Abstract
The commercial sexual exploitation (CSE) of children is a consequential public health and criminal justice problem, but no CSE prevention programs have been evaluated. The Boston-based My Life My Choice (MLMC) program offers a multisession psychoeducation group to girls who are identified as “at-disproportionate-risk” for CSE victimization and trains other agencies throughout the U.S. to offer this curriculum. The curriculum was designed to improve knowledge about the commercial sex industry and shift-related attitudes and behaviors. The current project was a multi-year, multi-site evaluation to assess the effectiveness of the MLMC prevention group. Using a one-group longitudinal design, changes in participant behavior and CSE knowledge were measured at baseline (n = 354), upon group completion (n = 296), and 3 months after group completion (n = 241). The sample was 95% female-identified, 28% Black/African American, 26% White/non-Hispanic, 25% Hispanic/Latina, and 22% other race. The mean age of participants was 15.6 years old. Approximately 28% identified as bisexual, and 10% identified as lesbian, asexual, pansexual, or other. In multivariable-adjusted models, participants reported fewer episodes of sexually explicit behavior at follow up as compared to baseline (relative risk [RR]: 0.52, 95% confidence interval [CI]: 0.37-0.72 at Follow-up 1, and 0.53, 95% CI: 0.35-0.82 at Follow-up 2). Participants were 24% less likely to report dating abuse at Follow-up 2 as compared to baseline (p = .06). In addition, as compared to baseline, participants were 40% more likely to have given help or information about CSE to a friend at Follow-up 2, and participants demonstrated increased knowledge and awareness about CSE and its harms over the follow-up period. Although additional evaluation using a comparison group and long-term follow up would increase confidence that observed changes are attributable to the group instead of other factors, results suggest that the MLMC curriculum may be effective in reducing the risk of CSE and improving other conditions for youth who are at-disproportionate-risk of CSE.
Keywords
Introduction
The commercial sexual exploitation (CSE) of children is a substantial global violence problem. In 2016, an estimated 4.8 million men, women, and children were victims of CSE, and 21% of these (~1 million individuals) were children (International Labor Organization, 2017). It is estimated that approximately 40,000 unique individuals younger than 18 years old were commercially sexually exploited in the Americas in 2016, which includes the United States, Canada, and South America (International Labor Organization, 2017). In the United States, it is challenging to assess the number of people who have been sexually exploited in part because the specific sexual acts that are codified as exploitation varies by state. Whether an adult may be counted as exploited can depend on whether they were engaged in stripping, pornography, and/or prostitution, and in which state. This is not true for minor children. The U.S. Trafficking Victims Protection Act of 2000 (TVPA) specifies that any inducement of commercial sex involving a person who has not attained 18 years of age qualifies as child trafficking (“TVPA,” 2000).
Although the precise number of minors who are commercially sexually exploited in the United States each year remains unknown and difficult to determine, results of self-report survey studies suggest that exploitative conditions are not rare. In 1996, approximately 3.5% of a nationally representative sample of school-attending youth in grades 8 to 12 reported they had exchanged sex for drugs or money at some time in their lives (Edwards, Iritani, & Hallfors, 2006), and in 2001 to 2002, 4.9% reported having done so (Ulloa, Salazar, & Monjaras, 2016). While exchanging sex for drugs or something else of value may not have qualified as CSE in every U.S. state prior to 2016, a change to federal law in 2016 means that all commercial sex transactions involving minors are considered instances of sex trafficking victimization (“Justice for Victims of Human Trafficking,” 2015). Risk factors for child victims of CSE include experiencing child neglect and abuse, involvement with the child welfare system, substance use, poor family support, limited education and employment opportunities, and running away (Fedina, Williamson, & Perdue, 2016; Gibbs, Henninger, Tueller, & Kluckman, 2018; Moynihan et al., 2018; Reid, Baglivio, Piquero, Greenwald, & Epps, 2017). Being lesbian, gay, bisexual, or transgender and homeless (lesbian, gay, bisexual, and transgender [LGBT]) is also a risk factor for CSE of children. The U.S. National Coalition for the Homeless reports that 59% of homeless LGBT youth in the United States experience CSE of children, while 33% of non-LGBT homeless youth experience do (Martinez & Kelle, 2013).
Child victims of CSE experience serious mental and physical consequences including depression, sexually transmitted infections, substance misuse, sexual assault victimization, and post-traumatic stress disorder (Greenbaum, Crawford-Jakubiak, & Committee on Child Abuse and Neglect, 2015; Landers, McGrath, Johnson, Armstrong, & Dollard, 2017; Le, Ryan, Rosenstock, & Goldmann, 2018). But as the report of the Committee on Commercial Sexual Exploitation and Sex Trafficking of Minors in the United States emphasizes: “Efforts to prevent the CSE and sex trafficking of minors in the United States are essential but largely absent” (Institute of Medicine [IOM] and National Research Council [NRC], 2013, p. 5). Importantly, that report, and other experts, has put out an urgent call for research to advance our understanding of the effectiveness of programmatic efforts to prevent CSE (Middleton, Gattis, Frey, & Roe-Sepowitz, 2018; Rothman et al., 2017).
Despite these calls for evidence-based prevention efforts, to our knowledge, there have been no prior evaluations of any CSE prevention strategies. There has been one prior evaluation of a CSE prevention website featuring a video for patients in a family planning clinic (Murphy, Bennett, Eaton, & Kottke, 2012), but “limited change in knowledge and attitudes was observed” (Murphy et al., 2012). There has also been one published description of an Atlanta, GA-based CSE prevention program (Taylor, 2018), and a second description about the creation of a middle school–based prevention program (Kruger et al., 2016), but neither have been evaluated.
My Life My Choice
My Life My Choice (MLMC) is a Boston-based program of the nonprofit Justice Resource Institute. It was founded in 2002 with the mission of combatting CSE through survivor empowerment, training and advocacy, and prevention solutions. The program has a central goal of preventing minors believed to be at-disproportionate-risk for exploitation from becoming exploited. In public health terms, this type of prevention strategy—one that targets individuals at high risk for a condition, rather than members of the general public—is called “secondary prevention” as opposed to primary prevention (U.S. Centers for Disease Control and Prevention. Picture of America, 2017). In service of their secondary prevention-related goal, MLMC created and offers a 10-session class (called a “group” herein) to teach girls who have been identified to be at risk for CSE how to recognize and avoid the recruitment tactics of exploiters. The authors of the MLMC curriculum, a survivor of CSE and a public health social worker, drew on the Health Belief Model when they created the program as well as information from dozens of child CSE survivors about their perceived vulnerabilities, pathways into and out of CSE, and what would have helped them to exit the commercial sex industry. The first group session was held in 2003 at a congregate care (i.e., residential treatment) facility for adolescent girls with severe behavioral and emotional problems.
Participants in the MLMC groups receive information designed to change their knowledge about the commercial sex industry and individuals who sexually exploit others, shift their attitudes about the commercial sex industry (i.e., “being in the life”), acknowledge barriers to making behavioral and safety-oriented changes (e.g., making new friends), become aware of how substance use and other risk factors (e.g., running away) are related to CSE, develop media literacy skills, participate in activities designed to build self-esteem and personal empowerment, build resilience, and assess their own vulnerability to exploitation. Furthermore, participants are encouraged to become peer leaders and share information about these topics with other youth who may be at risk. In addition to the group discussions and activities, MLMC prevention group participants keep journals in which they write about themselves and their evolving understanding of CSE; the journal is shared with the group facilitator, and participants receive direct feedback and encouragement from the adult who personalizes the material discussed in group and creates a private channel for direct communication about possible threats to safety or risky behavior.
All MLMC prevention groups are provided by trained professionals. Most are clinicians (social workers or counselors), although professionals with other training are also permitted to facilitate groups if they have been trained by MLMC. The 2-day training is conducted by MLMC staff and each participant receives a curriculum manual and suggested content for each session of the 10-week group.
One of the noteworthy aspects of the MLMC prevention program is that trained survivors of CSE often co-facilitate the groups. It has been established that peer education, as well as mentorship programs that match youth in need of support with positive role-model adults, can lead to a variety of positive outcomes—including positive social, academic, and health-related behavior (DeWit, DuBois, Erdem, Larose, & Lipman, 2016; Sieving et al., 2017), improved mental health, and sense of cultural identity (Crooks, Exner-Cortens, Burm, Lapointe, & Chiodo, 2017), and fewer behavior problems and less social anxiety (DeWit et al., 2016). For example, peer mentor interventions have been found to be effective for people with severe mental health disorders (O’Connell, Sledge et al., 2018), substance abuse (Dugdale, Elison et al., 2016), patients managing diabetes (Lachance, Kelly et al., 2018), spinal cord injury (Sweet, Michalovic et al., 2018), and to promote physical activity in adolescents (Corder, Brown et al., 2016), among others. The MLMC program does not permit anyone with CSE experience within the past 5 years to facilitate the groups because an unreliable and unstable facilitator, or one who becomes known to youth because she returns to street life, has the potential to make participants feel hurt, betrayed, and disillusioned. MLMC also has clear guidelines for facilitators about barring observers and trainees from sitting in on group sessions, maintaining the privacy of group members, and not using untrained substitutes when they need to miss a class session—these rules are to protect youths’ anonymity, and safeguard them from adults who may be underprepared to interact with them in a way that communicates acceptance and support. MLMC has trained facilitators from more than 30 U.S. states and regions of Canada to provide MLMC prevention groups and have established MLMC groups in congregate care facilities, juvenile detention centers, community-based agencies, child welfare offices, and public and private schools.
This research study was designed to evaluate changes in youth knowledge, attitudes, and selected behaviors over time that may be related to their participation in the MLMC prevention group.
Methods
Study Overview
The study used a one-group longitudinal design and collected self-reported survey data from youth at baseline (before the MLMC program started), at the final group session (~3 months after baseline) and 3 months after the final group session (~6 months after baseline). All procedures were reviewed and approved by the first author’s Institutional Review Board (IRB) and also approved by the IRBs at the Massachusetts Department of Children and Families (DCF), the Justice Research Institute (JRI), the Connecticut DCF, New Jersey DCF, and Florida DCF. Data collection took place between June 2015 and April 2017.
Recruitment Sites
Study participants were recruited through four organizations that use the MLMC prevention curriculum: MLMC (Boston, MA), Prevent Child Abuse New Jersey (New Brunswick, NJ), Selah Freedom (Sarasota, FL), and The Village for Families and Children (Hartford, CT). Program facilitators in New Jersey, Florida and Connecticut received 8 hours of curriculum training from Boston-based MLMC staff. The MLMC groups administered in each state included 10 sessions of 60 to 75 minutes each with young people who were determined to be “at-disproportionate-risk” for sexual exploitation by education or human service providers (e.g., DCF workers, residential home staff and school counselors). The criteria for deciding which youth are “at-disproportionate-risk” for CSE are not strictly defined by MLMC to encourage child protection workers, teachers, social workers, juvenile probation officers and others who are not typically trained to recognize risk factors for CSE to err on the side of referring someone to the group. As the group is a prevention group, and there are numerous factors that could predispose someone to CSE, it is not important to the model that the eligibility criteria for participating be strict. Each group is typically attended by five to nine participants. Groups are restricted to females, which includes transgender individuals who identify as female. Cis-gender males are not eligible for this group because adolescent girls may be helped when they experience the solidarity of an all-female group and because girls appear to MLMC to feel less shy, and more willing to be vulnerable, during discussions than if cis-gender males were present. Considering the groups that were included in this research study, the majority (55%) were held once per week for 10 weeks, but this was not always the case. Some groups were condensed, and the curriculum was delivered in fewer, lengthier sessions (e.g., five 2-hour sessions, or three 3.25-hour sessions). For this study, almost all of the groups (88%) were co-facilitated by a CSE survivor.
Participant Recruitment
Eligible youth were those participating in an MLMC prevention group, 11 years of age or older, and fluent in English. The age criterion changed during the course of the study. When the research began, the age eligibility criterion was 14 years old. As three children younger than 14 years old were referred into the MLMC prevention groups within the first few weeks of the study, this criterion was changed to 11 years old at 4 months into the study. Once eligible youth were identified as potential research participants, the MLMC program facilitator and a member of the research team worked together to determine if a parent or guardian needed to complete an informed consent form, to locate that person, and obtain their written, informed consent. In cases where parental rights had been terminated and youth were in the custody of DCF, consent forms were provided to DCF workers. In Massachusetts, in cases where youth were involved with DCF but parental rights had not been formally terminated by the state, the requirement to obtain parental consent was waived to protect children from the potential harm of disclosing their status as “CSE-involved” to parents or other guardians. All minors participating in the study were provided with an assent form.
At the outset of the study, the informed consent and assent documents were required to be returned to the study team prior to the first session of the MLMC group. However, the recruitment protocol was modified in July 2017 to specify that the research team had up until the fifth MLMC group session to locate the appropriate parent or guardian and collect the informed consent paperwork because youth at risk for CSE typically face numerous family challenges that made communicating and returning forms especially difficult. In October 2016, the protocol was modified to specify that the informed consent forms could be signed and collected by the research team at any time up until the final session of the MLMC group. This means that some youth participated in the baseline survey prior to the research team having the parental consent form in their possession, but the youth assent forms were always in-hand prior to their participation in the baseline survey. Youths’ completed baseline surveys were not examined, analyzed, or included in the research data set until the needed consent forms were in the possession of the research team. A member of the research team was present or on the phone in the first, and sometimes second, session of each MLMC group to go through the consent process with participants verbally and/or answer any questions about the study.
Data Collection
On the first day of the MLMC prevention group, all youth who had assented to participate in the research study completed a paper-based baseline survey before the first class started. The survey took approximately 20 to 30 minutes to complete. On the last day of the MLMC group, research participants completed the Follow-up 1 survey, which was identical to the baseline survey. Eighty-four percent (84%) of those who completed the baseline survey completed the Follow-up 1 survey; 90% completed the Follow-up 1 survey in-person on a paper-based survey. In a few cases (10%) where youth missed the last day of class, they completed the survey online or over the phone instead. Approximately 6 months after baseline, youth were re-contacted and asked to complete the Follow-up 2 survey, and 68% did so; of those who completed Follow-up 2, 61% completed Follow-up 2 online, 22% over the phone, and 16% selected to do in-person.
Measures
To our knowledge, there have not been prior evaluation studies of secondary prevention programs designed to address CSE for youth, so valid and reliable measures with previously established psychometrics were not available for several of the outcomes of interest. As a result, for some outcomes of interest, the research team adapted existing items from other research studies and created new items as needed. The process of developing new instruments was highly collaborative and involved multiple rounds of iterative feedback from all authors, staff of MLMC, and selected youth.
CSE and sexually explicit behavior
CSE and sexually explicit behavior in the past 3 months was assessed through a series of five original items. The first item, (a) “exchanged sex for money, food, a place to stay, drugs, gifts, or favors,” was adapted from a similar item on the Add-Health survey and was analyzed as a sole item representing CSE (Edwards et al., 2006). In addition, participants were asked if they had engaged in four sexually explicit behaviors that could have been CSE. These sexually explicit, potential CSE, experiences included: (b) “stripped or engaged in x-rated (naked) dancing,” (c) “taken x-rated (naked) selfies,” (d) “agreed to let their boyfriend (or someone else) take a video of them while they were naked or having sex,” or (e) “participated in x-rated (naked) or sexually explicit modeling or had a sexually explicit video taken.” Each of the items 2 to 5 were original. Participants were given response options of “yes” or “no.” Participants who gave any positive responses to the sole CSE item were categorized as having a CSE experience in the past 3 months. Participants who gave a positive response to any one of the four sexually explicit behavior questions were categorized as having engaged in sexually explicit behavior.
Dating abuse victimization
Physical, sexual, and psychological abuse victimization by a dating partner were assessed via five items. The first two items were adapted from an established measure of dating abuse (DAPAS), which has excellent reliability, α = 0.83 (Goncy & Rothman, 2019). Participants were asked if they were in a romantic or dating-type relationship with someone. If they were in a relationship, participants were asked how many times the other person in the relationship did the following: (a) pushed, grabbed, hit, beat, or hurt them physically, (b) forced them to have sex, (c) made them describe where they were every minute of the day, (d) would not let them do things with other people, and (e) made them feel like they could not say no to sex or made them have sex when they really did not want to. Cronbach’s alpha in this sample was α = 0.71. Due to an error, the referent period for this variable was the past 6 months at baseline, but the past 3 months (or since they started the MLMC program) at Follow-up 1, and the past 3 months (or since the MLMC program ended) at Follow-up 2. Response options were: never, 1 to 3 times, 4 to 9 times, or 10 or more times for each item. As the referent period for baseline was the past 6 months and we were able to construct a past-6-months estimate for Follow-up 2, but the referent period for Follow-Up 1 was only 3 months, data from Follow-Up 1 was not included in our multivariable-adjusted models.
Knowledge and attitudes about CSE topics
Participants answered 20 original Likert-type scale questions about their opinions about CSE to assess their knowledge and attitudes. Response options ranged from 1 (totally not true) to 6 (totally true). Sample items included: “Only girls who are addicted to drugs get pimped out,” and “If a guy spends a lot of money on a girl, she shouldn’t say no to sex.” Scores were summed across nonmissing items. The median score among the participants completing the baseline survey was used as a cut point to define “high” and “low/poor” knowledge levels at each of the follow-up periods. Participants with a total score greater than the median were considered to have poor knowledge and attitudes. Cronbach’s alpha in this sample was α = 0.70.
Gave help or information about CSE
Participants were asked four original questions about whether they had (a) told a friend about MLMC because they were trying to help the friend get out of a bad situation, (b) educated a friend about the myths and facts of selling sex, (c) warned somebody not to sell sex, and (d) helped a person their age get professional help (e.g., from a doctor or counselor) for a serious problem. Participants were given the response choices of “yes” or “no” for each question. The referent period for the questions was the past 3 months or since the prior assessment. Those who responded positively to one or more of these four questions were categorized as having given help or information about CSE. As Cronbach’s alpha for these four items in this sample was α = 0.58, we opted to analyze each of the four items separately rather than combining them into a scale.
Alcohol and marijuana use
Alcohol use in the past month was ascertained via four questions from the AUDIT which has a reliability of α = 0.86 for adolescents (Kelly & Donovan, 2001). Questions were modified to elicit the number of times each behavior had occurred in the prior month. Participants were asked how many times in the past 30 days they (a) had at least one drink of alcohol, (b) had four or more drinks of alcohol in a row, that is, within a couple of hours, (c) drank until they passed out, or (d) drank until they vomited. Response options included 0 days, 1 day, 2 to 3 days, 4 to 7 days, 8 to 14 days, 15 to 29 days, or all 30 days. Those who responded 0 days to all four questions were categorized as past month nonalcohol users, while those with any positive response to any of the four questions were categorized as past-month alcohol users. Past-month marijuana use was assessed via a sole item which asked about marijuana use in the past 30 days with the same response options as the alcohol questions.
Trust in police
Participants were asked two original items about their feelings about law enforcement over the past 3 months. They were asked “I felt like the police are not safe to trust,” and “I felt like I would ask police for help if I felt in danger.” Both questions had yes versus no response options.
Data Analyses
Descriptive statistics were calculated for demographic characteristics and survey items for all participants who completed the baseline survey as well as for the subset of participants who completed either the Follow-up 1 or Follow-up 2 surveys, or both. Means and standard deviations, and medians and interquartile ranges, were used for continuous variables, as appropriate, and frequency counts and percentages were used for categorical variables. For the subset of participants that attended all three visits, differences in mean scores across the three timepoints were assessed using a repeated measures analysis of variance (ANOVA). For the dichotomous outcome measures, differences over time were assessed using the Cochran’s Q chi-square statistics.
Relative risks (RRs) and 95% confidence intervals were calculated for the association between data assessment points and each outcome variable using a modified Poisson model with robust error variances combined with generalized estimating equations (GEEs) to account for within-participant clustering due to the repeated measures (Huta, 2014; Zou, 2004). We were specifically interested in examining changes in outcomes at two timepoints: immediately after group completion (Follow-up 1) and 6-months after completion (Follow-up 2). Therefore, we calculated RRs for each timepoint separately (baseline versus Follow-up 1, baseline versus Follow-up 2), rather than treating time as a continuous variable. Since our study was comprised of a nonrandom sample of states, we considered state residence as a fixed effect. Our primary model was unadjusted, and a secondary model was constructed with further adjustment for age, race/ethnicity, state, group location (school/residential facility/other), marijuana use (yes vs. no), alcohol use (yes vs. no), number of scheduled group sessions, percentage of sessions attended, and baseline prevalence of each outcome. All analyses were performed using SAS, version 9.4 (Cary, NC). A p-value of <.05 was considered statistically significant.
Results
Participant Characteristics
The 354 participants who completed the baseline survey ranged in age from 11 to 20 years old, and their mean age was 16 years old (Table 1). Almost all of the MLMC program participants in our study sample were female (95%), 5% were non-binary gender or other gender. The sample was approximately evenly divided in terms of how they defined their race/ethnicity: White/non-Hispanic (26%), Black/African American (28%), Hispanic/Latina (25%), and other race (22%). Most research participants attended groups in Massachusetts (60%), and 27% attended in New Jersey, 11% in Florida, and 3% in Connecticut. A sizable percentage of research participants attended MLMC groups in residential facilities (42%), while 29% attended groups in high schools, 9% in middle schools, and 20% in some other setting such as a social services agency or DCF office. A sizable minority of participants did not identify as heterosexual (Table 1). Approximately 28% identified as bisexual, and 10% identified as lesbian, asexual, pansexual, or other. Most MLMC prevention group participants attended half or more of their assigned group sessions (Table 2). While 37% attended 100% of their group sessions, approximately 89% attended 50% or more of the group sessions (Table 2).
Characteristics of the Study Sample Participating in Baseline, Follow-up 1, Follow-up 2, and All Three Timepoints.
Survey questions about gender and sexual orientation were added to the surveys part-way through the study period. Therefore, 213 participants were missing data on these questions at baseline, 170 missing at Follow-up 1, and 129 missing at Follow-up 2.
Intervention Characteristics and Session Attendance, My Life My Choice Prevention Groups.
Sessions attended versus being absent.
Participant Retention
Participant retention is described in Figure 1. Approximately 84% of participants who completed baseline surveys also completed the Follow-up 1 survey, and 68% completed the Follow-up 2 survey. A total of 213 participants (60%) completed both Follow-up 1 and Follow-up 2, and a small minority (9%) were lost to follow up after baseline. Retention rates did not vary substantially by state.

My life my choice prevention group study enrollment and retention diagram.
Outcomes
Changes in outcomes over the course of follow up are summarized in Table 3. At baseline, 4% of participants reported having experienced CSE in the past 3 months (Table 3). The percentage of those who had been exploited was low and no substantial changes were observed at Follow-up 1 or Follow-up 2 (2% and 3.5%, respectively). There were statistically significant changes in sexually explicit behavior; at baseline ~21% reported sexually explicit behavior in the past 3 months, and this percentage decreased to 9% at Follow-up 1% and 8% at Follow-up 2 (p < .0001; Table 3). Dating abuse victimization was prevalent in the baseline sample: 45% of participants who were in a recent relationship reported that in the past 6 months, they had experienced dating abuse (Table 3). By Follow-up 2, only 23% of participants who had a recent relationship reported dating abuse victimization in the past 6 months (p < .001; Table 3).
Summary of Outcomes Among Participants Who Completed Baseline, Follow-Up 1, and Follow-Up 2 Surveys (N = 213), MLMC Prevention Groups.
Note. MLMC = My Life My Choice.
Dating abuse questions were only given to the subset of participants who reported being in a dating relationship in the past 6 months, since group began, and since group ended (n = 91).
These questions were added once research was part-way concluded, thus the denominator for these questions was lower than for the other questions.
There were positive changes in knowledge and help-giving behavior. Higher scores on the knowledge/attitudes scale represented less knowledge and more favorable attitudes toward sexual exploitation. At baseline, the average score on the knowledge/attitudes scale was 2.12 (SD = 0.54), which decreased to 1.95 at Follow-up 1 (SD = 0.58), and 1.99 (SD = 0.73) at Follow-up 2 (F = 6.30, p < .01; Table 3). At baseline, 56% of participants reported that they had given help or information about CSE to a friend in the past three months. At Follow-up 1, the percentage of participants who gave help or information about CSE increased to 73% at Follow-up 1% and 79% at Follow-up 2 (p < .0001). Similarly, at baseline 57% reported that they would be able to tell a friend about three services that could help youth who were at-disproportionate-risk of CSE, while the percentage increased to 79% at Follow-up 2 (p < .0001).
Participants were more likely to report feeling in control over things in their lives over time (59% felt in control at baseline as compared to 71% at the end of follow up, p < .001). Most felt like they would ask the police for help at baseline (63%). At the end of the group, 73% reported that they would ask the police for help, and at Follow-up 2, 72% reported the same (p < .0001). There were no observed changes in the percentages of participants who reported using alcohol or marijuana in the past month over time. At baseline, approximately one-quarter reported alcohol or marijuana use in the past month (24% for both), and there were no statistically significant changes observed at the end of the group (25% and 20%, for alcohol and marijuana use, respectively), or at Follow-up 2 (21% and 23%, for alcohol and marijuana use, respectively).
Adjusted analyses
In multivariable repeated measures models adjusting for participants’ age, race/ethnicity, state, and the setting of the group (congregate care or other), number of scheduled group sessions, the percentage of group sessions that youth attended, and the baseline level of the outcome of interest, we observed statistically significant differences in our outcome measures at both Follow-up 1 and Follow-up 2 as compared to baseline (Table 4). Specifically, as compared to baseline, the risk of sexually explicit behavior was substantially lower at 3 months (RR: 0.52, 95% CI: 0.37-0.72) and at Follow-up 2 (RR: 0.53, 95% CI: 0.35-0.82), and the risk of having low knowledge and undesirable attitudes about CSE was lower at 3 months (RR: 0.75, 95% CI: 0.64-0.88), and at Follow-up 2 (RR: 0.73, 95% CI: 0.61-0.87; Table 4). There was a 23% decrease in the risk of dating abuse at Follow-up 2 as compared to baseline; however, this difference did not quite reach statistical significance in our multivariable-adjusted model (p = .06). Giving any CSE-related help or information to a friend increased substantially from baseline; at 3 months, the RR was 1.28 (95% CI: 1.10-1.48), and at Follow-up 2, it was 1.38 (95% CI: 1.20-1.59; Table 4).
Relative Risks for the Association Between Follow-Up Period and Within-Participant Change in Selected Outcomes of Interest, MLMC Prevention Groups (n = 354).
Note. RR = relative risk; CI = confidence interval; CSE = commercial sexual exploitation; MLMC = My Life My Choice.
Multivariable model adjusted for age, race, state, group location (i.e., school/residential facility/other), number of scheduled group sessions, percentage of sessions attended, and baseline level of outcome.
Relative risks for each of the dichotomous outcomes were calculated using a modified Poisson model with robust error variances combined with generalized estimating equations (GEEs) to account for the repeated measures.
Discussion
To our knowledge, this is the first evaluation of a CSE prevention program. Participants in the MLMC prevention groups demonstrated changes in knowledge, attitudes and behavior from the start of the program to its end at 3 months, and for many outcomes, the changes persisted an additional 3 months after the program had ended. Of particular note, the incidence of self-reported sexually explicit behaviors was reduced by half over time. Dating abuse victimization also decreased. Trust in the police increased substantially over the course of the program and was sustained, as was knowledge about CSE, feeling like one would be able to tell a friend about helping resources, and actually giving help or information about CSE to a friend. Participants’ sense of control over the conditions of their lives increased. Alcohol and marijuana use did not increase or decrease over the evaluation study period, despite expectations that it might have decreased because the intervention addressed substance use in one session.
Given the unacceptably high prevalence of CSE and the severity of its consequences, as well as the cost to society (IOM and NRC, 2013), it is a promising fact that this program may have protected girls. The program requires no expensive technology or materials, is easy to implement in a variety of settings, and professionals from a variety of disciplines can be trained to facilitate the program. While there are other CSE prevention curricula for youth being used in the United States, few of these prevention programs are multi-session, gender-specific, survivor-led, and focus on secondary prevention instead of primary prevention. To our knowledge, none have been evaluated.
We suspect that if the MLMC prevention program is influencing participants’ CSE-related knowledge, attitudes, and behaviors, that one of the “key ingredients” of the program that may produce this effect is the fact that nearly all of the prevention groups are co-facilitated by an adult who was formerly sexually exploited as a child (i.e., survivor-mentors). Anecdotally, many girls expressed to research staff that they enjoyed the program because it was facilitated by someone who seemed to genuinely understand their situation, the pressures that they faced, and the nature of the decisions that they had to make to become or stay safe. Importantly, incorporating survivor input from racially and ethnically diverse survivors means that the curriculum itself is not culturally homogeneous and is able to be implemented in a wide variety of settings with youth who are White, Black, Hispanic, Asian, Multiracial, and otherwise diverse in terms of gender-identity and sexual orientation. An interesting topic for future research would be to compare the effect of survivor-mentor-delivered groups to groups delivered by nonsurvivor-mentors. Evaluating the impact of the MLMC model program on cis-gender boys would also further the field.
In general, there is an acknowledged tension between the desire of participants in psychoeducational groups or therapy for a facilitator or therapist to self-disclose their own experience with an issue, versus the traditional psychotherapy practice of nondisclosure (or limited disclosure under certain conditions) (Mallow, 1998; Olarte, 2003). On one hand, there is a concern that facilitators’ disclosures about personal experiences could shift the focus away from the youth, burden youth with concern for the adult, or otherwise backfire by creating a distorted image of the adult that is detrimental to the youth’s personal growth. On the other hand, evidence appears to suggest that personal disclosures by therapists tend to enhance alliance with clients, and are frequently perceived as helpful (Hill, Knox, & Pinto-Coelho, 2018)—such that contemporary counselors cannot engage reflexively in nondisclosure (Panagiotidou & Zervas, 2014). In this case, analyses of qualitative data (under review) found that youth perceived their facilitators as flexible, honest, respectful, warm, and trustworthy because of their self-disclosures—which are key attributes that create a successful therapeutic and educational alliance (Hill et al., 2018).
An interesting direction for future research would be a dismantling design or research that permits inferences about which components of the MLMC prevention group were influential for which outcomes. Understanding precisely which elements of the MLMC model are the most important mechanisms for change could be useful for other programs that choose to adapt the model and implement it elsewhere or in different contexts (i.e., outside the United States or with specific cultural subpopulations in the United States). One problem is that funding for CSE prevention is extraordinarily limited—there is no governmental or nongovernmental entity that reliably awards funding for this type of research, and substantial resources will be needed to run a sufficiently large study that unpacks components of the intervention to isolate mechanisms of influence.
The finding that the MLMC group may have shifted knowledge, attitude, and behavior in the desired and anticipated direction for youth at-disproportionate-risk for CSE is very promising for the field of human trafficking prevention. Given the prevalence of CSE worldwide, and insidious consequences for the trafficked individuals, their families, friends, and communities, a protective intervention—if determined to be effective—could be adapted and disseminated in numerous languages, settings, and with new populations including cis-gender males and potentially even adults. Continued assessment of the impacts of the MLMC secondary prevention program, through larger-scale and experimental designs, will benefit the field.
Limitations
As this study did not use a comparison group, observed changes could be attributable to a factor other than group participation. However, the participants were enrolled at different times over a 2-year period, in different states, and different settings, reducing the chances that CSE knowledge, attitudes or behavior were affected by a national or state-level event, or some feature of the local environment. It is possible that girls who were identified as at-disproportionate-risk for CSE and referred to the group were referred to other services at the same time, and any changes could be attributable to these other services. Future studies should assess receipt of other services during the period of time girls are enrolled in the MLMC program. Anecdotally, our impression is that girls in congregate care environments were receiving complementary services while enrolled in the MLMC program, while girls in public school or other settings generally were not, but that the complementary services provided in congregate care did not explicitly educate girls about CSE. The very reason that MLMC is invited to implement their program in congregate care is because otherwise they lack the capacity to teach girls about these topics. We also did not measure, or analyze, the length of group sessions. Most youth received instruction in hour increments, but there were a few who received the program in lengthier day-long sessions. Future studies should assess whether shorter sessions over a period of 2 months or longer sessions delivered in a condensed timeframe influence intervention effect. A further limitation is that only 4% of the sample experienced CSE at baseline and is a relatively rare event, so examining the effect of the prevention group on this outcome would require a large sample. Finally, as with all studies that rely on self-report, there is a possibility that girls were not accurate reporters. This possibility is mitigated by the fact that many did report socially undesirable behaviors such as posing for sexually explicit photos or experiencing dating abuse. Moreover, if underreporting was an issue, it is more likely that our results would have been biased toward—and not away from—the null.
In conclusion, this one-group, longitudinal evaluation of a nationally used prevention program for CSE found that there were changes in participants’ knowledge, attitudes, and behavior over time that suggest that the program is likely having its intended effect. Additional, larger-scale studies that include a comparison group would increase confidence in these findings.
Footnotes
Authors’ Note
This manuscript has not been published elsewhere and is not under consideration by any other journal.
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: This project was supported by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice (grant no. 2014-R2-CX-0005).
