangladesh Violence Against Women Survey (Dhaka: BBS, 2016), https://asiapacific.unfpa.org/sites/default/files/pub-pdf/Bangladesh_VAW_survey_report_2015_compressed.pdf. 12 Nandita Bhatla, Pranita Achyut, Sancheeta Ghosh, Abhishek Gautam, and Ravi Verma, Safe Cities Free from Violence against Women and Girls: Baseline findings from the “Safe Cities Delhi Programme.” (New Delhi: UN Women and ICRW, 2013), https://tinyurl.com/y64nk9yh. 13 Code for Nepal, “#IWalkFreely—A Nepali Girl’s Dream,” 2016, https://codefornepal.org/2016/10/iwalkfreely-nepal/. 14 F. Anwar, Karin Osterman, and Kaj Bjorkqvist, “Three Types of Sexual Harassment of Females in Public Places in Pakistan.” Journal of Contemporary Medicine 9, no. 1 (2019): 65–73, doi: 10.16899/gopctd.468324. 15 Finance Division, Government of Pakistan 2018. “Pakistan Economic Survey. 2017-2018: Population, Labor Force and Employment.” Chapter 12, Government of Pakistan. https://www.finance.gov.pk/survey/chapters_18/12-Population.pdf 16 Pakistan Bureau of Statistics (PBS) 2017. Population and Housing Census, KP District 2017. Government of Pakistan, Ministry of Statistics. https://www.pbs.gov.pk/sites/default/files//population_census/KP%20District%20Wise.pdf 17 “City Population: Peshawar,” www.citypopulation.de, https://tinyurl.com/ybkvbsl6. 18 Arab Naz, Umar Daraz, Waseem Khan, Mohammad Hussain, and Qaisar Khan, “The Dormancy of Empowerment: An Analytical Study of Various Impediments to Women’s Education in Khyber Pakhtunkhwa Province of Pakistan” (June 12, 2012), available at SSRN: https://ssrn.com/abstract=2082988 or http://dx.doi.org/10.2139/ssrn.2082988. 3 norms.19 Though these norms exist among Pashtun in all of urban Pakistan (27.8 percent), surveyed by communities broadly, in Pakistan they tend to be the DHS. At the other end of the education spec- more pronounced in KP and Balochistan provinces trum, only 18 percent of girls and women surveyed than in other provinces of the country, perhaps be- by the PUHS complete or go beyond secondary cause of the higher concentration of Pashto popu- education; the PUHS estimate for men and boys in lations in these two provinces relative to other areas Peshawar is twice as high, at 44 percent completing of the country. secondary school or higher. However, women and girls in Peshawar have only slightly lower rates of Indicators for violence against women reflect secondary and higher schooling than do women of these norms. According to the latest Demographic the same age groups in the country as a whole (19.1 and Health Survey (DHS) conducted in Pakistan, 52 percent), surveyed by the DHS.21 percent of female respondents had experienced spousal violence in KP, a higher prevalence than in Peshawari women’s labor force participation is any other province except for Federally Adminis- one-fifth that of Peshawari men. The PUHS also tered Tribal Areas (FATA). Violence against women finds that only 15.5 percent of women in Peshawar is often justified under the guise of punishing vi- in this age group are in the labor force compared to olations of the existing norms of honor or of ex- over three-quarters (74.3 percent) of Peshawari men pected feminine behavior, and households prefer surveyed by the PUHS (see Appendix 2 for all sam- to keep incidents within the family as a private mat- ple characteristics). We cannot compare labor force ter. Two-thirds (63 percent) of women in KP agree participation rates between the PUHS and other sur- that a husband is justified in beating his wife for at veys because of methodological differences.22 least one of the following reasons: burning food, arguing with her husband, leaving the house with- The PUHS also shows that women and girls in out her husband’s permission, neglecting children Peshawar have very limited agency, in terms of or in-laws, or refusing sexual intercourse with her age at marriage, mobility, and decision making. husband.20 Girls marry early, with 50 percent of female respon- dents to our survey having been married by age 18, compared to 29 percent for Peshawar City Survey Results: Pakistan as a whole. Mobility 84% is highly restricted. Women Gender and Sexual Harassment who participated in the of women are always in the 2019 Peshawar Urban PUHS reported spending an accompanied when average of 23.1 out of 24 leaving home Household Survey hours per day inside the house, compared to an average of 13.7 hours for men. Women reported having gone outside the Women and girls in Peshawar city are more house for less than two days in the past week, and likely to have never attended school, and less in most cases, they went to visit family, friends, or likely to have completed secondary and higher neighbors. Most women who go out are accompa- levels of schooling, than are boys and men in Pe- nied by someone (84.1 percent). Even children are shawar, or women and girls in Pakistan overall. acceptable chaperones: among those who are al- The PUHS finds that almost 45 percent of women ways or sometimes accompanied, almost one-quar- and girls in Peshawar city between 15 and 64 years ter (21.5 percent) said they were accompanied by of age have never attended school, compared to children. Almost all women also observe some form 27 percent of men surveyed. Women and girls of purdah (94.3 percent) when they go outside the surveyed in the PUHS are also much more likely to house, with 81 percent covering their faces and have never attended school than are women in the their whole body. Finally, women have little say over same age group in all of Pakistan (35 percent) and the main decisions in their lives. Only 3 percent of 19 Qaisar Khan, Nighat Sultana, and Arab Naz. “The linguistic representation of gender identities in Pakhtu proverbs.” NUML Journal of Critical Inquiry 13, no.2 (2015): 73–87. 20 National Institute of Population Studies (NIPS) [Pakistan] and ICF, Pakistan Demographic and Health Survey 2017–18 (Islamabad, Pakistan, and Rockville, Maryland: NIPS and ICF, 2019). 21 For data on urban Pakistan, see NIPS and ICF, Pakistan Demographic and Health Survey 2017–18. 22 Methodological differences between the PUHS and other, standardized labor force surveys are explained in a different upcoming Note. 4 TABLE 1: REPORTED PREVALENCE OF SEXUAL HARASSMENT EVER EXPERIENCED (PERCENT) Inappropriate staring or comments 28.08 Gestures or actions of a sexual nature 11.03 Inappropriate use of your phone number, email, SMS, social networking (digital harassment) 7.52 Any type of sexual harassment 31.95 Note: Women could respond in the affirmative to having experienced more than one type of sexual harassment. N = 5,411. Responses are weighted. women reported having any say in whom they doing so. On the other hand, two-thirds of women marry, and a minority of women decided them- said they feel safe walking alone in their neigh- selves or with a spouse whether they can start or borhood at any time and 78 percent of those who continue an education (20.9 percent), seek profes- go outside the house using a mode of transport sional medical care (36.6 percent), or work outside said they feel that their chosen mode of transport the house for pay (12.9 percent). is safe at all times. These feelings of greater safety may be because they are within their neighbor- One in three women in Peshawar city reported hood and can choose what form of transport to having faced sexual harassment in public take, and thus these women are able to exercise spaces. About 32 percent of respondents said that some control over their external environment and they had experienced at least one form of sexual its perceived and actual risks. Outside their neigh- harassment about which the survey enquired. (Ap- borhood or in a public space in general, this may pendix 1 provides more detail on the survey meth- be harder, and women are likely to feel less safe. odology). Of the types of sexual harassment, inap- propriate staring or comments were reported most This fear of sexual harassment is likely, at least frequently, followed by sexual actions or gestures, in part, to fuel the preference that employed and digital harassment (Table 1). Survey results women work from home. Among both women are consistent with qualitative research conducted and men who feel that there are only certain con- last year by the World Bank’s Pakistan Gender and ditions under which women should be allowed to Social Inclusion Platform in Peshawar, in which work for pay, the majority (59.4 percent of men and women talked about the pervasiveness of sexual 80.5 percent of women) said that working from harassment in public spaces. This prevalence rate home or close to home is one such condition. While is lower than that reported in other studies in South women’s household responsibilities may be part of Asia (see Box 1). A key reason for this is likely the the reason for this preference, given Peshawar’s pa- extremely low mobility and access to public spaces triarchal norms described above, it is likely that the among Peshawari women, as we demonstrate later fear of exposure to sexual harassment is a potent in this Note, which lowers their exposure to the driver as well. possibility of sexual harassment in public spaces. In addition, several of the studies listed in Box 1 Most women are unaware of Pakistan’s laws (like those from Pakistan and from Afghanistan) against sexual harassment. Pakistan has several did not survey women in the community but, legal mechanisms in place to address sexual ha- rather, women already in public spaces such as ed- rassment. Section 509 of the Pakistan Penal Code ucational institutions or workplaces, which would (1860) establishes sexual harassment as a criminal also contribute to a higher prevalence reported in offence. In addition, in 2010 those studies. the Protection against Ha- 94% rassment of Women at the The fear of sexual harassment is also potent Workplace Act was enacted. of women are not and pervades women’s experiences and expec- However, only 6 percent of aware of the law on tations of safety in public spaces. For example, women in the survey said sexual harassment only one-third of survey respondents said they they knew about the 2010 feel safe walking around outside their neighbor- Act. The vast majority (92 percent) said they had not hood at any time. When women leave the house, heard of the law, while the remaining 2 percent said three-quarters of those who said they use a toilet they were unsure or had heard about it but did not outside of their house said they do not feel safe know what the law was about. 5 Factors Associated with Sexual ual harassment than those without such access. Similar Harassment in Public Spaces23 results were found for access to a cellphone, though of marginal statistical significance (Table 2). Descriptive statistics suggest that education, em- ployment, access to the outside world, and the Women’s knowledge of the law is not statisti- extent of unfettered mobility are all significantly cally significant after taking into account the re- associated with sexual harassment.24 A higher lationship of other socioeconomic characteristics percentage of women with post-primary levels of in multivariate analysis. In descriptive analysis, 45 education, who were employed and worked outside percent of respondents who knew about Pakistan’s the home, and who had access to the outside world law on sexual harassment said they had experienced via mobile phone and internet reported having been sexual harassment compared to less than one-third sexually harassed compared to women who had of those who were unaware of the law. This likely never gone to school, were outside of the labor force reflects that women who know about the law may or worked from home, and had little access to the be more likely to report harassment, rather than that outside world. Similarly, those who had unfettered women who know about the law are more likely mobility—or limited purdah and not always accom- to actually experience harassment, compared to panied when outside—were more likely to report women who are unaware of the law. Once a range having been sexually harassed in public spaces than of socioeconomic characteristics such as education were women who were always accompanied when and access to the internet and other means of com- they left their home, or who covered head and face, munication are taken into account, however, wom- or covered head, face, and body (Figure 1). en’s knowledge of the law was not significantly asso- ciated with reported sexual harassment. Logistic Multivariate Regression on Odds of Sexual Harassment in Public Spaces25 Employed women are significantly more likely to report having been harassed than women out Multivariate analysis confirms the association of of the labor force, multivariate analysis shows, reported sexual harassment with post-primary lev- though place of employment is no longer statis- els of education and access to the outside world tically significant. Employed women were 42 per- via cell phone and internet. Women who had gone cent more likely to report having been sexually ha- beyond primary schooling were almost 50 percent rassed in public spaces compared to women out of more likely to have said they experienced sexual ha- labor force, after controlling for other factors. Being rassment than those with no schooling. Those who unemployed did not have a statistically significant had regular access to or used the internet were 29 per- relationship with sexual harassment. While employ- cent more likely to say they had ever experienced sex- ment outside the home seemed to be associated FIGURE 1: FACTORS ASSOCIATED WITH SEXUAL HARASSMENT 60 40 Percent of Respondents 20 0 Secondary plus Primary None Employed Unemployed Out Yes No Yes No Sometimes Always None Head only Head+chest Total Schooling LFP Mobile phone Internet Accompanied Purdah 23 Because of small sample sizes, descriptive and multivariate analysis is conducted on all types of sexual harassment combined. 24 See Appendix 3 for the full descriptive statistics. 25 See Appendix 3 for the full multivariate results. 6 TABLE 2: SELECT RESULTS FROM MULTIVARIATE LOGISTIC REGRESSION Odds ratio Standard error p-value Education, employment, and communication access Attended/attending post-primary schooling or more (ref = no education) 1.46 0.181 0.002 Has access to cellphone (ref = no) 1.25 0.146 0.052 Has access to internet (ref = no) 1.29 0.156 0.033 Employed (ref = out of labor force) 1.42 0.225 0.027 Unemployed (ref = out of labor force) 1.64 0.517 0.113 Mobility Always accompanied when out (ref = no) 0.68 0.091 0.004 Cover head and bosom (ref = cover head, face, and body) 0.76 0.120 0.084 Cover head only (ref = cover head, face, and body) 0.90 0.194 0.611 No purdah (ref = cover head, face, and body) 1.69 0.595 0.139 Respondent knows about the anti-sexual harassment law 1.14 0.245 0.553 Total observations 5,406 Source: Authors’ analysis. Note: Full results presented in Appendix 4. Outcome: Reported having experienced sexual harassment in at least one way; pseudo-R² = 0.0533. with higher levels of reported sexual harassment sexual harassment than are single women. This re- than employment inside the home in descriptive sult likely reflects both exposure and confidence. analysis, once other related factors—such as mo- Under the strictures of Pashtunwali and, in fact, un- bility—are taken into account, the place of employ- der most of the patriarchal systems in Pakistan and ment is not statistically significant, while the fact of elsewhere in South Asia, getting married while still being employed remains important.26 a virgin is strongly associated with a family’s honor. Consequently, single women are likely to face much Multivariate analysis suggests that some aspects more rigid constraints on their behavior and access of mobility constraints may be more important to the outside world than those who are promised factors in reported sexual harassment than others. in marriage or who are currently or have been mar- After controlling for other factors, women who were ried. This, in turn, limits single women’s exposure always accompanied when they went outside were to public spaces and thus to harassment in these about a third less likely to say they experienced sex- spaces. Single women, whose honor and that of ual harassment than those who were never or were their families is tied closely to their virginity, may only sometimes accompanied. In contrast, descriptive also be more reluctant to report sexual harassment analysis suggested that the use of public transporta- than are women who are already married (although tion, and of no purdah, was associated with higher even married women have to continue to be careful reported experience of sexual harassment; however, of and safeguard their sexual honor). transport and type of purdah were no longer statisti- cally significant in the multivariate analysis, once over- The larger reported prevalence for women with all mobility (number of days outside the home) and higher education, labor force participation, expo- ability to be unaccompanied were taken into account. sure to communications, and mobility is likely a combination of reporting and prevalence. Women Marital status is also significantly associated with higher education, who participated in the with reported sexual harassment. Multivariate labor force, who have greater access to mobile analysis shows that women in any marital status phone and/or internet, and who have relatively except single are significantly more likely to report unfettered mobility may belong to less conserva- 26 Since there was no statistically significant difference in the regression coefficients between employment inside or outside the home, in the final model we combined the two to present one single category of “employed.” 7 tive families, given the overall patriarchal context to be harassed. Another 7 percent opined that they and low female schooling levels in Peshawar. Thus, are “good” women, that is, they observe all social they may be more aware of, and more willing to and gender norms expected of them. Specifically, report, any sexual harassment they may have faced women who gave this response opined that women when out in public spaces or via their communi- who don’t observe purdah, who dress up when they cation devices. They also have higher exposure go out, who are “unclean” or “bad,” or who don’t to the possibility of sexual harassment as a result lead a “pure” life are the ones who are harassed, of spending more time outside the home. In con- thus illustrating that they have internalized patriar- trast, women in more conservative households chal norms of “good” feminine behavior. Only a mi- that control and limit their mobility and communi- nority said they do not pay attention or that they cation to the outside world have less exposure to respond and challenge a harasser (Figure 2). These the possibility of—and thus may report lower levels findings echo prior qualitative research with women of—sexual harassment in public. Those from more in the city. As one young woman told us, “In my uni- conservative households may also be less likely to versity days, a man followed me in his car to my home. acknowledge to an enumerator any experience of When I tried to run away, he grabbed my dupatta sexual harassment even when privacy and confi- (scarf) in front of my neighbor, but my neighbor didn’t dentiality are ensured, because of the strong fear utter a single word to him. I dropped all my books.… of negative consequences should their experience I never walk alone now.” Fear of being blamed and of harassment become known to their family. The chastised by the family and community if they are fear of such consequences is likely a strong factor sexually harassed also pressures women to avoid influencing all respondents’ willingness to acknowl- public spaces. A working woman from Peshawar edge sexual harassment in a survey, given the con- described her experience as “We leave home [to sequences they may face because they live in the earn] for [our] children but are blamed that she has patriarchal system of the study area. We next turn become a bad woman. Our children are told that to these consequences. ‘Your mother is a bad woman. She is all dressed up while leaving home.’” Consequences of Sexual Harassment for Women FIGURE 2: HOW WOMEN CONSTRAIN THEIR OWN BEHAVIOR TO AVOID BEING SEXUALLY HARASSED The fear of sexual harassment and of being blamed and chastised by family for suffering 7.6 sexual harassment can pressure women to with- 8.1 draw themselves from public spaces and, when outside the home, do all they can to avoid being noticed or harassed. About one-quarter of the re- 20.1 64.3 spondents in our survey who said they had never experienced sexual harassment qualified their an- swers with an explanation of why they thought they had been able to avoid being harassed. Two- 64% thirds of these responses Shun public spaces Cover/look away/not attractive Don‘t notice/would respond Am a “good” woman of women who reported no reflected actions to with- draw from public spaces. Sample is based on women who reported that they had harassment explained that not experienced sexual harassment Respondents said that they shun public spaces they had never been sexually harassed be- Family members also control women’s move- cause they either never or rarely go out, never go ment or forbid women from going outside of out alone, are explicitly forbidden from going out- the house to prevent sexual harassment, thus side the house, or when they go out, they only go limiting their opportunities for school, work, or close by. Another 20 percent said they always cover other activities outside of the house. Activities themselves when they go out, look down and do related to the outside world—such as education and not look at anyone around them lest they be con- employment—are often restricted because a spouse strued as inviting unwanted attention, dress shab- or family forbid it. Among girls who reported never bily, or think they are not attractive or young enough having attended an educational institution, 30 per- 8 cent said the reason was lack of permission from a sexual nature, had reported these incidences to elders or other men in the house. Among women anyone. More had reported online or phone ha- who worked for pay from within the home, close to rassment, but unfortunately our data do not allow 46 percent said that they do so because they are us to determine why this kind of harassment was not allowed to leave home. Given the high value of more likely to be reported than other kinds. honor and its link with women’s sexual safety in the study area, it seems likely that the fear of sexual ha- Most women also do not report sexual harass- rassment—and thus the knowledge that a woman ment to anyone else because they feel no ac- was exposed to sexual harassment—is an important tion would be taken. Of all cases of harassment reason to refuse women permission to engage in documented in the survey, only 14 percent were activities outside of the home. Our qualitative re- reported either to authorities or to family. Among search is consistent with this hypothesis. When women who did not report harassment of any type asked why she does not work for pay outside the to anyone, the most frequently cited reason given home, one respondent said “…my husband does was “why bother.” Women felt that telling anyone not agree to me leaving home and going to work.… would not change the fact that men harass women He doesn’t like the outside and would not lessen their chances of being ha- environment”; the local rassed. Data from the few women who reported 81% terminology used (ma- either staring or sexual gestures show that the of women cover their haul bura hai) refers to the most common consequence of reporting was that threat of sexual harass- nothing changed, nor whole body and face ment for women outside was any action taken. More than half of the 115 before leaving home of the home. Another Once again, these women who said they told said, more clearly: “Family data echo themes hasn’t given me permission [to work] … [because from prior qualitative someone about their experi- the] environment outside is not good.… You know research in Peshawar, ences of visual sexual harass- about the harassment issues these days, bad things in which women said ment said no action was taken are being done.… Men stare at us when we walk there was no point in and nothing changed, and 45 past, they are watching our every move.” reporting incidents percent of the 52 women who of sexual harassment, reported sexual gestures said FIGURE 3: PERCENT OF WOMEN EXPERIENCING whether on the street or at work, because that no action was taken. SEXUAL HARASSMENT WHO REPORTED IT TO ANYONE it would not change anything for them. “This is men’s nature, which is never going to change,” said one 35-year-old woman. 54.3 91.1 89.0 Women also hesitate to report or discuss sexual harassment within the family because of the fear of being blamed by family or society, or of be- 45.7 ing forbidden by family to go outside for school, jobs, or other outside activities. Respondents 8.9 10.6 expressed fear that their family would no longer Staring/comments Sexual gestures Phone/email allow them to work outside the home, fear that they would be blamed for the incident, and fear Yes No that—when experienced in a workplace—bringing a complaint of sexual harassment to the authori- These dynamics can constrain reporting of sex- ties would mean dismissal for the women them- ual harassment, and women often do not report selves. Women interviewed in prior focus group sexual harassment to anyone. The vast majority of discussions echoed this concern. For instance, those who said they had experienced sexual ha- when asked why she didn’t tell her family about the rassment of any of the three types about which the catcalling she regularly faced on her way to and survey enquired had not reported the harassment from her educational institution, a focus group re- to anyone (Figure 3). Only about 10 percent of spondent said, “If I had told them [the family], there those who said they had experienced inappropri- would have been a huge fight.… I would have been ate staring or comments, or actions and gestures of asked to stay at home.” 9 TABLE 3: ADDRESSING CHALLENGES IN COLLECTING SEXUAL HARASSMENT DATA Challenge Measures taken Collecting data on sexual harassment as part of a large survey: Strong data quality protocols: Experts have long noted the pitfalls of collecting data on gender-based ü Frequent audits were employed to identify issues in real time and violence in multipurpose surveys, most notably the possibility of conduct refresher training as needed. underestimating the prevalence if a survey is not focused on, and the ü In addition, regular weekly meetings with field staff and the World enumerators not trained for, collecting data on the very sensitive issue of Bank research team helped to resolve issues and gauge progress in GBV.27 PUHS enumerators reported that it was often a challenge to get real time as well. full responses from all women in the household, partly because male household members could get suspicious of the long time it took for women to be interviewed for the overall survey, and partly because of respondent fatigue from a long survey. Enumerators were not specialized in conducting surveys on Specific, culturally sensitive training on gender and GBV, with gender-based violence: As the PUHS was a multipurpose survey, the follow-up: enumerators and other field staff were not experts in collecting data on ü Two of the six days of training for enumerators, other field staff, and GBV, including on sexual harassment in public spaces. senior project staff were devoted to understanding gender inequality, including sexual harassment and other forms of violence against women and girls, both in Pakistan overall and in Peshawar specifically. ü Where appropriate, sessions were held separately for female and male enumerators in accordance with local cultural norms so that fe- male enumerators (who would collect the data on sexual harassment) could discuss any concerns openly. ü Training was conducted in an interactive way so as to engage field staff unfamiliar with the issues. ü Privacy and confidentiality were stressed. ü Repeat training in the field was conducted as deemed necessary. Strong patriarchal norms: In most large surveys, violence against Training included discussion of patriarchy and what it means for data women is known to be underestimated because of the fear, stigma, collection: and norms surrounding the acceptance of such violence. In the case of ü Interactive sessions held with both male and female trainees dis- sexual harassment in public spaces in a conservative setting, where such cussed the structure of society and the difficulties that structure may harassment is emblematic of violating foundational cultural norms of pose for women to acknowledge sexual harassment. honor and thus can trigger familial retaliation, getting reliable, complete ü All women were interviewed by female enumerators fluent in the estimates is perhaps even more challenging. Fear of being blamed and local language and cultural norms. fear that their mobility and opportunities would be further restricted ü Women who had consented to the survey were asked again for made respondents even more reluctant to report instances of sexual consent before the sexual harassment module was implemented, so harassment. as to increase their comfort level. Maintaining privacy and confidentiality: The average household size All efforts were taken to maintain privacy and confidentiality, as this in Peshawar city is 7.5 individuals. As a consequence, it was very difficult was considered nonnegotiable in collecting data on sexual harass- to find a private space in which to conduct interviews. The strict control ment: of women’s access to outsiders made establishing privacy and confi- ü Each woman over age 15 in sampled households was separately and dentiality even more difficult. Often enumerators had to get permission individually interviewed in a private, confidential space. from multiple male family members to speak with a woman. Even after ü All female enumerators were accompanied by a male enumerator. male family members agreed, and the respondent herself consented, on While the female enumerator interviewed the respondent, her male occasion men in the household—including teenage boys—would show partner was responsible for engaging the men in the household to hostility and suspicion to the women being interviewed, or other women allay suspicion and interference. would want to hear what was being discussed. ü Any female supervisors accompanying the female enumerators would keep other women in the household at bay, allowing for maximum privacy for the respondent. ü Enumerators were trained to ask the questions about sexual harass- ment in the times of greatest privacy and to switch to other, more benign questions if someone entered the area in which the interview was being conducted. Respondents were told at the outset (in privacy) that this may occur. ü Female enumerator training emphasized the critical need to maintain confidentiality. 27 Ellsberg et al., “Researching Violence against Women,” https://tinyurl.com/yxs7vfvx. 10 TABLE 3: ADDRESSING CHALLENGES IN COLLECTING SEXUAL HARASSMENT DATA Challenge Measures taken Providing referral information for respondent: The lack of privacy and Use of a creative referral mechanism: association of harassment with honor made any admission of sexual ü Study investigators, advised by local staff, decided not to give women harassment more dangerous for women in this study, and thus made direct information about shelters or other resources to address violence providing referral information for sexual harassment equally dangerous. against women for fear of increasing respondents’ risk of violence. ü In each household, enumerators gave each of the women a business card with the name of the female supervisor for the study and gave the men a card with the name of the male supervisor. Men were told by male enumerators that they could call the number on the card if they remembered any further information they wanted to share, and that their wives had been given similar cards with the number of the survey’s female supervisor. Women were told by female enumera- tors—in private and after the sexual harassment module was imple- mented—that they could call the number on the card if they wanted help regarding the sexual harassment they had reported. Female respondents were also informed that the men in the household had been given a similar card, and that they had been told that women’s cards were for them to add any information that may have been missed. As far as the survey team is aware, this process prevented female respondents from experiencing any negative repercussions. Measures to Address Challenges in Rather, enumerators discovered that interviewed Data Collection women tended to give their cards to the men in their household, and in a few cases, men called the Experiences garnered in conducting this survey female supervisors to harass or berate them. Fur- offer important lessons for collecting data on ther, women may not have had access to a phone sexual harassment in public spaces, especially in to call the female supervisors even if they kept the very conservative settings. There is limited expe- referral card. Pakistan has the highest gender gap in rience in the global South on collecting quantitative mobile phone ownership in the world, with women survey data on sexual harassment. Thus, the chal- being 38 percent less likely to own a mobile phone lenges in collecting these data in Peshawar city, and than men;28 in urban Khyber Pakhtunkhwa, this dif- the measures taken to address these challenges, ference is close to 50 percent.29 Clearly, referral offer instructive information for future collection of mechanisms need to be changed in future exercises data on sexual harassment (Table 3). and contexts. Being aware of potential pitfalls and addressing Implications for Future Research them prior to fielding the survey as well as in real time during survey implementation allowed us and Intervention to get data that were as good quality as possible in the study context. In addition to the measures Additional stand-alone, mixed-methods re- described above, World Bank study staff were espe- search is urgently needed to estimate the prev- cially vigilant about the data being collected on sex- alence of sexual harassment faced by women ual harassment, paying particular attention to issues in Pakistan. Our qualitative research and studies that may arise when conducting field visits. The one from other parts of South Asia have established that mitigation measure that did not work as envisaged women and girls across the region are at high risk was the referral system put in place. The team did for some form of sexual harassment in any public not receive any calls from female respondents. space, including the street, transportation, schools, 28 Oliver Rowntree and Matthew Shanahan, “Connected Women: The Mobile Gender Gap Report 2020,” (GSMA, 2020), https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2020/05/GSMA-The-Mobile-Gender-Gap-Report-2020.pdf. 29 NIPS and ICF, Pakistan Demographic and Health Survey 2017–18, Tables 15.8.1 and 15.8.2. 11 and workplaces. Yet, quantitative estimates of the sexual harassment: radio, television, and (where prevalence of this kind of pervasive gender-based women have access to the internet) social media violence are scarce for Pakistan. Our experience are easily accessible conduits for such aware- also suggests, however, that embedding ques- ness raising. tions about sexual harassment in a larger survey is ü Raise awareness of digital and cyber harassment, likely to yield underestimates. Thus, stand-alone, which are relatively new forms of harassment mixed-methods studies are essential to bring to and may sometimes not be recognized as such. light this long-standing type of violence that ham- pers women’s engagement in public spaces. Improve safety in public spaces. Women’s lack of safety in public spaces is well recognized globally. Stand-alone experimental and behavioral re- Urban environments in particular are typically not search is also crucial to better understand how structured to be gender-inclusive, and, in fact, pub- to address sexual harassment and provide re- lic environments that are gender biased can exac- ferrals to survivors. Our experience with refer- erbate gender inequalities and prevent women and rals suggests that behavioral experiments could girls from contributing fully to a society’s economic help shed light on how best to structure referral and social growth. Thus safety in public spaces mechanisms to survivors in highly conservative and a broader view to urban planning that makes contexts such as Peshawar where providing referral public spaces more women-friendly are critical.31 itself can increase a woman’s risk of other types of There are increasing examples of interventions to gender-based violence, specially domestic abuse. make urban spaces more inclusive and to increase There are also huge gaps in rigorous intervention safety for women and girls in public spaces and in research to try and understand what can be done transport, including in Pakistani cities (see Appen- to address sexual harassment in public spaces that dix 3). Improving public safety—and the perception need to be addressed.30 of public safety—needs to be an urgent, explicit priority to address experiences and fear of sexual There are immediate policy and legal actions harassment and to enable women to use public that can begin to address sexual harassment in spaces more freely and fully. Pakistan. Pakistan has a very strong law against sexual harassment in the workplace with a paral- Address the digital gender gap so that women lel institutional structure to ensure that actions are have more options to report sexual harassment. taken. This and other laws related to sexual harass- Women’s lack of ownership of mobile phones can ment can be strengthened with measures such as: hamper efforts in an increasingly digital age to pro- ü Ensure implementation of the law: while insti- vide information and raise awareness on sexual ha- tutional structures are in place, our qualitative rassment among women, as well as limit women’s research showed that these are often not func- access to recourse, resources, and assistance. In tional or have not gained women’s trust. Thus, part, this digital gender gap arises from the same women are reluctant to report, even when they patriarchal norms that underlie sexual harassment, know the law. especially efforts to control women and their access ü Sensitize those in law enforcement agencies who to information: almost one-third of women in Paki- deal with sexual harassment complaints to en- stan in a recent survey cited family disapproval as gage appropriately with women who report such the main reason for not using mobile internet, com- harassment. pared to 2 percent of men.32 ü Increase women’s awareness of the institutional structure they can use to report harassment, In the medium and longer term, underlying gen- starting from provincial ombudspersons to work- der-unequal societal norms have to be tackled. place committees, and also counseling and legal The underlying patriarchal structure that perpetuates services available for women who have suffered and normalizes harassment while punishing women 30 Jennifer L. Solotaroff, and Rohini Prabha Pande, Violence against Women and Girls: Lessons from South Asia, South Asia Development Forum (Washington, DC: World Bank Group, 2014), https://openknowledge.worldbank.org/handle/10986/20153. 31 Terraza, Horacio, Maria Beatriz Orlando, Carina Lakovits, Vanessa Lopes Janik, and Anna Kalashyan, Handbook for Gender- Inclusive Urban Planning and Design (Washington, DC: World Bank, 2020), https://openknowledge.worldbank.org/handle/10986/33197. 32 Rowntree and Shanahan, “Connected Women: The Mobile Gender Gap Report 2020,” https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2020/05/GSMA-The-Mobile-Gender-Gap-Report-2020.pdf. 12 for suffering harassment and making it difficult for tunities they are missing because of the constraints women to learn about or report harassment needs that current norms and fear of harassment place to be addressed for sustainable change. While this on them, and are looking for change. Respondents is a daunting task indeed, especially in more conser- overwhelmingly said that they want a different world vative areas such as our study site, there are multiple with less harassment and more equality of oppor- global examples of how to address social norms that tunity for their daughters. It is incumbent on policy- can be adapted to the study area. Our qualitative re- makers and practitioners to use the many tools at our search shows that women are aware of their status disposal to move in that direction. as second-class citizens, know full well the oppor- This Note was co-authored by Rohini Prabha Pande, Sundas Liaqat, Noor Rahman, and Saman Amir from the Gender and Social Inclusion Platform for Pakistan. Maria Beatriz Orlando, Uzma Quresh, Tanya D’Lima, and Silvia Redaelli provided guidance and valuable comments. The data are from the Peshawar Urban Household Survey, a collaborative study between the Platform and the Poverty Global Practice (GP). Data were collected by the survey firm RCONS. Survey teams were trained on gender by the Center of Gender and Policy Studies (CGaPs). Ghazala Mansuri, Silvia Raedelli, Maria Qazi, Shiraz Hassan, and Fauzia Ali composed the team from the Poverty GP. For additional information, please contact: Maria Beatriz Orlando, morlando@worldbank.org and Uzma Quresh, uquresh@worldbank.org (co-Task Team Leaders of the Pakistan Gender and Social Inclusion Platform). 13 APPENDIX 1: METHODOLOGY The Peshawar Urban Household Survey (PUHS), uncomfortable or do not want to answer any ques- from which the data for this Note were drawn, was a tions, please let me know, and we will move to the multipurpose household survey designed to get a next question. Now, may I begin?” Enumerators re- statistically representative sample in order to study ceived extra training (including role-play) on asking the welfare of the population in Peshawar. The questions in this module, as well as on skipping the survey questionnaire included a range of themes, module and coming back to it in case the interview such as water and sanitation, urban poverty, labor was interrupted. market participation and economic empowerment, women’s status and gender inequality (including, The sexual harassment module was structured as but not limited to, sexual harassment), domestic a series of questions that captured information and international migration, as well as individual about different kinds of harassment. The first ques- aspirations. tions asked about whether respondents had ever experienced different kinds of sexual harassment, Fieldwork took place between February and April grouped into three: (a) inappropriate staring or 2018. Informed consent was taken from each re- comments; (b) gestures or actions of a sexual na- spondent. Data were collected on paper, via sep- ture; and (c) digital harassment, defined as inap- arate male and female questionnaires. The sample propriate use of the respondent’s phone number, was drawn at the household level, with a planned SMS, email, or social networking platform. For re- sample size of 2,400 households. The actual sam- spondents who responded in the affirmative to any ple was drawn from 2,357 households covering of these questions, follow-up questions were asked a total of 21,506 individuals. Data were collected about who (if anyone) they had told about their ex- separately from 5,870 working-age men and 5,629 perience, or if they had not told anyone, why they working-age women 15 to 64 years of age. had not done so. Four questions were included about women’s knowledge of Pakistan’s laws The module on sexual harassment included an against sexual harassment and, for those working in additional layer of consent over and above the formal institutions, about their knowledge and trust consent for participation in the survey overall. Enu- of the sexual harassment committees mandated by merators were trained to read the following text to law at their workplace. Finally, the enumerator was every eligible respondent, that is, all women 15–64 trained to report whether she was able to maintain years of age in participating households: “Now full privacy and whether, in her view, the respon- I would like to ask about your daily experience in dent felt visibly uncomfortable while answering places that you visit outside your home. This could questions in this module. be a workplace, market, school, college, etc. I want to remind you that everything you tell me will be The sexual harassment module had a response rate strictly confidential, that means no one will know of 96.1 percent (5,411 out of 5,629 women ages you have given me this information. If you are at all 15–64 responded). 14 APPENDIX 2: SAMPLE CHARACTERISTICS Sample characteristics Percentage Sample size Female respondents, ages 15–64, who answered harassment module 100 5,411 Afghan 7 373 Education (highest grade attained) Never attended school 44.52 2,409 Primary (grades 0–5) 15.01 812 Lower secondary (grades 6–10) 20.95 1,133 Higher secondary (grades 11, 12) 7.79 421 Lower tertiary (bachelor’s degree) 10.63 575 Higher tertiary (master’s degree, PhD) 1.09 59 Marital status Single 26.17 1,416 Married 65.13 3,524 Divorced/widowed/separated 4.31 233 Nikah/engaged 4.40 238 Relationship with head of household Head of household 0.50 27 Wife 40.40 2,186 Daughter 23.47 1,270 Daughter-in-law 17.37 940 Others 18.26 988 Labor force participation Out of labor force including training 83.83 4,532 Employed 14.59 789 Unemployed and discouraged 1.57 85 Access to ICT Use or have access to a mobile phone 67.30 3,644 Use or have access to internet 25.60 1,384 Mobility Always accompanied by someone 84.20 4,554 Transport Did not go out last week 34.60 1,875 Public transport 3.70 200 Taxi/rickshaw/carpool 35.50 1,921 Own/employer provided 11.10 603 Walking 14.90 811 15 Sample characteristics Percentage Sample size Level of purdah None 1.80 100 Cover head only 6.05 327 Cover head and bosom 11.50 623 Cover whole body and face 80.50 4,359 Harassment Have experienced some form of harassment 31.90 1,728 Know about law on sexual harassment 6.10 334 Feel safe In neighborhood: Yes 65.40 3,543 Outside neighborhood: Yes 36.50 1,980 Interview conditions Able to achieve comfort and privacy 23.40 1,268 Summaries Mean Sample size Age 32.00 5,411 Household size 9.80 5,411 Number of days went out last week 1.70 5,411 Income of women in household 3004.80 5,411 Source: Peshawar Urban Household Survey. 16 APPENDIX 3: DESCRIPTIVE ANALYSIS OF FACTORS ASSOCIATED WITH REPORTED SEXUAL HARASSMENT Sample: Women ages 15–64 Reported sexual harassment Percentage N Afghan 25 94 Pakistani 32 1,634 Education Never attended school 27.07 652 Primary attended (grades 0–5) 25.80 207 Post-primary attended (grades 6+) 39.54 869 Age 15–18 27.70 236 19–24 33.00 348 25–44 35.60 840 45–64 26.40 304 Marital Status Single 29.50 418 Married 32.30 1,140 Divorced/widowed/separated 32.80 77 Nikah/engaged 39.20 93 Number of Children 0 children 38.60 293 1 to 2 32.80 480 3 to 5 31.90 686 6 plus 25.70 269 Labor force participation Out of labor force including training 29.80 1,353 Employed 42.50 335 Unemployed and discouraged 43.40 37 Access to ICT Use or have access to a mobile phone 35.45 1,292 Don’t have access to mobile phone 24.72 437 Use or have access to internet 41.62 576 Don’t have access to internet 28.62 1,152 Mobility Always accompanied by someone when going out 29.50 1,344 Not always accompanied when going out 44.80 384 17 Sample: Women ages 15–64 Reported sexual harassment Percentage N Transport Did not go out last week 28.50 534 Public transport 39.90 80 Taxi/rickshaw/carpool 33.60 647 Own/employer provided 34.10 205 Walking 32.20 261 Level of purdah None 55.20 56 Cover head only 37.10 122 Cover head and bosom 32.40 203 Cover whole body and face 30.90 1,349 Harassment Know about law on sexual harassment 45.20 151 Don’t know about law on sexual harassment 31.10 1,577 Interview conditions Comfort and privacy obtained 25.50 324 Comfort and privacy were not obtained 33.90 1,404 Feel safe In neighborhood: Yes 33.92 1,202 In neighborhood: No 28.20 527 Outside neighborhood: Yes 36.70 727 Outside neighborhood: No 29.20 1,002 Summaries Mean Sample size Age 31.60 1457 Household size 8.80 1457 Number of days went out last week 2.00 1457 Income of women in household 3,558.50 1457 Source: Peshawar Urban Household Survey. 18 APPENDIX 4: MULTIVARIATE LOGISTIC ANALYSIS OF FACTORS ASSOCIATED WITH REPORTED SEXUAL HARASSMENT Odds ratio Standard error p-value Respondent’s age 0.99 0.005 0.004 Level of schooling (ref = none) Primary attended/attending 0.92 0.139 0.595 Post-primary attended/attending 1.46 0.181 0.002 Marital status (ref = single) Married 1.58 0.216 0.001 Divorced/widowed/separated 1.78 0.489 0.035 Nikah/engaged 1.52 0.338 0.059 Access to media Have access to a cellphone (ref = no) 1.25 0.146 0.052 Have access to the internet (ref = no) 1.29 0.156 0.033 Labor force participation (ref = none) Employed 1.42 0.225 0.027 Unemployed (including discouraged) 1.64 0.517 0.113 Respondent’s own income 1.01 0.016 0.594 Mobility Number of days went out last week 1.03 0.031 0.295 Always accompanied when out (ref = sometimes or never) 0.68 0.091 0.004 Feel safe within the neighborhood when outside of home (ref = no) 0.86 0.103 0.207 Feel safe outside the neighborhood when outside of home (ref = no) 1.23 0.142 0.077 Transport (ref = never goes out of the house) Public 1.36 0.337 0.218 Taxi/rickshaw/carpool 1.10 0.145 0.489 Own/employer provided 0.79 0.153 0.215 Walking 0.96 0.166 0.805 Level of purdah when outside (ref = full body) Cover head and bosom 0.76 0.120 0.084 Cover head only 0.90 0.194 0.611 None 1.69 0.595 0.139 Other covariates Household size 0.96 0.010 0.000 Know about law against sexual harassment (ref = no) 1.14 0.245 0.553 Comfort and privacy were achieved during interview 0.79 0.098 0.052 Constant 0.72 0.201 0.235 Pseudo-R² 0.0533 Note: Outcome is “Reported having experienced sexual harassment in at least one way.” 19