Peer Pressure and Risk-Taking Behaviors Among Adolescent Girls in a Region Impacted by HIV/AIDS in Southwestern Uganda

Abstract

Purpose:

This paper uses data from a 3-arm Cluster Randomized Control Trial, Suubi4Her (N = 1260; 14–17-year-old school-going girls) to (1) assess the relationship between peer pressure and adolescent risk-taking behaviors; and (2) test the mediating effect of peer pressure on an intervention on adolescent risk-taking behaviors.

Methods:

Students in the southwestern region of Uganda were assigned to three study arms: control (n = 16 schools, n = 408 students) receiving usual care comprising of sexual and reproductive health curriculum; and two active treatment arms: Treatment 1 (n = 16 schools, n = 471 students) received everything the control arm received plus a savings led intervention. Treatment 2 (n = 15 schools, n = 381 students) received everything the control and treatment arms received plus a family strengthening intervention. We used multilevel models to assess the relationship between peer pressure and risk-taking behaviors. We ran structural equation models for mediation analysis.

Results:

Using baseline data, we found that direct peer pressure was significantly associated with substance use risk behaviors, (β = 0.044, 95% CI = 0.008, 0.079). We also found a statistically significant effect of the intervention on acquiring STIs through the mediating effect of sexual risk-taking significant (β = −0.025, 95% CI: −0.049, −0.001, p = .045) and total indirect (β = −0.042, 95% CI: −0.081, −0.002, p = .037) effects. Also, there was a significant mediation effect of the intervention on substance use through peer pressure (β = −0.030, 95% CI: −0.057, −0.002, p = .033).

Discussion:

Overall, the study points to the role of peer pressure on adolescent girls’ risk-taking behaviors; and a need to address peer pressure at an early stage.

HIV and AIDS remain one of the deadliest pandemics of our time. Despite global commitments to reduce AIDS-related deaths, new HIV infections are still emerging [1]. In sub-Saharan Africa adolescent girls account for one in four new HIV infections [2]. In 2021, adolescent girls accounted for three-quarters of all new HIV infections among adolescents [3].Parents in Uganda and much of sub-Saharan Africa invest less in girls’ economic well-being, including education and economic resources ownership [4]. Taken as a whole, the lack of investment in women and young girls (including adolescent girls) impacts their overall health and well-being [5], hence robbing girls of the opportunity for an education, contributing to their risk of acquiring HIV and other risk-taking behaviors [6]. This gender disparity increases the need to pay attention to the fight against HIV/AIDS in the region among adolescent girls [7].

Adolescence, typically defined as ages 10–24 [8], is one of the most challenging periods as it involves multiple transitions: transition from childhood into adolescence, then from adolescence into young adulthood (ages 18 and above) with less parental supervision. Adolescents must juggle school expectations, parental guidance, the desire to fit in, and succumbing to peer pressure, defined as the influence of other people to act in a certain way [9]. Moreover, as children transition through adolescence, they spend less time with family and more time with their peers [10].

Peer Pressure can be Direct, defined as an explicit invitation to do something, such as offering alcohol, or Indirect–defined as the contribution toward forming an action [11]. Direct peer pressure may occur as an encouragement to act and offer alcohol. Indirect peer pressure influences can occur when youth associate with peers who drink or smoke [12], and being pressured to drink alcohol [13].

Peer pressure and risk-taking behaviors

Being a member of a peer group is one of the primary experiences of adolescence [14]. Generally, peer influence leads to engaging in risky behaviors such as substance use with some probability of negative consequences such as physical injury, social rejection, and financial loss [15]. Indeed, peer pressure is regarded as one of the strongest determinants of juvenile delinquency [16]. Peer pressure among boys and girls undermines healthy social norms and HIV prevention messages, including delaying sexual debut [17]. Moreover, peer pressure may lead to transactional sexual activities exposing young people to sexually transmitted infections (STIs) [18].

To illustrate, a study conducted in Indonesia [19] showed a statistically significant relationship between peer pressure and sexual behavior among teenagers. Specifically, in the Indonesia study, the proportion of adolescents who were motivated to have sexual intercourse due to peer influence was 26.6%. A similar study by Wubet [20] located similar results—reporting a direct relationship between peer pressure and the consumption of alcohol. Peer pressure might build upon a female teenager, and she ends up leaving school for early marriage, or she decides to seek a man to fit in a group with her peers, exposing them to STIs [21].

Peer pressure has also been linked to economic resources. Specifically, income inequality is associated with risk-taking [22]. Although not all adolescents from low-income families are likely to fall prey to negative peer influences, economic instability creates a “fertile ground” for adolescents to be influenced by peers. Adolescents with no resources may easily be encouraged to engage with financially stable peers to have a belonging. It is not uncommon in some peer relationships for peers who are relatively stable to provide the resources for a peer to acquire drugs and/or alcohol until one is stable to fund their acquisition [23]. Moreover, peers may be encouraged by fellow peers to engage in risk-taking behaviors such as early unprotected sexual intercourse to gain resources [24]. Thus, economic stability may play multiple roles in peer pressure influences.

Overall, as detailed above, if not well managed, peer pressure may lead to several negative consequences [25]. Thus, this study aims to understand peer pressure during adolescence and how it relates to risk-taking behaviors among school-going adolescent girls in Uganda.

Theoretical framework

Two theoretical frameworks guide this study. The first theory is Banduràs social learning theory, which posits that development in human cognition is explained by the interplay of social beings who hold a need/desire to interact with other individuals [26]. Specifically, social learning theory is based on the idea that people learn from interactions with others in a social context. This can be through observing the behaviors of others. In the process, people develop similar behaviors, and after observing, people assimilate or imitate [26]. In the case of adolescents, in the quest to experience adulthood, they engage in risk-taking behaviors. Social learning posits that emulating peers’ behavior starts as an experiment and ends in an addiction which, if not controlled, is problematic to adolescents.

The second theory is Asset theory [27], which posits that ownership of assets influences people’s behavior. When people are poor, they are more likely to engage in behaviors that they would otherwise not engage in if they were more financially stable. And when people own resources, they are more likely to be viewed positively. People want to identify with those who have. In the specific case of this study, one may posit that adolescents who grow up in poor conditions are more prone to peer influence. The reasons for this relationship may include low self-esteem resulting from poverty and the need for belonging among peers, who may be perceived to have more resources. In other words, asset ownership influences people’s behavior but also influences how people view and interact with others. In this case, adolescents may be attracted to peers with resources. In addition, the asset theory states that people with more assets at present will have more resources in the future. Therefore, if the family is empowered economically at present, the adolescents in the family will be more hopeful in the future, and the risk of engaging in risky behavior such as alcohol consumption, and unprotected sex would be reduced. Additionally, the psychological aspect of the theory states that when people have assets, they fear less and hope more. People develop a positive attitude towards life which provides a foundation for minimizing risk-taking. Moreover, family strengthening (FS) is shaped by the does and don’ts of a family that limit children’s behavior in a family.

Given the theoretical frameworks guiding this study, we specifically address two questions:

1. What is the role of peer pressure on risk-taking behaviors? This is an exploratory question for which we offer the following hypothesis: Adolescent girls who report having peers who engage in risk-taking behaviors (e.g., sexual risk-taking, drug use, and alcohol use) will more likely report engaging in the same behavior themselves.

2. What is the role of peer pressure in mediating the relationship between a combined intervention (FS plus economic empowerment (EE) components) and risky behaviors, including substance use and acquiring STIs? This question will be addressed using longitudinal data collected at 12- and 24-months postbaseline.

Methods

Sample

We use data from the longitudinal cluster randomized study (Suubi4Her study) 2017–2023, funded by the National Institutes of Mental Health under grant number R01MH113486. The overall aim of Suubi4Her is to examine an innovative combination intervention–combining FS and EE–on risk-taking behaviors among school-going adolescent girls (N = 1260; ages 14–17 years of age) in Uganda. The recruited girls were attending their first and second year of senior secondary education (an equivalent of high school in the US education system) from 47 schools. The schools were in five geopolitical districts in southern Uganda.

Participant recruitment

To begin, 111 secondary schools were assessed for eligibility (See Figure 1, consort flow diagram). Out of 111 schools assessed, 65 were excluded for not meeting the inclusion criteria, leaving a total of 47 Schools. The remaining 47 schools were included in the study based on their school size (total number of students enrolled), location (urban vs. rural), and overall performance based on the Uganda Certificate of Education examinations results. All included schools had to be registered by the Uganda Government’s Ministry of Education and Sports. For details on recruitment, see study protocol paper [7].

Figure 1.

Consort flow diagram.

Randomization.

Following school selection, students were randomly assigned to three study conditions at the school level (N = 47 secondary schools): a control arm (n = 16 schools, n = 408 students) receiving usual care comprising of sexual and reproductive health curriculum delivered in school and school supplies (textbooks and notebooks); and two active treatment arms. Treatment 1 (n = 16 schools, n = 471 students) received everything the control arm received plus a savings-led EE Intervention [(financial literacy training and a Youth Development Accounts (YDA)]. Treatment 2 (n = 15 schools, n = 381 students received everything the control and treatment arm one received plus a FS intervention (comprising a multiple family group (MFG) intervention). See details on the intervention below.

The first intervention arm consisted of YDA, with a 1:1 matched savings program. The matched saving account was opened in the adolescent’s name with her primary caregiver as a cosigner (required until age 18 years). The match was saved on a different account. At the time when payment for school fees is required, a direct wire transfer was made to the school, or a cheque using the matched funds were paid directly to the school. This process eliminates unwanted pressure to withdraw funds that were designated for the girl’s education and skills development. A maximum of 70% was used on education and skills development and 30% for investment in family-based income-generating activities.

The second intervention condition consisted of an innovative combination of YDA and FS intervention delivered through MFGs. This intervention was a manualized curriculum that integrated aspects of group therapy, family support, systemic family therapy, and behavioral parent training delivered through 16 sessions. Specifically, the intervention focused on the “4 Rs”: Rules, Responsibility, Relationships, and Respectful communication, plus factors related to family engagement in mental health services under the “2Ss”: Stress and Social support. Trained Community Health Workers and Parent Peers delivered the MFG intervention. The MFG approach opens an opportunity for adolescents and their families to share their experiences with others in similar situations, thus building hope by providing social support, normalization of similar experiences and struggles, and the sharing of effective solutions [28].

The third study condition consisted of Usual Care. Participants in all study conditions received the required Adolescent Sexual and Reproductive Health curriculum, which provides information relating to HIV and sexual risk-taking behavior.

Data was collected using a 90-minute interviewer-administered survey.

Study measures

Outcome variables.

Risk-taking behaviors were measured using two indicators: sexual risk-taking behaviors and substance use (including the use of alcohol, use of tobacco, and use of drugs). Sexual risk-taking behavior was assessed by: (1) self-reported question: “Have you ever had sexual intercourse?” with responses (yes = 1 vs. no = 0). (2). Biomarkers tests for pregnancy/HCG test (positive = 1 vs. negative = 0), and laboratory-confirmed STIs (positive = 1 vs. negative = 0) were captured at baseline. The STIs assessed in the study were Chlamydia, Gonorrhea, and Trichomonas. Because of the COVID-19 restrictions that included no personal contact, biomarkers were only captured at baseline and 12 months. The items were added to generate a total score for sexual risk-taking behavior, with higher scores showing more sexual risk-taking behaviors.

Substance use was assessed by asking the participants about their alcohol, cigarette smoking, and marijuana use behaviors. Specifically, participants were asked if they had ever had a drink of alcohol other than a few sips (yes = 1 vs. no = 0). 2) Cigarette smoking was also self-reported. Participants were asked whether they had ever tried cigarette smoking, even one or two puffs (yes = 1 vs. no = 0). Additionally, participants were asked if they had ever taken marijuana (yes = 1 vs. no = 0). The items were added to generate a score for substance use. Higher scores indicated that the adolescents indulged more in substance use.

Predictor variables

Sexual-risk taking attitudes.

Sexual-risk-taking attitudes were measured using a 5-item scale, with each statement rated on a 5-point Likert scale ranging from “Never = 1” to “Always = 5”. Participants were asked how each of the statements applied to them. Sample statements include: I believe it’s OK for people my age to have sex with someone they’ve just met; I believe it’s OK for people my age to have sex with someone they love; I believe it’s OK to have sex without protection with someone you know. Theoretical scores range from 5 to 25 (Cronbach alpha = 0.72), with a higher score indicating higher levels of sexual risk-taking attitudes.

Peer pressure.

We assessed both direct peer pressure and indirect peer pressure. Four questions measured direct peer pressure, including (1) How much do you feel peer pressure to smoke cigarettes? (2) How often do you feel peer pressure to drink alcohol? (3) How much do you feel peer pressure to have a romantic partner? and (4) How often do you feel peer pressure to smoke marijuana? Each question was rated on a 5-point Likert scale ranging from 1 = never to 5 = always. Items were recoded further to 1 = 0 and 2–5 = 1. Summated scores were obtained and then categorized into two: (1) participants who scored 0 were considered not to have peer pressure and were coded with 0 and (2) participants who scored one and above were considered to have peer pressure and were coded with 1. Four questions also assessed indirect peer pressure. These were: (1) How much peer pressure is there on people your age to smoke cigarettes? (2) How much peer pressure is there on people of your age to drink alcohol? (3) How much peer pressure is there on people of your age to have sex? and (4) How much peer pressure is there on people your age to smoke marijuana? Each question was rated on a 5-point Likert scale ranging from 1 = None to 5 = A great deal. Items were further recoded as 1 =0, and 2–5 = 1. Summated scores were obtained, with higher values suggesting more peer pressure.

Individual level factors.

These included the participant’s age, and orphanhood status (double orphan vs. single orphan vs. non-orphan). In addition, we include self-esteem in this block. We define self-esteem as the self-evaluation of one’s self-worth [29] and we measured it using the Rosenberg Self-Esteem Scale, a 10-item scale (1 = strongly disagree to 4 = strongly agree). Theoretical scores range from 10 to 40 (Cronbach alpha = 0.77), with higher scores indicating high levels of self-esteem.

Family level factors.

These included: (1) Primary caregiver (biological parents, grandparents, other relatives). (2) Socioeconomic status, measured using employment status by asking the following question: “Is the person who financially supports you currently employed in the formal sector and earning a salary or a wage?” (1 = yes vs. 0 = no); and total household assets—assessed by a 20-item index in which participants were asked, “Does the family you live in own the following: house, rentals, land, banana plantation, coffee plantation, car, bicycle, or television?”. The total number of assets was summed up and then categorized into two; low-asset possession and high-asset possession. We used the World Bank [30] suggested cutoff at the 40th percentile to represent low asset possession. (3) We also assessed (1) participant’s possession of essential household items using three variables: sets of clothes (1 or fewer = 0 vs. Two or more = 1); pairs of shoes (1 or fewer = 0 vs. Two or more = 1); and blanket ownership (yes = 1 vs. no = 0). The items were added and categorized into two categories: participants who owned one or fewer essential items were coded as 0 and, participants who owned two or more essential items were coded as one; (2) Food security was assessed using six questions: number of meals per day (1 or fewer = 0 vs. Two or more = 1); frequency of eating meat or _fish in the prior week (1 or fewer = 0 vs. Two or more = 1); frequency of eating egg(s) in the previous week (1 or fewer = 0 vs. Two or more = 1); frequency of taking milk in the previous week (1 or fewer = 0 vs. Two or more = 1); having breakfast on the day of interview yes = 1 vs. no = 0); and frequency of taking tea with sugar (1 or fewer = 0 vs. Two or more = 1). The same questions have been used in other studies to assess food security. (4) Number of people in the household; (5) Number of children in the household.

Data analysis procedure

Data were analyzed using Stata software version SE 17 (aim 1) and Mplus (aim 2). We summarized continuous variables using means, standard deviations, and categorical variables using percentages. The first study aim assessed for the association between peer pressure and two outcomes including sexual risk-taking behaviors and substance use risk behaviors. To answer this aim, we used baseline data and fitted a separate multilevel linear regression model for each outcome. In each model, the participants comprised level 1 of the model, and these were clustered under the schools, which we included in level 2. We examined the models to and determined that the residuals were normally distributed, there was homoskedasticity, and no multicollinearity was detected. We performed sensitivity analyses including allowing random slopes for several variables such as age and peer pressure, and cross-level interaction; however, these models reported comparable results. We reported β coefficients and their Huber-White cluster-adjusted confidence intervals (CIs), in which we adjusted for clustering at the school level. We also reported random effects including the variance of the school and participant random intercepts, variance of the residuals, and the intraclass correlation coefficients for the school and participant levels. Statistical significance was assessed at 0.05 level.

To assess the mediation pathways for the relationship between the intervention and the two outcomes (substance use behaviors and acquiring STIs), we fitted separate multilevel generalized structural equation models, by including the school as a clustering variable, using Mplus version 8.1 [31]. To preserve the temporal trend of events, we used direct peer-pressure (mediator) that was assessed at 12 months and outcomes—including substance use and acquiring STIs assessed at 24 months post intervention initiation. In all the models, we controlled for the participants’ age. During the Structural Equation Model (SEM) analysis, estimation of the model was conducted using Weighted Least Squares, Mean and Variance adjusted. This method is particularly effective for categorical and ordinal dependent variables. ‘STIs’ was classified as a categorical variable, while ‘Substance Use’ was considered a continuous variable. This decision was based on the inherent nature of these variables, thereby enabling us to maintain the accuracy and validity of our results. To deal with missing data, we employed the Full Information Maximum Likelihood estimation under the assumption of data missing at random. This approach makes use of all available data, resulting in unbiased parameter estimates when the data are missing at random. Furthermore, we used bootstrapping with 5000 draws to produce robust standard errors and CIs. We also implemented delta parameterization coupled with a probit link function to model the categorical nature of some variables in our models. In all the models, we controlled for the participant age.

We reported the standardized coefficients (β) with the corresponding 95% CIs based on α = 0.05. The goodness of fit of the SEM was assessed using (1) overall chi-square—where a nonsignificant p value would suggest good model fit, (2) Root Mean Square Error of Approximation below 0.06, (3) Standardized Root Mean Square Residual below 0.08 and (4) Comparative Fit Index above 0.90 (Gunzler et al., 2013) In the SEM, we estimated the total, direct, and indirect effect (through peer pressure) of the intervention on sexual risk-taking behaviors. In the models, the direct effect is denoted c’ and is the effect of the intervention on the outcome after controlling for the mediators. The indirect effect (a*b) is the effect of the intervention to the outcome through the mediators. The total effect, c, is the sum of the direct and indirect effects of the intervention on the outcome (c = c’ + a*b) [31] We also determined the total effect that was mediated. We also calculated the percent mediated as the ratio of the indirect effect to the total effect as shown. [Percent Mediated = (Indirect Effect/Total Effect) * 100%].

Ethical considerations

The study protocols were reviewed and approved by the Washington University in St. Louis Institutional Review Board (IRB) and by in-country local IRBs in Uganda, including the Uganda Virus Research Institute-Research Ethics committee and the Uganda National Council of Science and Technology. Written informed consent was obtained from the caretakers, and the children provided assent prior to participating in the study. To ensure the integrity of the consent process, the forms were translated and back translated between English and Luganda. To avoid any potential coercion, the consent process for adult caregivers and children was conducted separately. In accordance with the Uganda HIV disclosure policy, study participants received their HIV results independently of their caregivers.

Results

Sample characteristics

Table 1 presents the demographic characteristics of the sample. Specifically, the mean age was 15.4 years. About 83% of participants were nonorphans, and 76.6% identified a biological parent as their primary caregiver. The average number of people in the household was 7, with an average of three children living in the household. Only 23.2% of the participants reported that the person financially supporting them is employed in the formal sector. Almost 50% of the participating adolescents reported experiencing some form of peer pressure—specifically, 26% reported experiencing direct peer pressure, while 25.1% experienced indirect peer pressure to engage in risky behaviors.

Table 1.

Descriptive characteristics of study participants

Variable	Total sample (N = 1,260)
Age (min/max: 14–17) (mean, SD)	15.37 (0.87)
Orphan hood status
Double orphan	24 (1.90)
Single orphan	191 (l5.16)
Nonorphan	1,045 (82.94)
Primary care giver
Biological parent	965 (76.59)
Grandparent	140 (11.11)
Other relative	155 (12.30)
Household size (min/max: 2–31) (mean, SD)	7.00 (2.71)
Number of children in a household (min/max: 0–13) (mean, SD)	3.50 (2.10)
Employment of person supporting the family
No	968 (76.83)
Yes	292 (23.17)
Basic essential items
Owns one household item or less	82 (6.51)
Owns two or more household items	1,178 (93.49)
Food security (min/max: 0–6) (mean, SD)	3.41 (1.37)
Self-esteem (min/max: 10–40) (mean, SD)	33.00 (5.57)
Sexual risk-taking
Reported prior sexual intercourse	42 (3.33)
Confirmed pregnancy (positive HCG test)	14 (1.11)
Laboratory confirmed STIs
Chlamydia	7 (0.56)
Trichomonas	65 (5.16)
Substance use
Alcohol drinking	74 (5.87)
Smoking cigarette	15 (1.19)
Using marijuana	3 (0.24)

HCG = human chorionic gonadotrophin; SD = standard deviation; STIs = sexually transmitted infections.

We find that 3.3% of participants reported prior sexual intercourse, and 1.1% tested positive on the HCG test. Less than 1% of participants tested positive for Chlamydia, and 5.2% tested positive for Trichomonas. In regard to substance use, 5.9% of participants reported having had a drink of alcohol other than a few sips, 1.2% reported ever smoking cigarettes, and 0.24% had tried using Marijuana.

Research question #1.

What is the relationship between peer pressure and risk-taking behaviors (sexual risk-taking behaviors and substance use risk behaviors)? This question was addressed using baseline results.

Table 2 presents the relationship between peer pressure and risk-taking behaviors.

Table 2.

A regression table showing the relationship between peer pressure and risk-taking behaviors

	Sexual risk-taking behaviors		Substance use risk behaviors
	β (95% CI)	p value	β (95% CI)	p value
Participant age	0.052 (0.013, 0.090)	.009	−0.002 (−0.026, 0.021)	.834
Orphan hood status (ref: orphan)
Single orphan	−0.151 (−0.435, 0.113)	.298	−0.05 (−0.125, 0.114)	.932
Nonorphan	−0.173 (−0.451, 0.104)	.221	0.020 (−0.088, 0.128)	.719
Primary caregiver (ref: other relative)
Grandparent	0.052 (−0.091, 0.196)	.476	0.045 (−0.018, 0.108)	.165
Biological parent	0.068 (−0.044, 0.180)	.234	0.023 (−0.021, 0.067)	.303
Household size (number of people in a household)	0.001 (−0.016, 0.019)	.886	−0.006 (−0.014, 0.002)	.118
Number of children in a household	−0.015 (−0.036, 0.006)	.168	0.010 (−0.002, 0.021)	.113
Family assets (ref: low asset possession)
High asset possession	0.017 (−0.049, 0.084)	.610	0.017 (−0.015, 0.049)	.287
Employment of person supporting the family (ref: no)
Yes	−0.032 (−0.103, 0.039)	.372	0.039 (0.002, 0.075)	.037
Food security	−0.032 (−0.056, −0.008)	.010	0.004 (−0.007, 0.015)	.469
Possession of basic items (ref: one or less)
Two or more items	0.001 (−0.120, 0.121)	.990	0.029 (−0.026, 0.084)	.299
Self esteem	−0.002 (−0.009, 0.005)	.568	−0.003 (−0.007, 0.001)	.069
Direct peer pressure (ref: no)
Yes	0.033 (−0.044, 0.110)	.401	0.044 (0.008, 0.079)	.015
Indirect peer pressure (ref: no)
Yes	0.047 (−0.011, 0.105)	.115	−0.004 (−0.051, 0.042)	.864
Constant	−0.320 (−1.035, 0.395)	.380	0.119 (−0.324, 0.562)	.599
Random effects
Variance of school random intercept	0.022 (0.008, 0.060)		0.001 (<0.001, 0.036)
Variance of participant random intercept	0.251 (0.187, 0.338)		0.049 (0.040, 0.060)
Variance of the residuals	0.036 (0.023, 0.057)		0.025 (0.016, 0.039)
ICC (school)	0.070 (0.031, 0.152)		0.013 (<0.001, 0.330)
ICC (participant)	0.883 (0.864, 0.900)		0.671 (0.612, 0.726)

Bold values indicate statistically significant findings.

ICC = intraclass correlation coefficients; CI = confidence interval.

Sexual risk-taking behaviors

We found that direct and indirect peer pressure were not associated with sexual risk-taking behaviors. However, we found other factors that were associated with this outcome. For instance, increasing age was significantly associated with a higher likelihood of engaging in sexual risk-taking behaviors. Specifically, for every year increase in a participant’s age, the likelihood of engaging in sexual risk-taking behavior increases by 0.052 (β = 0.052; 95% CI = 0.013–0.090). We also find that for every unit increase in food security, the likelihood of engaging in sexual risk-taking behavior decreased by 0.032 (β = −0.032, 95% CI = −0.056 to −0.008).

Substance use risky behavior

Participants from a household where the person financially supporting the family was employed in the formal sector, were more likely to engage in substance use risky behaviors than the participants with a financial supporter who is not currently employed in a formal sector (β = 0.039; 95% CI = 0.002–0.075). We also found that the participants who experienced direct peer pressure were more likely to engage in in substance use risk-taking behaviors than participants who did not experience direct peer pressure (β = 0.044, 95% CI = 0.008–0.079).

Research question #2.

The second question explores the role of peer in mediating the relationship between Suubi4Her combination intervention and STI acquisition and substance use behaviors. This question was addressed using three waves of data collected at baseline, 12- and 24-months postbaseline. See Figure 2.

Figure 2.

SEM diagram for the mediators of the effect of an economic empowerment intervention on risky behaviors. In the first model (Model 1), the outcome was having a sexually transmitted infection, which was measured as an observed variable. In the second model (Model 2), the outcome was substance use behaviors, that was measured as a latent variable. Model two did not include sexual risk-taking attitudes as a mediator, which was only included in model 1. In the models, the mediators used including peer pressure and sexual risk-taking were measured at 12 months, while the outcomes were measured at 24 months.

Sexual risk-taking behaviors

In our analysis on mediators for the effect of the intervention on STIs, a significant mediation effect was observed through sexual risk-taking attitudes. More precisely, the intervention was effective in decreasing these attitudes, which subsequently led to a reduction in the likelihood of acquiring STIs. This mediation effect was evidenced by significant specific indirect (β = −0.025, 95% CI: −0.049, −0.001, p = .045) and total indirect (β = −0.042, 95% CI: −0.081, −0.002, p = .037) effects. Importantly, no significant direct effect or total effect of the intervention on STIs was detected. The mediation analysis revealed that 51.9% of the total effect was mediated through sexual risk-taking attitudes, further underlining the importance of this indirect pathway. However, we found that peer pressure did not significantly mediate the impact of the intervention on STIs. These findings are detailed in Table 3 and Figure 2.

Table 3.

Structural equation model results for the effects of the intervention on substance using and acquiring STIs

	STIs		Substance use
	β (95% CI)	p value	β (95% CI)	p value
Mediators on the intervention
Peer pressure	0.100 (0.014, 0.186)	.022	0.100 (0.038, 0.162)	.002
Sexual risk-taking attitudes	0.091 (0.014, 0.067)	.021	–
Sexual-risk attitudes on peer pressure	−0.044 (−0.073, −0.016)	.002	–
Outcome on the mediators
Peer pressure	−0.180 (−0.435, 0.075)	.167	−0.299 (−0.444, −0.153)	<.001
Sexual risk-taking attitudes	−0.275 (−0.354, −0.196)	<.001	–
Specific indirect effects
Peer pressure	0.001 (−0.048, 0.012)	.112	−0.030 (−0.057, −0.002)	.033
Sexual risk-taking attitudes	−0.025 (−0.049, −0.001)	.045	–
Peer pressure and sexual risk	0.018 (<0.001, 0.003)	.244	–
Effects
Direct effect	0.122 (−0.043, 0.288)	.148	0.295 (−0.246, 0.835)	.285
Total indirect effect	−0.042 (−0.081, −0.002)	.037	−0.030 (−0.057, −0.002)	.033
Total effect	0.081 (−0.077, 0.238)	.316	0.265 (−0.265, 0.795)	.327
Proportion of effect mediated	51.9%		10.2%
Model fitness
RMSEA (90% CI)	0.014 (<0.001,0.050)		0.056 (0.039, 0.075)
CFI	0.999		0.962
TLI	0.997		0.933
SRMR	0.033		0.082
Chi square	<0.001		<0.001

Bold values indicate statistically significant findings.

RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; SRMR = Standardized Root Mean Square Residual; STI = sexually transmitted infection; CI = confidence interval; TLI = Tucker-Lewis Index.

Substance use risky behavior

Regarding substance use, we found a significant mediation effect, indicating that the intervention influenced substance use indirectly through its impact on peer pressure. More specifically, our intervention was found to decrease peer pressure, which in turn, led to a reduction in substance use. The specific indirect effect of this mediating relationship was significant (β = −0.030, 95% CI: −0.057, −0.002, p = .033), as was the total indirect effect (β = −0.030, 95% CI: −0.057, −0.002, p = .033). These findings highlight the role of peer pressure in influencing substance use among the study participants. However, it’s worth noting that the direct effect of the intervention on substance use, as well as the total effect of the intervention on substance use, was not statistically significant. In this model, 10.2% of the total effect was mediated.

Discussion

This paper investigated the role of peer pressure on risk-taking behaviors and the role of peer pressure in mediating the relationship between a combined intervention (FS plus EE components) and risky behaviors.

Results indicated that as adolescents grow in age, the likely hood of engaging in sexual risk taking behaviors increase. Indeed as young people transition into adolescence, they tend to move more toward their peers—seeking acceptance by their peers. This may lead to young people being more susceptible to the influences of their peers, including increased risk-taking behaviors. Parental monitoring and supervision are powerful influences in the life of an adolescent. With reduced parental supervision, adolescents are more likely to engage in risk-taking behaviors, including having unprotected sex, alcohol use, and drug use. Moreover, the transition through adolescence comes with different physical changes, including changes in moods, hormonal levels, and increased curiosity. Adolescence is associated with poor mental health [32]. Some adolescents engage in risk-taking behaviors because it is considered a way of escaping challenges and depression.

Food insecurity is an essential driver of vulnerability. This study finds that adolescents reporting food security are less likely to engage in risk-taking behaviors. Indeed, adolescent girls tend to engage in risk-taking behaviors, including transaction sex work, because they desire to meet their basic needs, including food.

These results align with what Ogunsola mentions that having parents with full employment is a determinant of substance use in adolescents [33]. In addition, parents with tertiary education are another risk factor for engaging in substance use [34]. Adolescents from families with formal employment live in households with a stable flow of income, which may increase their chances of access to resources that could be used to engage in risky activities such as buying alcohol and drugs. Moreover, formal employment may mean that caregivers/families spend more time away from home, which may reduce parental-child supervision and monitoring, allowing adolescents to spend most of the time with their peers and being influenced by the same group. Again, these are simply logical speculations.

We found a statistically significant effect of the intervention in decreasing these attitudes sexual risk-taking attitudes and a decrease in peer pressure, which in turn, led to a reduction in substance use. The results align well with a study conducted by Ssewamala [35] that confirms the importance of economic intervention in protecting adolescents from sexual risk-taking behaviors, and in this case through reduced peer pressure. EE brings about a positive outlook for the future, Participants are less likely to be influenced by their peers to engage in behavior that can jeopardize their future, including avoiding risky sexual behaviors [24,36]. Moreover, Brathwaite [37], mentions that economic interventions are protective measures against alcohol and drug use.

The study employed a combination intervention, the EE component was complimented by a family-level intervention for effective interventions, this combination brought together families to pull resources together to save money and meet their specific needs. Sessions improved family communication and relations, and when families are empowered to work as a unit most of the emotional, social, and economic challenges are easily overcome. Hence, a buffer to engaging in sexual risk-taking behaviors.

Implications

Overall, this study points to the role of peer pressure on young adolescent girls’ risk-taking behaviors and the need to address peer pressure from an early childhood development stage.

Dealing with peer pressure is a skill that adolescents need. There should be systems to support adolescents through the challenges they face. In most cases, children will relate with peers on issues they find inappropriate to involve their parents. Nevertheless, caregivers/parents can create conducive environments to allow their children to gain trust and discuss issues they would otherwise not discuss with them. As adolescents find support in their parents and families, this serves as a buffer to engaging in risky behaviors and a control mechanism to check the kind of relationships and networks their children/adolescents relate with. Institutions like families and schools should talk about the negative components of peer pressure through peer education, and safe spaces. Peer education is a useful tool to deal with adolescent challenges [38] and can be implemented with cultural sensitivity in mind.

Additionally, intervention tools such as the friendship bench [39], where adolescents can speak freely to trained community members, similar to what has been done in Zimbabwe, Malawi, and Botswana, could be employed. The friendship bench is a culturally sensitive tool that appreciates local relationship dynamics. This could potentially make it a more effective tool for addressing peer education.

It is essential to increase interventions that increase economic security in young people because these serve as a buffer to sexual risk-taking behaviors and encourage youth to stay in school. Also, policymakers are called upon to integrate financial capability policies into youth development policies Further, research is needed to study the negative impacts of owning resources at an early age.

Limitations

The study conducted Biomarkers once a year, there is a possibility that we could have missed some infection windows, especially if participants received treatment and the illness is treated between the assessment intervals—although there is no free STI treatment especially for low-income school-going adolescents like the ones included in our study.

Further, not all peer pressure is negative; this research did not test the positive impact of peer pressure in terms of developmental influence. For example, when it comes to schooling, the positive aspect of peer pressure may posit that students learn a lot from their peers through discussion groups and may solve problematic questions in class [40]. Moreover, as one aspect of the positives of peer pressure, weak students (in this case, peers) may be supported by their colleagues to improve their class grades; and peers with good grades can also motivate others to perform better. Indeed, some students learn more from their peers than their teachers in class. Sometimes peer pressure on college students may positively affect them [41]. When they stay in a peer group, they will compare to others automatically. Thus, one cannot ignore the positive contribution of peer pressure on school and education attainment among adolescents.

Lastly, our findings are not generalizable to out-of-school adolescent girls especially at the time of recruitment. The study focused on in school adolescent girls and yet those out of school may be at higher risk of sexual risk-taking.

Overall, it is important to emphasize that the study is unique given its assessment of multiple outcomes of peer pressure in adolescent girls. Specifically, the study combines self-reports and biological data to link peer pressure and risk behaviors, strengthening the analysis. Despite the study’s limitations, the highlighted findings contribute to the existing literature on the relationship between peer pressure and risk-taking behaviors.

IMPLICATIONS AND CONTRIBUTION.

Overall, this study makes a scientific contribution to the role of peer pressure on young adolescent girls’ risk-taking behaviors and the need to address peer pressure from an early childhood development stage. This is important because adolescence is a very challenging period that requires very targeted adolescent specific interventions.