Non-Cognitive Skills Development and School-Based Violence Reduction in Central America.

This project aims to evaluate the impact of three interventions oriented to enhance participants’ non-cognitive skills and character. Our target audience will consist of students enrolled in participants and comparison schools in Central America, with the ages between 12 and 16 years—second and third educational levels.

The first intervention consists to participate in extra-curricular activities that include dance, sports, art, among others. In this alternative, children are not learning a specific curriculum, but remain under adult supervision and protected from their risky contexts during a couple of hours.

The second psychological curriculum we study, Construcción de Fortaleza de carácter (“Character Strengths Development Program” or CSD), aims to strengthen participants’ character and increase their development and psychological well-being. According to the theory of change for this intervention, positive psychological resources developed through CSD will drive beneficial behaviors such as staying in school and increasing academic performance; improvement in relationships; decision making and establishment of life goals; and taking actions to achieve the latter. These positive resources will help participants to prevent and reduce harmful behaviors such as school dropouts, poor academic performance, violent behaviors, involvement in illegal activities, and other risky behaviors.

CSD includes 32 sessions of training or reflection. The first type aims to reinforce the concepts and their importance in order for participants to achieve well-being. In the reflection sessions, participants are invited to perform self-analysis—that is, analysis of their history and environment. To achieve the desired level of reflection, implementers use an active learning methodology, which places the student at the center of the learning experience and motivates participation through individual and group activities. Participants actively practice the strengths, reflect their application in daily life, and acquire tools to adopt them easily.

The third intervention is a combination of extra-curricular clubs and a mindfulness program, Salon Tranquilo (“Calm Classroom®” or MF), a mindfulness-based and relaxation response program. The program includes directed meditation for stress and anxiety reduction and control of automatic responses; it uses thought techniques to help participants develop self-awareness, mental concentration, and inner calm. Through several activities, MF provides students and staff with tools to manage stress more effectively, to regulate their emotions, and at the same time to lead more productive and well-balanced lives. The MF program includes 42 different activities of breathing, stretching, relaxation, and focusing techniques. Each script technique takes three minutes to implement. Mentors conduct these scheduled practices during the clubs and encourage participants to use the techniques naturally when appropriate.

Participants will meet once per week during one academic year (between 7-8 months). Each of the sessions is implemented after-school time, with an approximate duration of 1-1.5 hours. For methodological reasons, club sizes are between 13-15 participants on average.

2019-03-11

2019-10-31

(i) Administrative data on misbehavior and school attendance, and (ii) academic performance using school records.

For each strand of data that we mentioned before, we will use the following instruments:
● Misbehavior at school. It will be measured from the student behavior reports in the schools, that we will obtain through administrative reports of the teachers. They are in scale [excellent; very good; well; average; below average]. These categories can be transformed into scores that can go from 0-10 points.

● Academic performance. Reports of math and student science notes on scales of 0-10 points will be standardized at the course level. These data will be collected from the administrative bases of the schools.

(i) Strengths and Virtues (Peterson and Seligman, 2004), (ii) socioemotional skills--impulsiveness, perseverance, and risk-taking behavior--collected using SoftGames App (Danon et al, 2018), (iii) socioemotional skills--grit, emotional regulation, and locus of control--using self-reported surveys, (iv) Emotional reaction and regulation, collected using an artificial intelligence-based algorithm.

For each strand of data that we mentioned before, we will use the following instruments:

● Strengths and Virtues.
We will employ the Values in Action Inventory of Strengths (Peterson and Seligman, 2004). It is a catalog of positive psychological traits, consisting of 240 items, ten for each of 24 different character strengths. Following Jordan and Rand (2018), we will use an adapted version that consists of only 24 items that ask directly about each character strength. Each item is scored on a Likert scale from 1 to 5 points. The outcome will be an index estimated as the sum of all validated items and can take the value of 1 to 80. The greater the index, the higher the participant’s strength of character.

● Non-cognitive skills.
1. GRIT: the combination of passion and perseverance for a singularly important goal. We will estimate this outcome using two measures:
• GRIT Scale: This variable will be the sum of validated items at the student level. This index will take a value between 1 and 60. A GRIT value close to 60 indicates that the student has a higher level of perseverance.
• Addition games (Alan and Ertact Grit Task): In this game, participants are presented with a set of additions, which can be easy or difficult to solve. After each round, they are asked to select the level of difficulty for the next one. Thus, we have two proxies for perseverance: (i) the proportion of difficult games chosen in all rounds and (ii) if the student persisted after failing in round 1, when her or his choice was forced. A greater value of any of these proxies indicates that the participant is more perseverant.

2. Impulsiveness: This trait is defined as the tendency of an individual to act suddenly without thinking carefully about what might happen because of what s/he is doing. To proxy for impulsiveness, we used the Go-noGo game. In this game, the participant is presented with a square on the screen for a very short period and is asked to touch the screen every time the square is not black (the noGo stimulus). The outcome is the proportion of times the participant correctly responds to the noGo stimulus. The higher the level of this proportion, the less impulsive the participant is.

3. Risk-taking behavior: This is defined as any consciously or unconsciously controlled behavior with a perceived uncertainty about its benefits or costs for the well-being of oneself or others (Trimpop, 1994). We measure this outcome using the Balloon Analogue Risk Task (BART) Game. It consists of pumping a balloon. Each pump earns points for the participant, but s/he will lose all earned points if the balloon pops. The outcome is measured as the mean of pumps on the balloons that did not pop. The more points the participant earns, the more risk loving s/he is.

4. Locus of Control: This is defined as the degree to which people believe they have power over the outcome of events in their lives. We use a four-item survey. This outcome will be the sum of all items and therefore takes a value between 1 and 20 points. The greater the index, the higher the expectation that they can control their lives.

● Emotion regulation and stress: We will proxy regulation of emotions and stress using two instruments.
a. Cognitive Emotion Regulation Questionnaire (CERQ): This instrument includes 36 items, each of them scored on a Likert scale from 1 to 5 points. The outcome will the sum of validated items at the student level. This index will take a value between 1 and 180. A higher index indicates the student’s increased ability to regulate her or his emotions.

b. Reactiva: To proxy for emotional regulation and stress, we will rely on the artificial intelligence-based (AI) algorithm that uses the analysis of videos captured from the front camera of smartphones or tablets to proxy for emotions. Then, we will use the specifications of Egana-delSol (2016) to construct arousal (proxy of stress) and valence (proxy of emotional regulation) indices at a resting state and in the onset of positive and negative stimuli. A greater level of arousal or valence indicates higher levels of stress or emotional regulation, respectively.

Randomization

The randomized control trial that will generate the data and evidence we are looking for will be implemented in 21 schools, including 7 in El Salvador, 7 in Honduras, and 7 in Guatemala. By stratifying at the country and school-risk level, all participating schools will be randomly assigned to three treatments:

● T1: Clubs: 1/3 of schools will be randomly assigned to the intervention of extracurricular clubs. Schools in this treatment will receive specific activities of the clubs including sports, languages, among others.

● T2: T1 + CS: 1/3 of schools will be randomly assigned to the intervention of CS + extracurricular clubs (T1). In each of the sessions, schools in this treatment will first receive the curricula that encourage CS and in the second part of the session, the specific activities of the clubs are developed.

● T3: T1 + Mindfulness: The remaining 1/3 of schools will be randomly assigned to the intervention of Mindfulness + extracurricular clubs (T1). In each of the sessions, schools in this treatment will first receive the curricula of Mindfulness and in the second part of the session, the specific activities of the extracurricular clubs.

In addition, we will identify 7 schools that will serve as a comparison group. Exploiting information from National Educational Censuses, we will use matching in propensity score approach to identify schools that are more like to those treated. From these comparison schools, we will gather information to characterize the potential demand for the Clubs.

Thus, when comparing the outcomes of T1 group with control schools, we will provide causal evidence of the protection mechanism provided by the extra-curricular clubs. Then, in the comparison of outcomes of T2 or T3 group with those of the control group, it will provide causal evidence of the effect of applying this type of interventions—character strengths and mindfulness—to individuals living in highly violent contexts, which is equivalent to both protection and learning mechanisms.

Finally, this experimental design and project structure allow us to disentangle learning and protection mechanisms. We argue that extra-curricular clubs mostly keep participants out of their violent environments. Then, those who also learn a specific curriculum on character strengths or mindfulness are learning life skills that these programs deliver while protected during a similar amount of time, i.e., school children in T1 receive only a protective service while children in T2 or T3 receive both protection and learning. Therefore, the difference in the results between T1 and any of the other interventions—CS (T2) and Mindfulness (T3)—can be interpreted as the separation between the mechanism of protection or learning.

Randomization

The randomized control trial that will generate the data and evidence we are looking for will be implemented in 21 schools, including 7 in El Salvador, 7 in Honduras, and 7 in Guatemala. By stratifying at the country and school-risk level, all participating schools will be randomly assigned to three treatments:

● T1: Clubs: 1/3 of schools will be randomly assigned to the intervention of extracurricular clubs. Schools in this treatment will receive specific activities of the clubs including sports, languages, among others.

● T2: T1 + CS: 1/3 of schools will be randomly assigned to the intervention of CS + extracurricular clubs (T1). In each of the sessions, schools in this treatment will first receive the curricula that encourage CS and in the second part of the session, the specific activities of the clubs are developed.

● T3: T1 + Mindfulness: The remaining 1/3 of schools will be randomly assigned to the intervention of Mindfulness + extracurricular clubs (T1). In each of the sessions, schools in this treatment will first receive the curricula of Mindfulness and in the second part of the session, the specific activities of the extracurricular clubs.

In addition, we will identify 7 schools that will serve as a comparison group. Exploiting information from National Educational Censuses, we will use matching in the propensity score approach to identify schools that are more like to those treated. From these comparison schools, we will gather information to characterize the potential demand for the Clubs.

Thus, when comparing the outcomes of T1 group with control schools, we will provide causal evidence of the protection mechanism provided by the extra-curricular clubs. Then, in the comparison of outcomes of T2 or T3 group with those of the control group, it will provide causal evidence of the effect of applying this type of interventions—character strengths and mindfulness—to individuals living in highly violent contexts, which is equivalent to both protection and learning mechanisms.

Finally, this experimental design and project structure allow us to disentangle learning and protection mechanisms. We argue that extra-curricular clubs mostly keep participants out of their violent environments. Then, those who also learn a specific curriculum on character strengths or mindfulness are learning life skills that these programs deliver while protected during a similar amount of time, i.e., school children in T1 receive only a protective service while children in T2 or T3 receive both protection and learning. Therefore, the difference in the results between T1 and any of the other interventions—CS (T2) and Mindfulness (T3)—can be interpreted as the separation between the mechanism of protection or learning.

Randomization Method: randomization done in office by a computer,

Unit of randomization: School

Yes

29 schools in total.

The final sample should be 1975 students, 556 from pure comparison schools and 1419 from treated schools, which accounts for an average of 49 students per school.

7 schools in T1, 7 schools in T2, 7 schools in T3, 8 schools in pure control group.

For power estimations, we use existing evidence on the effects of psychological interventions on socio-emotional skills and academic performance for at-risk adolescents and adults (Castro et al., 2019; Flook et al, 2015; Chernikoff, 2019; and Dinarte and Egana-Del Sol, 2019 (DE, 2019)). Considering that the results from DE (2019) are (i) the most conservative, (ii) the intervention analyzed is similar to T1, and (iii) this sample is the most similar to the students in our project, we use DE’s (2019) data as a reference for minimum detectable effects and for estimations of intra-cluster correlation and correlation between baseline and follow-ups. DE (2019) finds that an expected average effect of an intervention aimed at improving the socio-emotional skills of adolescents is around 0.17 and 0.13 standard deviations on misbehavior in school and academic performance, respectively. From that data, we also estimate correlations between baseline and follow-up of 0.81 and an intra-school correlation of 0.04 (the most conservative scenario). Moreover, Chernikoff (2019) estimates that a mindfulness program for adults can have an effect of approximately 0.3 standard deviations. We use this result as an MDE of reference for T2 and T3 treatments. Considering these effects and assuming a power of 0.80 and an alpha of 0.05, the most conservative estimation of sample size indicates that we would need a sample of 1,616 students from the total 29 schools—404 from comparison and 1,212 from treated schools. On average, we must recruit 56 students per school. The current sample of 1,975 students allows us to handle attrition of up to 10%, 3 percentage points higher than the attrition from DE (2019).