Pacifists and Revenge-Seekers in Response to Unambiguous Peer Provocation

Abstract

In order to better understand why some children retaliate when they feel provoked and others do not, the present study identified “pacifistically-oriented” children who made negative interpretations in response to unambiguous provocations, yet did not endorse revenge goals, and compared them to “revenge-seeking” children who also made negative interpretations but did endorse revenge goals. Groups were identified based on seventh-graders’ (N = 367; 54.77% male; 22.89% racial/ethnic minority) responses to hypothetical situations in which a peer excluded and insulted them. Comparing these groups revealed that Pacifists endorsed relationship-maintaining goals and emotion regulation goals more highly than Revenge-Seekers. Revenge-Seekers reported more anger and endorsed beliefs about negative reciprocity and aggression being legitimate more highly than Pacifists. Additionally, Revenge-Seekers were more disrespect-sensitive than were Pacifists, based on a measure of vigilance for signs of disrespect and expectations that others would disrespect them. Together these findings point to social-cognitive and emotion-related processes that may inhibit revenge-seeking in unambiguous provocation situations, even when children interpret the peer’s behavior quite negatively.

Introduction

It is generally more acceptable to retaliate in response to intentional harm-doing than to ambiguously intended or clearly accidental harm (Astor, 1994; Posada & Wainryb, 2008; Smetana, Campione-Barr, & Yell, 2003). Indeed, many parents instruct their children to stand up for themselves and get back at someone who has intentionally harmed them in some way (Farrell et al., 2010). Nevertheless, there are many youth who perceive negative intent in another child’s actions, but do not retaliate (e.g., Erdley & Asher, 1996; Quiggle, Garber, Panak, & Dodge, 1992). In studies of responses to ambiguous provocation, the correlation between attributing hostile intent to someone and then behaving aggressively toward that person is consistently significant but modest in magnitude (average r = .17; Orobio de Castro, Veerman, Koops, Bosch, & Monshouwer, 2002), suggesting that there are factors that weaken the association between viewing another person’s actions as negative toward self and responding with aggression.

The fact that many children who make negative interpretations do not act aggressively in response is not typically highlighted by researchers, nor are the reasons typically studied. One exception is Erdley and Asher (1996) who investigated factors that might alter the association between attributions of hostile intent and aggressive strategies in ambiguous provocation situations. For example, they found that relationship-maintaining goals were more likely to characterize children who, despite attributing hostile intent, indicated that they would engage in problem-solving strategies compared to children who also attributed hostile intent but indicated that they would engage in aggressive strategies.

It is one thing to identify factors, as Erdley and Asher (1996) did, that affect the association between negative interpretations of another person’s behavior and aggression toward that person in ambiguous provocation situations, but it would be another if one were to find evidence of this in more unambiguously provocative situations in which the protagonist, for example, simultaneously excludes and insults the child. To examine this possibility, the current study identified “pacifistically-oriented” youth who made negative interpretations in response to unambiguous provocation, yet did not endorse revenge goals, and compared them to “revenge-seeking” youth who also made negative interpretations but did endorse revenge goals. Because revenge-seekers are more aggressive in everyday life and are more likely to be rejected by peers (e.g., Lochman, Wayland, & White, 1993; McDonald & Lochman, 2012; Samson, Ojanen, & Hollo, 2012), identifying processes that characterize pacifists may help us to gain insight into how to better help youth who tend to desire revenge.

Prior research provides promising empirical and theoretical leads concerning the social-cognitive and affective processes that might distinguish pacifists from revenge-seekers in situations characterized by unambiguous provocation. With regard to social cognition, one strong possibility is that revenge-seekers are more likely to believe in the legitimacy of aggression. Children’s beliefs in the legitimacy of aggression have been found to predict aggressive strategy selection in response interpersonal conflict and to ambiguous provocation situations, as well as aggressive behavior in everyday life (e.g., Erdley & Asher, 1998; Huesmann & Guerra, 1997; Slaby & Guerra, 1988). For example, when Erdley and Asher (1998) asked children how they would respond to ambiguous provocation situations in which it was unclear whether a harmful action was intentional they found that children who more highly endorsed the legitimacy of aggression gave higher ratings to aggressive strategies than children who were lower in their endorsement of these beliefs. High endorsers of the legitimacy of aggression were also described by their peers as more aggressive in everyday life in school. Based on this research and other studies (e.g., Werner & Nixon, 2005), it could be expected that children who pursue revenge in response to a far less ambiguous and more explicit provocation are more likely than pacifistically-oriented children to believe that aggression is a legitimate response to harm-doing.

A second strong possibility is that pacifists and revenge-seekers differ in their endorsement of negative reciprocity beliefs. Reciprocity in social relationships is typically conceptualized as a belief or expectation that people should return positive treatment to others when positive treatment has been received. However, reciprocity norms can also pertain to how people should respond to the negative actions of others. Gouldner (1960) suggested that whereas norms favoring positive reciprocity may be a starting mechanism to encourage individuals to interact cooperatively, negative reciprocity beliefs would favor retaliation when another person behaves negatively or fails to cooperate.

Indeed, research with college students indicates that negative reciprocity beliefs predict hostile, retaliatory behaviors after being treated rudely (Eisenberger, Lynch, Aselage, & Rohdieck, 2004). However, research on negative reciprocity beliefs has not yet been conducted with youth. We hypothesized that negative reciprocity beliefs would more highly characterize youth who desire revenge in response to unambiguous peer provocation compared to youth who are pacifistic. As an exploratory question, we also investigated whether revenge-seekers and pacifists differed in their beliefs about positive reciprocity.

A third possibility is that revenge-seekers and pacifists differ in their sensitivity to disrespect. Evidence exists that when children are provoked by a peer, they are likely to make interpretations that the provocateur does not respect them (Miller, 2001). Ethnographers have described communities in which strong norms exist about treating people respectfully, and when these norms are violated, retaliation is acceptable or even expected. This set of shared beliefs that sanction revenge in response to being treated disrespectfully has been referred to as a “code of honor” or the “code of the street” (e.g., Anderson, 1999; Horowitz, 1983; Nisbett & Cohen, 1996). Interpretations of disrespect may lead to revenge, because retaliation serves as a means to communicate that one is worthy of respect and better treatment (Heider, 1958; McCullough, 2008). There is evidence that interpreting another person’s actions as disrespectful is correlated with revenge goals and aggressive strategies (DeBono & Muraven, 2014; MacEvoy, 2006; McDonald & Asher, 2013).

Building on this theoretical and empirical foundation, as well as on research on rejection sensitivity (Downey & Feldman, 1996), we created a new measure of sensitivity to disrespect as an individual difference characteristic and tested the hypothesis that revenge-seekers would be more disrespect sensitive than pacifists. Although a measure has been created to study individual differences in “code of the street” beliefs (Stewart & Simons, 2006), the measure has references to aggressive responding embedded in the items that ask about disrespect and negative reciprocity (e.g., “When someone disrespects you, it is important that you use physical force or aggression to teach him or her not to disrespect you;” “If someone uses violence against you, it is important that you use violence against him or her to get even”). Our goal was to develop a measure with items focused on two manifestations of disrespect sensitivity, hypervigilance to signs of disrespect from others (e.g., “If you are not careful, people will make you feel like you are no good”) and expectations of disrespectful treatment from others (e.g., “A lot of people can be very disrespectful”). As can be seen from the two illustrative items, this was accomplished without asking about aggressive behavior or negative reciprocity per se. This avoids the problem of confounding the assessment of disrespect sensitivity with the assessment of aggressive behavioral tendencies.

A fourth social-cognitive factor that may distinguish revenge-seekers and pacifists is the goals that children consciously or unconsciously pursue in social situations (for research and reviews, see Asher, MacEvoy, & McDonald, 2008; Crick & Dodge, 1994; Ojanen, Gronroos, & Salmivalli, 2005; Renshaw & Asher, 1982; 1983). In everyday social situations, individuals are likely to have multiple goals that they are trying to pursue and sometimes goals may complement one another. For example, when meeting someone new, the goals of getting to know the other person, trying to make an impression that one is kind, and getting the other person to like you, can be complementary in that they all could foster prosocial, friendly behaviors toward a new acquaintance. At other times, goals may conflict with one another. When meeting someone new, the goal of showing the other person that you are high in status and powerful may interfere with the goal being liked. Part of being a prosocial and well-liked person involves decisions about which goals to pursue in a given situation (Lochman et al. 1993; Renshaw & Asher, 1983; Rose & Asher, 1999). With regard to provocation situations, pacifists may pursue goals that run counter to revenge-seeking and, thereby, inhibit vengeful actions. This could include, for example, the goal of maintaining positive relationships and the goal of regulating emotional reactions when angry.

In addition to social-cognitive factors, affective processes may distinguish revenge-seekers from pacifists. Given the research on linkages between anger and aggression (e.g., Berkowitz, 1989; Eisenberg, Fabes, Nyman, Bernzweig, & Pinuelas, 1994), it is plausible that revenge-seekers would feel more angry about provocations. Research has found linkages between individuals’ emotional experiences in conflicts-of-interest and friendship transgression situations and their goals in those situations. For example, college students’ reports of anger have been positively associated with the endorsement of revenge goals in response to conflict-of-interest situations (McDonald & Asher, 2013), and children’s reports of anger have been linked to revenge goals in response to friendship transgression situations in which friends violate core expectations of friendship (MacEvoy & Asher, 2012). These studies suggest that revenge-seekers would feel more angry than pacifists in provocation situations.

Revenge-seekers and pacifists may differ in other types of negative affective experiences as well. Hurt feelings is a form of negative affect that often blends sadness and anger, with the latter emotion especially occurring when a person believes that the actions directed toward them were unjustified (Leary, Koch, & Hechenbleikner, 2001). In interviews with young adults about times when their feelings have been hurt, a majority reported that they retaliated against the offender by saying something nasty or critical in response (Leary, Springer, Negel, Ansell, & Evans, 1998). Based on this research, we predicted that revenge-seekers would report higher levels of hurt feelings in response to provocation than children who are more pacifistic.

The Current Study

The current study sought to identify revenge-seekers and pacifists in early adolescence. This period of development was chosen because concerns about peer popularity and acceptance increase during late childhood and peak in adolescence (Kloep, 1999; LaFontana & Cillessen, 2010). Additionally, the seventh-graders sampled for the study had just transitioned into a new middle school. The transition to middle school is often characterized by peer competition for social status and temporary increases in aggressive behavior (Eccles et al., 1993; Pellegrini & Bartini, 2000; Wang, Iannotti, & Nansel, 2009). As concerns for status and aggression increase during this time, this period may be one when youth are particularly likely to want to retaliate after peer provocation.

The current study’s primary goals were to test predictions about five social-cognitive and affective processes that we hypothesized, based on both theory and prior research, to distinguish Revenge-Seekers from Pacifists: a) beliefs about the legitimacy of aggression, b) beliefs about reciprocity, c) disrespect sensitivity, d) endorsement of various types of non-vengeful goals, and e) emotional responses to unambiguous peer provocations. To test our predictions, we identified Pacifists and Revenge-Seekers using hypothetical vignettes depicting unambiguous peer provocation (for examples of other studies that have used vignette methodology see Dirks, Suor, Rusch, & Frazier, 2014; MacEvoy & Asher, 2012; Recchia, Rajput, & Peccia, 2015). Using responses to the vignettes, we identified revenge-seeking youth who made negative interpretations and strongly endorsed revenge goals, and a group of pacifistically-oriented youth who also made negative interpretations in response to unambiguous peer provocations, yet did not highly endorse revenge goals.

In addition to the primary research questions, we also examined whether Revenge-Seekers and Pacifists differed in their behavioral strategies in response to the unambiguous peer provocation and whether they differed in their everyday behavioral characteristics as described by peers. Analyses of aggressive behavioral strategies and peer-nominated aggressive behavior serve as validity analyses in that we would expect that the Revenge-Seeker group would more highly endorse aggressive strategies in response to peer provocation and would be nominated by peers as more aggressive compared to the Pacifists. Other behavioral strategies assessed with the vignettes and other behavioral characteristics assessed via peer nominations provide additional useful information. For example, we compared whether Pacifists and Revenge-Seekers were equally likely to withdraw from conflict, seek teacher help, or be viewed by peers as submissive.

We also compared groups on their peer acceptance and popularity (initially referred to as “perceived popularity” by Parkhurst and Hopmeyer, 1998, to distinguish this construct from sociometric popularity/being liked). It is now well-documented that peer acceptance and popularity are two distinct aspects of status in the peer group (Cillessen & Rose, 2005; McDonald & Asher, in press; Parkhurst & Hopmeyer, 1998). Past research has demonstrated a negative association between revenge goals and peer acceptance (Lochman et al., 1993; Rose & Asher, 1999), therefore we hypothesized that Revenge-Seekers would be less well-accepted (liked) by peers compared to Pacifists. However, because popularity has been associated with both aggression and prosocial behavior (Parkhurst & Hopmeyer, 1998), we did not expect that groups would necessarily differ on popularity.

Finally, and also of considerable interest, we also examined whether Pacifists and Revenge-Seekers, who did not differ in the degree to which they made negative interpretations of the behavior of the peer who excluded and insulted them, did differ in other kinds of interpretations of the provocateur’s actions. Specifically, we explored whether Pacifists “held out more hope” for a future relationship with the provocateur and whether they were more likely to blame themselves for the provocation.

Methods

Participants

Participants were seventh-grade students from a suburban school in the Midwest. Students were members of one of three instructional teams in their grade. Teams ranged in size (from 61 to 172 students) and gender composition (from 42.1% to 60.5% male). Consent letters were sent by first-class mail to the parents of the 378 seventh-graders. Following procedures used in previous research (e.g., Ebesutani et al., 2012; Rose & Asher, 1999), consistent with Federal Regulation 45.46.116, and with approval by the (university name blinded) Institutional Review Board, parents were informed about the study and were asked to contact the researchers or the school if they had any questions or did not want their child to participate. One parent contacted the first author to ask questions about the study, and no parents declined permission for their child to participate. Children were also asked at the start of each session if they would like to participate and several chose not to participate in one or more sessions. For a more lengthy discussion of the Federal Regulations criteria for approving a waiver of written parental consent, see Rose and Asher (1999).

Of the 378 seventh-grade students, there were five who declined to participate in one or more sessions, one who was ill for an extended time and was not present during any of the data collection or make-up days, three who moved during the middle of the study and did not complete all data collection sessions, and two who did not complete 75% of the items on one or more of the measures. In total, 11 participants were removed from the sample, yielding a final sample of 367 seventh graders (54.77% male; 22.89% racial/ethnic minority) who participated in all three sessions (a participation rate of 97.09% of the total recruited sample). Comparing the students who completed all parts of the study to those who did not indicated no significant differences on gender (χ² = .0002), race/ethnicity (χ² = .59), or age (t = 1.74).

Of the 367 participants, 80.6% reported that they were 12 years old and 18% reported that they were 13 years old. Regarding race/ethnicity, participants self-identified as the following: 77.11% as White, Non-Hispanic, 12.81% as White, Hispanic, 3.81% as African-American, 1.91% as Asian-American, and 4.36% identified with more than one racial/ethnic group. School-level statistics indicated that students were primarily from working-class or middle-class backgrounds, with 8.31% of the students receiving free or reduced-price lunch.

Procedure

Participants completed measures during three, 40-minute-classroom sessions, spaced approximately a week apart. In the first session, participants responded to three of the six provocation vignettes. They also completed half of the peer nomination items. In the second session, students responded to the remaining vignettes and completed the remaining peer nomination items. In the third session, students responded to the Beliefs about Reciprocity scale, the Beliefs about the Legitimacy of Aggression scale, and the Disrespect Sensitivity measure. The order of measures within each session was randomized for each middle school team. To maximize participation, several make-up sessions were scheduled for students who were absent for one or more classroom sessions. As noted above, only one child was unable to participate due to repeated absenteeism.

Measures

Vignettes

Participants were presented with six hypothetical situation vignettes depicting unambiguous peer provocation that involved being excluded and simultaneously insulted or criticized (see Table 1). Participants were instructed to imagine that each situation really happened to them. Previous research has found that children’s responses to hypothetical situations, presented as carefully constructed vignettes, are meaningfully related to behavioral style, peer acceptance, and the quality of children’s friendships and sibling relationships in everyday life (e.g., Chung & Asher, 1996; Dirks et al., 2014; Recchia et al., 2015; Rose & Asher, 1999). The provocation situations used in this study were modeled after social psychology paradigms that have combined rejection with either insults or criticisms (e.g., Twenge & Campbell, 2003). All provocation situations were designed with ecological validity in mind, depicting realistic situations that youth might actually encounter in their daily school lives. Vignettes were also individually pilot tested with approximately 12 children, between the ages of 9 and 14, to ensure their ecological validity.

Table 1.

Provocation Vignettes

Picking Teams	There are two students that you know who are picking teams to play a game. This is a game you think you are really good at and you think that you can do really well for your team. There are an uneven number of people who want to play so it looks like someone will be left out. As people are being chosen, you realize that a lot of other players are being picked before you. Finally, there are only two players left: you and another kid. The other kid gets picked and you are left out and cannot play. One of the kids says “I am glad you aren’t on my team.”
Practice Time	You are trying out for a sports team and the coach chooses groups of players to practice together as a team before try-outs. One of the players is designated captain of your team by the coach. The captain assigns positions, but because there are extra people, one person will have to sit out while the others play. The captain keeps assigning you to sit out and you can’t figure out why. You ask to play but the team captain ignores you. Eventually, the team captain says “You can’t play because you are a bad player.”
The Party	A kid in your class is having a party for their birthday. The kid has invited a lot of people from your class. When you ask if you are invited to the party, the kid says “No. Only cool kids are invited to my party.”
The Play	You are auditioning for a play and you are really excited and really nervous. At the auditions, the director tells everyone that they need to pair up with a partner to read lines. You ask one of your classmates to pair up with you. The classmate says no and adds “You aren’t very good at this.”
Board Games	One day in class the teacher says you can have free time to play games. The kids in your class go over to grab the board games. You see some kids setting up the game you want to play. You go over to them and ask if you can play with them. One of the kids says no and laughs at you.
Group Work	You have a science project to do with two other people from your class. The teacher randomly assigned groups but you have worked with both of the other students before and you thought you got along well with them. While discussing the assignment, you attempt to give some input and share your ideas. However, one of your classmates does not listen to your ideas at all. When you share your ideas, the kid just talks about a different idea. After trying to tell them about your last idea, which you thought was really good, the kid says, “That’s a bad idea.”

During group testing sessions, researchers read the vignettes aloud to participants while they followed along. As in previous research using vignettes with children (e.g., Erdley & Asher, 1998; Rose & Asher, 1999), a practice item was included so that participants could practice using the response scales. In response to every vignette, participants rated the four emotions, seven interpretations, eight strategies, and ten goals on a scale ranging from 1 (strongly disagree) to 10 (strongly agree). See Table 2 for each emotion, interpretation, goal, and strategy that was rated for each vignette and internal reliabilities of the response scales across vignettes. Participants responded to each vignette by rating the likelihood that they would experience each emotion, make each interpretation, enact each behavioral strategy, and pursue each goal (see Chung & Asher, 1996, and Rose & Asher, 1999, for evidence that order of assessment does not affect how goals and strategies are related). To assess emotions, participants were asked, “How would you feel in this situation?” To assess interpretations, participants were asked, “What would you be thinking in this situation?” To assess strategies, participants were asked, “What would you do in this situation?” To assess goals, participants were asked, “What would your goals be in this situation?” Internal reliabilities were calculated for each of the emotions, interpretations, strategies, and goals across the 6 vignettes. Internal reliabilities ranged from .70 to .92, with the median coefficient alpha being .89. For each emotion, interpretation, strategy, and goal, participants received a single score based on the average rating they gave across the six vignettes.

Table 2.

Internal Reliabilities for the Emotions, Interpretations, Goals, and Strategies

		Alpha
Emotions	Item Wording
Anger	I would be angry.	.85
Hurt feelings	My feelings would be hurt.	.90
Sad	I would be sad.	.91
Fine	I would feel fine.	.83
Interpretations
Rejection	This person doesn’t like me.	.80
Disrespect	This person doesn’t respect me.	.83
Judgment of wrongdoing	This person’s behavior is wrong.	.86
Disassociation	This person doesn’t want to be with me.	.82
Relationship not viable	This person and I could never be friends.	.76
Resolvable	This person and I can easily make-up.	.84
Unintended emotional pain	This person did not mean to hurt me.	.70
Self-blame	What happened was my fault.	.79
Not worthy of being upset about	What this person did is not worth getting upset about.	.80
Goals
Revenge	I would be trying to get back at this person.	.91
	I would be trying to hurt this person like they hurt me.
Emotion regulation	I would be trying to not get upset.	.83
Getting Along	I would be trying to get along with this person.	.89
Interpersonal Control	I would be trying to keep myself from being pushed around.	.89
Harm avoidance	I would be trying to avoid getting hurt more.	.90
Forgiveness	I would be trying to forgive this person.	.92
Gaining partner acceptance	I would be trying to get this person to like me.	.91
Gaining partner respect	I would be trying to get this person to respect me.	.92
Gaining group acceptance	I would be trying to get other kids to like me.	.95
Gaining group respect	I would be trying to get other kids to respect me.	.95
Strategies
Leaving/walk away	I would leave or walk away.	.78
Ignoring the provocation	I would ignore what the person did.	.78
Verbal aggression	I would say something mean or insulting to this person.	.93
Passive aggression	I would act cold and unfriendly towards this person.	.91
Physical aggression	I would hit or push this person.	.92
Relational aggression	I would spread rumors or bad gossip about this person behind their back.	.92
Prosocial resolution	I would talk nicely to this person so that we could make up.	.91
Help-seeking	I would tell the teacher what happened and ask for help.	.91

Legitimacy of aggression beliefs

The Beliefs about the Legitimacy of Aggression scale is a 16-item measure that was adapted by Erdley and Asher (1998) from Slaby and Guerra (1988). Participants indicated how appropriate they thought it was to engage in particular forms of aggression under various circumstances (e.g., to get what you want; to protect yourself; if someone makes you angry). Participants responded to each item on a scale from 1 (strongly disagree) to 5 (strongly agree). Consistent with Erdley and Asher (1998), internal reliability was high for the measure (α = .93), and exploratory factor analysis using principal axis factoring and a promax rotation found that all items loaded on one factor, accounting for 54.05% of the variance in observed ratings.

Beliefs about reciprocity

The Beliefs about Reciprocity measure was based on Eisenberger et al.’s (2004) measure that included 24 items about positive and negative reciprocity. In the Eisenberger et al. measure, some items addressed beliefs about behavior (e.g., “If someone says something nasty to you, you should say something nasty back”) and some items were self-reports of behavioral reciprocity (e.g., “I always repay someone who has done me a favor”). For the present version, certain items were dropped or modified and all behavior items that were retained were transformed into belief items (e.g., “If someone does you a favor, you should make sure to do a favor for them in the future”) in order to assure that the beliefs measure did not have overlapping content with our behavioral measures. The final Beliefs about Reciprocity measure consisted of 18 items, equally divided between negative and positive reciprocity beliefs (see Appendix A in the Online Supplementary Material). Youth responded on to each item a scale ranging from 1 (strongly disagree) to 7 (strongly agree). Exploratory factor analysis using principal axis factoring and a promax rotation indicated that items fell into two scales, Positive Reciprocity Beliefs (α = .88) and Negative Reciprocity Beliefs (α = .96), and accounted for 66.03% of the variance. The two subscales were only moderately correlated (r = −.31, p < .05) pointing to their relative independence.

Disrespect sensitivity

The measure of Disrespect Sensitivity was newly developed for this study. Before being used in the present study, items on this measure were individually pilot tested with a snowball sample of approximately 12 children, between the ages of 9 and 14. Pilot testing was conducted to assure that items were relevant to the lives of older children and adolescents and that items were understandable. Based on the pilot testing, 18 items were chosen for the current study’s data collection. As indicated in Appendix B of the Online Supplementary Material, after factor analysis of the 18 items with our study sample of 367 students, eight items were chosen for the final measure. These eight items focused on either vigilance for signs of disrespect from others (e.g., “Signs that others disrespect you are everywhere”) or expectations that other people will respect/disrespect you (e.g., “People want to make you feel like they are better than you.”). Items in the final measure can be found in Table 3. Participants responded to items by rating how much they agreed with each statement on a scale ranging from “1” (strongly disagree) to “10” (strongly agree). Internal reliability for the eight-item scale was .82.

Table 3.

Items on the Disrespect Sensitivity Measure

Items
People want to make you feel like they are better than you.
If you are not careful, people will make you feel like you are no good.
People are always trying to make themselves look like they are better than you.
Signs that others disrespect you are everywhere.
Lots of people will treat you badly if you are not careful.
A lot of people can be very disrespectful.
People tend to respect you as much as you deserve to be respected.*
People tend to treat others in a kind and respectful way.*

Note. Items with a ^* are reverse-coded.

Peer assessments of classmates’ social behavior

Participants completed an unlimited peer nomination measure in which they were asked to circle the names of all of the classmates on a roster who fit particular behavioral descriptors (e.g., “starts fights”). The descriptors were based on the items used in the behavioral nomination measures of researchers such as Cillessen and Mayeux (2004), Coie, Dodge, and Coppotelli (1982), and Parkhurst and Asher (1992). The descriptor for each item was presented on a separate page, with the descriptor on the top of the page and the list of names of instructional team members below. Participants were asked to nominate classmates who fit descriptions for physical/verbal aggression (3 items; e.g., “starts fights”; α = .88), social aggression (3 items; e.g., “gossips or spreads rumors about other kids”; α = .89), prosocial behavior (3 items; e.g., “cooperates”; α = .80), and submissive behavior (3 items; e.g., “easy to push around”; α = .71). A child’s score for each item was computed as the proportion of nominations that the child received divided by the total number of classmates who knew the child and responded to that item for that child. Proportion scores were normalized with an arcsine transformation and then were standardized within team. The three standardized item scores for each category of behavior were then averaged to create the final scores used in analyses.

Since seventh-grade teams varied greatly in the number of students on each team (ranging from 61 to 171), each student’s nomination rosters had a number of students equal to the smallest team (i.e., 61 names). For larger teams, a computer program randomly selected 61 names for each particular child’s roster such that: (a) every participant had 61 names from which to nominate peers for each behavioral nomination item; (b) each participant was listed on the rosters of 61 peers for each behavioral nomination item; and (c) the lists of names were different for each behavioral nomination item (for a similar procedure see Parker, Low, Walker, & Gamm, 2005). For all items, including nominations for popularity and liking ratings (both of these measures are described below), participants were given the option of crossing out names of peers who they did not know. On average, participants knew 86% of the students on their lists.

Popularity

To measure popularity, participants were again given a roster of 61 randomly-selected peers’ names and were asked to circle the names of classmates who were “popular.” Again, youth could cross out names of peers who they did not know. A participant’s score was computed as the proportion of “popular” nominations that the child received divided by the total number of classmates who knew the child and responded to that item for that child. These scores were then normalized with the arcsine transformation and were standardized within team.

Peer acceptance

To measure peer acceptance, participants were asked to respond to a widely used rating-scale sociometric measure in which they rated peers on a roster of 61 randomly-selected names from 1 (I don’t like to) to 5 (I like to a lot) in terms of how much they liked to play with each child (see Hymel, Closson, Caravita, & Vaillancourt, 2014). Here too, participants were given the option of crossing out names of peers who they did not know. Participants were given a peer acceptance score based on the average rating they received, standardized within team.

Results

Data Reduction

First, a principal components factor analysis with a promax rotation was conducted to learn whether children’s ratings of various interpretations in response to the provocation vignettes could be meaningfully grouped into a smaller set of variables. Three factors emerged with eigenvalues over 1.00. The first factor had an eigenvalue of 4.52 and explained 50.27% of the variance, the second had an eigenvalue of 1.44 and explained 16.05% of the variance, and the third had an eigenvalue of 1.01 and explained 11.23% of the variance. The first factor consisted of negative interpretations (rejection, disrespect, judgment of wrongdoing, and disassociation; α = .94). The second factor comprised relationship-sustaining interpretations (unintended emotional pain, resolvable, not worth getting upset about, and relationship not viable, which was reverse-scored; α = .78). The last factor was comprised of only one interpretation, self-blame.

Second, a principal components analysis with a promax rotation was conducted on the goals that participants rated for the provocation vignettes. Participants’ ratings of revenge goals were not included in the factor analysis of goals because revenge goals were used in in the cluster analysis to learn whether Revenge-Seekers and Pacifists would be identified as distinct groups of youth (see below). Three factors with eigenvalues over 1.00 were revealed. The first factor yielded an eigenvalue of 4.87 and accounted for 54.16% of the variance, the second had an eigenvalue of 1.50 and accounted for 16.71% of the variance, and the third had an eigenvalue of 1.17 and accounted for 13.02% of the variance. The first factor comprised goals that were relationship maintaining (getting along, forgiveness, gaining partner acceptance; α = .93). The second factor comprised status-maintaining goals (gaining group respect, gaining group acceptance, and gaining partner respect; α = .89). The last factor consisted of interpersonal control, harm avoidance, emotion regulation goals (α = .82), however these goals were judged to represent meaningfully different goals in provocation situations and were therefore analyzed separately in subsequent analyses.

Finally, a principal components analysis with a promax rotation was used to analyze participants’ strategy ratings on the vignettes. Three factors emerged with eigenvalues over 1.00. The first factor had an eigenvalue of 4.19 and explained 52.36% of the variance, the second factor had an eigenvalue of 1.26 and explained 15.73% of the variance, and the third factor had an eigenvalue of 1.01 and explained 12.56% of the variance. The first factor comprised aggressive strategies (verbal aggression, passive aggression, physical aggression, and relational aggression; α = .93). The second factor consisted of disengaging strategies (leaving/walk away, ignoring the provocation; α = .79). The third factor combined help-seeking and prosocial resolution. However, these two strategies were judged to be meaningfully different responses to provocation and were left as separate strategies (see Renshaw & Asher, 1983, for a related discussion of help-seeking versus independent resourcefulness).

Data Analysis Plan

First, we examined gender differences with the entire sample because of previously observed gender differences on a number of variables assessed in this study (e.g., McDonald & Asher, 2013; Rose & Asher, 1999; Rose & Rudolph, 2006). Next, a cluster analysis was conducted to learn whether distinct clusters of Revenge-Seekers and Pacifists could be identified from children’s patterns of negative interpretations and revenge goal endorsement. Following identification of Revenge-Seekers and Pacifists, analyses were conducted to compare the two groups and to test for group × gender interactions. Finally, a logistic regression was conducted to compare how well various factors predicted to Revenge-Seeker and Pacifist group membership. Preliminary analyses also included the team of students as a possible covariate in analyses, however, team was not found to have a significant effect or moderate the other effects and was not included in the final analyses.

Gender Differences

A series of one-way MANOVAs were conducted with the entire sample to test for possible gender differences in behavioral nominations, interpretations, emotions, goals, strategies, and beliefs. When there was a significant MANOVA effect, follow up univariate ANOVAs were conducted. In addition, one-way ANOVAs were conducted testing for gender differences in acceptance, popularity, and disrespect sensitivity. The entire sample was used to test for possible gender differences since testing for gender differences with only Pacifists and Revenge-Seekers would provide a less complete and valid test than would be provided by using the entire sample. Table 4 shows descriptive statistics and results from univariate ANOVA tests.

Table 4.

Descriptive Statistics and Gender Comparisons

	Total Sample N = 367	Boys n = 201	Girls n = 166	Gender F (1, 365)	Partial η2
	M (SD)	M (SD)	M (SD)
Peer-Nominated Behaviors
Physical/Verbal Aggression	.00	.32 (.99)	−.38 (.65)	61.80***	.15
Social Aggression	.00	.02 (.93)	−.02 (.88)	.22	.001
Prosocial Behavior	.00	−.20 (.88)	.24 (.75)	25.61***	.07
Submissive Behavior	.00	.06 (.83)	−.07 (.75)	2.49	.01
Peer Acceptance	.00	−.14 (1.09)	.18 (.84)	9.32**	.03
Popularity	.01	.03 (1.03)	−.03 (.95)	.33	.001
Emotions
Anger	6.95 (2.09)	6.88 (2.04)	7.04 (2.16)	.53	.001
Hurt Feelings	6.22 (2.53)	5.70 (2.60)	6.83 (2.30)	19.06**	.05
Sadness	5.36 (2.72)	4.81 (2.73)	6.01 (2.56)	18.69***	.05
Fine	4.15 (1.97)	4.63 (1.93)	3.90 (1.99)	4.98*	.01
Goals
Revenge	3.34 (2.22)	3.69 (2.29)	2.92 (2.08)	11.07***	.03
Relationship Maintaining	5.17 (2.35)	4.89 (2.40)	5.50 (2.24)	6.05*	.02
Status Maintaining	7.11 (2.31)	7.14 (2.29)	7.08 (2.34)	.07	<.001
Interpersonal Control	7.86 (1.98)	7.91 (1.98)	7.80 (2.00)	.25	.001
Harm Avoidance	7.36 (2.26)	7.28 (2.34)	7.46 (2.15)	.60	.002
Emotion Regulation	6.78 (2.07)	6.76 (2.17)	6.80 (1.95)	.86	<.001
Strategies
Aggressive	3.12 (2.10)	3.50 (2.22)	2.65 (1.84)	15.50***	.04
Disengaging	6.06 (2.04)	5.77 (2.21)	6.41 (1.76)	8.95*	.02
Prosocial	4.47 (2.35)	4.23 (2.40)	4.77 (2.26)	4.84*	.01
Help-Seeking	4.28 (2.75)	4.26 (2.80)	4.30 (2.69)	.02	<.001
Beliefs
Legitimacy of Aggression	1.97 (.81)	2.14 (.82)	1.77 (.73)	20.62***	.05
Negative Reciprocity	2.99 (1.67)	3.33 (1.74)	2.57 (1.50)	19.56***	.05
Positive Reciprocity	6.49 (.69)	6.44 (.68)	6.54 (.71)	1.65	.005
Disrespect Sensitivity	5.97 (1.67)	6.16 (1.64)	5.73 (1.68)	6.12*	.02

Note.

^*p < .05;

^**p < .01;

^***p < .001.

A MANOVA indicated that there was a significant multivariate effect for gender on behavioral nominations, F (4, 362) = 43.43, p < .001, partial η²= .32. The univariate follow-up ANOVAs indicated that boys were nominated as being more physically/verbally aggressive than girls and that girls were nominated as being more prosocial than boys. There were not significant gender differences for social aggression or submissive behavior.

Focusing on how well-liked and popular children were in the peer group, two one-way ANOVAs were conducted to examine gender differences in acceptance and popularity. Results indicated that girls were more highly liked by peers than were boys, but there was no significant gender difference in popularity.

With regard to youth’s responses to the provocation vignettes, the multivariate effect of gender for interpretations was not significant. There was, however, a significant multivariate effect for gender on emotions in response to peer provocation, F (4, 362) = 6.56, p < .001, partial η²= .07. Follow-up ANOVAs revealed no significant gender difference for reported anger, but girls, compared to the boys, reported feeling significantly more hurt, more sad, and less “fine.” There was also a significant multivariate effect for social goals, F (6, 360) = 3.24, p = .004, partial η²= .05. Follow-up ANOVAs revealed that boys endorsed revenge goals more than girls, whereas girls endorsed relationship-maintaining goals more than boys. Finally, there was a significant multivariate gender effect for behavioral strategies, F (4, 362) = 4.45, p = .002, partial η²= .05. Follow-up ANOVAs revealed that boys, compared to girls, endorsed aggressive strategies more, whereas girls endorsed disengaging strategies and prosocial strategies more than boys.

There was also a significant multivariate effect for gender on beliefs, F (3, 363) = 7.28, p < .001, partial η²= .06. Follow-up ANOVAs indicated that boys more strongly endorsed legitimacy of aggression beliefs and negative reciprocity beliefs compared to girls. There was not a significant gender effect for positive reciprocity beliefs. Finally, the one-way analysis of variance conducted to examine gender differences in disrespect sensitivity indicated that boys were more disrespect-sensitive than girls.

Group Identification: Revenge-Seekers and Pacifists

K-means cluster analysis, a methodology for quantifying the structural characteristics of a set of observations (Magnusson, 1998), was used to learn whether Revenge-Seekers and Pacifists could be identified as distinct groups and whether one or more other distinct groups might be identified as well. The K-means cluster analysis was performed using standardized scores on negative interpretations (as identified in the factor analysis described above) and standardized scores on revenge goals in response to the provocation situations.

Latent Profile Analysis (LPA) can also be used to identify groups, however K-means analysis was chosen for several reasons. First, Hair and Black (2000) suggest that K-means cluster analysis is an appropriate technique when there is a theoretical rationale for a specific number of clusters. Second, LPA tends to overestimate the number of classes, especially when variables are non-normally distributed (Bauer & Curran, 2003; Tofighi & Enders, 2008), and revenge goal endorsement was positively skewed in our data, with most students giving relatively low ratings to this goal. Third, Wurpts and Geiser (2014) found that LPA models with less than five indicators were biased and had convergence difficulties — our hypothesized groups used two indicators to group children: negative interpretations and revenge goals. For these reasons, K-means cluster analysis was chosen as the most appropriate grouping strategy.

We expected that two groups of children would be identified from K-means cluster analysis: a) a group of children who were high on negative interpretations but low on revenge goals, and b) a group of children who were high on both negative interpretations and revenge goals. We were primarily interested in these two groups for theoretical reasons, however, for completeness we also tested for the possibility of a third or fourth group. Accordingly, K-means analyses were conducted to generate both three and four group solutions.

Two validation techniques were used to help select a solution derived from K-means cluster analysis. First, a Monte Carlo procedure was also used to validate the cluster solutions derived from the K-means method. A subset of 50% of the original sample was randomly selected and the K-means cluster technique was repeated. Cluster membership for the subset was compared with initial memberships in the full sample. This procedure of sampling a random 50% of the sample was repeated twice for both three- and four-group solutions. Average agreement between cluster memberships for the three-group solution was consistent for 94.99% of the sample, whereas the four-group solution was consistent for 91.10% of the sample.

Second, hierarchical cluster analysis, using Ward’s method with Squared Euclidean Distance, was used to examine convergence with the K-means method for the three and four group solutions (Hair & Black, 2000). For the three-group solution, the average agreement between cluster memberships for K-means and hierarchical analyses was consistent for 78.59% of the sample, whereas for the four-group solution, the average agreement between cluster memberships for K-means and hierarchical analyses was consistent for 49.32% of the sample. Together with the results from the Monte Carlo analysis, it was decided that a three-group solution was a better fit for the data.

The three-cluster solution derived from the K-means cluster analysis identified a group of revenge-oriented adolescents (Revenge-Seekers; N = 93; 25.34% of the sample) who, on the ten-point scale, rated negative interpretations highly (M = 8.15, SD = 1.36) and revenge goals relatively highly (M = 6.46, SD = 1.54), a second group of pacifistically-oriented youth (Pacifists; N = 150; 40.87% of the sample) who rated negative interpretations highly (M = 8.27, SD = .90) but revenge goals relatively low (M = 2.12, SD = 1.03), and a third group (N = 124; 33.78% of the sample) who rated negative interpretations near the midpoint of the scale (M = 5.23, SD = 1.18) and revenge goals (M = 2.36, SD = 1.33) at low levels. Since the present study was designed to compare the characteristics of the Revenge-Seekers and the Pacifists, the third group was not included in the remaining analyses. Interested readers can write to the first author for more information on this third group.

The next analyses had three purposes. First, we examined if the two groups differed in gender composition. Second, we compared groups on revenge goals and negative interpretations. Ideally, groups would not differ in their interpretations even though they would differ drastically on their revenge goals. Third, it was important to see if gender moderated group differences on revenge goals and interpretations. Consistent with the finding reported above that boys were more revenge-motivated than girls, there were more boys than expected by chance in the Revenge-Seeker group (67.7% male), whereas there were fewer boys in the Pacifist group (52.0% male) than would be expected by chance, χ² (1) = 5.89, p = .02. The 2 (gender) × 2 (group) ANOVA examining revenge goals indicated that there was the expected significant main effect for group, with Revenge-Seekers endorsing revenge goals more than the Pacifist group, F (1, 239) = 607.81, p <.001. Also as expected, the 2 (gender) × 2 (group) ANOVA examining negative interpretations indicated no significant group effect, F (1, 239) = .03, p = .86. In addition, gender did not moderate the group effect on revenge goals, F (1, 239) = .44, p = .51, but there was a significant interaction of group with gender for negative interpretations, F (1, 239) = 4.61, p = .03. However, probing this interaction revealed that there was not a significant group difference on negative interpretations for either girls, F (1, 100) = 1.88, p = .17, or boys, F (1, 139) = 3.17, p = .08. These findings regarding revenge goals and interpretations are important in that they document that any of the differences between Revenge-Seekers and Pacifists, explored below, are not the result of one group being significantly more negative in their interpretations of the provocateur’s actions in the vignettes.

Comparing Revenge-Seekers and Pacifists

Next, analyses were conducted to test hypotheses regarding Pacifists and Revenge-Seekers. Here, too, we tested for gender as a moderator of group differences. Following these analyses, we performed a logistic regression analysis to compare the relative and unique contributions of beliefs, disrespect sensitivity, non-vengeful goals, and emotions to the prediction of group membership.

Behavioral strategies on the vignettes

The 2 (group) × 2 (gender) MANOVA examining behavioral strategies revealed a significant multivariate effect for group, F (4, 236) = 89.05, p < .001, that was not moderated by gender, F (4, 236) = .92. As seen in Table 5, univariate ANOVAs revealed that Revenge-Seekers rated aggressive strategies significantly higher than did Pacifists, whereas the Pacifist group rated prosocial resolution strategies significantly higher than did Revenge-Seekers when responding to the unambiguous provocation situations. The Pacifist group also rated disengaging strategies significantly higher than Revenge-Seekers. Revenge-Seekers and Pacifists did not significantly differ on their ratings of seeking help from the teacher.

Table 5.

Descriptive Statistics and Group Comparisons for Behavioral Strategies, Alternative Interpretations, Peer-Nominated Behaviors, Peer Acceptance, and Popularity

	“Revenge-Seekers” n = 93	“Pacifists” n = 150	Group F	Partial η2
	M (SD)	M (SD)
Strategies
Aggressive	5.81 (1.79)	2.15 (1.08)	357.01***	.599
Disengaging	5.00 (1.84)	6.87 (1.96)	45.24***	.159
Prosocial	2.90 (1.42)	4.83 (2.33)	51.11***	.176
Help-Seeking	4.17 (2.74)	4.98 (2.87)	3.71	.015
Peer-Nominated Behaviors
Physical/Verbal Aggression	.24 (1.03)	−.23 (.76)	9.49**	.038
Social Aggression	.12 (.93)	−.15 (.88)	4.63*	.019
Prosocial Behavior	−.21 (.88)	.04 (.84)	3.10	.013
Submissive Behavior	.13 (.78)	.07 (.86)	.002	.000
Peer Acceptance	−.25 (1.07)	.06 (.95)	4.51*	.019
Popularity	−.04 (1.01)	−.13 (.89)	.34	.001
Alternative Interpretations
Relationship-Sustaining	3.75 (1.08)	4.75 (1.32)	36.40***	.132
Self-Blame	1.79 (.95)	1.83 (1.23)	.29	.001

Note.

^*p < .05;

^**p < .01;

^***p < .001.

Behavior with peers

Turning from strategy ratings on the vignettes to everyday behaviors, the 2 (group) × 2 (gender) MANOVA examining peer behavioral nominations revealed a significant multivariate effect for group, F (4, 236) = 3.07, p =.02, that was not moderated by gender, F (4, 236) =1.08. As seen in Table 4, univariate analyses revealed that Revenge-Seekers were viewed by peers as engaging in more physical/verbal aggression and more social aggression than Pacifists. There were no significant group differences for submissive behavior. The group effect for prosocial behavior approached, but did not reach, significance.

Peer acceptance and popularity

Two separate 2 (group) × 2 (gender) ANOVAs examining peer acceptance and popularity were conducted. Revenge-Seekers and Pacifists were similar in popularity, however Revenge-Seekers were liked less by peers than were Pacifists (Table 5). These effects were not significantly moderated by gender [acceptance F (1, 239) = 1.45; popularity F (1, 239) = .30]¹.

Relationship-sustaining and self-blame interpretations

A 2 (group) × 2 (gender) MANOVA was conducted to learn whether Revenge-Seekers and Pacifists differed on two types of interpretations that were not used in the cluster analysis to identify groups, namely relationship-sustaining and self-blame interpretations. This analysis revealed a significant multivariate effect for group, F (2, 238) = 18.77, p < .001, that was not moderated by gender, F (2, 238) = .34. Follow-up univariate analyses revealed that Pacifists endorsed relationship-sustaining interpretations in the vignettes more than Revenge-Seekers, but the two groups did not differ from one another on self-blame interpretations (Table 5).

Beliefs

A 2 (group) × 2 (gender) MANOVA examining beliefs revealed a significant multivariate effect for group, F (3, 237) = 60.13, p < .001, that was not moderated by gender, F (3, 237) = 1.40. Univariate tests revealed that Revenge-Seekers endorsed both legitimacy of aggression beliefs and negative reciprocity beliefs more than Pacifists, whereas Pacifists endorsed positive reciprocity beliefs more than Revenge-Seekers (see Table 6).

Table 6.

Descriptive Statistics and Group Comparisons for Emotions, Non-Vengeful Goals, Beliefs and Disrespect Sensitivity

	“Revenge-Seekers” n = 93	“Pacifists” n = 150	Group F	Partial η2
	M (SD)	M (SD)
Beliefs
Legitimacy of Aggression	2.74 (.76)	1.64 (.55)	167.59***	.412
Negative Reciprocity	4.54 (1.43)	2.32 (1.30)	143.08***	.374
Positive Reciprocity	6.33 (.84)	6.66 (.51)	16.38***	.064
Disrespect Sensitivity	6.64 (1.51)	5.95 (1.69)	7.70**	.031
Non-Vengeful Goals
Relationship Maintaining	3.74 (1.83)	5.74 (2.33)	45.74***	.161
Status Maintaining	7.10 (2.12)	7.73 (2.15)	3.47	.014
Interpersonal Control	7.96 (1.85)	8.45 (1.66)	3.47	.014
Harm Avoidance	7.44 (2.06)	8.16 (1.95)	4.71*	.019
Emotion Regulation	6.05 (1.96)	7.57 (1.83)	37.66***	.136
Emotions
Anger	8.04 (1.60)	7.66 (1.68)	4.66*	.019
Hurt Feelings	6.61 (2.43)	7.28 (2.35)	3.37	.014
Sadness	5.67 (2.64)	6.39 (2.70)	3.05	.013
Fine	3.73 (1.80)	3.54 (1.74)	.46	.002

Note.

^*p < .05;

^**p < .01;

^***p < .001.

Disrespect sensitivity

A 2 (group) × 2 (gender) ANOVA was conducted to address the question of whether Revenge-Seekers are more disrespect-sensitive than Pacifists. As seen in Table 6, there was a significant group effect for disrespect sensitivity, F (1, 239) = 4.64, p =.03, that was not moderated by gender, F (1, 239) = .001. Revenge-Seekers were more disrespect sensitive than Pacifists.

Non-vengeful goals

A 2 (group) × 2 (gender) MANOVA was conducted to examine whether Revenge-Seekers and Pacifists differed on their endorsement of non-vengeful goals. The analysis revealed a multivariate effect for group, F (5, 235) = 14.63, p < .001, that was not moderated by gender, F (5, 235) = 1.14. Univariate analyses revealed that Pacifists rated relationship-maintaining goals, emotion regulation goals, and harm avoidance goals more highly than Revenge-Seekers (see Table 6).

Emotions

A 2 (group) × 2 (gender) MANOVA was conducted to test the hypothesis that there would be differences between Revenge-Seekers and Pacifists in their emotional response to unambiguous peer provocation. The analysis revealed a significant multivariate effect for group, F (4, 236) = 3.99, p = .01, that was not moderated by gender, F (4, 236) = .99. As seen in Table 6, follow-up univariate analyses revealed a significant group difference for anger, with Revenge-Seekers reporting more anger than Pacifists. Revenge-Seekers and Pacifists did not differ from each other on reported “hurt feelings,” “sadness,” or feeling “fine.”

Predicting group membership

A logistic regression analysis was conducted to compare the relative and unique contributions of feelings of anger, several social-cognitive processes, and gender to predicting to whether children were classified as a Pacifist or a Revenge-Seeker. Each of the social-cognitive and affective processes used in this analysis was of theoretical interest and all of the variables, including gender, had significantly discriminated between Pacifists and Revenge-Seekers in the previous analyses. The social-cognitive variables in the logistic regression analysis were legitimacy of aggression beliefs, negative and positive reciprocity beliefs, disrespect sensitivity, and several types of non-vengeful goals, namely relationship maintaining goals, harm avoidance goals, and emotion regulation goals. All variables were standardized (except gender) and entered simultaneously into the logistic regression analysis.

As seen in Table 7, relationship maintaining and emotion regulation goals, legitimacy of aggression beliefs, and disrespect sensitivity contributed unique variance to the prediction of group membership. More specifically, a one SD increase in relationship maintaining goals increased the odds of being a Pacifist relative to a Revenge-Seeker by 1.68 times and a one SD increase in emotion regulation goals increased the odds of being a Pacifist by 1.85 times relative to a Revenge-Seeker. In contrast, higher scores on the legitimacy of aggression and the disrespect sensitivity measures increased the odds of being a Revenge-Seeker relative to a Pacifist. More specifically, for a one SD increase in legitimacy of aggression beliefs children were 4.35 times more likely to be a Revenge-Seeker than a Pacifist. For a one SD increase in disrespect sensitivity, children were 1.61 times more likely to be a Revenge-Seeker than a Pacifist. The effect of legitimacy of aggression beliefs was particularly strong compared to all other predictors.

Table 7.

Logistic Regression Predicting Membership in the Pacifist Group Relative to the Revenge-Seeker Group

	B (S.E.)	Odds Ratio	95% CI
Gender (Males = 1, Females = −1)	−.46 (.44)	.63	.27 – 1.50
Anger	−.53 (.30)	.59	.33 – 1.06
Relationship Maintaining Goals	.52 (.25)	1.68*	1.04 – 2.73
Harm Avoidance Goals	−.14 (.27)	.87	.52 – 1.46
Emotion Regulation Goals	.62 (.24)	1.85**	1.16 – 2.95
Legitimacy of Aggression Beliefs	−1.46 (.38)	.23***	.11 – .48
Negative Reciprocity Beliefs	−.31 (.36)	.73	.36 – 1.48
Positive Reciprocity Beliefs	.39 (.24)	1.47	.93 – 2.33
Disrespect Sensitivity	−.48 (.23)	.62*	.40 – .97

Note. All variables except gender were standardized for this analysis.

^*p < .05;

^**p < .01;

^***p < .001.

Discussion

Although youth are more likely to endorse retaliation in the face of unambiguous provocation compared to ambiguous provocation (e.g. Astor, 1994; Smetana et al., 2003), the current study documents that even in unambiguously provocative situations many children “decouple” their negative interpretations of provocation from the goal of seeking revenge. Cluster analyses revealed that a significant portion of youth (41%), whom we classified as Pacifists, made very negative interpretations of the behavior of the antagonist in the vignettes (e.g., thought that the provocateur did not like or respect them), yet endorsed revenge goals at low levels. By contrast, 25% of youth, whom we characterized as Revenge-Seekers, not only made negative interpretations of the antagonist’s behavior but also endorsed revenge goals at relatively high levels.

Pacifists and Revenge-Seekers were found to think about social situations and behave in distinct ways. Although Pacifists and Revenge-Seekers made negative interpretations of peer provocations at similar levels, Pacifists more highly endorsed the relationship-sustaining interpretations that they could resolve the issue, be friends with the provocateur, and that the situation was not worth getting upset about.

Pacifists were also less aggressive than Revenge-Seekers based on their lower endorsement of aggressive strategies on the vignettes measure and the fewer nominations they received from peers for physical, verbal, and social aggression. These findings serve to validate the group assignment based on the cluster analysis. Furthermore, Pacifists, despite their similar level of endorsement of negative interpretations, were more likely to endorse prosocial strategies and disengagement from the situation. In light of these strategy choices, it is also particularly interesting that Pacifists and Revenge-Seekers were nominated at similar rates by peers for submissive behavior and they endorsed help-seeking strategies and self-blame interpretations on the vignettes measure at similar levels. In other words, although Pacifists were less aggressive than Revenge-Seekers, they did not blame themselves more nor were they more passive. Pacifists were also liked by peers more than Revenge-Seekers and did not differ from Revenge-Seekers on perceived popularity. In short, the findings give little reason to believe that pacifistically-oriented children were “wimps” or low-status children.

The primary goals of the current study were to examine how Pacifists and Revenge-Seekers differed in their beliefs, disrespect sensitivity, non-vengeful goals, and emotions. Based on prior research linking anger and hurt feelings with retaliatory responses (Leary et al., 1998), we hypothesized that Pacifists would feel less angry and less hurt than Revenge-Seekers. Results revealed that Revenge-Seekers reported greater anger than Pacifists after provocation but they did not report greater hurt feelings or sadness. It is interesting, in light of MacEvoy and Asher’s (2012) study, that Pacifists were not significantly more sad than Revenge-Seekers. MacEvoy and Asher, in a study of children’s responses to friendship transgression situations, found that sadness was positively associated with the pursuit of relationship maintenance goals and negatively associated with the pursuit of revenge goals. Perhaps, sadness functions as a relationship-maintaining emotion primarily in close relationship contexts, such as in a friendship context, but does not function that way in other peer relationship contexts. It is also possible that sadness is differentially predictive of revenge depending on the situation; sadness may be protective in response to friendship transgressions, but is not when responding to unambiguous provocation.

Revenge-Seekers and Pacifists also differed on their endorsement of several goals that may compete with, or limit, revenge goals. Pacifists endorsed harm avoidance goals more than Revenge-Seekers, suggesting that they may be more aware of the risks involved in revenge-seeking. Additionally, Pacifists endorsed the goal of regulating their emotions more than Revenge-Seekers, suggesting that an important difference between these groups may be their motivation to monitor and control their negative affect and how it is expressed. This finding is consistent with research that finds it is the combination of negative affect and low effortful control that is more predictive of externalizing problems (Rothbart & Bates, 2006).

It is also interesting that Pacifists endorsed relationship maintenance goals more than Revenge-Seekers. The goal of working toward a sustained relationship with the provocateur likely inhibited the pursuit of revenge. Holding an entity theory of personality has been found to be positively related to revenge-seeking (Yeager, Trzesniewski, Tirri, Nokelainen, & Dweck, 2011). Perhaps pacifists are more likely to hold an incremental theory about others’ personalities, believing that provocateurs can change. It may be that holding an incremental theory of personality becomes increasingly linked with relationship maintenance goals over time and the goal of maintaining a relationship with the wrong-doer may help account for the association found by Yeager et al. (2011) between incremental theories and revenge-seeking.

Another main focus of the current study was to compare how Revenge-Seekers and Pacifists differed in beliefs that may contribute to revenge-seeking. Revenge-Seekers endorsed both legitimacy of aggression beliefs and negative reciprocity beliefs more than Pacifists. Indeed, believing in the legitimacy of aggression was a particularly strong predictor of being a Revenge-Seeker. This is consistent with the characterization of pacifists as people who have a moral objection to engaging in violence even when others might see violence as a legitimate and sanctioned behavior. It would also be interesting in future research to consider whether in some circumstances revenge-seekers morally justify revenge (Nisbett & Cohen, 1996) or morally disengage from their revenge-seeking (cf. Gini, Pozzoli, & Hymel, 2014). Additionally, given that children’s moral emotions are also related to antisocial and prosocial behavior (Malti & Krettenauer, 2013), it seems likely that pacifists might feel more guilty if they pursued revenge and acted aggressively.

A final goal of the study was to develop a reliable and valid measure of disrespect sensitivity. Prior ethnographic research and theoretical analyses about communities that sanction aggression and retaliation in response to signs of being disrespected (Anderson, 1999; Horowitz, 1983; Nisbett & Cohen, 1996) led us to hypothesize that individual differences in children’s sensitivity to disrespect would be associated with the tendency to seek revenge. This study presents a new and internally reliable measure of individual differences in disrespect sensitivity by asking youth questions about their expectations for disrespect from others and their vigilance for signs of disrespect from others. Boys were more disrespect-sensitive than were girls, which is consistent with previous evidence of their greater endorsement of social status goals and their greater inclination to interact in larger groups where dominance hierarchies may be more central to group functioning (see Rose & Rudolph, 2006 for a review).

Furthermore, as hypothesized, Revenge-Seekers were found to be more disrespect-sensitive than Pacifists. Perhaps, Revenge-Seekers’ tendency to expect and be vigilant for disrespect from others leads them to retaliate because they fear that they may be more likely to be disrespected (or victimized) in the future if they fail to retaliate or that their “tough” image may be tarnished (Farrell et al., 2010; Jaggi & Kliewer, 2016). In contrast, the fact that Pacifists are less likely to expect disrespect from others suggests that they may be less wary about negative treatment from others and believe that the way to increase the probability that subsequent treatment will be more positive is to treat a provocateur in a non-aggressive manner. Future research is needed to investigate the origins of disrespect sensitivity, especially the kinds of socialization experiences that give rise to disrespect sensitivity. We suggest that power assertive parenting, victimization by siblings or peers, and living in unsafe neighborhoods are some of the factors that give rise to disrespect sensitivity (for studies in the broader social development literature that highlight the importance of these variables see Anderson, 1999; Duncan, 1999; Pettit, Lansford, Malone, Dodge, & Bates, 2010; Stewart & Simons, 2006; Troop-Gordon & Ladd, 2005).

There is also a need to study the long-term effects of disrespect sensitivity and revenge-seeking for individual adjustment. Revenge-seeking was associated with aggression and peer rejection in this study as well as in past studies (Lochman et al. 1993; McDonald & Lochman, 2012; Samson et al., 2012), and being a revenge-seeker may be a precursor to later social relationship problems and certain forms of criminal behaviors. It will be important to learn whether revenge-seeking mediates the association between disrespect sensitivity and later difficulties including negative interactions with peers and close relationship partners. Youth who expect that others will disrespect them and endorse revenge goals may have difficulty over the long-term in forming high quality relationships with peers and with romantic partners because they may find themselves in cycles of negative interactions, retaliating to every perceived slight from relationship partners (cf Downey, Freitas, Michaelis, & Khouri, 1998; McDonald & Asher, 2013).

Although not a focus in the current study, cluster analyses discovered a third group of youth who made negative interpretations at moderate levels after provocations and did not express desires for revenge. These children, who constituted 34% of the sample, are an intriguing group to consider. Future research might investigate the factors that lead these children not to view the peer as treating them badly, even though the harsh exclusion was clear to most children. Perhaps these children employed an ego defense strategy by denying the possibility of another child being mean to them (see Zakriski & Coie, 1996) or had difficulty attending to and recalling the negative social cues embedded in the vignettes (Hoza, Waschbusch, Pelham, Molina, & Milich, 2000). There may be still other explanations for why these children reported moderate levels of negative interpretations about the unambiguous peer provocations, and it is also possible that this group consists of distinct subtypes.

Finally, there are several remaining questions from this study that future research should address. First, although vignettes were designed to study explicit provocations from peers and involved both an exclusion and an insult from a peer, youth sometimes encounter more severe forms of explicit provocations from peers (e.g., physical harm, more severe verbal aggression). Still, even in the face of a more severe provocation, there may be a sizable percentage of youth who do not desire revenge. However, in the face of a more extreme provocation, it may be that pacifists’ relationship-maintaining goals decrease because they may think it less likely that they could have a viable friendship with someone who treats them so poorly. The percentage of revenge-seekers may also increase because more extreme provocations are more anger arousing and it may be more justifiable to seek revenge if the provocation is more severe. Similarly, we may find that the third group of children, who only moderately endorsed negative interpretations in the current study, would decrease in number when the provocations were more severe. Additionally, there may be other contextual or situational factors that would affect whether a child would desire revenge, such as whether other peers are witness to the provocation and the social status or strength of the provocateur.

Second, it will also be important for future research to consider how affective and social-cognitive predictors interact in the prediction of revenge-seeking versus pacifism. For instance, it may be that emotion regulation skills (Eisenberg et al., 1994; Zeman, Cassano, Perry-Parrish, & Stegall, 2006) moderate how beliefs about the legitimacy of aggression or disrespect sensitivity predict revenge-seeking. It may be that youth who have more difficulty regulating anger would desire revenge more after provocation, even if their beliefs about the legitimacy of aggression or their disrespect sensitivity were relatively low.

Third, it will be important to examine if the school context affects the differences that were observed between revenge-seekers and pacifists. Researchers have described contexts in which adolescents and adults are expected to retaliate when offended as a means to defend their reputation or honor (Anderson, 1999; Guerra, 2012; Horowitz, 1983; Nisbett & Cohen, 1996), and it is plausible that these expectations or norms vary by neighborhood or by school. The children in the current study were from one school, and it may be that in other schools the proportion of youth who desire revenge after provocation would be smaller or larger than it was in the current study. On the other hand, we do not expect that that the linkages between variables would differ by school or neighborhood context. For example, no matter the school, it seems likely that, compared to pacifists, revenge-seekers would still endorse legitimacy of aggression beliefs at higher levels, would be less likely to endorse relationship maintenance goals, and would be more disrespect sensitive.

Fourth, drawing on findings from this correlational study, experimental designs could be employed to learn whether it is possible to alter children’s responses to unambiguous provocations by focusing on processes that played a role in distinguishing Pacifists from Revenge-Seekers. Our results suggest that revenge-seeking does not hinge on negative interpretations. Instead revenge-seeking may be more affected by interventions that decrease beliefs about the legitimacy of aggression and sensitivity to disrespect and promote emotion regulation goals and relationship maintaining goals. In addition, it may be interesting to examine if it is possible to change how youth conceptualize pacifism and retaliation. The findings from this study challenge the idea that pacifists are “wimps” (they were viewed as no more submissive than Revenge-Seekers) and instead suggest that pacifists are well liked by peers, and that they try to regulate their emotional responses and try to maintain positive relationships with peers. Sharing these findings with youth could challenge their scripts about what is a “strong response” to a peer provocation.

Conclusion

Early adolescence can be a time when youth more often encounter provocations from peers (Wang et al., 2009). To better understand how youth respond to these provocations, this study examined the social-cognitive and affective processes that may promote or limit the desire for revenge. The study identified distinguishing characteristics of Pacifist and Revenge-Seeking youth in a context where both groups of children viewed a provocative peer as treating them badly. In response to unambiguous provocation, Pacifists were less angry, more motivated to maintain their emotional composure, and more motivated to maintain positive relationships with provocateurs. They were less likely to believe that aggression was a legitimate behavior and were also less disrespect-sensitive (i.e., less expectant of and vigilant for signs of disrespect from others). Together these findings highlight the roles of several important cognitive and emotion-related processes that decrease the probability of youth being revenge-seeking when a peer behaves badly toward them.

Biography

Kristina L. McDonald is an Assistant Professor at the University of Alabama. Her research interests include peer relationships, social cognition, conflict, aggression, and friendships. Steven R. Asher is a Professor at Duke University. His research interests include peer relationships, friendships, social goals, loneliness, and relationship beliefs.