Objectives. To examine the impacts of a government-implemented cash plus program on violence experiences and perpetration among Tanzanian adolescents.

Methods. We used data from a cluster randomized controlled trial (n = 130 communities) conducted in the Mbeya and Iringa regions of Tanzania to isolate impacts of the “plus” components of the cash plus intervention. The panel sample comprised 904 adolescents aged 14 to 19 years living in households receiving a government cash transfer. We estimated intent-to-treat impacts on violence experiences, violence perpetration, and pathways of impact.

Results. The plus intervention reduced female participants’ experiences of sexual violence by 5 percentage points and male participants’ perpetration of physical violence by 6 percentage points. There were no intervention impacts on emotional violence, physical violence, or help seeking. Examining pathways, we found positive impacts on self-esteem and participation in livestock tending and, among female participants, a positive impact on sexual debut delays and a negative effect on school attendance.

Conclusions. By addressing poverty and multidimensional vulnerability, integrated social protection can reduce violence.

Public Health Implications. There is high potential for scale-up and sustainability, and this program reaches some of the most vulnerable and marginalized adolescents. (Am J Public Health. 2021;111(12):2227–2238. https://doi.org/10.2105/AJPH.2021.306509)

One billion children experience violence annually,1 and adverse effects include increased risk of injury, mental health problems, early pregnancy, sexual risk taking, and noncommunicable diseases.2–4 Violence experiences and witnessing of violence in childhood increase the risk of experiencing and perpetrating violence in adulthood.5,6

A consortium of international agencies developed INSPIRE, a set of strategies to reduce violence against children (VAC); strategies include income and economic strengthening (such as cash transfers) and life and social skills training.7

Poverty is a structural driver of violence, and there are several pathways through which economic-strengthening programs such as “cash plus,” defined as a combination of cash transfers with additional complementary support or linkages to services,8 may reduce the risk of childhood violence. “Cash plus” may also imact food security, financial empowerment and bargaining power, changes in time use activities (including schooling, labor, and domestic chores), caregiving behaviors, psychosocial well-being, child marriage, and time spent in high-risk settings.9,10

More evidence is still needed on VAC reduction interventions involving men and boys and around economic empowerment.3,11,12 A review of noncontributory social safety net programs, including cash transfers, found that social safety nets can contribute to VAC reduction.10 More recently, studies from Mali and Zimbabwe have demonstrated that household-targeted cash plus programs reduced violence experienced by children and youths.13,14

Violence impacts have also been examined in bundled interventions targeted to adolescents comprising components related to economic strengthening, life skills, and strengthening health capabilities, but the evidence is mixed. Interventions in sub-Saharan Africa show reductions in forced sex in Uganda (Livelihood Empowerment for Adolescents), physical or sexual violence in Zimbabwe (Shaping the Health of Adolescents in Zimbabwe) and Kenya (Adolescent Girls Initiative), and reduced participation in transactional sex in Zambia (Adolescent Girls Empowerment Program).15–18 Nevertheless, in Zambia, the Adolescent Girls Empowerment Program found no impacts on physical violence, intimate partner violence, or “unwanted sex,” and an intervention implemented in Kampala, Uganda, found that providing girls with a savings account but no other life skills training increased the risk of unwanted sexual touching and harassment.15,19 When replicated in Tanzania, the Livelihood Empowerment for Adolescents intervention failed to demonstrate protective impacts,20 while another intervention in Liberia (Girl Empower) did not lead to reductions in violence.21

The aforementioned interventions were implemented by nongovernmental organizations, and, thus, evidence is needed on government-implemented programs, which have greater potential for sustainability and scale-up. Recent studies have called for additional research of at-scale programs to prevent violence within households, as well as programs tailored to reach adolescent girls.22,23 We examined impacts of the “plus” components from a government-run, multisectoral cash plus intervention on male and female adolescents’ experiences and perpetration of violence.

We used data from a longitudinal cluster randomized controlled trial (n = 130 communities), which was implemented by United Nations Children’s Fund (UNICEF) Office of Research‒Innocenti, University at Buffalo, and EDI Global, in collaboration with Tanzania Social Action Fund (TASAF), Tanzania Commission for AIDS, and UNICEF Tanzania.


The cash plus intervention was piloted in 4 districts and targeted adolescents aged 14 to 19 years living in households participating in the Government of the Republic of Tanzania’s flagship social protection program, the Productive Social Safety Net (PSSN). The PSSN reaches 1 million households nationally, providing a cash transfer every other month, livelihoods enhancement, and public works.24 Cash transfer payments were variable depending on school enrollment and health-related coresponsibilities, but these averaged US $7.10 per month (maximum US $21.70 per month), equivalent to approximately 16% of household consumption.24,25 All study households (intervention and control arms) had been enrolled in the PSSN since 2015 and received cash transfers. The intervention that was randomized was the “plus” components targeted to adolescents. We used a sample size calculation to determine the number of clusters required (Appendix A, available as a supplement to the online version of this article at http://www.ajph.org).


Random assignment of 130 villages into study arms (65 intervention and 65 control) was conducted in July 2017, after completion of the baseline surveys, and was stratified by TASAF administrative areas (program administrative areas, corresponding to Mafinga and Mufindi districts in Iringa, and Rungwe and Busokelo districts in Mbeya) and village size (large vs small villages).


The “Ujana Salama” Cash Plus Model for Safe Transitions to a Healthy and Productive Adulthood intervention followed a capabilities approach26,27 to strengthen youth productive, human, and health assets (Conceptual Framework in Appendix B, available as a supplement to the online version of this article at http://www.ajph.org). Guiding principles included government ownership, implementation within the PSSN livelihoods enhancement strategy and existing government frameworks, linkages with other government services, and age- and gender-sensitive livelihoods interventions.

The intervention comprised (1) face-to-face livelihoods and life skills training delivered 2 hours per week over 12 weeks (January‒May 2018); (2) mentoring (occurring biweekly and then monthly between July 2018 and March 2019) and a productive grant (totaling US $80 disbursed in up to 2 payments between March and June 2019), conditional on having attended trainings and developing an approved educational or business plan; and (3) facilitated linkages to adolescent-friendly HIV and sexual and reproductive health services in government health facilities (July 2018‒March 2019; Appendix C, available as a supplement to the online version of this article at http://www.ajph.org).

The training included sessions on livelihoods, sexual and reproductive health, and HIV prevention and treatment, including a bundle of high-impact behavior change communication approaches and peer support groups (topics summarized in Appendix C). During mentoring, adolescents were provided with linkages to training and apprenticeship activities, input on business plans, peer education, and linkages to health facilities (Activities in Appendix D, available as a supplement to the online version of this article at http://www.ajph.org). Two mentors (1 male, 1 female) were selected per village. Addressing supply of health services, UNICEF worked with the Ministry of Health, Community Development, Gender, Elderly, and Children to perform a training in July 2019 with staff at government primary health care facilities in treatment villages to strengthen adolescent-friendly services.

We used TASAF administrative data on PSSN enrollment (in 2015) to identify households that would have adolescents aged 14 to 19 years in 2017. We approached all of these households and aimed to interview all eligible youths (including new youths not registered in the household in 2015; Appendix F, available as a supplement to the online version of this article at http://www.ajph.org). Three rounds of data were collected: baseline (April‒June 2017), round 2 (May‒July 2018), and round 3 (June‒August 2019). We used a split sample approach for modules on violence victimization and perpetration for male and female participants based on best practices guidance, in which male participants answered the violence module in one community and female participants in another, to protect confidentiality of participants and reduce the chance that a perpetrator‒victim pair would both be interviewed.28 We provided anonymized referral information to respondents containing contact numbers for district social welfare officers.29

Interviews were conducted in Swahili with same-sex enumerators in private settings and data were entered into SurveyBe version 5.10.210 (EDI Global Limited, Gerrards Cross, UK) software via tablets.


The primary outcomes were experienced violence (emotional, physical, sexual), help-seeking related to violence, and perpetration of physical and emotional violence in the past 12 months. We assessed these by using an adapted version of questionnaire items used in the World Health Organization (WHO) Multi-country Study on Women’s Health and Domestic Violence,30 which draws on the Conflict Tactics Scale.31 We adapted these items to include any perpetrator (not just intimate partners) for the current study. Binary violence indicators for emotional, physical, and sexual violence were created from multiple items per violence type (Appendix E, available as a supplement to the online version of this article at http://www.ajph.org). At baseline, because of an error in a data-entry skip pattern, sexual violence questions were only asked of individuals who reported ever having had sex. In subsequent rounds, sexual violence questions were asked to all individuals selected for the violence module.

Respondents experiencing any type of violence were then asked whether they had ever tried to seek help or tell anyone about the violence. In line with existing literature,32,33 we further categorized help seeking into formal and informal sources (Appendix E).

We added questions on violence perpetration at round 3. Binary perpetration indicators were created separately for 12-month emotional and physical violence based on items summarized in Appendix D.

Intermediate pathways examined included age at first sexual intercourse, entry into marriage or cohabitation, engaging in an age-disparate sexual relationship, self-perceived stress, self-efficacy, self-esteem, school attendance, and time use (Appendix E). Age-disparate sexual relationship was defined as having a sexual partner 5 or more years older, in line with previous studies.34 Self-perceived stress was measured using the Enhanced Life Distress Inventory, which measures distress across multiple areas of social life and functioning35 and results in an overall scale ranging from 0 to 39 and 3 subscales (economic and health-related well-being, risk and security, relationships). We measured self-efficacy by using a locus-of-control index36,37 and self-esteem by using 2 items from Rosenberg’s self-esteem scale.38 For time use, we examined participation in 5 types of economic activities performed in the past 7 days (farm work, livestock herding, fishing, nonagricultural business, paid work outside the household) and household chores performed in the previous day (collecting water; collecting firewood or other fuel materials; collecting nuts or other tree fruits; taking care of children, cooking, or cleaning; and taking care of the elderly or sick household members). Activities were not mutually exclusive, and multitasking was reported as separate activities. An additional pathway examined elsewhere was gender-equitable attitudes.39

Statistical Analysis

As policymakers are interested in understanding population impacts, we used intention-to-treat analyses to examine impacts of the program, including all clusters and adolescents, irrespective of program uptake (uptake was 48.5% among those eligible),40 among participants who were interviewed at all 3 rounds. We calculated intervention effects with data separately at each follow-up round by using ordinary least squares for continuous outcomes and linear probability models for binary outcomes. For continuous and binary outcomes, we reported b coefficients. We adjusted the regression models for age, sex, and stratum (district and village size), and we adjusted standard errors for clustering and heteroskedasticity by using the VCE (robust) command in Stata version 16 (StataCorp LP, College Station, TX). We further calculated average-treatment-on-the-treated estimates by using 2-stage least squares regressions with linear probability models, in which we predicted program take up in the first-stage regressions and treatment impacts on the treated in the second stage.

We tested program impacts on potential pathways of impact at rounds 2 and 3, and then tested pathway impacts by gender at round 3 only. We describe perpetrators of violence descriptively only, given low proportions in the categories of perpetrators and limited power to estimate program impacts.

Out of 3599 adolescents found eligible at baseline, a total of 2458 completed interviews (68%). Among this baseline evaluation sample (n = 2458), approximately half (n = 1165) were interviewed for the violence questions, and, among these, 988 and 1033 were followed up at rounds 2 and 3, respectively, representing the panel samples (Appendix F). Among the panel sample interviewed at all 3 rounds (n = 904), no adolescents had missing values for variables of interest. The average age of the sample at baseline was 16 years, 47.4% of the sample were female, and 57% were attending school (Table 1). All background characteristics were balanced at baseline between study arms.


TABLE 1— Sample Characteristics at Baseline by Treatment Status: Tanzania Adolescent Cash Plus Study, 2017, Iringa and Mbeya Regions of Tanzania

TABLE 1— Sample Characteristics at Baseline by Treatment Status: Tanzania Adolescent Cash Plus Study, 2017, Iringa and Mbeya Regions of Tanzania

Variables Full Sample (n = 904), Mean ±SD or No. (%) Cash Plus (Intervention; n = 440), Mean ±SD or No. (%) Cash Only (Control; n = 464), Mean ±SD or No. (%)
General characteristics
Female 429 (47) 204 (46) 225 (48)
Age, y 16.0 ±1.6 15.9 ±1.5 16.0 ±1.6
Age of household head, y 58.7 ±16.4 58.8 ±16.3 58.5 ±16.5
Female-headed household 603 (67) 288 (65) 315 (68)
Mufindi or Mafinga districts 475 (53) 239 (54) 236 (51)
Primary outcomes
Experiences of violence
 Emotional 318 (35) 136 (31) 182 (39)
 Physical 247 (27) 109 (25) 138 (30)
 Sexuala 8 (1) 4 (1) 4 (1)
Help seeking for emotional or physical violence (n = 402)
 Any 128 (32) 50 (28) 78 (35)
 Formal 34 (8) 14 (8) 20 (9)
 Informal 99 (25) 35 (20) 64 (28)
Intermediate pathway outcomes
Age at first sexual intercourse (n = 137), y 15.9 ±1.8 15.8 ±1.7 15.9 ±2.0
Age-disparate sex (female participants only; n = 68) 18 (26) 6 (19) 12 (33)
Locus of control index 3.2 ±0.5 3.3 ±0.5 3.2 ±0.5
Self-esteem index 3.9 ±0.8 4.0 ±0.8 3.9 ±0.8
ELDI (0–39) 3.3 ±4.7 3.3 ±4.7 3.3 ±4.6
 ELDI economic and health-related well-being subscale 2.7 ±3.5 2.6 ±3.5 2.8 ±3.5
 ELDI risk subscale 0.3 ±1.0 0.3 ±1.0 0.2 ±0.9
 ELDI relations subscale 0.3 ±1.2 0.4 ±1.3 0.3 ±1.1
Currently attending school 512 (57) 254 (58) 258 (56)
Has a spouse or cohabitating partner 1 (0) 0 (0) 1 (0)
Time use indicators (% participating)
 Any economic activities 709 (78) 347 (79) 362 (78)
 Paid work outside the household 121 (13) 52 (12) 69 (15)
 Farm work for the household, excluding livestock 598 (66) 298 (68) 300 (65)
 Livestock herding for the household 433 (48) 213 (48) 220 (47)
 Fishing for the household 14 (2) 7 (2) 7 (2)
 Household business 41 (5) 16 (4) 25 (5)
 Any chores 811 (90) 404 (92) 407 (88)
 Collecting water 607 (67) 316 (72) 291 (63)
 Collecting firewood 358 (40) 193 (44) 165 (36)
 Collecting nuts and other tree fruits 120 (13) 63 (14) 57 (12)
 Taking care of children, cooking, or cleaning 661 (73) 327 (74) 334 (72)
 Taking care of elderly or sick 211 (23) 109 (25) 102 (22)

Note. ELDI = Enhanced Life Distress Inventory. Economic activities included those conducted in the previous week, and time-use activities have a reference period of the previous day.

aThe sexual violence indicator included in this table differs from those used in impacts analyses in Tables 2 and 3 because of a skip pattern error at baseline, whereby only adolescents who reported having had sexual intercourse were asked questions about sexual violence. In subsequent waves, all youths in the violence module were asked questions about sexual violence, which may include other forced sexual acts.


At baseline, 35%, 27%, and 1% of adolescents reported having experienced emotional, physical, and sexual violence, respectively (Table 1). Physical and sexual violence outcomes were balanced at baseline in the pooled (male and female participants) sample, while emotional violence was not (39% of the treatment group vs 31% of controls had experienced emotional violence). Among those who had experienced physical or emotional violence (n = 404), 31.7% reported seeking help or disclosing to someone (8.4% to formal sources vs 24.5% to informal sources). Examining indicators by gender, we found evidence for baseline balance for all characteristics and outcomes, except farm work for the household excluding livestock among female participants and experience of physical violence and informal reporting of violence among male participants (Appendix G, available as a supplement to the online version of this article at http://www.ajph.org). By round 3, experiences of violence among the control (treatment) group were 33% (29%) for emotional violence, 13% (11.1%) for physical violence, and 6% (2%) for sexual violence (Table 2).


TABLE 2— Intervention Effects (Intent-to-Treat) on Violence at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

TABLE 2— Intervention Effects (Intent-to-Treat) on Violence at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

Variables Cash Plus, No. (%) Cash Only, No. (%) b (95% CI)
Experienced 440 464
 Emotional violence 126 (29) 155 (33) −0.05 (−0.11, 0.02)
 Physical violence 49 (11) 58 (13) −0.01 (−0.06, 0.03)
 Sexual violence 10 (2) 26 (6) −0.03 (−0.06, 0.00)
 Emotional, physical, or sexual violence 148 (34) 181 (39) −0.05 (−0.12, 0.02)
Sought help 148 181
 Any 61 (41) 76 (42) −0.02 (−0.13, 0.10)
 Formal 11 (7) 11 (6) 0.02 (−0.03, 0.07)
 Informal 54 (36) 65 (36) 0.00 (−0.11, 0.11)
Perpetrated 440 464
 Emotional violence 19 (4) 28 (6) −0.02 (−0.05, 0.02)
 Physical violence 16 (4) 30 (6) −0.03 (−0.06, 0.00)

Note. CI = confidence interval. Linear probability models, male and female participants. Models controlled for age, gender, and stratum (district and village size); only coefficients on treatment indicator are shown. Standard errors are clustered at the community level.

Intervention Impacts

By round 3, treatment adolescents had a 3-percentage-point reduction of experiencing sexual violence as a result of the intervention (b = −0.03; 95% confidence interval [CI] = −0.06, −0.00; Table 2). The intervention had no impact on emotional or physical violence experiences in the pooled sample, nor on help seeking (formal or informal sources). When examining impacts separately by gender, we saw that the sexual violence impacts were driven by the female sample, who experienced a 5-percentage-point reduction in sexual violence (b = −0.05; 95% CI = −0.10, −0.00; Table 3) as a result of the program (there were no impacts on this indicator among male participants, who reported low rates of sexual violence; Table 3). Impacts at round 2 are presented in Appendix H (available as a supplement to the online version of this article at http://www.ajph.org).


TABLE 3— Intervention Effects (Intent-to-Treat) on Violence at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

TABLE 3— Intervention Effects (Intent-to-Treat) on Violence at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

Variables Cash Plus, No. (%) Cash Only, No. (%) b (95% CI)
Male participants
Experienced 236 239
 Emotional violence 58 (25) 77 (32) −0.07 (−0.16, 0.01)
 Physical violence 23 (10) 30 (13) −0.01 (−0.07, 0.05)
 Sexual violence 5 (2) 11 (5) −0.03 (−0.07, 0.01)
Sought help 65 90
 Any 24 (37) 34 (38) 0.00 (−0.17, 0.16)
 Formal 6 (9) 7 (8) 0.01 (−0.07, 0.09)
 Informal 21 (32) 26 (29) 0.05 (−0.10, 0.20)
Perpetrated 236 239
 Emotional violence 10 (4) 13 (5) −0.01 (−0.05, 0.03)
 Physical violence 5 (2) 20 (8) −0.06 (−0.10, −0.02)
Female participants
Experienced 204 225
 Emotional violence 68 (33) 78 (35) −0.01 (−0.11, 0.09)
 Physical violence 26 (13) 28 (12) 0.00 (−0.06, 0.06)
 Sexual violence 5 (2) 15 (7) −0.05 (−0.10, 0.00)
Sought help 83 91
 Any 37 (45) 42 (46) −0.05 (−0.20, 0.10)
 Formal 5 (6) 4 (4) 0.02 (−0.06, 0.10)
 Informal 33 (40) 39 (43) −0.06 (−0.21, 0.10)
Perpetrated 204 225
 Emotional violence 9 (4) 15 (7) −0.02 (−0.08, 0.03)
 Physical violence 11 (5) 10 (4) 0.01 (−0.03, 0.06)

Note. CI = confidence interval. Linear probability models, by gender. Models controlled for age, gender, and stratum (district and village size); only coefficients on treatment indicator are shown. Standard errors are clustered at the community level.

In the pooled sample we found no impacts on emotional and physical violence perpetration. However, when we examined male and female participants separately, we found that male participants were less likely to report physical violence perpetration as a result of the intervention (b = −0.06; 95% CI = −0.10, −0.02; Table 3). There were no impacts on emotional violence perpetration among male participants nor on either outcome among female participants.

There were no impacts on help seeking among either gender (Table 3).

Pathways Analyses

We found positive impacts on participation in livestock herding for the household (b = 0.09; 95% CI = 0.02, 0.17; Table 4) and self-esteem (b = 0.19; 95% CI = 0.08, 0.29). There were no intervention impacts on self-perceived stress, self-efficacy (locus of control), marriage or cohabitation, school attendance, or age-disparate relationships. When we examined pathway impacts separately by gender (Appendix I, available as a supplement to the online version of this article at http://www.ajph.org), we saw that changes in livestock herding were driven by female participants (b = 0.14; 95% CI = 0.03, 0.24), while we observed increases in self-esteem for both female (b = 0.19; 95% CI = 0.05, 0.33) and male participants (b = 0.29; 95% CI = 0.05, 0.36). In addition, among female participants, we found a protective impact on delaying age at first sexual intercourse (b = −0.55; 95% CI = −1.02, −0.09) and a negative impact on school attendance (b = −0.10; 95% CI = −0.19, 0.00). This delay in sexual debut translates to approximately half a year (0.55 × 12 = 6.6 months).


TABLE 4— Intervention Effects on Intermediate Pathway Indicators at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

TABLE 4— Intervention Effects on Intermediate Pathway Indicators at Round 3: Tanzania Adolescent Cash Plus Study, 2019, Iringa and Mbeya Regions of Tanzania

Variables Cash Plus (n = 440), Mean ±SD or No. (%) Cash Only (n = 464), Mean ±SD or No. (%) b (95% CI)
Age at first sexual intercourse, y −0.27 (−0.61, 0.08)
 Mean ±SD 16.9 ±1.4 17.2 ±1.8
 Total no. 151 161
Age-disparate sex (female participants only) −0.10 (−0.23, 0.03)
 No. (%) 23 (30) 34 (40)
 Total no. 77 85
Locus of control index 3.3 ±0.5 3.3 ±0.5 0.03 (−0.04, 0.10)
Self-esteem index 3.9 ±0.8 3.8 ±0.8 0.19 (0.08, 0.29)
ELDI (0‒39) 3.5 ±4.9 4.1 ±5.3 −0.61 (−1.42, 0.21)
 ELDI economic and health-related well-being subscale 2.8 ±3.5 3.4 ±4.0 −0.59 (−1.12, −0.06)
 ELDI risk subscale 0.3 ±1.2 0.3 ±1.0 −0.01 (−0.19, 0.17)
 ELDI relations subscale 0.4 ±1.2 0.4 ±1.4 0.00 (−0.24, 0.23)
Currently attending school 152 (35) 174 (38) −0.04 (−0.10, 0.03)
Has a spouse or cohabiting partner 30 (7) 25 (5) 0.02 (−0.01, 0.05)
Time use 440 464
 Any economic activities 359 (82) 376 (81) 0.01 (−0.05, 0.06)
 Paid work outside the household 108 (25) 106 (23) 0.02 (−0.04, 0.07)
 Farm work for the household 265 (60) 261 (56) 0.04 (−0.03, 0.11)
 Livestock herding for the household 260 (59) 232 (50) 0.09 (0.02, 0.17)
 Fishing for the household 9 (2) 9 (2) 0.00 (−0.02, 0.02)
 Household business 69 (16) 68 (15) 0.01 (−0.04, 0.07)
 Any chores 393 (89) 412 (89) 0.01 (−0.04, 0.05)
 Collecting water 325 (74) 339 (73) 0.01 (−0.06, 0.08)
 Collecting firewood 148 (34) 122 (26) 0.07 (−0.01, 0.14)
 Collecting nuts and other tree fruits 36 (8) 23 (5) 0.03 (0.00, 0.07)
 Taking care of children, cooking, or cleaning 303 (69) 327 (70) −0.01 (−0.08, 0.06)
 Taking care of elderly or sick 81 (18) 83 (18) 0.01 (−0.05, 0.06)

Note. CI = confidence interval; ELDI = Enhanced Life Distress Inventory. Ordinary least squares and linear probability models (intent-to-treat estimates). Standard errors are clustered at the community level. Economic activities included those conducted in the previous week, and time use activities had a reference period of the previous day. Models controlled for age, gender, and stratum (district and village size); only coefficients on treatment indicator are shown.

Robustness Check

Difference-in-differences models confirmed findings from the single-difference models with respect to null impacts on emotional and physical violence (Appendix J, available as a supplement to the online version of this article at http://www.ajph.org).

Average-Treatment-on-the-Treated Analyses

Average-treatment-on-the-treated estimates were in the same direction and maintained the same significance levels but were generally 2 to 3 times larger in magnitude as compared with intention-to-treat estimates (Appendices K, L, and M, available as supplements to the online version of this article at http://www.ajph.org). We found impacts on the economic and health-related well-being stress Enhanced Life Distress Inventory subscale to be statistically significant (b = −1.40; 95% CI = −2.65, −0.14) in the average-treatment-on-the-treated models, but not in intention-to-treat models.


Among those who reported emotional or physical violence at baseline (n = 402), the most common perpetrators were peers (45%) and family members (41%), followed by partner or spouse (27%), and then authorities (21%; Appendix M). Other perpetrators were reported by only 3% of respondents. Subsequently, at round 3, those reporting physical or emotional violence (n = 316) reported the most common perpetrators as peers (44%), partner or spouse (34%), and family members (25%; Appendix N, available as a supplement to the online version of this article at http://www.ajph.org).

We examined the effects of a government-implemented cash plus program targeted to adolescents and found that the intervention reduced experience of sexual violence among female participants and perpetration of physical violence among male participants.

With respect to pathways, we found that the program increased self-esteem and led to changes in time use—namely, participation in livestock herding among female participants. We also found that the program delayed age at sexual debut by half a year and reduced school attendance among female participants. We found no impacts on other pathways. By delaying the window of exposure to sexual relationships, this may have prevented intimate partner sexual violence. Increases in self-esteem may have led to reduced sexual violence via improved assertiveness, communication, and interpersonal skills learned in the trainings. Previous research has found that a lack of these skills may contribute to sexual violence risk in southern Africa.41 This may, in part, explain why we found protective effects for sexual violence but no other forms of violence. Moreover, previous evidence demonstrates a link between low self-esteem and increased risk of both experiencing and perpetrating dating violence among adolescents,42 and given positive impacts on self-esteem in our study, this is a potential pathway for the observed reductions in violence experiences and perpetration.

In a context with limited formal employment opportunities, when older female participants were presented with the opportunity to start a business, they may have opted to do so, possibly explaining negative school attendance impacts. Changes in school attendance could have affected exposure to environments where adolescents are at risk for violence, but our data do not allow us to make more detailed conclusions about this pathway. Although we did not ask about perpetrators of sexual violence because of the sensitive nature of this information, peers and authority figures were common perpetrators of emotional and physical violence. These findings are supported by a global meta-analysis that found that student peers are the second-most-common perpetrators of VAC.22

The observed impact on livestock herding is likely a result of the training around business planning and starting a business, and some participants invested in small livestock with the aim of selling them for a profit. Some program mentors were agriculture extension workers, which may have also influenced choices. Another study from this sample found that engaging in paid work outside the household was associated with increased risk of experiencing sexual violence and emotional violence, while livestock herding was associated with increased risk of experiencing emotional and physical violence perpetrated by family members (T. P., written communication, July 15, 2020). Thus, more participation in livestock herding induced by the intervention may be protective against sexual violence when resulting from a substitution from paid labor outside the household. However, we found no simultaneous decreases in the latter.

The intervention led to an increase in equitable gender attitudes (including around violence) among male participants, examined elsewhere.39 This pathway may partly explain decreases in physical violence perpetration, as attitudes about sex, gender, normative behaviors, and hostile masculinity are learned in adolescence,41,43 and, thus, this is a key window to model equitable gender norms and healthy relationships.44 These findings may have implications for reducing future violence perpetration.

Our findings are consistent with those from an intervention in Uganda16 but are in contrast with studies from Uganda, Zambia, and Liberia that did not find protective effects against violence.15,19,21 These mixed findings underscore that context matters in violence-reduction interventions, as do program design and implementation. Among the cited studies, there is variation in program components, implementing agencies, and dosage of exposure to components. It is difficult to make conclusions about which combinations are most effective in reducing VAC because few studies have tested different combinations of components, and there have been limited examples of the same program being implemented in multiple contexts. In our own study, we were unable to distinguish impacts of the individual intervention components because we could not vary program components across villages.

Studies that have evaluated impacts of cash transfers (in contrast with our study, which only evaluated the “plus” components in cash plus) have found positive impacts on school attendance,45,46 reductions in intimate partner violence among adolescents,34 and reductions in other violence-related outcomes among children and adolescents.10 These studies are not in contrast with our findings, because the interventions studied were different (cash transfers vs “plus” components). Other studies examining effects of cash plus programs have found that these reduced violence against adolescents and youths,14 as well as harsh discipline or corporal punishment among young children.13,47,48


Generalizability of the results to all adolescents in Tanzania is limited. In addition, the productive grant was disbursed shortly before round 3, and, thus, full impacts of the intervention may not have fully materialized. During round 3 fieldwork, households experienced a delay in PSSN payments for the first time, and this may have mitigated impacts of the cash plus program. Underreporting of violence is not expected to be correlated with treatment status and therefore should not bias impact estimates. Perpetration of physical and emotional violence was not assessed at baseline so we could not assess balance between intervention arms before program roll-out. Our study design did not allow us to disentangle effects of the different program components, and, thus, we could not conclude which components may have contributed more to violence reduction. Finally, it is possible that our subanalyses were subject to type II errors, as power is reduced as sample sizes are reduced, and ability to make conclusions about heterogeneous treatment effects is limited.

A key factor in the successful implementation of this intervention was multisectoral coordination, including at the district level, where staff were based and activities were implemented. While most program components were administered through PSSN structures, health facilities strengthening was carried out by Ministry of Health, Community Development, Gender, Elderly, and Children, with technical support from UNICEF.


Multisectoral approaches are increasingly advocated to address the drivers of violence. Our study showed that a government-run, multisectoral cash plus intervention reduced violence experiences and perpetration. The broad potential reach of similar scalable interventions has implications for VAC reduction at the population level.

See also Galea and Vaughan, p. 2094.


Funding for this pilot and evaluation has been provided by Oak Foundation (OCAY-16-73) and UNICEF. Additional funding for the evaluation (2017‒2019) was provided by the United Kingdom’s Department of International Development (DFID 203529-102) and the Swedish Development Cooperation Agency (Sida G41102), both through a grant to UNICEF Office of Research‒Innocenti supporting the Transfer Project. Additional funding for program implementation activities (2018‒2020) was provided by Irish Aid (IA-TAN/2019/064).

Preliminary findings from this study were presented at the Population Association of America Annual Meeting virtually in May 2021 and at the AIDS 2020 23rd International AIDS Conference virtually in July 2020.

The authors would like to acknowledge the support of the Tanzania Social Action Fund (TASAF) and Tanzania Commission for AIDS (TACAIDS)—in particular, Ladislaus Mwamanga (TASAF), Amadeus Kamagenge (TASAF), and Fariji Mishael (TASAF) for the implementation of this evaluation. In addition, the UNICEF personnel instrumental to the initial planning stages of this pilot and study included Alison Jenkins, Beatrice Targa, Patricia Lim Ah Ken, Victoria Chuwa, Naomi Neijhoft, and Tulanoga Matwimbi. We would also like to acknowledge the hard-working field teams of EDI Global, who conducted the data collection for this study to the highest professional standards.

Members of the Tanzania Cash Plus Evaluation Team were as follows: University at Buffalo: Tia Palermo (co‒principal investigator); UNICEF Office of Research: Lusajo Kajula, Jacobus de Hoop, Leah Prencipe, Valeria Groppo, and Jennifer Waidler; EDI Global: Johanna Choumert Nkolo (co‒principal investigator), Respichius Mitti (co‒principal investigator), Nathan Sivewright, Koen Leuveld, Callum Taylor, and Bhoke Munanka; TASAF: Paul Luchemba and Tumpe Mnyawami Lukongo; TACAIDS: Aroldia Mulokozi; UNICEF Tanzania: Ulrike Gilbert, Paul Quarles van Ufford, Rikke Le Kirkegaard, Frank Eetaama, and Jennifer Matafu.

Note. The funders had no role in analysis or interpretation of data. The findings, interpretations, and conclusions expressed in this article are those of the authors and are not necessarily the position of UNICEF or the Government of the United Republic of Tanzania.


The authors have no competing financial and nonfinancial interests to declare.


Ethics approval for the study was granted by the National Institute for Medical Research (NIMR/HQ/R.8a/Vol.IX/2784) and the Tanzania Commission for Science and Technology. Informed assent and parental consent were obtained for all unmarried adolescents aged 14 to 17 years, and written informed consent was obtained directly for those aged 18 and 19 years and married adolescents aged 15 to 17 years. There were no married adolescents younger than 15 years, which was the legal age of marriage in Tanzania for girls. The study was retrospectively registered with the Pan African Clinical Trial Registry as PACTR201804003008116.


1. Hillis S, Mercy J, Amobi A, Kress H. Global prevalence of past-year violence against children: a systematic review and minimum estimates. Pediatrics. 2016;137(3):e20154079. https://doi.org/10.1542/peds.2015-4079 Crossref, MedlineGoogle Scholar
2. Ellsberg M, Jansen HAFM, Heise L, Watts CH, Garcia-Moreno C, WHO Multi-country Study on Women's Health and Domestic Violence Against Women Study Team. Intimate partner violence and women’s physical and mental health in the WHO Multi-country Study on Women’s Health and Domestic Violence: an observational study. Lancet. 2008;371(9619):11651172. https://doi.org/10.1016/S0140-6736(08)60522-X Crossref, MedlineGoogle Scholar
3. Hillis SD, Mercy JA, Saul JR. The enduring impact of violence against children. Psychol Health Med. 2017;22(4):393405. https://doi.org/10.1080/13548506.2016.1153679 Crossref, MedlineGoogle Scholar
4. Chiang LF, Chen J, Gladden MR, et al. HIV and childhood sexual violence: implications for sexual risk behaviors and HIV testing in Tanzania. AIDS Educ Prev. 2015;27(5):474487. https://doi.org/10.1521/aeap.2015.27.5.474 Crossref, MedlineGoogle Scholar
5. Abramsky T, Watts CH, Garcia-Moreno C, et al. What factors are associated with recent intimate partner violence? Findings from the WHO Multi-country Study on Women’s Health and Domestic Violence. BMC Public Health. 2011;11(1):109. https://doi.org/10.1186/1471-2458-11-109 Crossref, MedlineGoogle Scholar
6. Fulu E, Jewkes R, Roselli T, Garcia-Moreno C; UN Multi-country Cross-sectional Study on Men and Violence research team. Prevalence of and factors associated with male perpetration of intimate partner violence: findings from the UN Multi-country Cross-Sectional Study on Men and Violence in Asia and the Pacific. Lancet Glob Health. 2013;1(4):e187e207. https://doi.org/10.1016/S2214-109X(13)70074-3 Crossref, MedlineGoogle Scholar
7. INSPIRE: seven strategies for ending violence against children. Geneva, Switzerland: World Health Organization; 2016. Google Scholar
8. Roelen K, Devereux S, Abdulai AG, Martorano B, Palermo T, Ragno LP. How to make “cash plus” work: linking cash transfers to services and sectors. Florence, Italy: UNICEF Office of Research; 2017. Google Scholar
9. Buller AM, Peterman A, Ranganathan M, Bleile S, Hidrobo M, Heise L. A mixed-method review of cash transfers and intimate partner violence in low-and middle-income countries. World Bank Res Obs. 2018;33(2):218258. https://doi.org/10.1093/wbro/lky002 CrossrefGoogle Scholar
10. Peterman A, Neijhoft AN, Cook S, Palermo TM. Understanding the linkages between social safety nets and childhood violence: a review of the evidence from low- and middle-income countries. Health Policy Plan. 2017;32(7): 10491071. https://doi.org/10.1093/heapol/czx033 Crossref, MedlineGoogle Scholar
11. Ellsberg M, Arango DJ, Morton M, et al. Prevention of violence against women and girls: what does the evidence say? Lancet. 2015;385(9977): 15551566. https://doi.org/10.1016/S0140-6736(14)61703-7 Crossref, MedlineGoogle Scholar
12. Gibbs A, Willan S, Misselhorn A, Mangoma J. Combined structural interventions for gender equality and livelihood security: a critical review of the evidence from southern and eastern Africa and the implications for young people. J Int AIDS Soc. 2012;15(3 suppl 1):110. https://doi.org/10.7448/IAS.15.3.17362 MedlineGoogle Scholar
13. Heath R, Hidrobo M, Roy S. Cash transfers, polygamy, and intimate partner violence: experimental evidence from Mali. J Dev Econ. 2020;143:102410. https://doi.org/10.1016/j.jdeveco.2019.102410 CrossrefGoogle Scholar
14. Chakrabarti A, Handa S, Angeles G, Seidenfeld D. A cash plus program reduces youth exposure to physical violence in Zimbabwe. World Dev. 2020;134:105037. https://doi.org/10.1016/j.worlddev.2020.105037 Crossref, MedlineGoogle Scholar
15. Austrian K, Hewett PC, Soler-Hampejsek E, Bozzani F, Behrman JR, Digitale J. Adolescent Girls Empowerment Programme: research and evaluation mid-term technical report. Nairobi, Kenya: Population Council; 2016. https://doi.org/10.31899/pgy9.1005 CrossrefGoogle Scholar
16. Bandiera O, Buehren N, Burgess R, et al. Women’s empowerment in action: evidence from a randomized control trial in Africa. Am Econ J. 2020;12(1):210259. https://doi.org/10.1257/app.20170416 Google Scholar
17. Dunbar MS, Dufour MSK, Lambdin B, Mudekunye-Mahaka I, Nhamo D, Padian NS. The SHAZ! Project: results from a pilot randomized trial of a structural intervention to prevent HIV among adolescent women in Zimbabwe. PLoS One. 2014;9(11):e113621. https://doi.org/10.1371/journal.pone.0113621 Crossref, MedlineGoogle Scholar
18. Austrian K, Soler-Hampejsek E, Mumah J, et al. Adolescent Girls Initiative‒Kenya: midline results report. Nairobi, Kenya: Population Council; 2018. https://doi.org/10.31899/pgy14.1026 CrossrefGoogle Scholar
19. Austrian K, Muthengi E. Can economic assets increase girls’ risk of sexual harassment? Evaluation results from a social, health and economic asset-building intervention for vulnerable adolescent girls in Uganda. Child Youth Serv Rev. 2014;47(2):168175. https://doi.org/10.1016/j.childyouth.2014.08.012 CrossrefGoogle Scholar
20. Buehren N, Goldstein M, Gulesci S, Sulaiman M, Yam V. Evaluation of an adolescent development program for girls in Tanzania. Washington, DC: World Bank; 2017. https://doi.org/10.1596/1813-9450-7961 CrossrefGoogle Scholar
21. Özler B, Hallman K, Guimond MF, Kelvin EA, Rogers M, Karnley E. Girl Empower—a gender transformative mentoring and cash transfer intervention to promote adolescent wellbeing: impact findings from a cluster-randomized controlled trial in Liberia. SSM Popul Health. 2019;10:100527. https://doi.org/10.1016/j.ssmph.2019.100527 Crossref, MedlineGoogle Scholar
22. Devries K, Knight L, Petzold M, et al. Who perpetrates violence against children? A systematic analysis of age-specific and sex-specific data. BMJ Paediatr Open. 2018;2(1):e000180. https://doi.org/10.1136/bmjpo-2017-000180 Crossref, MedlineGoogle Scholar
23. Stöckl H, March L, Pallitto C, Garcia-Moreno C, WHO Multi-country Study Team. Intimate partner violence among adolescents and young women: prevalence and associated factors in nine countries: a cross-sectional study. BMC Public Health. 2014;14(1):751. https://doi.org/10.1186/1471-2458-14-751 Crossref, MedlineGoogle Scholar
24. Rosas N, Zaldivar S, Granata J, et al. Evaluating Tanzania’s Productive Social Safety Net: findings from the midline survey. Washington, DC: World Bank and Tanzania Social Action Fund; 2019. Google Scholar
25. Tanzania PSSN Youth Study Evaluation Team. Tanzania Youth Study of the Productive Social Safety Net (PSSN) impact evaluation: endline report. Florence, Italy: UNICEF Office of Research–Innocenti; 2018. Google Scholar
26. Gender and adolescence: why understanding adolescent capabilities, change strategies and contexts matters. London, England: GAGE Consortium, Overseas Development Institute; 2017. Google Scholar
27. Sen A. Human rights and capabilities. J Hum Dev. 2005;6(2):151166. https://doi.org/10.1080/14649880500120491 CrossrefGoogle Scholar
28. Tanzania UNICEF, US Centers for Disease Control and Prevention, Muhimbili University of Health and Allied Sciences. Violence against children in Tanzania: findings from a national survey, 2009. Summary report on the prevalence of sexual, physical and emotional violence, context of sexual violence, and health and behavioural consequences of violence experienced in childhood. Dar es Salaam, Tanzania: UNICEF Tanzania; 2011. Google Scholar
29. Researching Violence Against Women: Practical Guidelines for Researchers and Activists. Geneva, Switzerland: World Health Organization; 2005. Google Scholar
30. Garcia-Moreno C, Jansen HAFM, Ellsberg M, Heise L, Watts CH, WHO Multi-country Study on Women’s Health and Domestic Violence Against Women Study Team. Prevalence of intimate partner violence: findings from the WHO Multi-country Study on Women’s Health and Domestic Violence. Lancet. 2006;368(9543):12601269. https://doi.org/10.1016/S0140-6736(06)69523-8 Crossref, MedlineGoogle Scholar
31. Straus MA, Hamby SL, Boney-McCoy S, Sugarman DB. The revised Conflict Tactics Scales (CTS2) development and preliminary psychometric data. J Fam Issues. 1996;17(3):283316. https://doi.org/10.1177/019251396017003001 CrossrefGoogle Scholar
32. Palermo T, Bleck J, Peterman A. Tip of the iceberg: reporting and gender-based violence in developing countries. Am J Epidemiol. 2014;179(5):602612. https://doi.org/10.1093/aje/kwt295 Crossref, MedlineGoogle Scholar
33. Pereira A, Peterman A, Neijhoft AN, et al. Disclosure, reporting and help seeking among child survivors of violence: a cross-country analysis. BMC Public Health. 2020;20(1):1051. https://doi.org/10.1186/s12889-020-09069-7 Crossref, MedlineGoogle Scholar
34. Pettifor A, MacPhail C, Hughes JP, et al. The effect of a conditional cash transfer on HIV incidence in young women in rural South Africa (HPTN 068): a phase 3, randomised controlled trial. Lancet Glob Health. 2016;4(12):e978e988. https://doi.org/10.1016/S2214-109X(16)30253-4 Crossref, MedlineGoogle Scholar
35. Palermo T, Hall BJ, Cirillo C, Tanzania PSSN Youth Study Evaluation Team, Tanzania Adolescent Cash Plus Study Team, and Ghana LEAP 1000 Evaluation Team in Acknowledgement. Enhanced Life Distress Inventory: development and validation in two African countries. Br J Health Psychol. 2020;25(3):728753. https://doi.org/10.1111/bjhp.12448 Crossref, MedlineGoogle Scholar
36. Levenson H. Differentiating among internality, powerful others, and chance. In: Lefcourt HM. Research With the Locus of Control Construct. New York, NY: Academic Press; 1981:1563. https://doi.org/10.1016/B978-0-12-443201-7.50006-3 CrossrefGoogle Scholar
37. Laajaj R, Macours K. Measuring skills in developing countries. Washington, DC: The World Bank; 2017. https://doi.org/10.1596/1813-9450-8000 CrossrefGoogle Scholar
38. Rosenberg M. Society and the Adolescent Self-Image. Princeton, NJ: Princeton University Press; 1965. https://doi.org/10.1515/9781400876136 CrossrefGoogle Scholar
39. Chzhen Y, Prencipe L, Eetaama F, Luchemba P, Lukongo TM, Palermo T. Impacts of a cash plus intervention on gender attitudes among Tanzanian adolescents. J Adolesc Health. 2021;68(5):899905. https://doi.org/10.1016/j.jadohealth.2020.07.025 Crossref, MedlineGoogle Scholar
40. Tanzania Cash Plus Evaluation Team. A cash plus model for safe transitions to a healthy and productive adulthood: round 3 report. Florence, Italy: UNICEF Office of Research‒Innocenti: 2021. Google Scholar
41. Petersen I, Bhana A, McKay M. Sexual violence and youth in South Africa: the need for community-based prevention interventions. Child Abuse Negl. 2005;29(11):12331248. https://doi.org/10.1016/j.chiabu.2005.02.012 Crossref, MedlineGoogle Scholar
42. Pflieger JC, Vazsonyi AT. Parenting processes and dating violence: the mediating role of self-esteem in low-and high-SES adolescents. J Adolesc. 2006;29(4):495512. https://doi.org/10.1016/j.adolescence.2005.10.002 Crossref, MedlineGoogle Scholar
43. Levant RF, Brooks GR. Men and Sex: New Psychological Perspectives. New York, NY: John Wiley & Sons Inc; 1997. Google Scholar
44. John N, Stoebenau K, Ritter S, Edmeades J, Balvin N. Gender socialization during adolescence in low- and middle-income countries: conceptualization, influences and outcomes. Innocenti Discussion Papers. Florence, Italy: UNICEF Office of Research; 2017. Google Scholar
45. Baird S, Ferreira FHG, Özler B, Woolcock M. Conditional, unconditional and everything in between: a systematic review of the effects of cash transfer programmes on schooling outcomes. J Dev Effect. 2014;6(1):143. https://doi.org/10.1080/19439342.2014.890362 CrossrefGoogle Scholar
46. De Hoop J, Gichane MW, Groppo V, Zuilkowski SS, on behalf of the Productive Social Safety Net Youth Evaluation Team. Cash transfers, public works and child activities: mixed methods evidence from the United Republic of Tanzania. Working Paper Series (WP-2020-03). Florence, Italy: UNICEF Office of Research‒Innocenti; 2020. Google Scholar
47. Roelen K, Saha A. Pathways to stronger futures? The role of social protection in reducing psychological risk factors for child development in Haiti. World Dev. 2021;142:105423. https://doi.org/10.1016/j.worlddev.2021.105423 CrossrefGoogle Scholar
48. Roy S, Hidrobo M, Hoddinott J, Ahmed A. Transfers, behavior change communication, and intimate partner violence: postprogram evidence from rural Bangladesh. Rev Econ Stat. 2019;101(5):865877. https://doi.org/10.1162/rest_a_00791 CrossrefGoogle Scholar


No related items




Tia Palermo, PhD Leah Prencipe, MPH Lusajo Kajula, PhD for the Tanzania Cash Plus Evaluation Team Tia Palermo is with the the Department of Epidemiology and Environmental Health, School of Public Health and Health Professions, University at Buffalo, State University of New York. Leah Prencipe is a PhD candidate and is with the Department of Public Health, Erasmus MC, University Medical Centre, Rotterdam, the Netherlands, and the Social and Economic Policy Unit, United Nations Children’s Fund (UNICEF) Office of Research‒Innocenti, Florence, Italy. Lusajo Kajula is with the Social and Economic Policy Unit, UNICEF Office of Research‒Innocenti. “Effects of Government-Implemented Cash Plus Model on Violence Experiences and Perpetration Among Adolescents in Tanzania, 2018‒2019”, American Journal of Public Health 111, no. 12 (December 1, 2021): pp. 2227-2238.


PMID: 34878869