Displaced and non-displaced Colombian children's evaluations of moral transgressions, retaliation, and reconciliation

Abstract

In order to assess the effects of displacement and exposure to violence on children's moral reasoning, Colombian children exposed to minimal violence (non-displaced or low-risk) (N = 99) and to extreme violence (displaced or high-risk) (N = 94), evenly divided by gender, at 6-, 9-, and 12 - years of age, were interviewed regarding their evaluation of peer-oriented moral transgressions (hitting and not sharing toys). The vast majority of children evaluated moral transgressions as wrong. Group and age differences were revealed, however, regarding provocation and retaliation. Children who were exposed to violence, in contrast to those with minimum exposure, judged it more legitimate to inflict harm or deny resources when provoked and judged it more okay to retaliate for reasons of retribution. Surprisingly, and somewhat hopefully, all children viewed reconciliation as feasible. The results are informative regarding theories of morality, culture, and the effects of violence on children's social development.

Little is known about the effects of displacement and exposure to violence on the development of children's moral reasoning, particularly regarding evaluations of transgressions involving peer conflicts, such as unprovoked hitting and denial of resources (e.g., not sharing toys). The aim of this study was to draw on the social cognitive domain model (see Smetana, 2006; Turiel, 1998) to extend the current literature and investigate how exposure to violence and displacement affects Colombian children's evaluations of moral transgressions as well as their reasoning about the provocation, retaliation, and reconciliation of moral transgressions. The children in this study came from two groups: one group constituted children living in intact families in a relatively peaceful town outside of Bogotá; the other group constituted children who had been displaced by the war and were living in shantytowns, with high levels of violence, in another town in the outskirts of Bogotá. While these two cultural groups vary on several dimensions, exposure to violence was (and remains) a very salient dimension in which their daily lives are different. Thus, this was the main variable of focus for this project.

Exposure to violence has been shown to influence how children solve conflicts and disagreements (Arsenio & Lemerise, 2004; Astor, 1994; Astor & Behre, 1997; Lemerise & Arsenio, 2000). Experiencing the stress associated with exposure to violence in cultural contexts with high conflict has been shown to affect children's emotional and cognitive development as well (Boxer, Edwards-Leeper, Goldstein, Musher-Eizenman, & Dubow, 2003; Garbarino, Dubrow, Kostelny, & Pardo, 1992; Garbarino, Kolstelny, & Dubrow, 1998; Guerra, Huesmann, & Spindler, 2003).

Yet, extensive research on children's moral judgments has shown that children from a wide range of cultures evaluate moral transgressions, such as hitting and the denial of resources, as wrong due to the negative intrinsic consequences affecting another person (Killen, Lee-Kim, McGlothlin, & Stangor, 2002; Killen, Margie, & Sinno, 2006; Smetana, 2006; Turiel, 2006; Wainryb, 2006). This has been demonstrated by showing that children judge moral transgressions as wrong even when authority condones it (the act is not a matter of authority jurisdiction), and when in other contexts and cultures (the principle underlying the evaluation is generalizable). Moral transgressions are wrong because of the intrinsic negative consequences to another rather than as a function of authority mandates, punishment avoidance, or cultural expectations. These criteria, authority jurisdiction and generalizability, have been used in many investigations of children's moral reasoning, and constitute part of the standard criteria used to assess moral reasoning (see Smetana, 1995; Tisak, 1995, for reviews).

Research using these criteria to assess moral reasoning has been conducted in a wide range of cultures (see Wainryb, 2006), including countries such as Brazil (Howe, Kahn, & Friendman, 1996; Nucci, Camino, & Milnitsky-Sapiro, 1996), Israel (Cole et al., 2003; Wainryb & Turiel, 1994), China (Helwig, Arnold, Tan, & Boyd, 2003), Jordan, Palestine (Brenick et al., 2007, under review), Nigeria (Hollos, Reis, & Turiel, 1986), India (Shweder, Mahapatra, & Miller, 1987), Japan (Killen, Crystal, & Watanabe, 2002), and Korea (Song, Smetana, & Kim, 1987) and varying by rural, urban, high- and low-SES status (see Smetana, 1995, 2006). This literature has provided evidence to support the claim that children's evaluations of moral transgressions follow a similar developmental trajectory in a wide range of cultures (Arsenio & Lemerise, 2004; Turiel, 1998, 2006). At the same time, very few studies have examined moral reasoning in children living in war-torn areas, such as Colombia, and only a few studies, that we know of, have examined South American's children's social reasoning. Nucci et al. (1996) examined personal reasoning, Howe, Kahn, and Friedman (1996) examined social and moral environmental reasoning in Brazilian children, Ardila-Rey and Killen (2001) examined Colombian preschool children's moral reasoning, and Posada and Wainryb (under review) studied Colombian children's judgments about revenge. In addition, a few studies have been conducted on South American children's prosocial emotions and judgments (for Brazilian children and adolescents, Eisenberg, Zhou, & Koller, 2001; Carlo, Koller, & Eisenberg, 1996, 1998).

A number of studies have analyzed the various effects of violence, however, on children's social development. Studies by Fox and colleagues, utilizing a violence exposure measure (VEX, Fox & Leavitt, 1995), have analyzed the relations between exposure to violence and levels of distress and behavioral problems (Shahinfar, Fox, & Leavitt, 2000; Stein et al., 2001). This variable has been examined in several countries affected by war and violence such as Colombia, Ireland, Israel, and South Africa (see Leavitt & Fox, 1993, for a review) exploring the effects of violence on levels of stress and coping abilities of children as reflected on personality and behavioral changes (Ardila, 2004; Kostelny & Garbarino, 1994; Llanos, Amar, & Botto, 2001; Shahinfar et al., 2000; Silva, 1999), exposure to violence and child aggressiveness (Brook et al., 2003; Liddell, Kvalsvig, Qotyana, & Shablala, 1994), and long term effects of violence (Franco Agudelo, 1997; Groves, 1996; Rieck, 1994). The findings of these studies have shown that children in violent communities are more likely to become involved in aggression (Guerra, Huesmann, & Spindler, 2003; Liddell et al., 1994), that adolescent boys become hostile toward authority, and girls develop symptoms of anxiety (Toner, 1994), that children develop negative perceptions of their abusive parents (Sternberg et al., 1994), and that, with age, children increase their aggressive cognitions (Guerra, Huesmann, & Spindler, 2003). Additionally, work with Colombian children finds that exposure to familial violence is negatively related to school attendance and completion (Knaul & Ramírez, 2005). Very little research, however, has been conducted on the effects of violence on children's moral reasoning.

One exception has been research by Astor (1994) who found that extremely aggressive children, from poor inner city environments in the United States, judged it more all right to retaliate when provoked than did non-aggressive inner city children. Non-aggressive children judged hitting in response to provocation as wrong, and in most cases condemned violence as a form of retribution. All children stated that it was wrong, however, to commit acts of violence in unprovoked situations. While Astor (1994) and other researchers studied children who had been diagnosed as “violent,” a handful of studies have examined moral reasoning in the context of intergroup exclusion among preschool-aged children living amidst violence and conflict (Brenick et al., 2007; Cole et al., 2003, these focus on a media intervention as well) finding these children used moral reasoning for straightforward transgressions. They did not examine issues of retaliation or reconciliation, nor look at children in middle childhood, as in this study.

The present study differed from prior studies (Astor, 1994; Brenick et al., 2007; Cole et al., 2003) by determining how exposure to violence and displacement influences different aspects of moral reasoning (e.g., evaluations of transgression, authority jurisdiction, generalizability, retaliation, retribution, and reconciliation), and for both aggressive and non-aggressive types of moral transgressions, rather than a sole focus on how being violent or is related to evaluating aggressive provocation. Based on previous research, one of the goals of this study was to investigate how children, with different levels of exposure to violence, weigh context considerations such as provocation, retaliation, and retribution, when evaluating moral transgressions such as hitting someone (aggression) and taking away toys (refusing to share resources).

In addition to using standard moral judgments assessments, we included a new measure referred to as reconciliation, based on current findings in social development. Reconciliation refers to a method for resolving conflicts that occur between individuals (Verbeek, Hartup, & Collins, 2000), and while few studies in the area of moral reasoning have focused on children's evaluations of reconciliation, this is a key aspect of how peer-oriented moral transgressions are resolved (Fry, 2006). Verbeek and de Waal (2001) found that the domain of the conflict, the level of aggression, and the relationship and previous interactions between the opponents affected the reconciliation strategies used by the children. Thus, in addition to the context parameters described above, children were interviewed about whether they believed that children involved in a negative exchange could reconcile after a transgression had occurred. Based on previous findings, we predicted that non-displaced, low-risk children would judge that peers could be friends after the termination of a negative exchange. How displaced, high-risk children would evaluate reconciliation was an open question.

Before describing our hypotheses, it is important to provide a few points about the context regarding where this study was conducted, and to define our population groups. Colombia is in the midst of an internal armed conflict and displacement has driven large numbers of people away from their homes, particularly in rural areas (Garfield & Llanten Morales, 2004). It is estimated that more than 3.5 million Colombians have fled the rural areas since 1985 to seek refugee in the cities, increasing the population of the poverty belts that surround most of the larger cities (Organization of American States (OAS), 1999; Consultoría para los Derechos Humanos y Desplazamiento (COHDES), 2005). According to the OAS, the majority (about 70%) of the people internally displaced during the 90's were minors. This situation is far from improving. Statistics from the Colombian Government's Social Solidarity Network show 128,590 families were displaced between January 2001 and November 2002 (Red de Solidaridad Social, Presidencia de la República de Colombia, 2002). A reported peak number of 412,553 people (1,144 a day) became displaced during 2002 (COHDES, 2003, 2005). Displacement has left countless children and families with significant social and economic resource losses (Garfield & Llanten Morales, 2004; Ibáñez & Vélez, 2003; Moser, 2000). Moreover, the displacement process has exceeded the capacity of the receiving cities' economy to absorb the population involved, and also the capacity of the State to meet their basic needs -- particularly the needs of the children-- and to provide solutions to the resulting conflicts and tensions (CINEP, 2001; Moser, 2000).

As a result, it is vital to study the displaced children in Colombia and how violence and extremely disadvantaged living conditions are affecting children's social and moral development. A recent survey study found that at least one third of the population in Bogotá has been either a victim or a perpetrator of violence (Klevens, Duque, & Ramirez, 2002). Two recent studies have pointed to risk factors for Colombian adolescents (Brook, et al., 2003), and familial factors that contribute to depression in adolescents (McClellan, Heaton, Forste, & Barber, 2004). While these studies identify associated factors that bear on violent behaviors, no studies have examined Colombian children's moral evaluations of peer interactions, and to what extent, exposure to violence and displacement bears on their judgments.

Ardila-Rey and Killen (2001) investigated middle-class Colombian children's evaluations of personal, moral, and social-conventional interactions in the classroom setting. These children evaluated moral transgressions as wrong independent of the teacher's viewpoints, or the existence of a rule, consistent with the findings documented for U.S. middle-class children. Though violence in Colombia is pervasive (Ardila, 2004), the children in the study conducted by Ardila-Rey and Killen (2001) were of middle-class backgrounds from a relatively peaceful city, and exposure to violence and its effects on the children's reasoning were not assessed, but were expected to be minimal. Thus, the present study extended the previous work by interviewing Colombian children from two communities surrounding Bogotá, Colombia, using similar story scenarios as was used by Ardila-Rey & Killen (2001).

For the present study, we selected two communities with divergent living conditions and levels of risk for exposure to violence to allow for a broader assessment of the impact of violence among Colombian children. In order to determine how pervasive violence at political, community, and family levels affects children's moral reasoning, we employed the VEX, a measure of exposure to violence developed by Fox and Leavitt (1995) for use with children and adolescents. We used this measure to chart the level of exposure for our two population groups, as well as to relate it to different moral judgment assessments. While our two groups of children varied on a number of dimensions, the overall characterization was “low risk for exposure to violence” (living in non-displaced, middle class families) and “high risk for exposure to violence” (living in displaced communities outside of Bogotá (see the Methods for more details).

We hypothesized that the children from these two groups would not differ in their evaluations of straightforward moral transgressions, as was shown by Astor (1994) and Brenick and colleagues (2007; Cole et al., 2003), but would differ in their evaluations of more complex issues. Specifically, we expected that children in both groups would view hitting someone or taking toys from someone as generalizably wrong due to the negative intrinsic consequences, independent of authority jurisdiction and cultural contexts. We predicted that children from the two groups would differ, however, in their evaluations of provocation, retaliation, and retribution.

Regarding exposure to violence, we expected that displaced children with a high exposure to violence would judge it as more permissible to commit a moral transgression if provoked (someone else called you a bad name) or to retaliate (someone else did it first) than would children with a minimal exposure to violence. We predicted that children with a high exposure to violence would base these judgments on retribution unlike children with a minimal exposure to violence. Our expectations about reconciliation were open-ended given very little prior research on this topic. On the one hand, children displaced from their communities due to violence may have given up on hopes of reconciliation. On the other hand, children may hold onto the last chance of reconciliation as a possible way out of violent conditions. Finally, our expectations about age and gender differences were based on our past research with middle-class children in Colombia. We expected few gender differences in moral reasoning or evaluations of the context parameters. Yet, we predicted that age differences would exist, with a greater proportion of older children condoning reactions to provocations and retaliation, given their adoption of survival strategies in a very stressful environment.

Methodology

Participants

A total of 193 Colombian children from low and high-risk for violence exposure backgrounds were interviewed in the outskirts of Bogotá, the capital city of Colombia. The low-risk group was composed of non-displaced children (n = 99, 19 girls and 20 boys at 6 years of age, 15 girls and 15 boys at 9 years of age, and 16 girls and 14 boys at 12 years of age), with an expected low level of exposure to violence. This group of children was drawn from a local school in the town of Chía, a semi-rural area to the north of Bogotá. These children came mostly from middle and lower-middle class families (strata 4 and 5, according to Colombian housing/SES classifications). The low-risk group was labeled “non-displaced school” for the purpose of this paper.

The high-risk group consisted of displaced children (n = 94, 16 girls and 16 boys at 6 years of age, 16 girls and 15 boys at 9 years of age, and 16 girls and 15 boys at 12 years of age), with an expected high level of exposure to violence. This group of children was drawn from a small school in a community in the town of Soacha, located to the south of Bogotá. These children were part of a large group of displaced children and families that relocated to shantytowns in the mountains that surround the town of Soacha. The displaced families came from rural areas that have been under guerrilla and paramilitary attack, and have fled to larger cities to protect their lives. The SES of these children was working to lower class (strata 1 and 2). The high-risk group was labeled “displaced school” for the purpose of this paper.

The age range for the youngest group was 5.0 to 7.3 (Displaced, M = 5.78, non-displaced M = 6.28, total M = 6.06, SD = .71), for the middle group, the age range was from 7.93 to 10.0 years (Displaced, M = 8.81, non-displaced, M = 8.45, total M = 8.64 years, SD = .73), and for the oldest group, the age range was from 10.8 to 14 years (Displaced M = 12.16, Non-displaced M = 11.00, total M = 11.59 years, SD = 1.12).

Data collection for this study was conducted with the collaboration of faculty and students from the School of Medicine of the University San Martin in Bogotá, who are involved in community service with displaced populations. The Dean of the Medical School at the University San Martin contacted the authors of this paper for assistance with training his medical school students in conflict resolution techniques. The University San Martin collaborates in social interest programs with the health center and the community center that serve both displaced and non-displaced families in the town of Soacha, in the outskirts of Bogotá, the capital city of Colombia. The school in the town of Chia was also contacted through the same university. As part of an exchange, the medical school students were extensively trained to the point of reliability in the methodology of this project by the first author, and assisted with the administration of the interviews to the participants. In return, general information about the findings of this project has been provided to the medical staff. The medical school students were blind to the hypotheses of this study, however, and were unaware of any explicit analyses or comparisons being conducted for the empirical investigation.

The town of Chía is located in a fertile area of the Bogotá savanna, with a relatively low population density. The area is mostly peaceful and for this reason it has attracted increasing numbers of middle class professionals from Bogotá who commute to the capital city for work and live in the urban or the rural areas of Chía. Children were drawn from a small K-12 private school, located in a farm-like setting, with a total enrollment of 380 students and a maximum of 25 children per classroom. All teachers have at least bachelor's degrees in their areas of specialization, and children have access to a variety of resources and to several extra-curricular sports and activities. Most of the parents of the participating children were employed, many held professional degrees (medical doctors, veterinarians, engineers, teachers), and several were local artisans or small business owners or employees.

The town of Soacha is located in a highly populated marginal area south of Bogotá, with a low SES level (strata 1 to 2) and high deficiencies in the areas of health and education. There is one paved road that crosses the neighborhood, which is served by several bus lines to Bogotá. According to a social and health risks evaluation study conducted by the University San Martin simultaneously with this study (Delgado, Pacheco, Rodriguez, & Pardo, 2001), the neighborhood has been developing for the last 15 to 20 years through a system of self-construction, thus the majority (70%) of people own their homes. Thousands of displaced families have arrived to Soacha during the last 20 years. The area has high indices of crime, violence, and social conflict. During the two weeks of data collection time in this area, two bombs exploded in the town, one in front of a school's entrance.

The chair of the community association, the director of the health center, and the commander of the army post, facilitated data collection at a small school. Most parents were sub-employed (street vendors, per-day workers, domestic services, gardeners, car washers). About 30% of the parents were unemployed, and only 5% had some kind of permanent jobs. Thirty five percent of the mothers did not complete elementary education. Parents or guardians agreed to participate in the study and the community provided access to a second community center --located about one and a half miles from the school-- where the interviews could be conducted. In order to collect the data, interviewers and children walked escorted by soldiers (who also served as teachers at the school) from the school to the community center.

It is important to note these differences (e.g. socioeconomic status, community type, access to resources) in the two communities when interpreting the results. Children's exposure to violence was likely confounded with the broader community and historical context. We sought to examine ecologically valid groups that exist in Colombia and thus we examined the effects of the broader environmental conditions and specifically children's exposure to violence on their moral reasoning understanding the results should not be solely attributed to differences in exposure to violence. However, the findings remain relevant and generalizable because these other factors typically accompany exposure to violence in real world settings.

Procedure

Children were individually interviewed (40-45 minutes) at a community center in Soacha and at the school in Chia by trained Colombian medical students and the first author. The interviews were tape-recorded and later transcribed and coded for analysis. Only children whose parents or legal guardians agreed to participate were interviewed.

Measures

Three instruments were used in this study: A demographics questionnaire, the Violence Exposure Scale for Children -Revised (VEX-R), Spanish/Preschool Version (Fox & Leavitt, 1995), and the Moral Judgment Interview designed by the authors. The interview was a modification of an instrument developed for a previous study conducted with Colombian middle-class children (Ardila-Rey & Killen, 2001).

Violence Exposure Scale for Children- Revised (VEX-R)

The VEX–R preschool version is a child self-report measure that has previously been administered to children from high violence areas -- such as inner city children (Shahinfar et al., 2000) and Israeli children (Raviv, Raviv, Shimoni, Fox & Leavitt, 1999). A modified version of the VEX-R was used for this study in order to standardize the questionnaire across the age range of the children involved in the study, and to reduce the number of variables involved. The instrument consisted of 18 questions about witnessing or being a victim of aggressive acts presented with picture cards in a cartoon format. For the purpose of the present study, two questions were added to the original preschool instrument: being a victim of stabbing and shooting. These two questions were previously used in a study where children came from violent neighborhoods (Stein et al., 2001) and were included based on the characteristics of the communities in this study. No picture cards were presented to children for these two questions to avoid unnecessary distress. Two of the original VEX questions were not included (arrest, drug-deal) to avoid placing children in a difficult position if a family member had been involved in these acts (see Table 1 for VEX-R questions).

Table 1. Violence exposure incidents assessed by the VEX-R (Spanish preschool version- revised).

EVENT	WITNESSING	VICTIMIZATION	LEVEL
GROUP 1: Mild violence witness		GROUP 2: Mild violence victim
Beat-up	X	X	Mild
Chase	X	X	Mild
Push/Shove	X	X	Mild
Slap	X	X	Mild
Spank	X	X	Mild
GROUP 3: Severe violence witness		GROUP 4: Severe violence victim
Robbery	X	X	Severe
Weapon-threat	X	X	Severe
Shoot	X	X	Severe
Stab	X	X	Severe

Responses to the VEX-R were recorded during the interview on a paper and pencil scoring form. Children reported the frequency rating of their violence exposure, as well as the context where the event occurred, when it happened, and who was involved. Each VEX-R question was coded as a 1 if child answered “Yes” regardless of the frequency (or recurrence) of exposure, and a 0 if the child had answered “No, never”. Children rated the frequency of the events on a 4-point scale (never = 0, once = 1, a few times = 2, lots of times = 3). Their responses were coded according to the context of exposure (coded as home, school, street, or political violence); however, the context was coded as political violence only when children identified the participants as guerrilla, army, or paramilitary. If a child knew the victim or perpetrator, his or her relationship was categorized by the context: family or household member, school peer or teacher, street violence, or political violence (paramilitary, guerrilla, army). One 6-year-old child refused to answer the questionnaires and was excluded from the study.

Moral Judgment Interview

All interviews were audio-recorded and transcribed for analyses. The interview consisted of two scenarios that were described to the children and presented with colorful picture cards. The two “moral transgression” scenarios were: (1) aggression: unprovoked hitting, and (2) refusal to share resources (see Ardila-Rey & Killen, 2001; Killen & Smetana, 1999; Salinas, Posada, & Isaza, 2002; Smetana, 1995). For each of these scenarios six two-part questions (judgment and justification for the judgment) were asked.

The first three questions were (1) Evaluation (Is it all right or not? Why?); (2) Authority Jurisdiction (Would it be okay if the parents said that it was all right? Why?), and (3) Generalizability (Would it be okay in another country? Why?). The next set of questions referred to the context parameter: (4) Provocation (“Would it be okay to do it if they had teased or hurt her first? Why?), (5) Retaliation (Would it be okay to hit her back? Why?), and (6) Reconciliation (Can they still be friends? Why?).

Design

A between- and within- subjects design was used in the analysis of the interview. Age, gender, and displacement status (school) were the independent between-group variables. All children responded to all of the stories and questions, thus the scenario (story) was the independent within-group variable. For the analysis of judgments, the question or assessment was the dependent variable. For the analyses of justifications, the questions or assessments became within-subject independent variables, and justification was the dependent variable. Level of violence exposure, as measured by the VEX-R questionnaire, was used for correlational analyses with the independent between-group variables (displacement status, age and gender) as well as with the within-groups dependent judgments variables. Stories were presented in a randomly assigned, counterbalanced order. Preliminary analysis showed no story order effects.

Data Coding

Moral Judgments Interview

Children's judgments about the interview assessments (e.g., “Is it okay or not okay?”) were recorded on a checklist to be coded and analyzed at a later time. In order to obtain proportional data for judgments, all judgments were coded dichotomously (0 = transgression OK, 1 = transgression Not OK). The reconciliation assessment was reverse-coded as 0 = Not OK to reconcile, 1= OK to reconcile. Thus, responses supporting the validity of the act, authority legitimacy, generalizability, provocation, and retaliation, and rejecting reconciliation, were assigned a score of 0, and responses denying the validity of the act, supporting lack of authority legitimacy, denying the validity of provocation, retaliation, and generalization, and approving reconciliation were assigned a score of 1. This method of coding has been used in over 100 prior empirical studies (for reviews, see Smetana, 2006; Turiel, 1998).

Justifications for judgments were coded with a reliable coding system developed for previous studies and modified accordingly based on the interview responses (Ardila-Rey & Killen, 2001; Smetana, 2006) (See Table 2). Coding children's justification responses into categories aids in assessing what aspects of social and moral situations children attend to when making their evaluations. Children's justifications were coded into five categories: (1) Moral (Fairness, Equality, Rights, Others' Welfare); (2) Prosocial and Friendship; (3) Social-conventional (Authority, Rules, and Punishment Avoidance); (4) Retribution and Self-Defense; and (5) Other (undifferentiated responses) (see Table 2 for examples). Data were entered as 1 if the type of justification was used (.5 each if two types of justifications were used), and 0 if it was not used. Thus, justifications were the proportions of the moral, prosocial, social-conventional, and retribution categories used by children.

Table 2. Justification Coding Categories.

Types of Justifications
1. Moral: Fairness, Equality, Rights, and Others' Welfare: Appeals to equal treatment for all, fair distribution of resources (“All the children need to get toys”), and avoiding physical or psychological harm (e.g., “It would hurt them”, “She would get upset, sad”).
2. Prosocial and Friendship: Appeals to being nice, being friends, and being kind to others (e.g., “Friends love each other and treat each other special”; “They should talk instead, so they don't fight anymore”).
3. Social Conventional: Authority, Rules, and Punishment Avoidance: Appeals to teachers' or parent's authority (e.g. “The parent knows best” “She is in command”, “You always have to obey your parents”), adhering to conventional rules (e.g. “It's bad manners”, “It's not what's done” “It's not polite”) and assigning punishment for transgressors. (e.g. “She would be sent to time-out if she hits the other girl”; “ I don't want him to get punished”).
4. Retribution and Self-defense: Appeals to the need for retribution and self-defense: (“One shouldn't let anybody take advantage”, “She has to defend herself, “If they hit her, she has to hit back”)
5. Other: Undifferentiated. (“Because.”)

Reliability and power analyses

Two coders, blind to the gender, age, and social group of the interview respondents, independently coded 25% of the responses to the Moral Judgments Interview. Coding reliability percent agreement and Cohen's kappa coefficients were 90%, and .87, respectively. Power analyses were conducted to ensure the robustness of the design and computations were high. With the exception of one analysis at .50, power ranged from .70 to 1.0 for judgments and from .81 to 1.0 for justifications.

Results

Exposure to Violence

Plan for Analysis

The 18 VEX variables were classified into four categories corresponding to the violence level and type of exposure of the aggressive acts represented in the instrument (see Table 1). These categories were labeled “mild violence witness”, “mild violence victim”, “severe violence witness” and “severe violence victim”. Analyses were conducted on the number of children that reported being exposed at least once to each level and type of violence, and on the mean proportion of times children in general reported being exposed to violent events within each category (has this happened to you/have you seen this happen?). The context of the episodes (where/who?) and the rate of recurrence for each event reported within each category (how many times has this happened?) were also analyzed.

Number of children exposed to violence

Overall, 98% of the children reported having witnessed or been a victim of at least one incidence of violence indicating that the vast majority of all children in the study had some level of exposure to violence.

Proportion of violent events reported

A total of 1,147 events were reported, 56% of these events reported by displaced children, and 44% reported by non-displaced children. Individual children reported an average of 5.9 incidents of exposure to violence, ranging from 0 to 17 incidents reported by child, with a standard deviation of 3.4 (Mode = 4). The mean number of mild violent events reported per child was 4.8 (SD = 2.6), and the mean for severe violent events was 1.05 (SD = 1.4). (see Figure 1).

Figure 1. Frequency of mild and severe violence reported by context.

Context of violence occurrences

Children described the context of occurrence (“Where did it happen?” “Who was involved?”) for 98.4% of the violent events reported. The data in Figure 1 show the frequency of and context in which the violence occurred for mild and severe violence. Overall, 45 % (N = 503) of the incidents referred to street violence. A total of 36% (N = 406) of the incidents occurred in the home/family context, 19 % (N = 211) in the school context, and only 0.8% (N = 9) were explicitly referred to as occurring in the context of political violence (e.g. children identified perpetrators as either army, guerrilla, or paramilitary). The incidence of home/family violence was similar for both groups, with 69% of the children reporting being spanked at least once. Displaced children reported a higher incidence of street violence than non-displaced children.

Rate of recurrence

Each event was ranked for recurrence from 0 = never to 3 = many times. Small but significant negative correlations between rate of recurrence and age indicated that even though more of the older children reported exposure to different violent events overall, younger children were more likely to report recurrent violence exposure. (For age and mild violence for displaced children, r = .23, p < .05; for age and severe violence for non-displaced children, r = .22, p < .05).

To further analyze exposure to violence by age, gender, and school or risk group, a 2 (school: displaced, non-displaced) X 2 (gender: girl, boy) X 3 (age: 6, 9, 12) X 2 (violence level: mild, severe) X 2 (role: victim, witness) ANOVA with repeated measures on the last two factors was conducted on the mean proportions of reported exposure events data. Significant findings were analyzed further using 2 (school: displaced, non-displaced) X 2 (gender) X 3 (age) univariate ANOVAs for between group effects, and paired samples t-tests for within group effects. The results revealed a significant between-groups effects for school, F (1,181) = 19.643, p < .001, and gender, F (1, 181) = 6.981, p < .043. Displaced children reported more instances of exposure to violent events (M = .42, SD = .27) than did non displaced children (M = .35, SD = .23). Gender differences were due to boys reporting more incidents (M = .39, SD = .25) than did girls (M = .30, SD = .21). No significant age main effects or interactions were found.

A significant Violence Level X School interaction, F (1, 181) = 7.367, p <. 01 revealed that although displaced children reported significantly more incidences of violence across both levels than did non-displaced children, the difference was greater for the severe violence events (displaced M = .30, SD = .35, non-displaced M = .11, SD = .17), F (1, 181) = 25.441, p < .001, than for the mild violence events (displaced M = .53, SD = .27, non-displaced M = .45, SD = .23), F (1, 181) = 5.492, p < .02.

In sum, displaced children were exposed to more violence than were non-displaced children, and boys were exposed more than were girls. Further, while all children reported more mild than severe violence, displaced children reported more severe violence than did non-displaced children.

Moral Judgments Interview

Plan for Analysis

To test our hypotheses about children's judgments about moral transgressions, we initially conducted separate 2 (school: displaced and non-displaced) X 2 (gender of participant: female, male) X 3 (age group of participant: 6, 9, 12) X 2 (scenarios: aggression, unfair distribution of resources) MANOVAs with repeated measures on the last factor on children's judgments. Only one gender main effect (at .05 level) was found for all of the analyses, and thus all analyses were rerun without gender as a variable. Thus, we report here results from the 2 (school) X 3 (age group) X 2 (scenarios) ANOVAs with repeated measures that were conducted separately for each assessment. Post hoc comparisons were performed using one-way ANOVAs with Tukey's HSD for between subjects differences, and paired-samples t-test for within subjects differences.

To test hypotheses regarding children's use of justifications, separate 2 (school: displaced and non-displaced) X 3 (age group of participant: 6, 9, 12) X 2 (scenarios: aggression, unfair distribution of resources) X 3 (justification: Moral, Prosocial, Social-Conventional,) MANOVAs with repeated measures on the last 2 factors were conducted for the three Domain questions (Evaluation, Authority, Generalizability), and separate 2 (school: displaced and non-displaced) X 3 (age group of participant: 6, 9, 12) X 2 (scenarios: aggression, unfair distribution of resources) X 4 (justification: Moral, Prosocial, Social-Conventional, Retribution) MANOVAs with repeated measures on the last 2 factors were conducted for the three Context questions (provocation, retaliation, reconciliation). Only justification categories with a proportion larger than .05 were used for analyses. Follow-up tests were conducted as described above. All analyses of justifications were conducted on the proportions of justification categories used by participants, as has been done in previous research using justification data (see Smetana, 2006). ANOVA-based procedures are robust when used with dichotomous data (for details, see Wainryb, Shaw, Laupa, & Smith, 2001, footnote 4).

Evaluation, Authority Jurisdiction, and the Generalizability of Moral Transgressions

We first report the findings for children's responses for the evaluation (okay or not okay?), authority (okay if parent says it's okay?) and generalizability (okay in another country?) assessments. Confirming our hypotheses, all of the non-displaced (1.00) and the vast majority of the displaced children (.97) viewed aggression (hitting) and the denial of resources (not sharing toys) as wrong (see Table 3). Analyses for children's reasoning about their evaluations revealed a significant school by age effect, F (2,187) = 4.13, p < .05. A small minority of the youngest displaced children viewed moral transgressions as all right; all other children viewed it as wrong. Similarly, the vast majority of the children viewed hitting and not sharing toys as wrong even when a parent condoned it, thus viewing moral transgressions as wrong independent of authority mandates (see Table 3). A school by age effect on the authority assessment, F (2, 187) = 5.47, p < .01, indicated that children from the youngest age group in the displaced school were more likely to view moral transgressions as okay when parents approved it than were older children in the displaced school or for any of the non-displaced children. There were no significant effects for generalizability; the vast majority of all children (M = .82) viewed moral transgressions as wrong in a different country (see Table 3). Even though nearly all children evaluated aggression as wrong and only a small minority condoned the transgression when happening in another country, a main effect for question revealed that the minority of children who condoned the transgression in another country was statistically significantly larger than those who condoned the original aggression, F (2, 186) = 37.61, p < .001.

Table 3. Proportions of Judgments for Evaluation of Transgressions, Authority, and Generalizability by Assessment and Scenario.

Assessment by Scenario
	Aggression			Distribution of Resources
	Eval	Auth	Gener	Eval	Auth	Gener
Displaced
6 yr-old	.91	.84	.78	.91	.78	.77
	(.30)	(.37)	(.42)	(.30)	(.42)	(.42)
9 yr-old	1.00	.97	.87	1.00	.97	.84
	(.00)	(.18)	(.34)	(.00)	(.18)	(.37)
12 yr-old	1.00	1.00	.84	1.00	.97	.90
	(.00)	(.00)	(.37)	(.00)	(.18)	(.30)
Total	.97	.94	.83	.97	.90	.84
	(.18)	(.25)	(.38)	(.18)	(.30)	(.37)
Non-Displaced
6 yr-old	1.00	.97	.87	1.00	.97	.79
	(.00)	(.16)	(.34)	(.00)	(.16)	(.41)
9 yr-old	1.00	.97	.70	1.00	.90	.67
	(.00)	(.18)	(.47)	(.00)	(.31)	(.48)
12 yr-old	1.00	.93	.77	1.00	.87	.80
	(.00)	(.25)	(.43)	(.00)	(.35)	(.41)
Total	1.00	.96	.77	1.00	.92	.76
	(.00)	(.20)	(.41)	(.00)	(.27)	(.43)
Total	.98	.95	.81	.98	.91	.80
	(.12)	(.22)	(.40)	(.12)	(.28)	(.40)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. 0 = OK. 1 = Not OK. Eval. = Evaluation of Conflicts. Auth. = Authority. Gener. = Generalizability.

The analyses of justifications for children's evaluations revealed a significant justification effect, F (2, 374) = 641.61, p < .001, confirming our expectation that the vast majority of all children would use moral justifications more often than prosocial or social-conventional justifications to evaluate transgressions as wrong. A justification by age group effect, F (4, 374) = 5.98, p < .01, indicated that the use of moral reasons increased with age, while the use of social-conventional justifications decreased. There were no significant school differences, as shown in Table 4. There were no age or school differences for the use of justifications for the authority jurisdiction or generalizability assessments, and thus these means are not reported as there were no hypotheses associated with justifications for these assessments.

Table 4. Proportions of Justifications for Evaluation of Transgressions by Scenario.

Justifications by Scenario
	Aggression			Distribution of Resources
	Moral	Pros	S-Con	Moral	Pros	S-Con
Displaced
6 yr-old	.64	.14	.06	.80	.07	.09
	(.48)	(.34)	(.25)	(.40)	(.20)	(.30)
9 yr-old	.97	.03	.00	.90	.06	.03
	(.18)	(.18)	(.00)	(.30)	(.25)	(.18)
12 yr-old	.85	.10	.05	.87	.13	.00
	(.35)	(.30)	(.20)	(.34)	(.34)	(.00)
Total	.82	.09	.04	.86	.08	.04
	(.38)	(.28)	(.18)	(.35)	(.27)	(.20)
Non-Displaced
6 yr-old	.76	.05	.19	.81	.09	.05
	(.43)	(.19)	(.39)	(.39)	(.28)	(.22)
9 yr-old	.98	.00	.02	.88	.08	.03
	(.09)	(.00)	(.09)	(.31)	(.27)	(.18)
12 yr-old	.93	.03	.03	.80	.17	.00
	(.22)	(.18)	(.13)	(.41)	(.38)	(.00)
Total	.88	.03	.09	.83	.11	.03
	(.31)	(.16)	(.27)	(.37)	(.31)	(.17)
Total	.85	.06	.06	.84	.10	.04
	(.35)	(.23)	(.23)	(.36)	(.29)	(.19)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. Pros = Prosocial. S-Con = Socio-Conventional. Only justification categories with a proportion larger than .05 were used for analyses

Exposure to Violence and Straightforward Moral Judgments

Analysis of the correlations between violence exposure scores and the initial evaluations of moral transgressions revealed significant results for the aggression scenario. A significant positive correlation was found between agreement with the initial transgression in the aggression scenario and being a victim of severe violence. When looking within each social group separately, this correlation was significant only for the displaced children (r = .22, p < .05) (See Table 5). As Table 5 also shows, judging it legitimate to condone unfair distribution of resources in another country (failing to recognize a criterion of morality) was correlated with witnessing mild (r = .24) and severe violence (r = .21), (ps < .05) for the displaced children only. These results seem to indicate that, independent of other factors inherent to the population group, exposure to violence was significantly related to the straightforward moral judgments.

Table 5. Correlations Between Exposure To Violence and Moral Judgment Assessments.

Level and Type of Violence Exposure by Scenario
Assessment	Mild Violence			Severe Violence			Total
	Witness	Victim	Total	Witness	Victim	Total
All Children
Aggression Scenario
Initial Transgression	ns	ns	ns	ns	.20 *	ns	ns
Distribution of Resources Scenario
Provocation	.15 *	ns	ns	.29 **	.17 *	.30 **	.19 **
Displaced Children
Aggression Scenario
Initial Transgression	ns	ns	ns	ns	.22 *	ns	ns
Distribution of Resources Scenario
Generalizability	.24 *	ns.	ns	21 *	ns	ns	ns
Provocation	.21 *	ns	ns	.35 **	ns	.34 **	.28**

Note. N = 193. Displaced N = 94. Non-Displaced N = 99

Only groups with significant correlations are shown. ns = non-significant correlation

^*= correlation significant at the .05 level,

^**= correlation significant at the .01 level.

In sum, as hypothesized, the vast majority of all children thought that unprovoked transgressions were wrong even when approved by parental authority. A small minority of children were more likely to agree with transgressions, however, after they were told the transgressor lived in another country with different or no rules. The level and role of exposure to violence was correlated to children's assessments of transgressions. Children who had been victims of severe violence were more likely to agree with the aggressive transgression than were children had not been victims.

Provocation, Retaliation, and Reconciliation

As hypothesized, displaced children were more likely to condone moral transgressions in provoked situations than were non-displaced children. This was revealed by a significant school effect, F (1, 187) = 12.47, p < .001 (the means are shown in Table 6). There were no significant effects for scenario or age. Analyses for children's use of justification categories demonstrated a significant Justification by School effect, F (3, 561) = 4.82, p < .01, which indicated that justifications for provocation varied by school, as shown in Table 7. As expected, displaced children justified their judgments by using retributive reasons more often than did non-displaced children, F (1,191) = 10.61, p < .01, and non-displaced children gave prosocial reasons more often that did displaced children, F (1, 191) = 5.85, p < .05. The following is an illustration of two participants' responses for the provocation assessment: A 12-year-old displaced girl used a prosocial justification for rejecting provocation in the aggression scenario: “It would be wrong. They should (keep) respect (to) each other, and talk to find out if they have ill feelings for each other” [“Estaria mal, Se deben guardar respeto entre las dos, y hablar a ver si tienen bronca entre las dos.”] An 8-year-old displaced boy accepting provocation in the distribution of resources scenario said: “It is ok that he takes all the toys, because, why would they make fun of him if he didn't do anything to them? That's why he was mad, that's why he took all the toys. THEN, IS IT WRONG OR RIGHT? Wrong for them to make fun of him. SO, IS IT OK OR WRONG FOR HIM TO TAKE ALL THE TOYS? It's Ok.” [“Esta bien que el coja todos los juguetes, porque, por qué se burlan de el si el no les hizo nada? Por eso es que el esta bravo, por eso es que se cojió todos los juguetes. ENTONCES, ESTA BIEN O MAL HECHO? esta mal hecho que ellos se burlen de el. ENTONCES ESTUVO BIEN O ESTUVO MAL QUE EL SE COGIERA TODOS LOS JUGUETES? estuvo bien.”]

Table 6. Proportion of Judgments for Provocation, Retaliation and Reconciliation by Scenario.

Assessment by Scenario
	Aggression			Distribution of Resources
	Provoc	Retal	Recon	Provoc	Retal	Recon
Displaced
6 yr-old	.66	.50	.81	.72	.68	.94
	(.48)	(.51)	(.40)	(.46)	(.48)	(.25)
9 yr-old	.74	.81	.77	.74	.94	.90
	(.44)	(.40)	(.42)	(.44)	(.25)	(.30)
12 yr-old	.77	.68	.90	.65	.90	.94
	(.42)	(.48)	(.31)	(.49)	(.30)	(.26)
Total	.72	.66	.83	.70	.84	.93
	(.45)	(.48)	(.38)	(.46)	(.37)	(.26)
Non-Displaced
6 yr-old	.87	.79	.90	.85	.90	.87
	(.34)	(.41)	(.31)	(.37)	(.31)	(.34)
9 yr-old	.90	.90	1.00	.83	.97	.97
	(.31)	(.31)	(.00)	(.38)	(.18)	(.18)
12 yr-old	.90	.83	.97	.93	.97	1.00
	(0.31)	(.38)	(.18)	(.25)	(.18)	(.00)
Total	.89	.84	.95	.87	.94	.94
	(.32)	(.37)	(.22)	(.34)	(.24)	(.24)
Total	.89	.84	.95	.87	.94	.94
	(.32)	(.37)	(.22)	(.34)	(.24)	(.24)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. For Provocation and Retaliation 0 = OK. 1 = Not OK; for Reconciliation 0 = not OK. 1 = OK. Provoc=Provocation. Retal = Retalation. Recon = Reconciliation.

Table 7. Proportion of Justifications for the Provocation Assessment by Scenario.

Justification by Scenario
	Aggression				Distribution of Resources
	Moral	Pros	S-Con	Retrib	Moral	Pros	S-Con	Retrib
Displaced
6 yr-old	.47	.09	.11	.22	.34	.12	.22	.19
	(.51)	(.27)	(.30)	(.42)	(.48)	(.34)	(.42)	(.40)
9 yr-old	.37	.21	.13	.29	.26	.32	.08	.29
	(.48)	(.40)	(.34)	(.46)	(.43)	(.46)	(.26)	(.46)
12 yr-old	.35	.31	.10	.21	.39	.19	.00	.35
	(.47)	(.46)	(.27)	(.40)	(.50)	(.40)	(.00)	(.49)
Total	.40	.20	.11	.24	.33	.21	.10	.28
	(.48)	(.39)	(.30)	(.43)	(.47)	(.40)	(.30)	(.45)
Non-Displaced
6 yr-old	.37	.27	.17	.10	.60	.12	.11	.15
	(.47)	(.44)	(.37)	(.31)	(.49)	(.31)	(.31)	(.37)
9 yr-old	.40	.30	.23	.07	.33	.42	.12	.13
	(.48)	(.47)	(.41)	(.25)	(.46)	(.50)	(.31)	(.35)
12 yr-old	.22	.48	.13	.13	.40	.43	.05	.08
	(.39)	(.48)	(.32)	(.35)	(.50)	(.50)	(.20)	(.27)
Total	.35	.34	.18	.10	.46	.30	.09	.13
	(.47)	(.47)	(.37)	(.30)	(.50)	(.46)	(.28)	(.33)
Total	.37	.27	.15	.17	.40	.26	.10	.20
	(.47)	(.44)	(.34)	(.37)	(.48)	(.43)	(.29)	(.40)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. Pros = Prosocial. S-Con = Social-Conventional. Retrib = Retribution.

Similarly, the hypothesis that displaced children would view retaliation as more legitimate than would non-displaced children, was confirmed by a significant school effect, F (1, 186) = 11.72, p < .001 (for means, see Table 6). There was also a significant effect for type of transgression, F (1, 186) = 20.54, p < .001, with more displaced children viewing retaliation as all right in the aggression story than in the unfair distribution story, as shown in Table 6. Follow-up tests for the significant age effect, F (2, 186) = 7.13, p < .01, indicated that the youngest group was more likely to support retaliation than were the two older groups, ps < .01.

Children's justifications for the retaliation assessment paralleled the findings for the provocation assessment. The displaced children used different reasons for retaliating than did the non-displaced children, as indicated by a significant Justification by School effect, F (3, 561) = 3.77, p < .05. Follow-up tests indicated that retribution was used as a reason by the displaced children more often than was used by the non-displaced children, F (1,191) = 5.40, p < .05, and prosocial reasons were used by non-displaced more often that by displaced children F (1, 191) = 5.93, p <.05. Follow-up to a story by justification interaction, F (3, 561) = 7.754, p <.001, revealed that retribution was used more frequently for the aggression story than for the unfair distribution story, (p <.001), whereas moral justifications were used more frequently for the unfair distribution story (p < .001), as shown in Table 8. A significant justification by age effect, F (6, 561) = 5.85, p < .001, indicated that younger children used different reasons than did older children. Follow-up tests indicated that young children were more likely to use retribution justifications than were older children, F (2, 190) = 3.87, p < .05, and that older children used more prosocial reasons than did younger children, F (2,190) = 12.55, p < .001.

Table 8. Proportion of Justifications for the Retaliation Assessment by Scenario.

Justification by Scenario
	Aggression				Distribution of Resources
	Moral	Pros	S-Con	Retrib	Moral	Pros	S-Con	Retrib
Displaced
6 yr-old	.27	.11	.09	.31	.42	.14	.11	.20
	(.44)	(.30)	(.30)	(.47)	(.49)	(.34)	(.30)	(.40)
9 yr-old	.37	.23	.18	.19	.44	.26	.27	.03
	(.48)	(.42)	(.38)	(.40)	(.50)	(.44)	(.44)	(.18)
12 yr-old	.23	.32	.06	.32	.48	.27	.15	.10
	(.40)	(.48)	(.21)	(.48)	(.47)	(.44)	(.32)	(.30)
Total	.29	.22	.11	.28	.45	.22	.18	.11
	(.44)	(.41)	(.30)	(.45)	(.48)	(.41)	(.36)	(.31)
Non-Displaced
6 yr-old	.32	.19	.18	.23	.59	.13	.17	.09
	(.47)	(.39)	(.39)	(.43)	(.48)	(.32)	(.37)	(.28)
9 yr-old	.30	.37	.27	.07	.45	.33	.15	.00
	(.46)	(.49)	(.45)	(.25)	(.50)	(.48)	(.35)	(.00)
12 yr-old	.17	.50	.20	.13	.22	.63	.03	.07
	(.38)	(.51)	(.41)	(.35)	(.41)	(.47)	(.18)	(.25)
Total	.27	.34	.22	.15	.43	.34	.12	.06
	(.44)	(.47)	(.41)	(.36)	(.49)	(.47)	(.32)	(.22)
Total	.28	.28	.16	.22	.44	.28	.15	.08
	(.44)	(.45)	(.37)	(.41)	(.48)	(.44)	(.34)	(.27)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. Pros = Prosocial. S-Con = Social-Conventional. Retrib = Retribution.

A 9-year-old displaced boy gave the following moral and prosocial justifications for rejecting retaliation in the aggression scenario: “Wrong, she was calm. WHY IS IT WRONG? Because that would be revenge and revenge is not good. She should forgive, but, it wouldn't be fair if Diana did that to Maria. First, Maria wasn't doing anything to her, and second, who knows which mean thing she did to Maria for her to take revenge on Diana.” For another example, a 6 year-old non-displaced boy agreeing with retaliation in the aggression scenario used the following “retribution” reason: “It's OK to hit her back”.” WHY?” “Because if she hit her first, she should also hit her so that she feels it.”

Our expectations for children's evaluations of reconciliation were open-ended. The vast majority of all children, displaced and non-displaced, judged that the children could reconcile their conflict after the transgression had taken place (“Yes, they could still be friends”). Analyses revealed that there were significant effects for school, F (1, 186) = 5.36, p < .05, which were qualified by a significant interaction effect for scenario by school, F (1, 186) = 4.21, p < .05. As shown in Table 6, displaced children were less likely to judge reconciliation feasible when the act was about aggression than when it was about the unfair distribution of resources. Analyses for justifications for reconciliation found no effects for school. Children in both groups justified reconciliation using primarily prosocial reasons (M = .73, SD = .44) (see Table 9).

Table 9. Proportion of Justifications for the Reconciliation Assessment by Scenario.

Justification by Scenario
	Aggression				Distribution of Resources
	Moral	Pros	S-Con	Retrib	Moral	Pros	S-Con	Retrib
Displaced
6 yr-old	.09	.66	.03	.03	.03	.52	.11	.03
	(.30)	(.48)	(.18)	(.18)	(.18)	(.50)	(.30)	(.18)
9 yr-old	.06	.71	.00	.10	.08	.84	.00	.00
	(.25)	(.46)	(.00)	(.30)	(.26)	(.35)	(.00)	(.00)
12 yr-old	.03	.71	.03	.06	.03	.77	.00	.03
	(.18)	(.46)	(.18)	(.25)	(.18)	(.42)	(.00)	(.18)
Total	.06	.69	.02	.06	.05	.71	.04	.02
	(.25)	(.46)	(.15)	(.25)	(.21)	(.45)	(.18)	(.15)
Non-Displaced
6 yr-old	.05	.71	.06	.03	.13	.77	.03	.00
	(.22)	(.46)	(.23)	(.16)	(.34)	(.43)	(.16)	(.00)
9 yr-old	.02	.78	.07	.00	.07	.82	.03	.00
	(.09)	(.41)	(.25)	(.00)	(.25)	(.38)	(.18)	(.00)
12 yr-old	.07	.82	.03	.00	.00	.73	.00	.00
	(.25)	(.38)	(.18)	(.00)	(.00)	(.45)	(.00)	(.00)
Total	.05	.76	.06	.01	.07	.77	.02	.00
	(.20)	(.42)	(.22)	(.10)	(.25)	(.42)	(.14)	(.00)
Total	.05	.73	.04	.04	.06	.74	.03	.01
	(.22)	(.44)	(.19)	(.19)	(.23)	(.43)	(.16)	(.10)

Note. N = 193. Displaced N = 94. Non-Displaced N = 99. Proportions cannot exceed 1.00. Standard deviations are in parenthesis. Pros = Prosocial. S-Con = Social-Conventional. Retrib= Retribution.

Exposure to Violence and Contextual Moral Judgments

The level and role of exposure to violence were correlated to children's reasoning regarding provocation. Pearson correlation analyses showed that children who had witnessed more mild violence (r = .15, p < .05), or who had witnessed (r = .29, p < .001) or been victims of more severe violence (r = .17, p < .05), were more likely to approve of provoked transgressions in the distribution of resources scenario (thus, they approved that the child took all the toys in response to the provocation). When analyzing each group separately, though, these correlations turned out to be significant only for the displaced children (see Table 5). There were no significant correlations between level and type of violence exposure and approval of retaliation. While more displaced children were likely to support provocation and retaliation as a legitimate cause or response to aggression and unfair distribution of resources, the majority of all children rejected these forms of response (see Table 6 for means). In similar fashion, no significant correlations were found for level and type of violence exposure and acceptance of reconciliation.

Discussion

The pervasive violence in Colombia has taken its toll on Colombian children, and according to the results of this study, negatively affect children's moral development. Our measure of exposure to violence (VEX; Leavitt & Fox, 1995) revealed that the vast majority of children interviewed had witnessed or been a victim of violence. We found that boys had been exposed to violence more often than had girls in our sample. Displaced children in this study were more likely to report suffering severe violent events than did middle-class children, indicating that being a displaced child in Colombia, particularly a young child from 5 to 7 years of age, is synonymous of living with a higher risk of becoming a victim of violence. Exposure to violence levels were clearly linked to other variables (e.g., levels of stress, SES, access to resources) and thus the findings have to be interpreted within these contextual conditions.

Our hypotheses regarding exposure to violence and moral judgments between children who were displaced by the war and their peers not displaced by the war were confirmed in this study. Colombian children from high-risk environments, exposed to extreme violence and who were displaced by the war, did not differ in how they evaluated an unprovoked moral transgression from Colombian children from low-risk environments, exposed to minimal violence, and who were not displaced. At the same time, children exposed to violence and displaced by the war, were more likely to condone moral transgressions (such as hitting or not sharing toys) when provoked or for reasons of retaliation than were non-displaced children who had low exposure, that is, living in intact families in more sheltered environments.

Surprisingly, there were no major differences between the two groups regarding reconciliation; the majority of all children judged that the transgressor and the recipient could be friends following the conflict. Yet, caution should be exercised in the interpretation of these results. Even though the statistical significance was high for most of our findings, other factors not specifically explored in this study such as family structure, parent education, or school curriculum, or more detailed aspects of reconciliation may also be contributing to the differences found (for example, Forero-Pineda & Escobar-Rodriguez (2002) found that for Colombian children, schools acted as buffers from the effects of outside violence). Moreover, the results of this study have to be interpreted within the constraints of the two communities as they differed in ways beyond the level of violence exposure. Nevertheless, these complex patterns of findings are important to document, and offer promise for intervention.

As predicted, children in both groups evaluated moral transgressions as wrong. Children clearly stated that it would be wrong to hit others or not share toys even when an adult condoned it (moral principles are not under authority jurisdiction) and that it would be wrong for people in another country to condone it (the principle is generalizable). The findings in the present study, then, extend the current evidence (see Smetana, 1995; Smetana & Turiel, 2003) regarding the use of moral criteria, such as authority jurisdiction and generalizability, to a sample of children exposed to violence in a war-torn region of the world. Still, a slightly larger minority of children condoned a moral transgression when it happened in another country than the minority of children who condoned the unprovoked transgression. In addition, despite cultural portrayals of Colombian culture as rule and authority oriented (see Triandis, 1990), children did not evaluate the transgressions as under authority jurisdiction; the acts were wrong for reasons pertaining to the negative intrinsic consequences (as evidenced by their use of moral justifications), similar to what has been found in previous research elsewhere (see Laupa, Turiel, & Cowan, 1995). These results extend the findings by Ardila-Rey and Killen (2001) who documented the coexistence of collectivistic and individualistic judgments in middle-income Colombian children. In the present study, children in diverse economic conditions also displayed a range of orientations regarding authority and morality. A similar coexistence of divergent beliefs or values among Colombian adults has been described by Ardila (2004), and by Killen, Ardila-Rey, Barakkatz, and Wang (2000).

The youngest group of children, for both displaced and non-displaced groups, used less moral reasoning than did the older children when justifying their evaluation of the transgressions. A minority of the 6 year olds in the displaced group used moral criteria significantly less often than did older children. They viewed these acts as contingent on parental approval, and were more likely to use prosocial rather than moral reasons when assessing moral situations than were the older children (and non-displaced younger children). At the same time, about 20% of the 6 year olds in the non-displaced group used social conventional reasoning more often than did the older children. Yet, previous studies on moral reasoning in the United States have found that even preschool children are able to provide domain appropriate justifications when reasoning about moral situations (Killen et al., 1994, Nucci, 1984). It is possible that for the younger children in these two communities the higher use of social-conventional reasoning than would be expected may be due to the strong authority presence in their lives. Parents and teachers in Colombia are concerned about the precarious situations that children often confront in a war-torn area, and young children may be more likely to defer to authority mandates, even when the mandate violates a moral principle. This interpretation of the findings remains to be tested. The positive aspect of the findings in this study is that, with age, children used moral criteria and moral reasoning to evaluate the infliction of harm on another, and the unfair distribution of toys, indicating that despite the difficult life circumstances these children face, they seem to be able to maintain a basic recognition about what makes hitting and not sharing resources wrong.

The areas of divergence for the displaced and non-displaced children's evaluations of transgressions were with their responses to provocation and retaliation. Displaced children justified provocation using retributive reasons more often than did non-displaced children. Given their exposure to violence, it was surprising to find that while more displaced than non-displaced children relied on retribution, the majority of all children used moral and prosocial reasons rather than using non-moral reasons to reject retaliation. How it is that some of the displaced children were resilient to their stressful environment requires further study. It may be that these children have adults in their communities who provided the social support that has been shown to be so necessary to help children's resilience to stress from violence exposure (Garbarino et al., 1992, 1998). Clearly, children whose families are stable and offer a secure home environment are better able to cope with violent experiences than children without this support. Most likely, this may account partially for the differences between the two groups. Often, adults in communities where violence is prevalent usually present symptoms of anxiety, depression and fear (Groves, 1996; Osofsky, 1995) lessening their ability to provide social support to their children. These factors and other variables should be explored as a way of interpreting why it is that some children are more resilient and less vulnerable to exposure to violence than are other children.

Consistent with previous findings that showed that the type of provocation and the intention of the act affect children's acceptance of retaliation (Ferguson & Rule 1988), children considered several factors of the situations when making judgments about moral transgressions. First, in this study, provocation was presented as a moral offense (hurtful teasing). Because of the antecedent of this moral offense, children may have perceived that the victim, who was initially presented as innocent, became the perpetrator of an act in some cases worse than the transgression that followed, thus, this transgression was justified. Second, retaliation was presented as physical harm (should they hit her for doing that?) thus, children may have perceived this again as the initial transgressor becoming a victim of a worse wrongdoing than his or her acts. Children were more likely to consider that it was okay to retaliate in the case of the aggressive scenario, however, thus it may be that they viewed this as an issue of retributive justice. Piaget (1932) believed that retaliation represents an early form of reciprocity in retributive justice that leads to the development of ideas of equality –distributive justice. Yet, children's justifications for judgments reflected what in Piaget's view would be a more developed concept of justice, and indicate that children consider the type of transgression when assessing the use of retaliation as a form of justice.

Similar to Astor's findings (1994), children in this study used moral justifications to explain their rejection of the provoked transgressions and retaliation. Children, however, also used other types of justifications when rejecting or agreeing with these concepts. Children rejected the transgressions not only as moral issues (“Because it hurts”; “Because it's not fair”) but also due to preference for prosocial interpersonal interactions (“It's better to forgive”, “They should try to talk it out instead”). On the other hand, children who agreed with the provoked transgressions and retaliation explained their agreement with reasons that reflected a hostile view of social interactions or a need for self-defense (“You should not let others step on you”, “If they hit you, you have to hit back”). The difference in use of justifications was also reflective of the children's exposure to violence. Non-displaced children used prosocial justifications more frequently than displaced children, who used retribution justifications more frequently.

Studies have shown that children who live in aggressive or violent environments have the tendency to develop aggressive behavior (Liddell et. al, 1994; Patterson, Kuperdsmit, & Vaden, 1990), and that subsequent exposure to violence will affect children who display aggressive behaviors more adversely than children who are not involved in aggression (Cummings, Iannoti, & Zahn-Waxler, 1985). Similarly, Brook et al (2003) found that Colombian adolescents who had been the object of violence were more likely to display violent behaviors than did adolescents who were in violent communities but had not been victims themselves. These children are more likely to develop negative views of the world (Garbarino et.al., 1998) and possibly less sympathy and perspective taking, factors which affect prosocial reasoning (Eisenberg, Zhou, Koller, 2001). Franco Agudelo (1997) explains that living amidst frequent and pervasive violence can lead to an acceptance of violence as a trivial matter of little significance which leads individuals to take on an attitude that violence is a fact of life that cannot be stopped and to have muted, if any, responses to the continued violence. The similarity on the assessment of provocation between displaced children in this study and aggressive children in Astor's study (1994) indicates that displaced children may be developing negative and aggressive views of the world, and that their perception of social interactions is tinted by their frequent negative experiences with pervasive violence.

What makes children in our study different from aggressive children studied elsewhere, however, is that the Colombian children viewed it feasible for the transgressor and the recipient to be friends after the conflict, confirming prior studies with normative samples in the United States (Verbeek & de Waal, 2001). A small minority of the displaced children rejected the possibility of reconciliation for the aggressive transgression scenario, but not for the unfair distribution scenario. Research has shown that while increased accounts of aggressive behavior in children's lives have long-term negative consequences, increased accounts of object disputes do not (see Shantz, 1987). Unlike aggression, object disputes can be “undone”; object disputes serve as an important source of experience for children to learn to negotiate, compromise, and bargain (see Hay & Ross, 1982; Ross, 1996).

These findings also relate to studies from social information-processing perspectives, which examine hostile attributional bias in children (see Arsenio & Lemerise, 2004; Lemerise & Arsenio, 2000). Similar to what has been found for highly aggressive children (Crick & Dodge, 1994; Dodge, 1985; Dodge & Crick, 1990; Waas, 1988; Zelli, Dodge, Lochman, & Laird, 1999) children with higher exposure to violence in this study seemed to infer hostile motives in the actions of others more often than did children with lower exposure to violence. Like highly aggressive children, children who have been exposed to high levels of violence favored the use of retaliation for instances of direct provocation. It is possible that due to their negative social experiences with violence, these children are making more generalized hostile attributions to the intentions of others from behavioral or informational cues.

While boys were exposed to more violence than were girls, there were almost no significant gender differences for the evaluations of moral transgressions. While the youngest boys, more than the youngest girls, used retribution, the sample was not large enough to document this difference statistically. Further research focusing on the youngest age group should explore these potential gender differences with a larger sample and should examine more in depth how the intersection of the characteristics of the environment and the child's individual characteristics are reflected on the formation of moral judgments.

Although other factors, such as emotional adjustment or personality were not assessed in this study, these findings also concord with Kostelny and Garbarino's finding (1994) that 5 to 6 year old children in Israel suffered more personality and behavioral changes after the Intifada than did older children. Younger children in this study were more likely to provide hostile responses than were older children. Even though measures of cognitive development were not assessed in the present study, nor was children's agreement with the different actors of the violence in Colombia (e.g., guerrilla), it is possible that, as found by Kostleny and Garbarino (1994), older children's cognitive competence is helping them make better sense of the violence, or that they might be resorting to ideology as a buffering factor in coping with violence and aggression. Most studies on violence have explored its effects on children's behavior, personality, and emotional disturbances. This study is one of the first to account how violence may affect the development of moral reasoning.

There were several limitations of the study, which need to be addressed. First, the populations studied diverged not only in displacement status, but also in socioeconomic status (SES) and other related factors. Previous research (Eisenberg, Zhou. & Koller, 2001) has shown that Brazilian adolescents from lower SES backgrounds were lower in prosocial reasoning which also may have contributed to our results. However, given the current circumstances of violence and conflict in Colombia that affected data collection, it was not possible to obtain a matched sample of displaced and non-displaced children controlling for all other variables. Thus, violence and SES, were highly correlated in this study, which is often inevitable when studying special populations, particularly in high stress contexts. Although analyses of correlations between moral judgments and violence were conducted within each group separately, it is difficult to differentiate effects of the various factors related to SES from those of violence on children's moral reasoning. These dimensions require further examination in future research.

Second, the measure of violence exposure, VEX-R, is a self-report measure and children's accounts were not confirmed with parents, nor observations were conducted on the actual experiences of children in the two communities. Future research should include multiple measures of exposure to violence, such as questionnaires for parents and teachers. Third, no measure of children's individual characteristics or of the quality of family interactions was obtained, thus it was not possible to examine family factors that contribute to moral judgments. Further study should be conducted to assess these variables and identify the supportive and coping factors that may foster resilience in children exposed to the violence and displacement.

In sum, the novel findings in this study were that children living in a high-risk environment with a high level of exposure to violence and poverty evaluated moral transgressions differently from children who were not exposed to the same levels of violence and life stress. The results in this study show that the degree of displacement and exposure to violence makes a difference, even in a war-torn country like Colombia. Extremely stressful conditions influence how children evaluate moral transgressions and how they view provocation and retaliation. The other novel and more encouraging findings were that almost all children, displaced or non- displaced, considered reconciliation possible. As reported by Forero-Pineda and Escobar-Rodriguez (2002), schools can foster positive behaviors despite violent external-environment conditions. Research by Hewstone and colleagues has also shown that positive intergroup contact in contexts of high societal conflict can help individuals to view reconciliation as more feasible (Hewstone & Brown, 1986). An important next step will be to design interventions that help displaced children understand the wrongfulness of provocation and retaliation, and that promote the use of reconciliation strategies.. Understanding how extreme environmental conditions influence children's social and moral development is essential information for structuring and facilitating positive social environments for all children.