639581

research-article2016
JIVXXX10.1177/0886260516639581Journal of Interpersonal ViolencePopkin et al

Article
Journal of Interpersonal Violence
2019, Vol. 34(1) 27­–49
Coercive Sexual © The Author(s) 2016
Article reuse guidelines:
Environments: sagepub.com/journals-permissions
DOI: 10.1177/0886260516639581
https://doi.org/10.1177/0886260516639581
Development and journals.sagepub.com/home/jiv

Validation of a Scale

Susan Popkin, PhD,1 Chantal Hailey, BA,2

Janine Zweig, PhD,1 Nan Astone, PhD,1
Reed Jordan, BA,1 Leah Gordon, MPH3,
and Jay Silverman, PhD3

Abstract
In this article, we present the results of our efforts to develop and test
a scale to operationalize and measure a neighborhood-level indicator of
coercive sexual environments (CSEs), a construct emerging from our
earlier work on safety and sexual threats among young girls living in
chronically disadvantaged neighborhoods. Data for this study come from
a survey of 124 adult and 79 youth respondents living in public housing
in Washington, D.C., and participating in the Housing Opportunities and
Services Together Demonstration, a multisite project testing the feasibility
and effectiveness of place-based, dual-generation case management models
to improve outcomes for vulnerable families. Our psychometric analysis
indicates that the CSE scales we developed for adults and youth have high
internal consistency. Together with our analyses of construct validity, the
present findings suggest that CSE is a unitary construct that may be an
important factor to include in models of neighborhood processes and risk.

Keywords
community violence, sexual assault, sexual harassment, violence exposure

1Urban Institute, Washington, DC, USA

2New York University, New York City, NY, USA
3University of California San Diego School of Medicine, CA, USA

Corresponding Author:
Reed Jordan, Urban Institute, 2100 M St NW, Washington, DC 20037, USA.
Email: [email protected]
28 Journal of Interpersonal Violence 34(1)

In this study, we describe the construction of two scales designed to measure

perceptions of and exposure to a coercive sexual environment (CSE). The
CSE construct emerged from our earlier work on safety and sexual threats
among young girls living in chronically disadvantaged neighborhoods
(Popkin, Leventhal, Weismann, 2010). In the present analysis, we present a
theoretical framework for the construct, assessments of the scales’ psychomet-
ric properties and initial tests of construct validity.

Background
A long history of de facto and de jure segregation and systemic underinvest-
ment in social services, infrastructure, and economic development in the
United States has created a myriad of hypersegregated communities with
concentrated poverty (Massey & Denton, 1993; Rothstein, 2015; Turner,
Popkin, & Rawlings, 2009; Wilson, 1987). These sociopolitical factors have
also led to these neighborhoods being mired in chronic disadvantage and suf-
fering a range of social ills, including high rates of social disorder and chronic
violence (Fox & Benson, 2006; Hannon, 2005; Kawachi, Kennedy, &
Wilkinson, 1999; Sampson, 2012; Sampson, Raudenbush, & Earls, 1997).
Neighborhood poverty attenuates residential stability, crippling connections
between neighbors, community attachment and, in turn, social cohesion
(Sampson, 2012; Shaw & MacKay, 1969). In addition, residents’ precarious
economic conditions and strained relationships with social institutions, like
public service and police departments, foster a sense of “perceived power-
lessness” to avoid harm and enact informal social control (Ross, Mirowsky,
& Pribesh, 2001; Sampson, 2012). In other words, economic segregation and
community violence in these neighborhoods both stems from and helps to
perpetuate low levels of collective efficacy, or “social cohesion among neigh-
bors combined with their willingness to intervene on behalf of the common
good” (Sampson, 2012; Sampson et al., 1997, p. 918). Although Sampson
et al.’s (1997) original study focused on the negative association between
collective efficacy and community violence, the effects of collective efficacy
also span to relationship violence. Research has shown neighborhood collec-
tive efficacy can reduce both intimate homicide rates and nonlethal partner
violence (Browning, 2002).
Focusing on neighborhood risk does not undercut individual’s ability to
cope within and have resiliency despite those situations. Yet, despite indi-
vidual resilience, there is a great deal of evidence that concentrated pov-
erty and disadvantage pose well-established risk factors to youth:
developmental and cognitive delays; poor physical and mental health; and
the likelihood of dropping out of school, engaging in risky sexual
Popkin et al 29

behavior, and becoming involved in delinquent and criminal activities; and

social mobility (Brooks-Gunn & Duncan, 1997; Chetty & Hendren, 2015;
Ellen & Turner, 1997; Leventhal & Brooks-Gunn, 2004; Sampson, 2012;
Sampson, Morenoff, & Gannon-Rowley, 2002; Sampson, Sharkey, &
Raudenbush, 2008; Wodtke, Harding, & Elwert, 2011). Prior research also
supports the idea that girls and boys experience the effects of chronic dis-
advantage in very different ways, especially as they enter adolescence. In
the 1990s, Anderson argued that young men in inner-city neighborhoods
felt pressured to act tough to maintain respect, following the “code of the
street,” and girls gained status and respect through getting pregnant
(Anderson, 1999). In a more recent example, one study of African
American youth growing up in high-crime communities found that young
men focus on maintaining respect and avoiding the risk of gun violence,
whereas young women focus on the fear of being the object of predatory
behavior (Cobbina, Miller, & Brunson, 2008). In her graphic portrayal of
life for African American girls living in low-income, urban neighbor-
hoods, Miller (2008) noted that other people often believe that the girls are
to blame for their own neighborhood risks because of the way they behave
or dress, rather than understanding the context in which these young girls
live (Miller, 2008).
We have theorized that when disadvantage and violence are prevalent and
collective efficacy is low, societal sexual and gender norms can bolster a
neighborhood characteristic that has differential effects on male and female
youth (Smith et al., 2014). To be specific, some communities develop what
we have termed a CSE, wherein sexual harassment, sexual violence, and
sexual exploitation of women and girls, even very young girls, become part
of everyday life (Popkin, Acs, & Smith, 2010; Popkin, Leventhal, &
Weismann, 2010; Popkin & McDaniel, 2013).
This concept came to light as part of the U.S. Department of Housing and
Urban Development’s (HUD’s) experimental Moving to Opportunity (MTO)
for Fair Housing Demonstration. It found strikingly different outcomes for
adolescent girls and boys whose families received special vouchers to allow
them to move from distressed public housing to lower poverty communities.
Girls in the experimental group fared unexpectedly better in terms of mental
health and engagement in risky behavior (Ludwig et al. 2011; Sanbonmatsu
et al., 2011). This result first appeared at the MTO Interim Evaluation (Orr
et al., 2003); we conducted subsequent qualitative studies to explore this
unexpected finding. That work suggested key differences in how neighbor-
hood safety matters for male and female adolescents, with girls in high-
poverty, high-crime communities also coping with sexual harassment and
fear of sexual violence—in essence, a CSE (Briggs, Popkin, & Goering,
30 Journal of Interpersonal Violence 34(1)

2010; Popkin, Leventhal, & Weismann, 2010). We conducted additional

qualitative studies and used data from the MTO Final Evaluation Survey
(Sanbonmatsu et al., 2011) to explore the relationship between perceptions
of neighborhood violence and disadvantage, reports of unwanted sexual
attention, and mental health outcomes for girls. This research revealed that,
in neighborhoods of concentrated disadvantage, young women live with
chronic fear of sexual harassment and intimate partner violence, including
rape, which has negative consequences for both their behavior and their
mental health (Briggs et al., 2010; Popkin, Leventhal, & Weismann, 2010;
Smith et al., 2014). We hypothesize that relief from these environmental
threats to girls’ sexual safety and the fear related to these threats account for
the female specific-positive effect of moving away from distressed
neighborhoods.
However, the MTO survey data did not allow us to develop a robust mea-
sure of CSE to assess how this may operate relative to other neighborhood
mechanisms like collective efficacy. The MTO survey contained only a
small number of items on women and girls’ perceptions of sexual harass-
ment and unwanted sexual attention; the data were sufficient for us to
develop a hypothesized model, but not to develop a reliable measure that
could be tested in other research. In this article, we present the results of our
efforts to develop and test a scale to operationalize and measure neighbor-
hood-level CSE.

Conceptual Model
The conceptual model of CSE that emerged from our previous work and
guided the development of the measure presented herein is illustrated in
Figure 1 (Smith et al., 2014). Though factors associated with a CSE appear
across a variety of contexts (e.g., college campuses, workplaces), our model
posits that CSE results from not just broader cultural norms around gender-
based abuse and harassment but also combines with high levels of chronic
violence and poverty, low levels of collective efficacy, and residents’ “per-
ceived powerlessness” present in highly distressed communities. In these cir-
cumstances, sexual harassment, coercion, and gender-based violence become
normalized, perhaps undermining the life chances of young women and girls.
CSE does not occur in isolation. Instead, it is born out of broader historic
and ongoing racial segregation, combined with cultural and societal norms
that support gender-based sexual harassment, threats, and violence. It begins
with societal norms that form the way girls and women are perceived and
treated no matter where they live. However, institutions and systems can cre-
ate social and economic isolation of people of color and frame opportunities
Popkin et al 31

Figure 1. Conceptual model.

Note. CSE = coercive sexual environment.
aSegregation and discrimination in federal, state, and local policies generated communities

with concentrated disadvantage.

for low-income children of color. These forces combine in racially segre-

gated neighborhoods of concentrated disadvantage. In these contexts, pov-
erty, violence, and social disorganization take root, allowing CSE to emerge
as a place-based and gender-specific phenomenon. CSE then affects the lives
and well-being of the girls and women who live there.
The legacy of segregation and discrimination in federal, state, and local
policies and practices means that a considerable portion of low-income
African Americans are concentrated in high and medium-poverty neighbor-
hoods (Massey & Denton, 1993; Rothstein, 2015; Turner et al., 2009; Wilson,
1987). Although African Americans and other people of color are less starkly
segregated today than in the past, nearly all American neighborhoods that
have concentrated poverty have populations for which the majority are peo-
ple of color. Patterns of systemic injustice have created and sustained neigh-
borhoods of concentrated disadvantage where high poverty and distress
blight the life chances of those who live there (Sampson, 2012). Distressed,
central-city neighborhoods, like those in which the families in this study live,
are some of the most racially and economically segregated communities in
the nation, where the worst consequences of concentrated disadvantage are
plainly evident—physical decay, a dearth of quality social services like
schools, limited economic opportunities, the absence of even the most basic
32 Journal of Interpersonal Violence 34(1)

amenities, such as grocery stores and laundromats (LaVeist, 2005; LaVeist,

Thorpe, Mance, & John Jackson, 2007; Turner et al., 2009). In turn, many
adults who live in these communities are disconnected from the labor market
and suffer from high rates of physical and mental illness; many of the chil-
dren and youth are in danger of injury and educational failure (Popkin, Acs,
& Smith, 2010; Popkin et al., 2000). These deleterious circumstances com-
bine to create communities with high rates of violent crime, drug trafficking,
drug and alcohol addiction among adults, incarceration, and social disorder
and community members with “perceived powerlessness” and low levels of
collective efficacy (Ross et al., 2001; Sampson, 2012).
Our model also posits that beyond these neighborhoods there exists a pat-
tern of behaviors, on the part of both males and females that broadly support
gender-based abuse and a lack of sanctions against those perpetrating such
abuse. This “culture of gender-based abuse” includes pervasive use of speech
that demeans and sexually threatens women and girls (e.g., speech that claims
social status based on male perpetration of sexual violence, that blames
female sexual victimization on female behavior, and that infers male com-
modification of women’s and girls’ sexuality), lack of speech condemning
perpetration of sexual harassment and violence, and emulation of celebrities
based on their speech and behavior that demeans and threatens women and
girls.
Although Gordon and Riger (1989) demonstrated that all women experi-
ence this female fear to some degree, we hypothesize that women in neigh-
borhoods with high levels of chronic violence and disadvantage are
particularly vulnerable (Popkin, Leventhal, & Weismann, 2010). We argue,
and our qualitative work suggests, that the culture of gender-based abuse—
which exists in many contexts—combines with community violence in these
contexts to create an environment where norms supportive of sexual violence
and coercion emerge because low levels of collective efficacy, a sense of
vulnerability, and fear of verbal or physical retaliation makes it difficult for
adult residents of the neighborhoods who find these norms unacceptable to
challenge them. This is consistent with Sampson’s notions that violence
undermines collective efficacy. Adult residents living in low-income com-
munities often react to their neighborhood-level “culture of gender-based
abuse” by isolating themselves (thus reducing collective efficacy even more)
and restricting their own and their daughters’ mobility in ways that limit the
opportunities for recreation and enrichment (Popkin, Leventhal, & Weismann,
2010).
Some research suggests that in high-poverty communities, nonphysical
sexual harassment is often tolerated because of the prevalence of violence in
general (Dominguez & Menjivar, 2014). Another qualitative study suggests
Popkin et al 33

that high levels of sexual harassment might be associated with drug traffick-
ing, specifically drug dealers hanging out in public spaces and the sexualiza-
tion of women in the drug trade (Cobbina et al., 2008). Burton’s (2014)
ethnographic study, in addition, found that women living in poverty may be
willing to accept certain relational situations they might not otherwise to
maintain their partners’ vital contributions to their and their family’s financial
stability. More generally, research shows that women report greater fear of
harassment in socially isolated, public housing communities with high levels
of disorder (Alvi et al., 2001), as well as in public spaces that are poorly lit or
marked by vandalism (Weatherburn, Matka, & Lind, 1996).
The result of societal “culture of gender-based abuse,” neighborhood
chronic violence and social disorder, and residential “perceived powerless-
ness” is a CSE where there are high levels of sexual harassment, violence,
and exploitation of women. This sexual harassment puts young women at risk
of a range of poor outcomes, including a higher likelihood for experiencing
intimate partner violence and emotional abuse, substance abuse, and delin-
quency (Chiodo et al., 2009). In addition, sexual harassment has negative
consequences for girls’ school performance (Hand & Sanchez, 2000).
Our goal in this article is to develop a robust indicator of the box in white
in Figure 1, neighborhood CSEs. In doing so, we use techniques whereby we
develop procedures to measure neighborhood mechanisms and ecological
settings, and improve upon the accuracy and validity of those measures,
through the integration of tools from psychometrics (Sampson, 2012;
Sampson & Raudenbush, 1999).

Data and Method

Data
Housing Opportunities and Services Together (HOST) demonstration. The data
for this study are from an evaluation of a demonstration project called HOST
(Popkin et al., 2012)). The goals of this project are to use two strategies for
intervention with families living in poverty in public housing located in dis-
advantaged neighborhoods. The two strategies are (a) using public housing as
a platform to provide services and (b) using a two generation approach to
intervening with adults and children from disadvantaged families (Popkin
and McDaniel, 2013; Popkin et al., 2012).
The HOST Demonstration was fully implemented in three public and
assisted housing communities in Chicago; Portland, Oregon; and Washington,
D.C. Each participating housing authority selected target participants from its
list of leaseholders at the target site. Eligibility for the HOST program
34 Journal of Interpersonal Violence 34(1)

required the household to have children, and depending on the site, additional
risk factors such as failure to comply with agency work requirements; an
unemployed head of household, lease violations, or at risk of eviction (Popkin
et al., 2012; Popkin & McDaniel, 2013). The number of households targeted
for HOST varied across sites, from more than 200 in Chicago to approxi-
mately 140 in Portland. In the first year of the demonstration at each site, we
attempted to complete surveys with an adult and target youth in each HOST
household to capture baseline measures for each target household; across the
sites, response rates for adults exceeded 80% and for youth, 90% (Scott et al.,
2013).

Survey data. During the first HOST implementation year, we fielded two sur-
veys—an adult survey and a youth survey—to capture baseline characteris-
tics of HOST families and their communities. The adult survey asked heads
of household about themselves and up to two focal children—one between
the age of 6 and 11, and another between the age of 12 and 18. Parents with a
child in the older age range could then consent for that child to participate in
a separate youth survey. The youth survey asked adolescents between the age
of 12 and 18 about themselves.
Surveys were fielded in three HOST sites—Chicago’s Altgeld Gardens
public housing development and Portland’s New Columbia and Humboldt
Gardens mixed income developments in the summer 2012 and in
Washington, D.C.’s Benning Terrace public housing development in sum-
mer 2013. All Portland and Chicago interviews (adult and youth) and D.C.
adult interviews were conducted onsite in the homes or apartments of
respondents, using Computer-Assisted Telephone Interviewing (CATI).
Due to the sensitive sexual experience questions included in the D.C. youth
survey, we adopted a bimodal method for conducting the youth interview.
This approach entailed a CATI interview supplemented by a hardcopy com-
pletion of the sensitive sexual experiences questions. The survey questions
are drawn from validated measures from national or large-scale surveys
including the National Health Interview Survey, the National Longitudinal
Survey on Youth, the National Longitudinal Study on Adolescent Health
(ADD HEALTH), and the Project on Human Development in Chicago
Neighborhoods.
Our study focuses on adult and youth respondents from the Washington,
D.C. HOST site because the survey measures in D.C. related to CSE bene-
fited from substantial revisions made after it was fielded in Chicago and
Portland. We measure exposure to CSE for youth because youth are more
likely able to appropriately identify sexually exploitive acts rather than gen-
eral perceptions of neighborhood problems related to sexual activity, which
adults are more likely to be able to identify and understand.
Popkin et al 35

Table 1. D.C. HOST Baseline Survey Sample.

Variable Name Adult Survey Youth Survey

Respondents 124 79
Race and gender (%)
African American, non-Hispanic 96.5
Female 96.7 40.5
Age
M (SD) 39.9 (9.6) 15.6 (1.9)
Household income (%)
<US$5,000 37.9
US$5,000-US$14,999 25.8
US$15,000-US$29,999 21.0
≥US$30,000 7.3
Missing 8.1
Worked in past year
% 45.2

Source. HOST D.C. Baseline Survey.

Households were eligible for participating in HOST if they had at least

one youth between the ages of 9 and 18. We attempted to survey all eligible
households, conducting interviews with an adult and one youth in our target
age range. If there was more than one eligible youth in the household, we
selected a focal youth at random. For youth to participate in the survey, both
their responsible adult and the youth themselves needed to participate. Adults
and youth were provided written and informed consent materials. For youth
to participate, both their responsible adult and the youth themselves were
required to consent. Youth and adults were provided US$25 and US$50 gift
cards, respectively, as compensation for participating in the survey
Our response rate for the survey was 81% of the eligible adults in Benning
Terrace and 87% of the eligible youth. We describe the 124 adult and 79 D.C.
youth respondents in Table 1. Like the other residents in Benning, the adult
respondents are African American women (97%) with very low incomes. The
average adult respondent’s age is 40. The youth respondents are all African
American teenagers who are an average 15 years old.

Variables
Scale variables. We developed a set of seven items to be used with adults to
assess perceptions of CSE1 and eight items to be used with youth to assess
exposure to CSE.2 The items are listed in Tables 2 (adult) and 3 (youth).
36 Journal of Interpersonal Violence 34(1)

Table 2. Items Used to Construct Adult CSE Scale.

Introductory Text Item Text Item No.

How much of a problem in is rape or other sexual attacks? 1
your neighborhood:
Big problem are women or girls trading sex for money? 2
Some problem are men or boys making unwanted 3
sexual comments or gestures toward
girls or women?
No problem are men or boys hurting women or girls? 4
How often in the past year: did someone make unwanted sexual 5
comments, jokes, or gestures toward
you?
Happened every day did someone touch, grab, or pinch you in 6
a sexual way that you did not want?
Happened once or twice were you afraid to go places because you 7
a week were worried about being touched,
harassed, hurt in these ways?
Happened a couple of times
each month
Happened a couple of times
in the past year
Never happened

Note. CSE = coercive sexual environment.

Construct validity variables. Table 4 lists the variables from the survey that we
used in our analysis of construct validity. They are measures of adult percep-
tions of levels of violence, neighborhood victimization, neighborhood social
disorder, adult and youth perceptions of neighborhood trust and engagement
with neighbors (collective efficacy), and youth exposure to neighborhood
violence and victimization.

Control variables. In the multivariate regression models for adults, we control

for adult sex, age, union status (married or in a marriage-like relationship or
not), and employment status. In the youth multivariate models, we control for
the same characteristics of the youth’s parent as in the adult models, as well
as youth gender.

Method
We used principal components analysis, using a varimax rotation to identify
subscales in our measures. All factors with eigenvalues greater than one were
Popkin et al 37

Table 3. Items Used to Construct Youth CSE Scale.

Introductory Text Item Text Item No.

During the past 12 months, You saw women or girls who were 1
how often did each of the trading sex or oral sex for money, drugs
following things happen: or other things they wanted.
Never You saw men or boys making unwanted 2
sexual comments or gestures toward
girls or women.
Once You saw men or boys touching, grabbing, 3
or pinching girls in a sexual way that
they didn’t want.
More than once You saw men or boys hitting, pushing, 4
slapping, choking, or otherwise
physically hurting women or girls.
How often in the past year: did someone made unwanted sexual 5
comments, jokes, or gestures toward
you?
Happened every day did someone touched, grabbed, or 6
pinched you in a sexual way you that did
not want?
Happened once or twice did someone spread sexual rumors about 7
a week you?
Happened a couple of times did someone email or texted you sexual 8
each month pictures, photographs, or messages that
you did not want?
Happened a couple of times
in the past year
Never happened

Note. CSE = coercive sexual environment.

considered. Items were retained if they loaded .4 or higher on the factor. We

calculated internal consistency of the items using Cronbach’s alpha, as a mea-
sure of reliability. Regression methods were used to test their construct valid-
ity and relationship to other relevant constructs. When doing so, one looks to
see whether relationships with other constructs are within the expected direc-
tion and the effect sizes of the relationships are moderate to high.

Results
Scale Identification
To identify an indicator of CSE for adults, we subjected the seven items in
Table 2 to a principal components analysis with varimax rotation. Some items
38 Journal of Interpersonal Violence 34(1)

Table 4. Descriptions of Variables for Construct Validity Analysis.

Variable Name Wording or Description Scale Items

Collective Index of whether people in Index ranges from 10
Efficacy Scale neighborhood are willing to help, 1 (respondent
(Adult) share the same values, are close does not agree
knit, can be trusted, and generally or it is unlikely)
get along with each other, and to 4 (respondent
the likelihood that neighbors do strongly agrees or it
something if they saw children is very likely)
skipping school, spray-painting
graffiti, showing disrespect toward
an adult, or if a fight breaks out
in front of their home or the fire
station closest to their homes was
going to be shut down
Social Index of how big of a problem were Index ranges from 1 4
Disorder groups of people hanging out, (respondent believes
Scale (Adult) people selling drugs, people using it is no problem at
drugs, and gangs all) to 3 (respondent
believes it is a big
problem)
Violence Scale Index of how big of a neighborhood Index ranges from 1 2
(Adult) problem are shootings and (respondent believes
violence, and people being attacked it is no problem at
or robbed all) to 3 (respondent
believes it is a big
problem)
Adult perception of neighborhood violence
How big of a neighborhood problem Dummy variable, equals 1
are shootings and violence? 1 when problem is
considered big.
How big of a neighborhood problem Dummy variable, 1
are people being attacked or equals 1 when
robbed? problem is
considered big.
Youth exposure to neighborhood violence
During the past 12 months, how Dummy variable, 1
often did you see someone shoot equals 1 when once
or stab another person? or more.
During the past 12 months, how Dummy variable, 1
often have you heard gun shots? equals 1 when once
or more.
(continued)
Popkin et al 39

Table 4. (continued)
Variable Name Wording or Description Scale Items
Youth neighborhood violence victimization
During the past 12 months, how Dummy variable, 1
often did Someone pull a knife or equals 1 when once
gun on you? or more.
During the past 12 months, how Dummy variable, 1
often did someone shoot you? equals 1 when once
or more.
During the past 12 months, how Dummy variable, 1
often did someone cut or stab you? equals 1 when once
or more.
During the past 12 months, how Dummy variable, 1
often were you jumped? equals 1 when once
or more.
Youth exposure to neighborhood social disorder
During the past 12 months, how Dummy variable, 1
often did you see someone dealing equals 1 when once
drugs out in the open? or more.
During the past 12 months, how often Dummy variable, 1
did you see drug paraphernalia on equals 1 when once
the ground/in public? or more.
During the past 12 months, how Dummy variable, 1
often did you see gang activity equals 1 when once
(graffiti, selling drugs, violence)? or more.
Youth perceptions of neighborhood trust
People in this neighborhood look Dummy variable 1
out for each other. representing true/
false response.
Youth engagement with neighbors
You know most of the people in Dummy variable 1
your neighborhood. representing true/
false response.
In the past month, you have stopped Dummy variable 1
on the street to talk with someone representing true/
who lives in your neighborhood. false response.
Head of What is your/his/her sex? Dummy variable 1
household equals 1 for female.
sex
Head of What is your marital status? Dummy variable 1
household equals 1 for
union status married or living
in a marriage-like
situation
(continued)
40 Journal of Interpersonal Violence 34(1)

Table 4. (continued)
Variable Name Wording or Description Scale Items
Head of Respondent worked in the past 12 Dummy variable 1
household months representing yes/no
employment response.
status
Youth gender Focal child sex Dummy variable, 1
equals 1 for female

were reverse coded in order that high and low scores indicated the same
direction for all variables. The analysis produced two eigenvalues above 1,
which indicates that there that there were two underlying dimensions to the
items. The factor pattern that emerged is summarized in Table 5.
Items 1 through 4 all loaded positive and high on Factor 1, while Items 5
and 6 loaded positive and high on Factor 2. Item 7 did not load high on either
factor. We identified Factor 1 as indicating the extent to which the adult per-
ceived that there was a CSE, which was the measure we were after.
To construct the CSE perception scale, the four items that loaded high on
Factor 1 were coded as 0 if the respondent indicated that the subject of the
item (e.g., rape for Item 1) was not a problem in their neighborhood, 1 if it
was somewhat of a problem and 2 if it was a big problem. The four items
were then summed together to create a scale that ranges from 0 (respondent
does not perceive their neighborhood as having a problem with any actions
mentioned in the items) to 8 (respondent perceives their neighborhood as
having a big problem with all the actions mentioned in the items). The scale
has a Cronbach’s alpha of .86 indicating it is internally reliable. The scale has
a mean of 3.9 and a median of 4 (indicating little skew).
We regard Factor 2 as an indicator of CSE victimization. These two fac-
tors are indicative that perception of the environment characterized as CSE is
distinct from experiencing forms of sexual victimization that are associated
with CSE, which is in accordance with our theoretical framework. We use
this indicator in the construct validity tests we describe below.
To identify an indicator of CSE for youth, we subjected the eight items in
Table 3 to a principal components analysis with varimax rotation. The analy-
sis produced only one eigenvalues above 1 which indicates that there that
there was only one underlying dimension to the items. The factor pattern that
emerged is summarized in Table 6. Items 1 through 4 loaded high onto the
factor. We call the scale made from these items the CSE exposure scale. The
Popkin et al 41

Table 5. Factor Loadings for Adult Items.

Item No. Loadings Loadings for

(From Table 2) Abbreviated Item Text for Factor 1 Factor 2
1 Rape/sexual attacks—how 0.576 −0.238
much of a problem in your
neighborhood?
2 Women/girls trading sex for 0.827 −0.084
money—how much of a
problem in your neighborhood?
3 Men/boys making unwanted 0.816 −0.158
gestures/comments—how
much of a problem in your
neighborhood?
4 Men/boys hurting women/girls— 0.836 −0.03
how much of a problem in your
neighborhood
5 Unwanted sexual comments/ −0.325 0.695
jokes/gestures (reversed)
6 Someone touched/grabbed/ −0.012 0.871
pinched you in a sexual way
(reversed)
7 Past 12 months, how often were −0.259 0.117
you afraid to go places because
you were worried about being
touched/harassed/hurt in these
ways? (reversed)

other items that did not load onto a factor (Items 5 through 8) are victimiza-
tion items and are treated in the same way we treated the CSE victimization
items for adults. Although those items may not represent a unitary construct,
they have face validity as indicators of victimization.
The responses to the CSE exposure items were coded 0 (if the respondent
never saw the CSE action), 1 (if the respondent saw the CSE action once),
and 2 (if the respondent saw the CSE action more than once). The youth CSE
exposure scale is a sum of the responses and ranges from 0 (no exposure to a
CSE in their neighborhood; respondent did not see any CSE actions in the
past year) to 8 (high exposure to a CSE in their neighborhood; respondent
saw all four CSE actions more than once in the past year).
The scale has a Cronbach’s alpha score of .76 indicating internal reliability.
The scale has a mean of 2.0 and a median of 1 (indicating some positive skew).
42 Journal of Interpersonal Violence 34(1)

Table 6. Factor Loading for Youth Items.

Item No. Abbreviated Loadings for Loadings for

(From Table 3) Item Text Factor 1 Factor 2
1 You saw women or girls who 0.27192 0.59174
were trading sex or oral sex
for money, drugs, or other
things they wanted—how
often in your neighborhood?
2 You saw men treating women 0.28587 0.56655
badly (verbally)—how often
in your neighborhood?
3 You saw men treating women 0.25367 0.8895
badly (touching, grabbing,
pinching)—how often in your
neighborhood?
4 You saw men physically 0.15477 0.71911
abusing (hitting, pushing,
slapping) women—how often
in your neighborhood?
5 Someone made unwanted 0.77772 0.22662
sexual comments, jokes, or
gestures toward you?
6 Someone touched, grabbed, or 0.78562 0.23595
pinched you in a sexual way
you that did not want?
7 Someone spread sexual 0.42687 0.1892
rumors about you?
8 Someone email or texted you 0.80234 0.26257
sexual pictures, photographs,
or messages that you did not
want?

Construct Validity
Our theoretical model (Figure 1) posits that a CSE will emerge from the
social characteristics of very disadvantaged neighborhoods—specifically,
high levels of community violence, high levels of social disorder, and low
collective efficacy. These characteristics themselves emerge from conditions
of high residential segregation and concentrated poverty.
To assess the construct validity of the two scales that we have identified,
we examined both bivariate and multivariate associations between the scales
and well-established indicators of community violence, social disorder, and
collective efficacy (listed in Table 4).
Popkin et al 43

Table 7. Correlations Between Neighborhood Characteristics and CSE

Perceptions Among Adults.

Variable Unadjusted Adjusteda

Adult believes shooting and violence are big 0.48*** −0.34
neighborhood problems
Adult believes people being attacked or robbed is a 0.54*** 1.70***
big neighborhood problem
Social Disorder Scale 0.65*** 2.50***
Sampson’s Collective Efficacy −0.39*** −0.76*

Note. CSE = coercive sexual environment.

aFrom an ordinary least squares regression of the CSE Perception Scale on the variable and

adult sex, age, union status, and employment status.

*p < .05. **p < .01. ***p < .001. (two-tailed)

Table 7 contains the results for adults. The table shows that our indicator
of adult perceptions of CSE is positively correlated with their perceptions of
community violence (for beliefs about shootings and violence, r = .48, p <
.001, and for beliefs about being attacked or robbed, r = .54, p < .001) and
social disorder (r = .65, p < .001) and negatively correlated with their percep-
tions of neighborhood collective efficacy (r = −.39, p < .001), as predicted in
our theoretical model. In the multivariate model, which controls for all the
indicators of neighborhood characteristics simultaneously as well as for the
control variables, the significant correlation between perceptions of shooting
and violence disappears but all the other associations hold (p < .05).
The results for youth are in Table 8, which shows the estimate of the asso-
ciation between our indicator of youth exposure to CSE and most of the indi-
cators we have of the youths’ perceptions or experience of community
violence, social disorder, and collective efficacy is significantly different
from zero. More specifically, experience of community violence was posi-
tively related and included the following: seeing someone stab or shoot
another person (r = .25, p < .05), hearing gun shots (r = .35, p < .01), having
a gun or knife pulled on them (r = .22, p < .05), being cut or stabbed in the
past 12 months (r = .30, p < .01), and being jumped in the past 12 months (r
= .25, p < .05). Social disorder was positively related and included: seeing
someone doing drugs (r = .25, p < .05), seeing drug paraphernalia on the
ground (r = .40, p < .001), and seeing gang activity (r = .45, p < .001).
Collective efficacy as measured by people look out for each other was nega-
tively related (r = −.34, p < .01). In the multivariate model, some of these
associations disappear, but the estimate of the association between at least
one indicator each of community violence (cut or stabbed in past 12 months),
social disorder (drug paraphernalia), and collective efficacy (engagement
with neighbors) and CSE is significantly different from zero (p < .05).
44 Journal of Interpersonal Violence 34(1)

Table 8. Correlations Between Neighborhood Characteristics and CSE Exposure

Among Youth.

Variable Unadjusted Adjusteda

Saw someone shoot or stab another person 0.25* −0.02
Heard gun shots 0.35** 0.43
Gun or knife pulled on youth 0.22* 0.89
Cut or stabbed in past 12 months 0.30** 2.91*
Jumped in past 12 months 0.25* 0.99*
Saw someone dealing drugs out in the open 0.25* 0.85*
Saw drug paraphernalia on the ground 0.40*** 1.09*
Saw gang activity 0.45*** 0.94*
People look out for each other −0.34** −1.41*
Knows most people in their neighborhood 0.06 −0.35
Stopped to talk to someone on the street 0.10 1.12*

Note. CSE = coercive sexual environment.

aFrom an ordinary least squares regression of the CSE Perception Scale on the variable and

adult sex, age, union status, and employment status.

*p < .05. **p < .01. ***p < .001. (two-tailed)

Discussion
In sum, past research by the current team suggests that the reason moving
away from neighborhoods of chronic disadvantage proved positive for
young women and neutral for young men in the MTO experiments was that
young women were, thereby, able to escape a CSE which had pernicious
effects on their mental health. This research led us to formulate a theoretical
model that posits a CSE emerges from the high levels of community vio-
lence, social disorder, and low collective efficacy pervasive in neighbor-
hoods of concentrated disadvantage. The current study was focused on
developing an indicator that could be used to measure CSE. Informed by
qualitative findings, we developed a set of items and tested them on young
people and adults who reside in a Washington, D.C., public housing project.
We have three findings.
First, our principal components analysis showed that while CSE and
sexual victimization were related, among both youth and adults, the envi-
ronmental measure is distinct from the measure of victimization. Second,
our psychometric analysis indicated that the CSE scales we developed for
adults and youth have acceptable Cronbach’s alpha values and therefore
are internally consistent. Third, our analyses of construct validity suggest
that CSE fits into our model of neighborhood processes as we
hypothesized.
Popkin et al 45

Despite the potential for this measure to be useful for future research, our
study, as with all research, had limitations. First, though the measure was
developed iteratively by using it in two different housing developments first
and then revising it, the study is based on data from one housing development
in Washington, D.C. Thus, it is not clear whether our results are applicable to
other settings. For instance, it is not yet clear whether CSE is a unitary phe-
nomenon across contexts or whether varying types of CSE might exist across
different neighborhoods and settings. We also cannot speak to geographic or
regional differences in measurement. Second, the sample size of the study is
quite small, though high proportions of those identified for the study actually
participated. Third, the data herein are collected at one time point. As such,
we also cannot speak to two important issues that longitudinal data would
afford: (a) How does the concept of CSE vary across time, and (b) how might
CSE be perceived differently by youth at varying developmental stages?
Future research should examine the utility of this measure using larger sam-
ples, across different contexts, and across time.
Nevertheless, we regard our findings as supporting our hypothesis that there
is an element of social disorder in chronically disadvantaged neighborhoods
that constitutes a CSE which is distinct from, although correlated with, other
indicators of social disorder. This aspect of neighborhood quality can be mea-
sured with a scale that has good psychometric properties. Although the partici-
pants in our research were living in a disadvantaged neighborhood in the United
States, we anticipate that the measure we have developed and tested here will
be relevant to a wide variety of settings such as military bases, communities
with poverty in other countries, and potentially, college campuses. The next
step is for data on the scale we have developed to be collected in other places
with other samples so our findings can be replicated and extended.

Declaration of Conflicting Interests

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

Funding
The author(s) disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article: Funding for the research study came
from an R24 award from the National Institute of Health Minority Health and Health
Disparities, the Kellogg Foundation, and the Open Societies Foundations.

Notes
1. The items for adults were originally developed to be fielded as part of Wesley
Skogan’s (1992) measure of neighborhood social disorder.
46 Journal of Interpersonal Violence 34(1)

2. The items for youth were originally developed to be incorporated into a scale of
exposure to violence for the National Longitudinal Survey of Adolescent Health
(AddHealth).

References
Alvi, S., Schwartz, M. D., DeKeseredy, W. S., & Maume, M. O. (2001). Women’s
fear of crime in Canadian public housing. Violence Against Women, 7(6),
638-661.
Anderson, E. (1999). Code of the street: Decency, violence, and the moral life of the
inner city. New York, NY: W.W. Norton.
Briggs, X. S., Popkin, S. J., & Goering, J. (2010). Moving to opportunity: The story of
an American experiment to fight ghetto poverty. Oxford, UK: Oxford University
Press.
Brooks-Gunn, J., & Duncan, G. J. (1997). The effects of poverty on children. The
Future of Children, 7, 55-71.
Browning, C. R. (2002). The span of collective efficacy: Extending social disorgani-
zation theory to partner violence. Journal of Marriage and Family, 64, 833-850.
Burton, L. M. (2014). Seeking romance in the crosshairs of multiple partner fertility:
Ethnographic insights on low-income urban and rural mothers. The Annals of the
American Academy of Political and Social Science, 654, 185-212.
Chetty, R., & Hendren, N. (2015). The impact of neighborhoods on intergenerational
mobility: Childhood exposure and county-level estimates. Cambridge, MA:
National Bureau of Economic Research.
Chiodo, D., Wolfe, D. A., Crooks, C., Hughes, R., & Jaffe, P. (2009). Impact of sexual
harassment victimization by peers on subsequent adolescent victimization and
adjustment: A longitudinal study. Journal of Adolescent Health, 45(3), 246-252.
Cobbina, J., Miller, J., & Brunson, R. (2008). Gender, neighborhood danger and risk
avoidance strategies among urban African American youth. Criminology, 46,
501-538.
Dominguez, S., & Menjivar, C. (2014). Beyond individual and visible acts of vio-
lence: A framework to examine the lives of women in low-income neighbor-
hoods. Women’s Studies International Forum, 44, 184-195.
Ellen, I. G., & Turner, M. A. (1997). Does neighborhood matter? Assessing recent
evidence. Housing Policy Debate, 8, 833-866.
Fox, G. L., & Benson, M. (2006). Household and neighborhood contexts of intimate
partner violence. Public Health Reports, 121, 419-427.
Gordon, M. T., & Riger, S. (1989). The female fear: The social cost of rape. New
York, NY: Free Press.
Hand, J. Z., & Sanchez, L. (2000). Badgering or bantering? Gender differences in
experience of, and reactions to, sexual harassment among US high school stu-
dents. Gender & Society, 14(6), 718-746.
Hannon, L. E. (2005). Extremely poor neighborhoods and homicide. Social Science
Quarterly, 86, 1418-1434.
Popkin et al 47

Kawachi, I., Kennedy, B. P., & Wilkinson, R. G. (Eds.). (1999). Income inequality
and health (Vol. 1). New York, NY: New Press.
LaVeist, T. A. (2005). Disentangling race and socioeconomic status: A key to solving
health disparities. Journal of Urban Health, 82, 26-34.
LaVeist, T. A., Thorpe, R., Mance, G., & John Jackson, J. (2007). Overcoming con-
founding of race with socioeconomic status and segregation to explore race dis-
parities in smoking. Addiction, 102, 65-70.
Leventhal, T., & Brooks-Gunn, J. (2004). A randomized study of neighborhood effects
on low-income children’s educational outcomes. Developmental Psychology, 40,
488-507.
Ludwig, J., Sanbonmatsu, L., Gennetian, L., Adam, E., Duncan, G., Katz, L., &
Thomas McDade, T. (2011). Neighborhoods, obesity, and diabetes—A random-
ized social experiment. New England Journal of Medicine, 365, 1509-1519.
Massey, D. S., & Denton, N. A. (1993). American apartheid: Segregation and the
making of the underclass. Cambridge, MA: Harvard University Press.
Miller, J. (2008). Getting played: African American girls, urban inequality, and gen-
dered violence. New York, NY: New York University Press.
Orr, L., Feins, J. D., Jacob, R., Beecroft, E., Sanbonmatsu, L., Katz, L. F., & Kling,
J. R. (2003). Moving to opportunity for fair housing demonstration program:
Interim impacts evaluation (Report prepared by Abt Associates Inc. and the
National Bureau of Economic Research for the U.S. Department of Housing and
Urban Development, Office of Policy Development and Research). Washington,
DC: U.S. Department of Housing and Urban Development.
Popkin, S. J., Acs, G., & Smith, R. (2010). Understanding how place matters for kids.
Community Investments, 22, 23-26.
Popkin, S. J., Buron, L. F., Levy, D. K., & Cunningham, M. K. (2000). The gautreaux
legacy: What might mixed-income and dispersal strategies mean for the poorest
public housing tenants? Housing Policy Debate, 11, 911-942.
Popkin, S. J., Leventhal, T., & Weismann, G. (2010). Girls in the hood: How safety
affects the life chances of low-income girls. Urban Affairs Review, 45, 715-774.
Popkin, S. J., & McDaniel, M. K. (2013). HOST: Can public housing be a platform
for change? Washington, DC: Urban Institute.
Popkin, S. J., Scott, M. S., Parilla, J., Falkenburger, E., McDaniel, M., & Kyung, S.
(2012). Planning the housing opportunity and services together demonstration:
Challenges and lessons learned. Washington, DC: Urban Institute.
Ross, C. E., Mirowsky, J., & Pribesh, S. (2001). Powerlessness and the amplifica-
tion of threat: Neighborhood disadvantage, disorder and mistrust. American
Sociological Review, 66, 568-591.
Rothstein, R. (2015). From Ferguson to Baltimore: The fruits of government-spon-
sored segregation. Washington, DC: Economic Policy Institute.
Sampson, R. J. (2012). Great American City: Chicago and the enduring neighbor-
hood effect. Chicago, IL: University of Chicago Press.
Sampson, R. J., Morenoff, J. D., & Gannon-Rowley, T. (2002). Assessing “neighbor-
hood effects”: Social processes and new directions in research. Annual Review of
Sociology, 28, 443-478.
48 Journal of Interpersonal Violence 34(1)

Sampson, R. J., & Raudenbush, S. W. (1999). Ecometrics: Toward a science of

assessing ecological settings, with application to the systematic social observa-
tion of neighborhoods. Sociological Methodology, 20, 1-41.
Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and violent
crime: A multilevel study of collective efficacy. Science, 277, 918-924.
Sampson, R. J., Sharkey, P., & Raudenbush, S. W. (2008). Durable effects of con-
centrated disadvantage on verbal ability among African-American children.
Proceeding of the National Academy of Sciences, 105, 845-852.
Sanbonmatsu, L., Ludwig, J., Katz, L. F., Gennetian, L. A., Duncan, G. J., Kessler, R.
C., & Tessler Lindau, S. (2011). Moving to opportunity for fair housing demon-
stration program—Final impacts evaluation. Washington, DC: U.S. Department
of Housing and Urban Development.
Scott, M., Popkin, S. J., McDaniel, M., Saxena, P., & Jordan, R. (2013). Serving
HOST families: The challenges to overcome. Washington, DC: Urban Institute.
Shaw, C. R., & McKay, H. D. (1969). Juvenile delinquency and urban areas: A study
of rates of delinquency in relation to differential characteristics of local commu-
nities in American cities. Chicago, IL: University of Chicago Press.
Skogan, W. G. (1992). Disorder and decline: Crime and the spiral of decay in
American neighborhoods. New York, NY: Free Press.
Smith, R., Gallagher, M., Popkin, S. J., Mireles, A., & George, T. (2014). Coercive
sexual environments: What MTO tells us about neighborhoods and sexual safety.
Cityscape: A Journal of Policy Development and Research, 16, 85-112.
Turner, M. A., Popkin, S. J., & Rawlings, L. (2009). Public housing and the legacy of
segregation. Washington, DC: Urban Institute.
Weatherburn, D., Matka, E., & Lind, B. (1996). Crime perception and reality:
Public perceptions of the risk of criminal victimisation in Australia. Crime and
Justice Bulletin No. 28. Sydney, Australia: NSW Bureau of Crime Statistics and
Research.
Wilson, W. J. (1987). The truly disadvantaged: The inner city, the underclass, and
public policy. Chicago, IL: University of Chicago Press.
Wodtke, G. T., Harding, D. J., & Elwert, F. (2011). Neighborhood effects in temporal
perspective: The impact of long-term exposure to concentrated disadvantage on
high school graduation. American Sociological Review, 76, 713-736.

Author Biographies
Susan Popkin, PhD, is both director of The Urban Institute’s Program on
Neighborhoods and Youth Development and senior fellow in the Metropolitan
Housing and Communities Policy Center. A nationally recognized expert on public
and assisted housing, her work focuses on the ways neighborhood environments
affect outcomes for youth, and in assessing comprehensive community-based inter-
ventions. She is the coauthor of the award-winning Moving to Opportunity: The Story
of an American Experiment to Fight Ghetto Poverty, lead author of The Hidden War:
Crime and the Tragedy of Public Housing in Chicago, and coauthor of Public Housing
Transformation: The Legacy of Segregation.
Popkin et al 49

Chantal Hailey received her BA in sociology from Howard University and is cur-
rently a PhD sociology student at New York University. Her research interests include
urban sociology, education, neighborhood effects, public housing, adolescents, social
policy, and mixed methods. She will focus her graduate research on how neighbor-
hoods affect children’s educational and behavioral outcomes and how public housing
and educational institutions collectively affect child well-being through “silo-bust-
ing” initiatives. The National Science Foundation Graduate Research Fellowship
Program generously funds her doctoral studies. Prior to pursuing her doctoral degree,
she was a research associate with the Metropolitan Housing & Communities Policy
Center at the Urban Institute.
Janine Zweig is a senior fellow for the Urban Institute’s Justice Policy Center. Her
work addresses issues related to violent victimization, primarily intimate partner and
sexual violence. It includes both applied and basic research with an eye toward con-
tributing useful information to policymakers and practitioners to solve problems and
better assist victims, as well as to the body of knowledge. She has conducted several
studies that have addressed provisions of the Violence Against Women Act and the
Prison Rape Elimination Act.
Nan Astone is a senior fellow in the Center on Labor, Human Services, and Population
at the Urban Institute, which she joined in 2013 after serving 24 years on the faculty
of the Johns Hopkins Bloomberg School of Public Health. She is a demographer with
expertise on reproductive health, the family, adolescence, and the transition to adult-
hood. Astone, a former William T. Grant Scholar, received her PhD from the
University of Chicago.
Reed Jordan is a research associate in the Policy Advisory Group at the Urban Institute.
Since joining Urban in 2012, he has been involved in various national and metropolitan
projects related to affordable housing, place-based initiatives, community development,
and performance management. His research portfolio includes providing technical
assistance to grantees of the U.S. Department of Education’s Promise Neighborhood
program, evaluating the early implementation of the U.S. Department of Housing and
Urban Development’s Choice Neighborhoods initiative, and performing housing
affordability assessments for local governments and national organizations.
Leah Gordon, MPH, currently serves as Oregon Health & Science University’s proj-
ect director for the National External Evaluation of AHRQ’s EvidenceNOW Initiative.
She previously served as a program manager at the University of California, San
Diego, coordinating domestic and international research projects on sex trafficking,
gender-based violence, and HIV risk among vulnerable populations. Her work focuses
on health disparities and mixed-methods approaches to evaluation.
Jay Silverman, PhD, is the director of research for the Center on Gender Equity and
Health and a professor of Medicine and Global Public Health at UCSD. He is a lead-
ing global researcher on understanding and preventing gender-based violence against
adolescent and adult women (e.g., intimate partner violence, sexual violence, sex
trafficking).