Table S1. Propensity Score Matching: Bias Correction

Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Table S1. Propensity Score Matching: Bias correction

U=unmatche Mean % reduct. | t-test d bias| Variable M=matched SCC, IL-1 SCC %Bias t p>|t| Individual Characteristics U 15.99 16.28 -23.0 -3.490 0.000 Age (mean) M 16.032 16.04 -0.3 9.88 -0.040 0.860 U 0.42 0.43 -2.8 -0.430 0.670 Sexo (male) M 0.44 0.40 9.3 -234.30 1.120 0.260 U 0.04 0.03 6.7 1.030 0.300 Speaks an indigenous language M 0.04 0.02 10.0 -49.4 1.170 0.350 U 0.41 0.38 7.0 1.070 0.280 Has any work experience M 0.38 0.39 -2.9 58.70 -0.350 0.770 U 7.56 7.61 -5.4 -0.830 0.410 Offical grades (mean) M 7.58 7.56 2.0 63.90 0.230 0.790 Currently has student status U 0.7 0.73 -5.1 -0.770 0.440 only M 0.74 0.73 3.9 22.60 -0.480 0.740 Perception of School environment U 0.95 0.95 0.7 0.110 0.920 My school is unsafe M 0.95 0.94 6.6 -856.30 0.730 0.380 U 0.01 0.01 -1.6 -0.240 0.810 My school is dangerous M 0.01 0.01 0.0 100.00 0.000 1.000 Household Characteristics Lives with his/her mother and U 0.61 0.64 -4.9 -0.740 0.460 father M 0.65 0.61 9.5 -95.60 1.130 0.250 U 0.07 0.05 9.0 1.380 0.050 Lives with his/her father M 0.05 0.07 -1.5 83.00 -1.220 0.220 U 0.26 0.24 3.0 0.450 0.650 Lives with his/her mother M 0.24 0.27 -7.4 148.30 -0.870 0.380 U 0.36 0.32 9.2 1.410 0.160 Head of househols Sex M 0.33 0.37 -9.7 -5.40 -1.150 0.250 U 44.65 44.81 -1.9 -0.290 0.770 Head of household Age M 44.5 44.69 -1.8 9.20 -0.210 0.830 U 0.74 0.72 3.4 0.520 0.600 Pipied water M 0.75 0.72 6.4 -87.10 0.760 0.330 U 3.63 3.74 -6.0 -0.920 0.360 Room (mean) M 3.67 3.67 0.4 93.80 0.050 0.920 U 2.63 2.67 -3.4 -0.510 0.610 Bedrooms (mean) M 2.67 2.71 -3.4 -1.00 -0.380 0.700 U 0.8 0.89 -13.1 -2.000 0.050 Cell phone M 0.86 0.89 -3.7 74.80 0.400 0.600 Note: Kernel (0.001)

Sample Ps R2 LR chi2 P>chi2 MeanBias MedBias B R %Var Unmatched 0.023 30.22 0.025 6.2 5.1 36.1* 0.84 50 Matched 0.009 7.21 0.981 4.6 3.4 22.6 0.98 17 Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Table S2. Description of dating violence at baseline and follow-up by exposure

Full sample Difference SCC, IL-1a SCC (SCC, IL-1 vs. SCC)a n 292 407 PANEL A: Dating violence- % of students Experienced Psychological Baseline 27.30 25.40 1.90 Follow-up 21.52 21.63 -0.11 Difference (Baseline vs. Follow-up)c -5.78 -3.77 Physical Baseline 9.71 12.90 -3.19 Follow-up 7.87 9.72 -1.85 Difference (Baseline vs. Follow-up)c -1.84 -3.18 Sexual Baseline 5.99 5.08 0.91 Follow-up 4.20 5.18 -0.98 Difference (Baseline vs. Follow-up)c -1.79 0.09 PANEL B: Dating violence- % of students

Perpetrated Psychological Baseline 38.20 33.42 4.78 Follow-up 30.53 29.16 1.37 Difference (Baseline vs. Follow-up)c -7.67 -4.26** Physical Baseline 13.11 17.38 -4.27 Follow-up 11.07 13.62 -2.55 Difference (Baseline vs. Follow-up)c -2.04 -3.76* Sexual Baseline 6.37 7.22 -0.85 Follow-up 4.96 5.18 -0.22 Difference (Baseline vs. Follow-up)c -1.41 -2.04* a Attended at least one session of True Love Curriculum could participate or could not participate as agents of change in enhancing the school climate through schoolyard activities. Kruskal-Wallis ANOVA Tests to identify signifiant differences. b Difference test between SCC, IL-1 and SCC **<0.01 *<0.05 c Difference test between baseline and follow-up **<0.01 *<0.05 Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Table S3. Description of attitudes and beliefs toward violence at baseline and follow- up, by exposure

Full sample Differenc e (SCC, SCC, IL-1a SCC IL-1 vs. SCC)b n 381 504 PANEL A: Attitude and beliefs toward Mean of score violence (a) (range 1-4) Acceptance and justification of violence index (a) mean Baseline 1.27 1.29 -0.02 Follow-up 1.23 1.23 0.00 Difference (Baseline vs. Follow-up)c -0.04** -0.06* Acceptance of sexist attitudes index (a) mean Baseline 1.23 1.25 -0.02 Follow-up 1.28 1.32 -0.04 Difference (Baseline vs. Follow-up)c 0.05 0.07* Acceptance of sexist attitudes in dating index (a) mean Baseline 1.23 1.22 0.01 Follow-up 1.17 1.17 0.00 Difference (Baseline vs. Follow-up)c -0.06** -0.05* PANEL B: Participation in activities to % of students prevent dating violence Index (a) Baseline 24.40 24.95 -0.55 Follow-up 74.60 33.20 41.4** Difference (Baseline vs. Follow-up)c 50.2** 8.25* PANEL C: Knowledge of institutions which provide support to address dating % of students violence Index (a) Baseline 38.58 33.53 5.05 Follow-up 51.44 47.22 4.22* Difference (Baseline vs. Follow-up)c 12.86** 13.69** a Attended at least one session of True Love Curriculum could participate or could not participate as agents of change in enhancing the school climate through schoolyard activities. Kruskal-Wallis ANOVA Tests to identify significant differences (a) Index range: 1-4 b Difference test between SCC, IL-1 and SCC **<0.01 *<0.05 c Difference test between baseline and follow-up **<0.01 *<0.05 Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Table S4. Intervention impact: Dating violence. Fixed-effects models. Subsample of students who both were exposed to the True Love curriculum and actively engaged in schoolyard activities

Full sample Female Male n 692 325 367

Dating Violence SCC SCC SCC Coeff. Follow Relative Coeff. Follow- Relative Coeff. Follow- Relative [95%CI] -upa Effectb [95%CI] upa Effectb [95%CI] upa Effectb PANEL A: dating violence- experienced Psychological -0.05 0.22 -0.23 0.05 0.34 0.15 -0.19* 0.26 -0.73 [-0.36;-

[-0.15;0.06] [-0.09;0.18] 0.02] Physical 0.04 0.12 0.33 0.03 0.12 0.25 0.04 0.12 0.33 [-0.05 ;0.12] [-0.09;0.14] [-0.10;0.17] Sexual -0.03 0.05 -0.60 -0.03 0.07 -0.43 -0.04 0.04 -1.00 [-0.08;0.02] [-0.09;0.03] [-0.12;0.04] PANEL B: dating violence- perpetrated Psychological -0.05 0.29 -0.17 0.04 0.33 0.12 -0.18* 0.25 -0.72 [-0.34;-

[-0.15;0.06] [-0.09;0.17] 0.02] Physical 0.03 0.14 0.21 0.01 0.15 0.07 0.04 0.12 0.33 [-0.06;0.11] [-0.11;0.12] [-0.10;0.18] Sexual -0.02 0.05 -0.40 -0.03 0.05 -0.60 -0.01 0.06 -0.17 [-0.07;0.04] [-0.10;0.04] [-0.10;0.08] Impact is estimated by the coefficient of the interaction between the variables exposure (0=SCC 1=SCC, IL-2) and time (0=baseline 1=follow up). The models are adjusted by: age, official grades, first sexual intercourse, currently has boyfriend/girlfriend, has had sexual activity with current or the most recent partner, time in months of current or the most recent relationship, school group, mean of Rosenberg self-esteem scale, and cluster at school level * p<0.05 **p<0.01 a Prevalence among SCC group at follow-up b Effect size relative to SCC group at follow-up Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Table S5. Intervention impact: Attitude and beliefs towards violence. Fixed-effects models. Subsample of students who both were exposed to True Love curriculum and actively engaged in schoolyard activities

Full sample Female Male n 817 361 456 Effec Effec Coeff. SCC Coeff. SCC Coeff. SCC Effect t t [95%CI] Follow- [95%CI] Follow- [95%CI] Follow- Sizeb up Meana Sizeb up Meana Sizeb up Meana

Acceptance and -0.03 1.23 -0.02 -0.03 1.15 -0.03 -0.03 1.29 -0.02 justification of violence index (a) [-0.08;0.03] [-0.09;0.04] [-0.13;0.06] Acceptance of sexist -0.02 1.32 -0.02 -0.05 1.26 -0.04 0.01 1.36 0.01 attitudes index (a) [-0.07;0.03] [-0.12;0.02] [-0.06;0.09] Acceptance of sexist -0.06* 1.17 -0.05 -0.06* 1.10 -0.05 -0.06 1.23 -0.05 attitudes in dating index [-0.15 - [-0.11;-0.00] [-0.10;-0.01] (a) 0.03]

Impact is estimated by the coefficient of the interaction between the variables exposure (0=SCC 1=SCC, IL-2) and time (0=baseline 1=follow up). The models are adjusted by: age, official grades, first sexual intercourse, currently has boyfriend/girlfriend, school group, mean of Rosenberg self-esteem scale, and cluster at school level * p<0.05 **p<0.001 (a) Score range: 1-4 a Mean index score among SCC group at follow-up b Effect size relative to SCC group at follow-up

Table S6. Program impact1: Participation in activities to prevent violence and knowledge of institutions that provide support to address violence. Fixed-effects models. Subsample of students who both were exposed to True Love curriculum and actively engaged in schoolyard activities

Full sample Female Male n 817 361 456 Non Non Non Exp Relativ Exp Relativ Exp

Coeff. Follow- e Coeff. Follow- e Coeff. Follow- Relative [95%CI] upa Effectb [95%CI] upa Effectb [95%CI] upa Effectb

Participation in activities 0.68** 0.33 2.06 0.68** 0.32 2.13 0.69** 0.34 2.03 to prevent dating violence. Index [0.60;0.76] [0.57;0.79] [0.57;0.80]

Knowledge of 0.12* 0.56 0.21 0.14* 0.64 0.22 0.12 0.51 0.29 institutions which [0.03;0.21] [0.02;0.27] [-0.01;0.25] provide support to address dating violence. Index Impact is estimated by the coefficient of the interaction between the variables exposure (0=SCC 1=SCC, IL-2) and time (0=baseline 1=follow up). The models are adjusted by: age, official grades, first sexual intercourse, currently has boyfriend/girlfriend, school group, mean of Rosenberg self-esteem scale, and cluster at school level * p<0.05 **p<0.001 a Percentage among SCC group at follow-up b Effect size relative to SCC group at follow-up Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City.

Figure S1. Behavioral Theoretical Framework of the targeted risk factors in True Love Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City.

Figure S2. Analytical Sample Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City. Figure S3. Propensity Score Matching: graphical results Supplementary material: True love: effectiveness of a school-based program to reduce partner violence among teenagers in Mexico City.

Figure S4. Timeline of the intervention, evaluation and school calendar

2014 Feb Mar May June Aug Sep Nov Dec Intervention: School climate component Intervention: Schoolyard activities Summer Intervention: True Love Module 1+ Break Curriculum n=353 (92%)* Module 3+ n=342 (89%)* School Calendar 2nd semester of 10th grade 1st semester of 11th grade Evaluation measurements Follow- up survey +Each module included four 1-hour sessions. Sixteen 1-hour sessions in total * Attended at least one session of True Love Curriculum (paired sample) Note: In the True Love Curriculum the percentage of attendance to at least one session of each module was 92%, 91% and 89% for the first, second and third modules, respectively, attendance to the fourth module was much lower at 52%