Adolescents’ attributions about aggression: an initial investigation

Adolescents’ attributions about aggression: an initial investigation

ARTICLE IN PRESS Journal of Adolescence Journal of Adolescence 26 (2003) 559–573 www.elsevier.com/locate/jado Adolescents’ attributions about aggres...

165KB Sizes 0 Downloads 50 Views

ARTICLE IN PRESS

Journal of Adolescence Journal of Adolescence 26 (2003) 559–573 www.elsevier.com/locate/jado

Adolescents’ attributions about aggression: an initial investigation Paul Boxer*, Marie S. Tisak Bowling Green State University, USA Accepted 22 July 2003

Abstract This study examined causal attributions about aggression made by 362 participants from three age groups: early (138 7th and 8th graders), middle (121 11th and 12th graders), and late (103 college students through age 22) adolescence. Participants read a brief vignette describing a peer who displayed either proactive (non-emotional, instrumental) or reactive (emotional, undercontrolled) aggression. Participants then rated the extent to which they attributed each of 48 descriptors to be causes for the peer’s aggressive behaviour. Confirmatory factor analyses of the attribution questionnaire supported the hypothesized model of causal beliefs. The strength of endorsements of internally oriented causal factors increased with age. Findings are discussed with regard to social–cognitive development in adolescence and implications for future research on aggression in the peer context. r 2003 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.

1. Introduction A range of causal influences on aggressive behaviour in adolescents has been identified, including harsh parental discipline, exposure to media violence, and temperamental predisposition (Geen & Donnerstein, 1998; Hill & Maughan, 2001). However, adolescents themselves have typically not been asked about the causes of aggression. Adolescents are often exposed to aggression among their peers (Kingery, Coggeshall, & Alford, 1998; Singer, Anglin, Song, & Lunghofer, 1995), and very little developmental research tracks causal understandings or attributional abilities into this developmental phase. Adolescents’ beliefs about the causes of *Corresponding author. Research Center for Group Dynamics, 5034 ISR, 426 Thompson Street, Ann Arbor, MI 48106, USA. E-mail address: [email protected] (P. Boxer). 0140-1971/$30.00 r 2003 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S0140-1971(03)00056-3

ARTICLE IN PRESS 560

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

aggression in their peers are important to investigate given the potentially deleterious effects of negative peer reputations. The current study represents an initial examination of the beliefs held by adolescents about the causes of aggressive behaviour. 1.1. The development of attributional abilities Considerable theory and empirical research have established changes and progressions in children’s social cognitive abilities throughout childhood (Rholes, Newman, & Ruble, 1990; Terwogt & Harris, 1993; Tisak, 1995). Most relevant to the current investigation, researchers have generally documented shifts in children’s understandings of the causes of behaviour from external, situational explanations to internal, dispositional explanations (Rholes et al., 1990). Basic attributional abilities appear to emerge at a very young age and develop over time. Researchers have found that children as young as preschool age have the capacity to make attributional inferences—or judgments about ‘‘perceived causality’’ (Fincham, 1983, p. 118)—of others’ behaviour (Friedberg & Dalenberg, 1990, 1991). Although many researchers have studied and tracked changes in attributional abilities, however, developmental attribution research has generally focused only on the childhood years (i.e. between preschool and early adolescence). Few studies have examined developmental differences in attributions made during and across adolescence. 1.2. Attributions about aggression Attributional ability is a subcomponent of general social–cognitive ability. Contemporary models of aggressive behaviour development have emphasized the critical role of social–cognitive factors (e.g. hostile attributions, normative beliefs, outcome expectancies) in the emergence and maintenance of aggressive responding over time (e.g. Crick & Dodge, 1994; Huesmann, 1998). These models have placed attributional abilities and understandings in a key position with regard to aggression. Dodge and colleagues (e.g. Dodge & Frame, 1982; Crick & Dodge, 1994) have underscored the critical effects of hostile attributional biases among aggressive youth. Studies conducted by Dodge and others have demonstrated that aggressive children and adolescents are more likely that are their non-aggressive counterparts to attribute hostile intent to an ambiguous provocation. Despite the considerable research attention directed to the attributions made by aggressive youth themselves, little research has attended to the attributions made by youth about the behaviour of aggressive peers. This is surprising, especially for research with adolescents, given that peer relationships and behaviours can have considerable influence on the expression and persistence of aggression (e.g. Dishion, McCord, & Poulin, 1999; Henry et al., 2000). Further, considering the influence that beliefs and causal biases can have on peer relations, it is important to understand the ways in which adolescents think about their aggressive peers. Hymel (1986) suggested that although intervention programs might be able to attenuate aggressive behaviour, they might not change the pariah status already earned by aggressive individuals. Examining the causal beliefs adolescents hold for aggression might provide an initial basis for developing ways to support interventions for aggressive youth by intervening in peer groups to change beliefs and

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

561

biases. In the current study, we examined beliefs held by adolescents about the reasons for aggression in their peers. Past findings have been consistent in demonstrating that youth may behave aggressively for a number of reasons, spanning both internally based (e.g. temperament) and externally based (e.g. environmental exposure to violence) influences. Prior research has established several specific domains of causality for aggression, including: undercontrolled impulses (Lennings, 1991); underlying conduct disorder or developing antisocial personality disorder (McMahon & Estes, 1997); poor emotion regulation (Eisenberg et al., 1996); harsh parenting (Mahoney, Donnelly, Lewis, & Maynard, 2000); deviations in social cognition or social reasoning (Crane-Ross, Tisak, & Tisak, 1998; Dodge & Price, 1994); peer rejection or victimization (Schwartz, 2000); and exposure to aggression in peers (Dishion et al., 1999), parents (Mahoney, Donnelly, Boxer, & Lewis, 2003), and the media (Huesmann, Moise-Titus, Podolski, & Eron, 2003). These domains have been emphasized repeatedly in broad reviews of research on the development of aggressive behaviour through childhood and adolescence (Huesmann, 1994; Coie & Dodge, 1998; Geen & Donnerstein, 1998; Hill & Maughan, 2001). Whether adolescents recognize and endorse these potential causes for aggression among their peers is the focus of the current investigation. As noted, very few studies have assessed directly the causal attributions made by children or adolescents about others’ aggressive behaviour. Studies in which this has occurred have typically framed aggressive behaviour as ‘‘deviant,’’ or as the behaviour of problematic peers. That is, aggressive behaviour has not been considered as an isolated form of behaviour, independent of stigmatizing labels or undesirable consequences. Attributions about this kind of behaviour have generally been studied within the rubric of peer acceptance. That is, researchers have examined whether children’s social preferences for abnormal, deviant children covary with their beliefs about the causes of such children’s behaviour. However, some research has examined children’s and adolescents’ beliefs about the causes and nature of other forms of deviant behaviour, such as school phobia (Chassin & Coughlin, 1983), withdrawal (Graham & Hoehn, 1995), hyperactivity (Johnston, Patenaude, & Inman, 1992), and paranoia (Coie & Pennington, 1976). In an oft-cited study, Coie & Pennington (1976) presented children at four age levels (7, 10, 13, and 17 years) with two vignettes describing hypothetical peers displaying ‘‘loss of control’’ or ‘‘distorted perceptions.’’ Children were then asked to rate how different from most children the story actors were, and explain what it was about the actors that made them so different. The authors found that aggression (e.g. physical aggression) and social norm violations (e.g., disruptive acts) were the most frequently mentioned descriptors of the deviant behaviour alluded to in the vignettes. Further, the authors identified some age trends in children’s responses. Specifically, the oldest group made more normatively based judgements (e.g. more references to social norm violations) about the level of deviance displayed in the vignettes and therefore paid more attention to the social impact of deviant behaviour. Coie and Pennington (1976) speculated that adolescence brings with it a new type of social judgement—that others can fail to ‘‘match up to a social criterion for psychological stability’’ (p. 412). In other words, adolescents are able to consider the behaviour of others with regard to how people generally behave, and recognize deviations from those norms. The researchers suggest an irony in the commonly held belief that with development,

ARTICLE IN PRESS 562

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

children’s understandings of persons progress towards viewing others as stable and predictable. They propose that advances in social cognition also lead to the understanding that behaviour can in fact be highly unstable and unpredictable—that behaviour can represent irrationality or disorder. The current study was an initial investigation into the causal attributions made by adolescents at three developmental levels (early, middle, and late adolescence). This study focused on the testing of a hypothesized model of different domains of causal beliefs for aggression (e.g. parenting, peer rejection). The fit of this model was examined across all respondents, as well as within age group. Thus, it was hypothesized that the different causal domains would be salient and differentiable for adolescents at different ages. It was further hypothesized that increases by age would emerge in the strength of endorsement for all domains. This hypothesis was based on the idea that with age, adolescents acquire more experience and familiarity with aggressive behaviour in their peers and thus greater recognition that aggression can be the result of a wide variety of causes. Adolescents considered one of two types of aggressive behaviour: proactive and reactive (Dodge, 1991). Proactive aggression is learned aggressive behaviour, typically non-emotional, emitted to achieve a purposeful goal—for example, one child shoving another to cut in line. Reactive aggression is aggressive behaviour provoked or influenced by a negative emotional reaction to a situation or event—for example, a bullied child lashing out from fear. Proactive and reactive aggression are thought to have differing aetiologies, where proactive aggression appears to be more closely linked to persistent exposure to aggressive models, and reactive aggression more closely associated with childhood histories of abuse and trauma (Dodge, Lochman, Harnish, Bates, & Pettit, 1997). Adolescent observers might also believe that proactive and reactive aggression have dissimilar origins. It was hypothesized that the strength of attributional endorsement would vary by type of aggression, such that proactive aggression would be attributed more strongly to stable, dispositional factors, while reactive aggression would be attributed more strongly to unstable, situational factors.

2. Method 2.1. Participants Data were obtained from students at two middle schools, one high school, and a mediumsized state university in Northwest Ohio. Participants in the study were 362 adolescents (mean age=15.81 years, sd=2.71; 260 female, 102 male). There were 138 early adolescents (7th and 8th graders; mean age=12.75 years, sd=0.71; 62% female), 121 middle adolescents (11th and 12th graders; mean age=16.55 years, sd=0.59; 70% female), and 103 late adolescents (college students through age 22; mean age=19.06 years, sd=0.96; 86% female) in the sample. The majority of all participants were Caucasian (92%), living in two-parent homes (81%). Sixty-two per cent of participants reported that their fathers (or father figures) had obtained college-level or higher educations, and 53% reported this for their mothers (or mother figures).

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

563

2.2. Materials 2.2.1. Stimulus vignettes Two vignettes were written to provide stimulus information for measuring attributions about proactive and reactive aggression. These vignettes were intended to present clear descriptions of affect and behaviour, independent of any larger contextual or mitigating circumstances. One vignette describes a hypothetical peer emitting proactively aggressive behaviour; the other describes a hypothetical peer emitting reactively aggressive behaviour. Content across the two vignettes was similar: Mike is a student in your school. He is about your age and is interested in the same kinds of things as you. Mike gets into a lot of fights where he yells at or hits other people. He is almost always the one who starts these fights. Mike NEVER seems angry before these fights. The italicized terms in the last two sentences of the vignette were modified depending upon the type of aggression represented. For proactive aggression, the character was ‘‘almost always’’ the one starting fights and ‘‘never’’ seeming angry. For reactive aggression, the character was ‘‘almost never’’ the one starting fights and ‘‘always’’ seeming angry. The gender of hypothetical peers was matched to the gender of participants (i.e. characters named ‘‘Mike’’ or ‘‘Nicole’’). 2.2.2. Attributions about aggression questionnaire This 48-item questionnaire was designed to tap a broad array of attributional domains for the causes of adolescent aggressive behaviour. Items on the measure were written to assess some of the traditional attributional domains measured in prior research, such as internal causes (e.g. impulsivity) as well as other areas identified by empirical research as potentially causal for aggressive behaviour (e.g. social learning from the media). Items were presented in randomized order. Participants were asked to ‘‘decide how much you think each description is a reason why (the character in the vignette) acted that way.’’ A six-point response scale (ranging from 1=‘‘definitely not a reason’’ to 6=‘‘definitely a reason’’) was utilized for items in the following nine domains: Impulsivity (e.g. ‘‘He can’t control himself’’; 5 items); Antisociality (e.g. ‘‘He enjoys hitting other people’’; 6 items); Emotional instability (e.g. ‘‘He blows up easily’’; 5 items); Parenting (e.g. ‘‘He doesn’t get along with his parents’’; 6 items); Social cognitive deficits (e.g. ‘‘He thinks other people try to hurt him’’; 5 items); Social rejection (e.g. ‘‘He just doesn’t have many friends’’; 5 items); Social learning from peers (e.g. ‘‘He thinks it’s cool to act that way’’; 5 items), Parents (e.g. ‘‘His parents act that way too’’; 6 items), and the Media (e.g. ‘‘He watches a lot of violent TV’’; 5 items). These domains followed directly from those consistently identified in reviews of research as important causal influences on aggression. Confirmatory factor analysis (CFA) using maximum-likelihood estimation computed by the AMOS program (Arbuckle, 1999) was used to determine whether participants’ responses on the attribution questionnaire fit the hypothesized nine-factor structure of causal beliefs about aggression. These factors were each examined independently across all participants. Prior to conducting CFA for any factor, correlation matrices generated with listwise deletion were prepared independently for the items on each factor (by full sample, age group, and aggression type). Only data from participants who responded to every item in a factor were included. This

ARTICLE IN PRESS 564

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

procedure maximized the number of participants available for each factor, thus enhancing the statistical power of the CFA. Sample sizes available for each factor ranged from 346 to 356. Model fit was assessed using the w2 statistic (where good fit is indicated by non-significant w2 values) and the root mean square error of approximation (RMSEA) fit index (where good fit is indicated by RMSEA values in the 0.000–0.050 range). Confirmatory analyses were conducted for three separate samples on each factor: (1) The full sample on which factor models were calibrated; (2) simultaneously across the three age groups; and (3) simultaneously across the two types of aggression. In tests of simultaneous fit, fit attempts began with all parameters free across groups, and then constraints were added across groups to examine measurement equivalence: First the item slopes and intercepts were constrained together to be equal across groups. If this model yielded a poor fit to the data, then a model constraining item slopes only was tested. At each step, a w2 difference test was utilized to determine whether the more restricted model could be retained. The results of the model-fitting procedures indicated that the hypothesized model of causal domains was generally supported by the data with few items needing to be excluded. An interesting exception emerged with regard to the parenting domain, which yielded two factors subsuming the parent–child relationship and emotionally/physically abusive practices. Table 1 displays the items comprising the final models retained for each factor. Model fitting statistics for the attribution questionnaire are presented in Appendix A1. As shown, the fit statistics obtained suggest that the data conformed well to the hypothesized model2. 2.3. Procedures School principals provided permission for the investigators to solicit participants under the age of 18 through letters sent to parents. Parental consent and participant assent were obtained for these participants prior to any data collection. Rates of participation varied somewhat by school: in the middle schools, rates ranged from 30% to 75% by homeroom class; in the high school, the rate was over 75%. College student participants (age 18 and older) were recruited through introductory psychology courses, and provided their own consent. For all participants, questionnaires were administered in a group format. Basic instructions for questionnaire completion (e.g. try to complete all items) were read before each administration by research staff or homeroom teachers; questionnaires were otherwise self-administered. Each participant was randomly assigned to read and respond to either a proactive (n=184) or reactive (n=178) 1

For the purpose of brevity, complete model-fitting information and item-level data (item means, standard deviations, slope and intercept estimates, and item reliability) from the confirmatory factor analyses are not presented. This material may be obtained by contacting the authors. 2 Modifications to factors during model-fitting procedures typically involved item deletion. For two factors (emotional instability and social rejection), additional modifications were necessary to obtain adequate fit. The final model for emotional instability required that the error variances of ‘‘Has emotional problems’’ and ‘‘Has mood swings’’ be correlated. Though this is not an ideal constraint, the similarity in content of the two items makes the constraint a reasonable one. The final model for social rejection required that the error variances of ‘‘Others students don’t like to hang out with him/her’’ and ‘‘Other students just don’t like him/her’’ be constrained as equal. Though this is also not an ideal constraint, the almost exact content of the two items makes such a constraint acceptable.

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573 Table 1 Items retained for final factor models Factor

Items retained

Impulsivity

Can’t control self. Acts before thinks. Makes quick decisions. Does the first thing that comes to mind. Does things without thinking.

Antisociality

Enjoys Enjoys Enjoys Enjoys

Emotional instability

Blows up easily. Has emotional problems. Gets upset very easily. Has mood swings.

Social cognitive deficits

Thinks people try to hurt him/her. Doesn’t know how to say what s/he feels. Gets confused about what other people say or do. Doesn’t understand what people want from him/her. Doesn’t know what else to do.

Social rejection

Other students don’t like to hang out with him/her. S/he just doesn’t have many friends. Other students just don’t like him/her.

Parent-child relationship

Parents don’t spend much time with him/her. Doesn’t get along with parents. Parents don’t talk much with him/her.

Emotional/physical abuse

Parents yell at him/her for no reason. Parents often hit him/her for no reason. Parents punish him/her extremely hard.

Peer social learning

Sees other kids act that way. All the people s/he hangs out with act that way. Sees friends act that way. Thinks it’s cool to act that way. Popular kids act that way.

Media social learning

Watches a lot of violent TV. Hears about a lot of violent things on the news. Watches a lot of violent movies. Plays a lot of violent video games.

Parent social learning

Sees parents act that way at home. Parents get into a lot of fights with each other. Parents act that way too. Learned to act that way from parents.

hitting people. hurting other people. scaring people. fighting.

565

ARTICLE IN PRESS 566

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

stimulus vignette (vignette conditions were distributed approximately evenly within each age group). As noted, vignette character gender was matched with respondent gender; participants each read a single vignette. Participant questionnaire packets did not differ in any other way except for the colour of the paper used to discriminate male (tan) from female (blue) stimulus characters. Individual questionnaires were presented in the order described above. Survey administration typically lasted no more than 30 min.

3. Results As illustrated above, the primary hypothesis of this study was confirmed in that the data obtained fit a hypothesized model of causal beliefs about aggression in a sample of adolescents. The factors confirmed through the model-fitting procedures detailed above were utilized for subsequent inferential analyses, with exceptions as described below. However, we first examined the causal domains with regard to the extent to which each was endorsed overall as a cause of aggressive behaviour. Such an inspection of the data provides an interesting broad look at the causes adolescents believe to be salient with regard to aggression. We converted the observed means of each factor to z-scores in order to examine the percentages of participants who endorsed each factor above the true (latent) mean of 3.5 (i.e. exact mean on 1–6 scale). Table 2 presents these percentages. As can be seen from this table, emotional instability and impulsivity were the most consistently endorsed across respondents as causes of aggression, while antisociality and media social learning were the least endorsed. As described, confirmatory factor analyses were conducted to obtain factors fitting across the three age groups and the two types of aggressive behaviour. This was done because developmental (age) and typological (aggression types) differences were of central concern in the present analysis. However, we were also interested in ensuring that analyses could be performed across both genders. This issue emerged for three reasons. First, given the preliminary nature of the

Table 2 Percentages of respondents endorsing attributional factors as reasons for aggressive behaviour across full sample Factor

Percentage

Emotional instability Impulsivity Social learning from parents Parent-child relationship Social cognitive deficits Emotional/physical abuse Social rejection Social learning from peers Antisociality Social learning from media

83.2 78.5 69.7 64.9 63 60.5 48.9 33.7 27.6 22.3

Note: Percentages calculated using z-score conversion of construct (manifest) means to obtain actual percentage of scores above true factor (latent) means of 3.5.

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

567

investigation, we wanted to emphasize developmental and typological differences in the construction of a measure that would be useful for investigating causal attributions. Second, although there is much evidence demonstrating that males and females engage in different . forms and frequencies of aggression (e.g. Bjorkqvist, Lagerspetz, & Kaukiainen, 1992), there is little to suggest that males and females understand the causes of physically aggressive behaviour in different ways. Third, given that the majority of our sample was female (n=260), we were interested in maintaining the statistical power gained by collapsing across gender in our analyses. Prior to examining developmental and typological differences, we conducted t-tests (a ¼ 0:05) by gender on each of the causal factors, for participants with complete data on each factor. These analyses revealed that among the ten causal factors identified, gender differences were present in two: antisociality (males M=3.06, sd=1.39; females M=2.67, sd=1.12) and social cognitive deficits (males M 3.63, sd=1.06; females M=3.89, sd=0.92). We thus included gender as a covariate in subsequent analyses of these two factors.

3.1. Developmental and typological differences in causal attributions about aggression To examine the hypothesis that endorsement of causal domains would increase with age and vary by aggression type, factorial analyses of variance (ANOVA) were conducted individually on all attribution factors. In these analyses, age group (early, middle, and late) and aggression type (proactive, reactive) served as the independent variables by which attribution factor scores were compared. For three factors, only one-way ANOVAs were conducted: antisociality and emotional instability (fit across age group only; typological differences not examined); and parent social learning (fit across aggression type only; age differences not examined). Analysis of covariance (ANCOVA) procedures including gender as a covariate were used for the antisociality and social cognitive deficit factors. Factor scores were derived by calculating the mean of the indicators for each factor, for subjects who had complete data on that factor. Planned pairwise contrast analyses were conducted to test differences between group means for main effects of age group. To control for Type I error, alpha level for overall ANOVAs was set at a ¼ 0:01: Alpha level for age group contrasts within each factor (three pairwise contrasts per factor) was derived using the Bonferroni correction (Keppel, 1991, p. 167) as applied to a maximum acceptable familywise error of aFW ¼ 0:05; yielding a ¼ 0:02: There were no interaction effects of age group with aggression type for any of the factors. Significant main effects of age group in the predicted direction were observed on four of the attribution factors. Age increases were observed on: impulsivity (F [2, 346]=7.72, po0.01); emotional instability, (F [2, 347]=9.09, po0.001); antisociality (F [2, 346]=7.12, po0.01); and social cognitive deficits (F [2, 352]=9.74, po0.001). On the media social learning factor, a curvilinear effect was found, F (2, 354)=14.70, po0.001. The youngest and oldest age groups endorsed this causal domain more strongly than did the middle age group. Table 3 displays the results of the contrasts for these analyses, and the means for all factors by age group. No significant differences were observed as the function of aggressive vignette type.

ARTICLE IN PRESS 568

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

Table 3 Means (standard deviations) and cell sizes for attributional factors by age group Factor

Early adolescent n

Mean (sd) Impulsivity Antisociality Emotional instability Social cognitive deficits Social rejection Peer learning Media learning Parent-child relationship Emotional/physical abuse

4.03 2.53 4.22 3.56 3.50 3.06 2.92 3.77 3.59

Middle adolescent

a

(1.23) (1.25)a (1.14)a (1.05)a (1.27) (1.15) (1.36)a (1.18) (1.25)

125 133 129 132 133 129 135 135 132

Mean (sd) 4.44 3.05 4.53 3.81 3.51 3.15 2.10 3.99 3.88

b

(0.92) (1.24)b (1.03)ab (0.91)a (1.05) (0.87) (1.05)b (1.09) (1.13)

Late adolescent n 120 120 118 120 119 120 118 120 117

Mean (sd) 4.53 2.79 4.79 4.16 3.40 2.99 2.57 4.04 4.03

b

(0.82) (1.05)ab (0.76)b (0.79)b (0.99) (1.03) (1.13)a (0.99) (1.05)

n 101 97 100 100 100 100 101 101 100

Note: n=Cell size. Means with different superscripts are significantly different at po0.02. Means based on items fitting final factor models for participants with complete data available. Items scaled from 1=‘‘Definitely not a reason’’ for aggressive behaviour to 6=‘‘Definitely a reason’’ for aggressive behaviour. Parent social learning not included in this table because factor model did not fit simultaneously across age groups.

4. Discussion Aggressive behaviour is determined by an array of influences, including temperamental predisposition, early learning history, family functioning, and social cognitive mediation (Geen & Donnerstein, 1998; Hill & Maughan, 2001). The most prominent finding from the present study was that adolescent respondents of all ages recognized and endorsed this multiplicity of causes for aggressive behaviour. Although some causal factors (e.g. emotional instability) were endorsed more strongly overall than others (e.g. media social learning), no single factor was dismissed as being completely causally unrelated to aggression. Related to this finding, the confirmatory factor analyses of the attribution measure yielded several useful scales exploring causal beliefs in a variety of domains. A second major finding from the present study was that some variability in causal attribution was found across adolescent age groups. For many factors, increases in the strength of attributional endorsements were observed from early through late adolescence. One notable exception was media social learning, which was endorsed by early and late adolescents more strongly than by middle adolescents. 4.1. Attributions about aggression This study contributes to past research by assessing attributions made about the causes of aggressive behaviour across different age groups of adolescents. In the context of research on various aspects of social cognition, aggression has been viewed as deviant (Coie & Pennington, 1976) and more attributable to physical and psychological features of the actor, relative to causes such as peer or parental treatment (Chassin & Coughlin, 1983). Hypothetical aggressive children have been rated as less deserving of sympathy and more deserving of anger, and considered more responsible for their behaviour, than hypothetical withdrawn children (Graham & Hoehn, 1995). Aggression is often associated with social rejection and other negative peer reactions (Coie,

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

569

Dodge, & Coppotelli, 1982; Fabes, Eisenbeng, Smith, & Murphy, 1996). Taken together, studies of peer perceptions of child aggression (both real and imagined) suggest that aggressive behaviour tends to receive internal, controllable attributions. That is, observers seem to think and behave as though the aggression is due to deliberate actions of the actor him- or herself, thus resulting in negative reactions to the actor. Consistent with this interpretation, the results of the present study demonstrated that internal attributions were endorsed more heavily as causes of aggressive behaviour than other types of causes. As Table 2 indicates, 83.2% of respondents believed emotional instability to be a cause of aggression, and 78.5% of respondents believed this of impulsivity. Both of these constructs represent causal factors located squarely in the individual. Interestingly, however, both of these constructs also represent causal factors for aggression that are somewhat out of the control of individuals. That is, these constructs suggest that aggressive behaviour is the function of impulse without forethought, or of unsteady affective states. The attribution factor most indicative of an internal, controllable cause, antisociality, was the second least-endorsed cause of aggression, at 27.6%. This possibly reflects the motivation of participants to view aggression as the product of unpredictable cognitive or emotional states as opposed to a stable trait. It is safer and more selfprotective to believe that aggression can be avoided so long as one does not provoke it, rather than to think that there are actually those who enjoy hitting and harming others. Participants seemed to place the least responsibility for aggression on the factors in which they could be the most complicit—i.e. the peer environment in which they might exist as observers or models of aggression, and the media violence which they most likely watch and enjoy. If adolescents recognize that peer rejection and peer modelling cause aggressive behaviour, they essentially acknowledge their own potential culpability for it. Further, if adolescents endorse the causal effects of media violence on aggression, they essentially admit that they too have the potential to become aggressive when exposed to violent entertainment, or that what they do in their spare time can be harmful to them. 4.2. Development in causal attributions Developmental differences on the order of increasing endorsement with age were observed on only four of the attributional factors: impulsivity, antisociality, emotional instability, and social cognitive deficits. The hypothesis that increasing age would be associated with increasing endorsements for all attributional factors was therefore not confirmed. Rather, it appears that with age comes the tendency to place more blame on those causal factors for aggression most closely related to the internal characteristics of the actor. Attributions of causality for aggressive behaviour to the internal domain are consistent with the correspondence bias. This is a social psychological construct in which observers tend to draw dispositional inferences about behaviour even when situational factors are present and credible (Gilbert & Malone, 1995). In the current study, a relative strengthening in this bias is evident across ages. This finding provides some support for the common belief in developmental psychology that internal, as opposed to external, causal attributions become more common with increasing age (e.g. Rholes et al., 1990). Adolescents in the current study did not seem to demonstrate an increasingly exclusive preference for internal over external causes, but rather an increasing confidence in the explanatory power of internal causes.

ARTICLE IN PRESS 570

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

One finding presented an interesting twist with regard to developmental differences. For the media social learning factor, a curvilinear effect emerged. Middle adolescents, as opposed to early and late adolescents, provided the lowest endorsement of exposure to media violence as a cause of aggression. This might be due to the high school age group being the most actively engaged in routine exposure to (and enjoyment of) media portrayals of violence. It also possible that, because of the process of identity development, high school students may be the least willing to connect media habits and interests to a negative behavioural construct. Believing that media violence causes aggressive behaviour might require those who regularly view it to view themselves as aggressive people. 4.3. Aggression typology Given substantial research demonstrating the validity of the proactive/reactive typology (e.g. Dodge et al., 1997), it is surprising that the hypotheses concerning attributional differences with regard to aggression type were largely not confirmed. A possible explanation of this result is that although aggression may appear to take on different forms from the perspective of the scientific observer, this may not be the case for those exposed directly to it. Adolescents may conceptualize the aggression in their peer environments based on more distal causes (e.g. parenting), or solely based on the consequences of aggression (e.g. getting punched) regardless of immediate causality or motivation. A distinction highlighting reactive and proactive proximal causes may not have any practical application in peer interactions, even while it applies to intervention approaches (Vitaro, Gendreau, Tremblay, & Oligny, 1998) and more enhanced understandings of the forms and functions of aggressive behaviour (Poulin & Boivin, 2000). 4.4. Limitations and future directions The great majority of participants in this study were Caucasian and middle-class, and subsequent studies should therefore examine causal attributions for aggression in more culturally and socioeconomically diverse samples. Given that disadvantaged, minority groups of children and adolescents display higher levels of problematic aggression on average than the type of population sampled in this study (Guerra, Huesmann, Tolan, Van Acker, & Eron, 1995), it is possible that attributional constructs might be more pragmatically researched in those groups. Also, females were over-represented (72%) relative to males in the current study. Although this did not greatly impact the results of developmental analyses, future studies of attributional constructs could utilize a more gender-balanced sample in the service of investigating gender differences in causal attribution in greater detail. Methodological limitations might also have accounted for the lack of causal differentiation made by participants for proactive and reactive aggression. For example, the stimulus vignette used to create the impression of proactive versus reactive aggression might not have been strong enough to provoke different understandings of causality. Alternatively, a greater number of vignettes read by each participant might have been necessary to stimulate differences in causal attribution by providing a range of examples of proactive and reactive aggression (Wells & Windschitl, 1999). It should be emphasized that the current study utilized a cross-sectional design. Subsequent investigations could track the development of causal attributions using longitudinal methods.

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

571

However, the results of the current study do suggest that adolescents consider the causes of peer aggression in conceptually distinct domains. Future studies in this area could incorporate additional domains of causality for aggression, such as biological predispositions, overt peer victimization, and neighborhood violence exposure. It should also be noted that a critical goal of the current study was to develop an adequate measure of causal domains using quantitative techniques. However, a qualitative study (e.g. open-ended interviews) of these constructs might also yield important and interesting information. Additional research could also consider the ways in which the study of causal attributions for aggression might be applied, for example by investigating the social–cognitive processes that lead adolescents to reject or accept aggressive peers. Acknowledgements This paper is based on a pre-dissertation project conducted by the first author under the second author’s supervision. We are grateful to the participating principals, teachers, and students of Sylvania Northview High School (Sylvania, OH), McCord Middle School (Sylvania, OH), and Eastwood Middle School (Pemberville, OH) for their time and assistance. We also wish to acknowledge the insightful feedback provided by Eric F. Dubow, Dara Musher-Eizenman, and Julie Burke on earlier versions of this work, as well as the useful editorial guidance provided by Nancy Galambos and two anonymous reviewers. Shannon Brinker and Jennifer VanScoyoc assisted with data collection. Appendix A Model fit statistics for attribution factors by full sample, age group, and type of aggression are given below: Model fit statistics Factor

Full sample w2

Impulsivity Antisociality Emotional instability Social cognitive deficits Social rejection Parentinga Peer learning Media learning Parent learning

By age group

By aggression type

df

p

RMSEA w2

df

p

RMSEA w2

df

p

RMSEA

6.884 1.837 0.077

5 2 1

0.229 0.399 0.782

0.033 0.000 0.000

28.01 11.208 11.017

23 12 9

0.215 0.511 0.275

0.025 0.000 0.026

22.309 — —

19 — —

0.269 — —

0.023 — —

8.826

5

0.116

0.047

29.899

23

0.152

0.029

17.286

14

0.241

0.026

1.572 10.183 7.883 2.716 3.089

1 8 5 2 2

0.210 0.252 0.163 0.257 0.213

0.040 0.028 0.041 0.032 0.039

14.136 45.879 30.773 9.982 —

13 32 5 12 —

0.364 0.053 0.009 0.618 —

0.016 0.036 0.055 0.000 —

10.871 29.994 13.774 12.532 17.750

7 26 14 11 12

0.144 0.268 0.467 0.325 0.123

0.040 0.021 0.000 0.020 0.042

Note. RMSEA=Root mean square error of approximation. a Two-factor model.

ARTICLE IN PRESS 572

P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

References Arbuckle, J. L. (1999). AMOS 4.0 user’s guide. Chicago: SmallWaters Corporation. . Bjorkqvist, K., Lagerspetz, K. M. J., & Kaukiainen, A. (1992). Do girls manipulate and boys fight? Developmental trends in regard to direct and indirect aggression. Aggressive Behaviour, 18, 117–127. Chassin, L., & Coughlin, P. (1983). Age differences in children’s attributions for deviant behaviours. Psychiatry, 46, 181–185. Coie, J. D., & Dodge, K. A. (1998). Aggression and antisocial behaviour. In W. Damon (Series Ed.), N. Eisenberg (Vol. Ed.), Handbook of child psychology, Vol. 3: Social, emotional, and personality development (5th ed.) (pp. 779–862). New York: Wiley. Coie, J. D., Dodge, K. A., & Coppotelli, H. (1982). Dimensions and types of social status: A cross-age perspective. Development Psychology, 18, 557–570. Coie, J. D., & Pennington, B. F. (1976). Children’s perceptions of deviance and disorder. Child Development, 47, 407–413. Crane-Ross, D., Tisak, M. S., & Tisak, J. (1998). Aggression and conventional rule violation among adolescents: Social-reasoning predictors of social behaviour. Aggressive Behaviour, 24, 347–365. Crick, N. R., & Dodge, K. A. (1994). A review and reformulatioin of social information-processing mechanisms in children’s social adjustment. Psychological Bulletin, 115, 74–101. Dishion, T. J., McCord, J., & Poulin, F. (1999). When interventions harm: Peer groups and problem behaviour. American Psychologist, 54, 755–764. Dodge, K. A. (1991). The structure and function of reactive and proactive aggression. In D. Pepler, & K. Rubin (Eds.), The development and treatment of childhood aggression (pp. 201–218). Hillsdale, NJ: Erlbaum. Dodge, K. A., & Frame, C. L. (1982). Social cognitive biases and deficits in aggressive boys. Child Development, 53, 620–635. Dodge, K. A., Lochman, J. E., Harnish, J. D., Bates, J. E., & Pettit, G. S. (1997). Reactive and proactive aggression in school children and psychiatrically impaired chronically assaultive youth. Journal of Abnormal Psychology, 106, 37–51. Dodge, K. A., & Price, J. M. (1994). On the relation between social information processing and socially competent behaviour in early school-aged children. Child Development, 65, 1385–1397. Eisenberg, N., Fabes, R. A., Guthrie, I. K., Murphy, B. C., Maszk, P., Holmgren, R., & Suh, K. (1996). The relations of regulation and emotionality to problem behaviour in elementary school children. Development and Psychopathology, 8, 141–162. Fabes, R. A., Eisenbeng, N., Smith, M. C., & Murphy, B. C. (1996). Getting angry at peers: Associations with liking of the provocateur. Child Development, 67, 942–956. Fincham, F. D. (1983). Developmental dimensions of attribution theory. In J. Jaspars, F. D. Fincham, & M. Hewstone (Eds.), Attribution theory, and research: Conceptual developmental, and social dimensions (pp. 117–164). London: Academic Press. Friedberg, R. D., & Dalenberg, C. J. (1990). Causal attributions in young children. Child Study Journal, 20, 139–151. Friedberg, R. D., & Dalenberg, C. J. (1991). Attributional processes in young children: Theoretical, methodological, and clinical considerations. Journal of Rational-Emotive and Cognitive-Behaviour Therapy, 9, 173–183. Geen, R. G., & Donnerstein, E. (Eds.). (1998). Human aggression: Theories, research, and implications for social policy. San Diego, CA: Academic Press. Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias. Psychological Bulletin, 117, 21–38. Graham, S., & Hoehn, S. (1995). Children’s understanding of aggression and withdrawal as social stigma: An attributional analysis. Child Development, 66, 1143–1161. Guerra, N. G., Huesmann, L. R., Tolan, P. H., Van Acker, R., & Eron, L. D. (1995). Stressful events and individual beliefs as correlates of economic disadvantage and aggression among urban children. Journal of Consulting and Clinical Psychology, 63, 518–528. Henry, D., Guerra, N., Huesmann, L. R., Tolan, P., Van Acker, R., & Eron, L. D. (2000). Normative influences on aggression in urban elementary school classrooms. American Journal of Community Psychology, 28, 59–81. Hill, J., & Maughan, B. (2001). Conduct disorders in childhood and adolescence. New York: Cambridge University Press.

ARTICLE IN PRESS P. Boxer, M.S. Tisak / Journal of Adolescence 26 (2003) 559–573

573

Huesmann, L. R. (Ed.) (1994). Aggressive behaviour: Current perspectives. New York: Plenum. Huesmann, L. R., Moise-Titus, J., Podolski, C., & Eron, L. D. (2003). Longitudinal relations between children’s exposure to television violence and their aggressive and violent behaviour in young adulthood: 1977–1992. Developmental Psychology, 39, 201–221. Hymel, S. (1986). Interpretations of peer behaviour: Affective bias in childhood and adolescence. Child Development, 57, 431–445. Johnston, C., Patenaude, R. L., & Inman, G. A. (1992). Attributions for hyperactive and aggressive child behaviours. Social Cognition, 10, 255–270. Keppel, G. (1991). Design and analysis: A researcher’s handbook. Upper Saddle River, NJ: Prentice-Hall. Kingery, P. M., Coggeshall, M. B., & Alford, A. A. (1998). Violence at school: Recent evidence from four national surveys. Psychology in the Schools, 35, 247–258. Lennings, C. J. (1991). A modification of the Kendall–Wilcox self-control scale for delinquents. International Journal of Offender Therapy and Criminology, 35, 83–91. Mahoney, A., Donnelly, W. O., Boxer, P., & Lewis, T. (2003). Marital and severe parent-to-adolescent physical aggression in clinic-referred families: Mother and adolescent reports on co-occurrence and links to child behaviour problems. Journal of Family Psychology, 17, 3–19. Mahoney, A., Donnelly, W. O., Lewis, T., & Maynard, C. (2000). Mother and father self-reports of corporal punishment and severe physical aggression toward clinic-referred youth: Prevalence, demographic risk factors, and links to externalizing behaviour problems. Journal of Clinical Child Psychology, 29, 266–281. McMahon, R. J., & Estes, A. M. (1997). Conduct problems. In E. J. Mash & L. G. Terdal (Eds.), Assessment of childhood disorders (3rd ed.) (pp. 130–196). New York: Guilford. Poulin, F., & Boivin, M. (2000). The role of proactive and reactive aggression in the formation and development of boys’ friendships. Developmental Psychology, 36, 233–240. Rholes, W. S., Newman, L. S., & Ruble, D. N. (1990). Understanding self and other: Developmental and motivational aspects of perceiving persons in terms of invariant dispositions. In E. T. Higgins, & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behaviour (Vol. 2) (pp. 369–407). New York: Guilford. Schwartz, D. (2000). Subtypes of victims and aggressors in children’s peer groups. Journal of Abnormal Child Psychology, 28, 181–192. Singer, M. I., Anglin, T. M., Song, L., & Lunghofer, L. (1995). Adolescents’ exposure to violence and associated symptoms of psychological trauma. Journal of the American Medical Association, 273, 477–483. Terwogt, M., & Harris, P. L. (1993). Understanding of emotion. In M. Bennett (Ed.), The development of social cognition: The child as psychologist (pp. 62–86). New York: Guilford. Tisak, M. S. (1995). Domains of social reasoning and beyond. Annals of Child Development, 11, 95–130. Vitaro, F., Gendreau, P., Tremblay, R. E., & Oligny, P. (1998). Reactive and proactive aggression differentially predict later conduct problems. Journal of Child Psychology and Psychiatry and Allied Disciplines, 39, 377–385. Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social psychological experimentation. Personality and Social Psychology Bulletin, 25, 1115–1125.