Change in school attendance will be measured by school attendance reports. Number of missed days will be collected

Change in self-concept was measured by the Tennessee Self-Concept Scale Short Form (TSCS). The 20-item short version of the original 100-item TSCS scale assesses adolescents' perception of self-identity and self-satisfaction. The items are rated on a 5-point Likert scale ranging from 1 = always false to 5 = always true. The theoretical range for the TSCS is 20-100, with higher scores representing a more positive self-concept.

change in life satisfaction was measured by Multidimensional Student Life Satisfaction Scale (MSLSS) (Wilson, 2015). The scale includes 40 items assessing life satisfaction on 5 specific aspects while maintaining an overall life satisfaction score (Huebner et al., 1998). Responses are rated on a 6-point Likert scale with 1 = strongly disagree, 2 = moderately disagree, 3 = slightly disagree, 4 = slightly agree, 5 = moderately agree, and 6 = strongly agree (theoretical range 40-240). To avoid the potential for type I error, the total score was summed up and used in the analysis, with the higher values representing higher satisfaction in student life.

Change in stress levels was measured by the Adolescent Stress Questionnaire (ASQ) (Byrne et al., 2007). The ASQ consists of 48 items, each measuring different stressors on a 5-point Likert scale ranging from 1 (not at all stressful) to 5 (very stressful). The questionnaire includes eight subscales: stress of home, stress of school performance, stress of school attendance, stress of peer pressure, stress of teacher interaction, stress about future uncertainty, stress of school or leisure conflict, and stress of financial pressure. We calculated a composite score by summing the responses, with higher scores indicating greater stress levels (theoretical range 48-240).

We assessed family cohesion using seven items (theoretical range: 7-35, Cronbach alpha = 0.84) adapted from both the Family Environment Scale (Moos, 1994) and the Family Assessment Measure (Skinner et al., 1983). The items measured the degree of commitment, help and support family members provide for one another. Participants were asked to rate how often each item occurred in their family using a 5-point scale (with 1 = 'never' and 5 = 'always'). Items included 'Do your family members ask each other for help before asking nonfamily members for help?', and 'Do you listen to what other family members have to say, even when you disagree?' Summary scores were created, with higher scores indicating higher levels of family cohesion.

The perceived child-caregiver relationship scale was adapted from the Family Assessment measure and assessed relationships on two dimensions: (1) acceptance and warmth - the extent to which the caregiver perceives as involved in their child's life; and (2) psychological autonomy - the extent to which the caregiver employs a non-coercive, democratic discipline and encourages the child to express individuality within the family. Participants were asked to rate the adults they live with, on each of the 16 items (range: 16-80), on a 5-point scale (1 = 'never' and 5 = 'always'). Sample items include: "Can you count on your parents to help her if she has a problem?" and "Do your parents keep challenging you to do the best in whatever you do?" Summary mean scores were created, with higher scores indicating a more positive child-caregiver relationship.

Change in social support was measured by the Social Support Behavior Scale. The SS-B measure consists of 45 items designed to tap five modes of support: emotional support, socializing, practical assistance, financial assistance, and advice/guidance. The Likert scale is from 1=strongly disagree to 5=strongly agree. The theoretical range is 45-225, with higher scores indicating higher social support.

Change in perceived social support was measured by the Multidimensional Scale of Perceived Social Support. The MSPSS is a self-report measure to assess participants' social support. The 12-item scale had statements such as, "there is a special person who is around when I am in need," rated on a 5-point Likert scale ranging from strongly agree to strongly disagree, with 5=strongly agree, 4=agree, 3=neutral, 2=disagree, 1=strongly disagree.

Responses were coded and added up. The theoretical range was 12-60, with higher scores indicating greater social support.

We assessed gender norms among caregivers using a 10-item scale adapted from the Gender Norm Attitudes Scale (Waszak et al., 2001), which measures participants' understanding of the appropriateness of behaviors as they relate to being female and male. The items in the scale encompass aspects related to educational performance, future expectations for both genders, family support, encouragement, decision-making, and involvement in intimate relationships and behaviors. The scale items featured binary responses (Agree = 1 and Disagree = 0). To create a summation score, items in the reverse direction were appropriately reverse-coded (theoretical range 0-10). Higher scores denote more egalitarian gender norms and beliefs.

the average bank savings as measured by Bank statements

Change in future orientation was measured by two items evaluating adolescents' expectations and optimism regarding their educational goals. The first item asked participants, How sure are you that you will achieve your educational plans? with response options ranging from 1 = Not at all sure to 5 = Extremely sure. The second item asked, How hopeful are you that you will achieve your educational plans? with parallel response options ranging from 1 = Not at all hopeful to 5 = Extremely hopeful (theoretical range 2-10). For both items, higher scores indicated greater perceived likelihood of achieving educational goals and stronger optimism about the future. A composite future orientation score was created by summing the two items, with higher scores reflecting greater optimism and confidence in achieving educational aspirations. Only participants with non-missing values for both items were included in the composite measure.

Self-esteem was measured using the Rosenberg Self-Esteem Scale (RSES) (Rosenberg, 1965). The scale is a common measure of self-esteem (Sinclair et al., 2010) and has been used across different cultural contexts with high internal consistency, including in Ghana (α = 0.83-0.85) (Ahulu et al., 2020; Glozah, 2014). The scale comprises 10 statements about general feelings of self-worth or self-acceptance rated on a 4-point Likert scale response option (strongly agree to strongly disagree, with 4 = strongly agree, 3 = agree, 2 = disagree, 1 = strongly disagree). The items were scored on a theoretical range of 10-40 and summed, with higher scores representing higher self-esteem.

Emotional self-efficacy was assessed using the Emotional self-efficacy scale (ESE; Valois & Zulig, 2013), an 8-item scale to measure an individual's perceived ability to manage emotions effectively. This scale has been widely used to evaluate emotional self-regulation across different contexts. Participants responded to each item on a 5-point Likert scale ranging from 1 (not at all well) to 5 (very well) (theoretical range 8-40). Sample items included, "How well do you succeed in cheering yourself up when an unpleasant event has happened?" and "How well do you succeed in becoming calm again when you are very scared?" Scores were summed, with higher scores reflecting greater emotional self-efficacy

Social self-efficacy was measured using the Social self-efficacy scale (Zullig, Teoli, Valois, 2011), a 8-item scale to evaluate an individual's confidence in handling social situations and interactions. Participants rated their responses on a 5-point Likert scale, ranging from 1 (not at all well) to 5 (very well). Sample items included "How well can you express your opinions when other classmates disagree with you?" and "How well can you become friends with other children?" The items were summed, with higher scores indicating greater social self-efficacy (theoretical range 8-40).

In addition to bolstered care, participants in this arm will receive the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG).

Anzansi Family Program: the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG). Family EE includes: 1) Workshops on asset building, future planning, and protection from risks; 2) Child Development Account (CDA); and 3) Family income-generating/microenterprise promotion (IGA) component:

MFG a family-centered, group-delivered, evidence-informed intervention designed for children and adolescents whose families struggle with poverty and associated stressors. The MFG is based on building family support through opportunities for parents and children to communicate in a safe setting with other families who have shared experiences, and allow each family to learn from one another. MFG builds protective factors for healthy parent-child relationships while addressing familial, social and community stressors and barriers to adolescent girls' well-being.

Total of all reporting groups

The total number is the sum of caregivers and adolescent girls.

Race and Ethnicity were not collected from any participant.

In addition to bolstered care, participants in this arm will receive the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG).

Anzansi Family Program: the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG). Family EE includes: 1) Workshops on asset building, future planning, and protection from risks; 2) Child Development Account (CDA); and 3) Family income-generating/microenterprise promotion (IGA) component:

MFG a family-centered, group-delivered, evidence-informed intervention designed for children and adolescents whose families struggle with poverty and associated stressors. The MFG is based on building family support through opportunities for parents and children to communicate in a safe setting with other families who have shared experiences, and allow each family to learn from one another. MFG builds protective factors for healthy parent-child relationships while addressing familial, social and community stressors and barriers to adolescent girls' well-being.

Change in attitudes towards school were measured by School attitude assessment survey (SAAS; McCoach, 2002). The survey consists of 20 items rated on a 5-point Likert scale (1 = not at all, 2 = a little bit, 3 = pretty well, 4 = well, and 5 = very well; range 20 to 100). The scale measures aspects of students' lives that predict their academic achievement, including peer attitudes, attitudes toward school, self-motivation, and self-regulation. The items were coded and summed, with the higher values representing higher positive attitudes toward school (Cronbach's alpha = 0.92).

Adolescent girls. One participant was lost to follow-up at 9 months.

Change in school attendance will be measured by school attendance reports. Number of missed days will be collected

This outcome only applies to adolescent girls.

Change in self-concept was measured by the Tennessee Self-Concept Scale Short Form (TSCS). The 20-item short version of the original 100-item TSCS scale assesses adolescents' perception of self-identity and self-satisfaction. The items are rated on a 5-point Likert scale ranging from 1 = always false to 5 = always true. The theoretical range for the TSCS is 20-100, with higher scores representing a more positive self-concept.

Adolescent girls. One participant was lost to follow-up at 9 months.

change in life satisfaction was measured by Multidimensional Student Life Satisfaction Scale (MSLSS) (Wilson, 2015). The scale includes 40 items assessing life satisfaction on 5 specific aspects while maintaining an overall life satisfaction score (Huebner et al., 1998). Responses are rated on a 6-point Likert scale with 1 = strongly disagree, 2 = moderately disagree, 3 = slightly disagree, 4 = slightly agree, 5 = moderately agree, and 6 = strongly agree (theoretical range 40-240). To avoid the potential for type I error, the total score was summed up and used in the analysis, with the higher values representing higher satisfaction in student life.

Adolescent girls. One participant was lost to follow-up at 9 months.

Change in stress levels was measured by the Adolescent Stress Questionnaire (ASQ) (Byrne et al., 2007). The ASQ consists of 48 items, each measuring different stressors on a 5-point Likert scale ranging from 1 (not at all stressful) to 5 (very stressful). The questionnaire includes eight subscales: stress of home, stress of school performance, stress of school attendance, stress of peer pressure, stress of teacher interaction, stress about future uncertainty, stress of school or leisure conflict, and stress of financial pressure. We calculated a composite score by summing the responses, with higher scores indicating greater stress levels (theoretical range 48-240).

Adolescent girls. One participant was lost to follow-up at 9 months.

We assessed family cohesion using seven items (theoretical range: 7-35, Cronbach alpha = 0.84) adapted from both the Family Environment Scale (Moos, 1994) and the Family Assessment Measure (Skinner et al., 1983). The items measured the degree of commitment, help and support family members provide for one another. Participants were asked to rate how often each item occurred in their family using a 5-point scale (with 1 = 'never' and 5 = 'always'). Items included 'Do your family members ask each other for help before asking nonfamily members for help?', and 'Do you listen to what other family members have to say, even when you disagree?' Summary scores were created, with higher scores indicating higher levels of family cohesion.

Adolescent girls. One participant was lost to follow-up at 9 months.

The perceived child-caregiver relationship scale was adapted from the Family Assessment measure and assessed relationships on two dimensions: (1) acceptance and warmth - the extent to which the caregiver perceives as involved in their child's life; and (2) psychological autonomy - the extent to which the caregiver employs a non-coercive, democratic discipline and encourages the child to express individuality within the family. Participants were asked to rate the adults they live with, on each of the 16 items (range: 16-80), on a 5-point scale (1 = 'never' and 5 = 'always'). Sample items include: "Can you count on your parents to help her if she has a problem?" and "Do your parents keep challenging you to do the best in whatever you do?" Summary mean scores were created, with higher scores indicating a more positive child-caregiver relationship.

We analyzed adolescent participant data. One participant was lost to follow-up at 9 months.

Change in social support was measured by the Social Support Behavior Scale. The SS-B measure consists of 45 items designed to tap five modes of support: emotional support, socializing, practical assistance, financial assistance, and advice/guidance. The Likert scale is from 1=strongly disagree to 5=strongly agree. The theoretical range is 45-225, with higher scores indicating higher social support.

Adolescent data were analyzed from baseline to 15 months. One participant was lost to follow-up at 9 months.

Change in perceived social support was measured by the Multidimensional Scale of Perceived Social Support. The MSPSS is a self-report measure to assess participants' social support. The 12-item scale had statements such as, "there is a special person who is around when I am in need," rated on a 5-point Likert scale ranging from strongly agree to strongly disagree, with 5=strongly agree, 4=agree, 3=neutral, 2=disagree, 1=strongly disagree.

Responses were coded and added up. The theoretical range was 12-60, with higher scores indicating greater social support.

Adolescent girls. One participant was lost to follow-up at 9 months.

We assessed gender norms among caregivers using a 10-item scale adapted from the Gender Norm Attitudes Scale (Waszak et al., 2001), which measures participants' understanding of the appropriateness of behaviors as they relate to being female and male. The items in the scale encompass aspects related to educational performance, future expectations for both genders, family support, encouragement, decision-making, and involvement in intimate relationships and behaviors. The scale items featured binary responses (Agree = 1 and Disagree = 0). To create a summation score, items in the reverse direction were appropriately reverse-coded (theoretical range 0-10). Higher scores denote more egalitarian gender norms and beliefs.

Caregiver data

the average bank savings as measured by Bank statements

We analyzed bank savings information from the ANZANSI group. The control group had no bank data. We report the average bank savings for only the ANZANSI Family Program at 9 months in Ghana Cedis.

Change in future orientation was measured by two items evaluating adolescents' expectations and optimism regarding their educational goals. The first item asked participants, How sure are you that you will achieve your educational plans? with response options ranging from 1 = Not at all sure to 5 = Extremely sure. The second item asked, How hopeful are you that you will achieve your educational plans? with parallel response options ranging from 1 = Not at all hopeful to 5 = Extremely hopeful (theoretical range 2-10). For both items, higher scores indicated greater perceived likelihood of achieving educational goals and stronger optimism about the future. A composite future orientation score was created by summing the two items, with higher scores reflecting greater optimism and confidence in achieving educational aspirations. Only participants with non-missing values for both items were included in the composite measure.

Adolescent Change in future orientation will be measured by the Thinking about the future measure. One participant was lost to follow-up at 9 months

Self-esteem was measured using the Rosenberg Self-Esteem Scale (RSES) (Rosenberg, 1965). The scale is a common measure of self-esteem (Sinclair et al., 2010) and has been used across different cultural contexts with high internal consistency, including in Ghana (α = 0.83-0.85) (Ahulu et al., 2020; Glozah, 2014). The scale comprises 10 statements about general feelings of self-worth or self-acceptance rated on a 4-point Likert scale response option (strongly agree to strongly disagree, with 4 = strongly agree, 3 = agree, 2 = disagree, 1 = strongly disagree). The items were scored on a theoretical range of 10-40 and summed, with higher scores representing higher self-esteem.

Adolescent girls. One participant was lost to follow-up at 9 months.

Emotional self-efficacy was assessed using the Emotional self-efficacy scale (ESE; Valois & Zulig, 2013), an 8-item scale to measure an individual's perceived ability to manage emotions effectively. This scale has been widely used to evaluate emotional self-regulation across different contexts. Participants responded to each item on a 5-point Likert scale ranging from 1 (not at all well) to 5 (very well) (theoretical range 8-40). Sample items included, "How well do you succeed in cheering yourself up when an unpleasant event has happened?" and "How well do you succeed in becoming calm again when you are very scared?" Scores were summed, with higher scores reflecting greater emotional self-efficacy

At 9 months follow-up, one person was lost during interviews in the treatment arm. However, this participants was interviewed at 15 months follow-up.

Social self-efficacy was measured using the Social self-efficacy scale (Zullig, Teoli, Valois, 2011), a 8-item scale to evaluate an individual's confidence in handling social situations and interactions. Participants rated their responses on a 5-point Likert scale, ranging from 1 (not at all well) to 5 (very well). Sample items included "How well can you express your opinions when other classmates disagree with you?" and "How well can you become friends with other children?" The items were summed, with higher scores indicating greater social self-efficacy (theoretical range 8-40).

One participant was lost to follow-up at 9 months

In addition to bolstered care, participants in this arm will receive the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG).

Anzansi Family Program: the ANZANSI that combines Family Economic Empowerment (EE) with Multiple Family Groups (MFG). Family EE includes: 1) Workshops on asset building, future planning, and protection from risks; 2) Child Development Account (CDA); and 3) Family income-generating/microenterprise promotion (IGA) component:

MFG a family-centered, group-delivered, evidence-informed intervention designed for children and adolescents whose families struggle with poverty and associated stressors. The MFG is based on building family support through opportunities for parents and children to communicate in a safe setting with other families who have shared experiences, and allow each family to learn from one another. MFG builds protective factors for healthy parent-child relationships while addressing familial, social and community stressors and barriers to adolescent girls' well-being.