kokko_2006_jra

Upload: mesondanes

Post on 04-Jun-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/13/2019 Kokko_2006_JRA

    1/26

    Trajectories of Prosocial Behavior andPhysical Aggression in Middle

    Childhood: Links to Adolescent SchoolDropout and Physical Violence

    Katja Kokko

    University of Jyvaskyla

    Richard E. Tremblay and Eric Lacourse

    University of Montreal

    Daniel S. NaginCarnegie Mellon University

    Frank Vitaro

    University of Montreal

    Trajectories of prosocial behavior and physical aggression between 6 and

    12 years of age were identified for a sample (N5

    1,025) of males. The tra-jectories were then used to predict school dropout and physical violence atage 17. Using a group-based semi-parametric method, two trajectories ofprosociality (low and moderate declining) and three trajectories of physicalaggression (low, moderate, and high declining) were obtained. Only a smallminority (3.4%) of the boys were characterized by both high aggression andmoderate prosociality. Physical aggression predicted both school dropoutand physical violence, but contrary to expectations, prosocial behavior didnot have additive or protective effects.

    JOURNAL OF RESEARCH ON ADOLESCENCE, 16(3), 403428Copyrightr 2006, Society for Research on Adolescence

    Requests for reprints should be sent to Katja Kokko, Department of Psychology, PO Box35, 40014 University of Jyvaskyla, Finland. E-mail: [email protected]

  • 8/13/2019 Kokko_2006_JRA

    2/26

  • 8/13/2019 Kokko_2006_JRA

    3/26

    aggressive and prosocial behaviors are independent individual charac-teristics, residing in the same individual. First, prosocial behavior has been

    studied as a protective factor against the possible harmful effects of ag-gression. For example, prosocial tendencies have been shown to buffer anaggressive child against peer rejection (Bierman, Smoot, & Aumiller, 1993;Nangle & Foster, 1992; Volling, MacKinnon-Lewis, Rabiner, & Baradaran,1993), criminal behavior (Pulkkinen & Tremblay, 1992), and long-termunemployment (Kokko & Pulkkinen, 2000a). There also exists some ev-idence that child prosocial behavior might be negatively linked to latercriminality, independently of aggression (Hamalainen & Pulkkinen, 1995,1996).

    Second, distinct correlates and etiologies of prosociality and aggressionhave been found. Krueger et al. (2001) show in their study of adult twinsthat altruism, a specific form of prosocial behavior, and antisocial behaviordid not correlate with each other, had different personality correlates, anddifferent heritability estimates. Antisocial behavior correlated with neg-ative emotionality and a lack of constraint, whereas prosciality was relatedto positive emotionality. Furthermore, antisocial behavior was linked togenetic factors and non-shared environments, whereas altruism was as-sociated with shared and non-shared environments (for a review of con-

    tradictory findings, see Hay, 1994). These findings are in line with theproposition that prosociality and aggression are independent behavioralstrategies, rather than representing opposite ends of the same personalitytrait (Pulkkinen, 1984).

    The present study had two main objectives: the first was to investigatethe links between the developmental trajectories of prosocial behaviorand physical aggression during middle childhood (612 years of age)for males. A group-based semi-parametric method was used for theidentification of developmental trajectories of these behaviors (Nagin,

    1999). The second was to investigate to what extent these two typesof developmental trajectories during middle childhood complementeach other when used to predict male problem behavior duringadolescence.

    The study of developmental trajectories has recently been applied to thedevelopment of physical aggression from childhood to adolescence.1

    1 The same sample as used in this investigation has been previously used for the study ofdevelopmental trajectories of aggression in several publications but their focus has been dif-

    ferent from the present attempt. The Brame et al. (2001) paper focused on the joint trajectories ofphysical aggression in childhood and late adolescence, that is, between ages 6 and 13 (note thatages 612 were analyzed here) and between ages 13 and 17. Nagin (1999) and Nagin andTremblay (1999, 2001a, 2001b) have analyzed the development of physical aggression from age6 to 15 and have linked its trajectories to the development of other types of externalizing

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 405

  • 8/13/2019 Kokko_2006_JRA

    4/26

    These studies have consistently shown, first, that meaningful subgroupsof individuals who follow distinctive developmental trajectories can be

    identified (Brame, Nagin, & Tremblay, 2001; Nagin, 1999; Nagin & Tremb-lay, 1999, 2001a)a finding that is replicated using several different lon-gitudinal studies conducted in different countries, as shown by Broidyet al. (2003). Using the same sample as in the present study but for aslightly different age range, Brame et al. found that at age 613 years, therewere three aggression trajectories: low, medium, and high. Second, thesestudies consistently show that about 510% of the individuals can becharacterized as chronically physically aggressive during middle child-hood. This means that their level of aggression at school entry is higher

    than that among the other children, and that they remain more aggressivethan the others throughout their childhood and adolescent years. Theresults from these analyses also indicate that the frequency of physicalaggression generally decreases with age, and that there is no significantsubgroup of children with an onset of chronic physical aggression afterschool entry. As discussed by Tremblay (2003), these results lead to asocialization theory of aggression that suggests that children do not learnto aggress physically, but rather learn alternatives to the physical aggres-sion they use spontaneously in early childhood. From this perspective,

    children who are on a chronic physical aggression trajectory are childrenwho have not learned to use alternative strategies to achieve their aims.

    Much less is known about the developmental trajectories of prosocialbehavior. Age differences in prosocial behavior have been studied mainlywith cross-sectional data. In their review of research on prosocial behavior,Eisenberg and Fabes (1998) concluded that there has been little consensuson whether or not there are age-related changes in the development ofprosocial tendencies (p. 744). Their meta-analysis of age differencesshowed that as children grow older, they usually display more prosocial

    behavioralthough this link seems to be affected by the specific agecomparisons made, the way prosocial behavior is investigated (e.g.,whether by self- or other reports or observations, and which type of pro-social behavior is examined), and the type of analysis of the age differ-ences used. One would expect that with increasing cognitive capacities,continued emotional development, and an increasing willingness andability to empathize with other childrens problems, combined with a

    behavior problems (including later physical violence) or to different types of predictor variables.The Broidy et al.s (2003) multisite study also used trajectories of physical aggression based onthe 615-year-old Montreal boys. The main interest of the present study was in links between thedevelopment of aggression and prosociality and in the possible compensatory or protectiveeffects of prosocial behavior.

    406 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    5/26

    widening social environment, children will likely develop higher levels ofprosocial behavior (e.g., Carlo et al., 2003). Furthermore, the theory of

    moral development by Kohlberg, Levine, and Hewer (1983) postulatesthat children progress through different (and gradually more complex)stages of moral reasoning as they mature.

    A few recent studies, based on preschoolers and school-aged children,support the notion of complicated relations between different types ofprosocial behavior over time. For example, Hay et al. (1999) showed that,from age 18 months to 3 years, spontaneous sharing behavior decreasedwhereas sharing by request increased. Furthermore, Jackson and Tisak(2001) found in their cross-sectional study of 712-year-old children that

    younger children showed more sharing than older children, whereas bothcooperating and comforting behaviors had more curvilinear associationsover this time period. In helping behavior, there were no differences be-tween the age groups in this cross-sectional study. Eisenberg et al. (1999),on the development of prosociality from age 4 to 5 years into youngadulthood, report complex associations over time: early spontaneoussharing behavior was related to young adult prosocial dispositions viasympathy, but early prosociality, not reflecting an other orientation, didnot show rank-order continuity into young adult behavior. She and her

    colleagues argue that generally prosocial dispositions show substantialrank-order continuity from childhood to adulthood (Eisenberg et al.,2002). However, as highlighted by Fabes et al. (1999), still more informa-tion is needed about prosocial development, particularly in early adoles-cence. Especially important would be to study longitudinally thedevelopmental trajectories of prosocial behavior. The rank-order continu-ity cannot reveal to what extent and at which period in time the frequencyof behavior is increasing or decreasing.

    To our knowledge, only one longitudinal study has attempted to trace

    the developmental trajectories of prosocial behavior. Cote, Tremblay,Nagin, Zoccolillo, and Vitaro (2002), using a large representative sampleof over 1,800 kindergarten children in Canada, reported that, for bothboys and girls, prosocial behavior was stable from 6 to 12 years ofage. About one half of the participants (54% of the boys and 58% of thegirls) had a stable moderate level of prosocial behavior. About 2% of theboys and 27% of the girls followed a stable high trajectory, and about44% and 15%, respectively, followed a low trajectory of the behavior inquestion.

    Sparse data on the developmental trajectories of prosocial behavior, aswell as the lack of prosocial and antisocial behavior assessments on thesame individuals over several childhood years, have precluded thestudy of developmental links between prosocial behavior and antisocial

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 407

  • 8/13/2019 Kokko_2006_JRA

    6/26

    behavior. From a socialization theory perspective (e.g., Eisenberg & Fabes,1998; Tremblay, 2003), it is expected that, with age, most children will

    decrease the frequency of their physical aggressions and increase thefrequency of their prosocial behavior. There is good evidence that physicalaggression declines with age, and there are theoretical indications, butno clear empirical data, suggesting that prosocial behavior increases. Theaim of this study was to investigate the developmental links betweenphysical aggression and prosocial behavior in a large sample of maleparticipants. As discussed above, there are many claims that antisocialand prosocial tendencies may reside in the same individual. Consequent-ly, one would expect to find a group of children with both frequent

    prosocial behavior and physically aggressive behavior. For preventiveand intervention purposes, it is important to identify this subgroupof children, because prosociality may protect them against latermaladjustment.

    We also examined whether the trajectories for physical aggression andprosociality contributed to the prediction of adjustment problems duringadolescence. Adjustment problems were measured by physical violenceand school dropout at age 17. There exists an extensive body of literatureshowing that early aggressive behavior is strongly associated with later

    criminal behavior and deviant peer relations (e.g., Cairns & Cairns, 1994;Coie & Dodge, 1998; Loeber & Hay, 1997; Rutter, Giller, & Hagell, 1998).Nagin and Tremblay (1999) have previously shown that trajectories ofphysical aggression, based on assessments between ages 6 and 15, arelinked to juvenile delinquency, that is, physical violence at age 17. Boyswho followed a trajectory of chronic physical aggression were signifi-cantly more likely to report episodes of physical violence in adolescencethan boys on the other trajectories. In this study, we investigated whetherprosocial behavior protected some of these chronically aggressive boys

    against later violence. It was assumed that the level of violence mightdecrease as a function of accumulated positive experiences associatedwith increased prosocial behavior (Kokko & Pulkkinen, 2000b).

    Furthermore, it has also been shown that child and adolescent aggres-sion is associated with poor school achievement (e.g., Bergman & Mag-nusson, 1997; Brook & Newcomb, 1995; Caspi, Elder, & Bem, 1987; Kokko& Pulkkinen, 2000a) and school dropout (e.g., Cairns, Cairns, & Necker-man, 1989; Ensminger & Slusarcick, 1992; Fergusson & Horwood, 1998),which are further related to later problems in the domain of work such as

    unemployment (Caspi, Wright, Moffitt, & Silva, 1998; Fergusson& Horwood, 1998; Kokko & Pulkkinen, 2000a). However, prosocial ten-dencies buffered highly aggressive children against long-term unemploy-ment in adulthood (Kokko & Pulkkinen, 2000a). In the present study, we

    408 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    7/26

    examined whether the protective effects of prosociality could already beseen in the rates of school dropout among the aggressive boys.

    The protective effects of prosociality were tested as compensatory ad-ditive effects and protective effects. Bycompensatory effects,it is meant thatthe risk of a negative effect (physical aggression) is reduced by a positiveeffect (prosocial behavior; Fergusson, Beautrais, & Horwood, 2003) in theentire sample. Byprotective effects,it is meant that favorable influences of aprotective factor (prosocial behavior) exercise their influence only on high-risk individuals (the aggressive; Rutter, 1987). The compensatory effectcorresponds to a main effect opposite to a risk factor, whereas a protectiveeffect corresponds to an interaction between the protective factor and a

    risk factor. In summary, the present study of a large sample of boys at-tempted for the first time to identify the links between the developmentaltrajectories of prosocial behavior and physical aggression during middlechildhood, and to identify to what extent these two types of develop-mental trajectories complement each other when used to predict problembehavior during adolescence.

    METHOD

    Participants

    Participants were drawn from the ongoing Montreal Longitudinal and Ex-perimental Study. This sample consists of boys who were originally selectedfrom French-speaking low socioeconomic areas of Montreal and who werefirst assessed in 1984 at the age of 6 (mean age 6.2 years, SD5 .3) by theirkindergarten teachers. Eighty-seven percent of the teachers responded,yielding information on 1,161 boys from 53 schools. In order to control forethnic background, those boys whose parent(s) were not born in Canada or

    did not speak French as a mother tongue were excluded from the sample.After taking these criteria into account, as well as those who refused toparticipate, the final sample of white, homogeneous French-speaking boystotaled 1,037. For further information, see Nagin and Tremblay (1999).

    At age 6 (in 1984), about 69% of the 1,025 boys lived with bothparents, whereas the rest lived either in a nonintact family (25%) or in areconstructed family (6%). The mean age of the mothers at the time of thebirth of the first child was 23.3 years (SD5 4.1), and the respective meanage of the fathers was 26.4 years (SD55.0). On average, the mothers had

    completed 10.5 years of schooling (SD52.8) and fathers 10.6 years(SD5 3.3). On the Canadian socioeconomic index for occupations (Blis-hen, Carroll, & Moore, 1987), the mean score was 38.2 (SD5 12.1) formothers and 39.4 (SD512.9) for fathers.

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 409

  • 8/13/2019 Kokko_2006_JRA

    8/26

    In the present study, information yielded by teacher-ratings from ages6 to 12 and self-ratings as well as registers (school-dropout) at age 17 was

    used. Data on physical aggression and prosociality were available as fol-lows: 94% of the age 6 sample at age 10; 91% at age 11; and 85% at age 12.For the trajectory analyses of physically aggressive and prosocial behav-iors, we selected those boys who were assessed at least twice over theperiod from age 6 to 12, yielding a sample size of 1,025. Thus, 12 boys wereexcluded from the trajectory analyses. These boys did not differ from the1,025 boys included in the analyses in terms of social behavior at age 6 orfamily characteristics. At age 17, 76% of the boys filled out self-reports andinformation about school dropout was available for 96% of the boys.

    Measures

    Trajectories of physical aggression and prosocial behavior from age6 to 12. Physical aggression and prosocial behavior were assessed by theteacher most knowledgeable about the child at ages 6 and 10, 11, and12 with the Social Behavior Questionnaire (SBQ; Tremblay et al., 1991).The SBQ consists of 38 items that have been subjected to a factor

    analysis, producing four broader factors: disruptive behavior, anxiety,inattentiveness, and prosocial behavior. A subscale for disruptivenessincludes three items for physical aggression. In the present study, we usedinformation about physical aggression and prosociality.

    Physical aggressionwas a sum score of the following three items: Kicks,bites, hits other children; fights with other children; and bullies orintimidates other children. Teachers were asked to rate each child on ascale from 0 (does not apply) to 2 (certainly applies). The reliabilitiesranged from .79 to .88. For further information on this aggression scale, see

    Tremblay et al. (1991), Nagin and Tremblay (1999), and Brame et al. (2001).Prosocial behavior was a sum score of the following ten items: If there is a

    quarrel or dispute will try to stop it; will invite bystanders to join in agame; will try to help someone who has been hurt; spontaneouslyhelps to pick up objects which another child has dropped (e.g., pencils,books, etc.); takes the opportunity to praise the work of less ablechildren; shows sympathy to someone who has made a mistake; of-fers to help other children who are having difficulty with a task in theclassroom; helps other children who are feeling sick; comforts a child

    who is crying or upset; and volunteers to help clear up a mess someoneelse has made. Teachers rated the children using a scale from 0 to 2. Thereliabilities (Cronbachsas) of the sum scores ranged from .87 to .92. Forfurther information about this scale, see Tremblay et al. (1992).

    410 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    9/26

    Measures of problem behaviors at age 17. We followed Vitaro,Larocque, Janosz, and Tremblays (2001) definition ofschool dropout: A

    dropout is an individual who stops attending school before high schoolgraduation, whether or not he or she reenters at a later time (p. 407).Information about school dropout was collected from the computerizedlists of the Montreal school board and the Ministry of Education and theparticipants themselves. If a participant was not listed on the annualSchool board list, the Ministry of Education was contacted in order to findout whether the participant had moved to another school board within theProvince of Quebec. The participants themselves were also asked toconfirm whether they had left school during the annual data collection

    period. At 17, 16.4% of the boys had dropped out of school.Physical violence. Physically violent behaviorwas assessed using seven

    self-reported items, on a scale from 1 (never) to 4 (often) (Nagin &Tremblay, 1999). These items were as follows: fist fighting, gang fighting,carrying or using a deadly weapon, threatening or attacking someone, andthrowing an object at someone. The reliability (Cronbachsa) of the sumscore was .78.

    The sum score for physical violence was dichotomized, using a 75thpercentile cut-off (05no behavior in question, 15yes behavior in ques-

    tion), for three reasons: first, we were interested in the frequencies of thehighly physically violent individuals among different trajectory groups.Second, the associations between antecedents (trajectories) and outcomeswere indicated as odds ratios (OR). As argued by Farrington and Loeber(2000), the use of OR results in a more understandable interpretation of theobserved associations and, in some cases, a more realistic interpretationthan the use of more conventional statistics allows for (e.g., productmo-ment correlation). Third, the distribution of the sum score for physicalviolence indicated that there were a large number of participants (55.4%)

    who never showed any indication of violence. Consequently, even logtransformations of the sum score would have failed to transform it intonormal distribution. Farrington and Loeber also provide some evidencefor the one-quarter/three-quarters split as an acceptable way of the di-chotomization.

    Data Analysis

    Descriptive statistics were calculated using Pearsons productmoment

    correlations for aggression and prosociality items at different ages and forthe associations between aggression and prosociality. Developmentaltrajectories of physical aggression and prosocial behavior were estimatedusing a group-based semi-parametric method (Jones, Nagin, & Roeder,

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 411

  • 8/13/2019 Kokko_2006_JRA

    10/26

    2001; Nagin, 1999; Nagin & Tremblay, 2001a). With this method it is pos-sible to (a) identify distinct subgroups among the study population; (b)

    estimate the proportion of the study population following each trajectory;and (c) assign individuals to this trajectory group, where they most likelybelong.

    The software used to estimate trajectories was an SAS-based procedure(Jones et al., 2001). Because the data used to estimate the trajectories werebased on psychometric scales with censoring at the scale minimum andmaximum, trajectory estimation used the censored normal option in theSAS procedure for model estimation (Nagin, 1999).

    Determination of the optimal number of groups and trajectory shapes

    was guided by the Bayesian information criterion (BIC), which is calcu-lated in the following way:

    BIC logL 0:5 logn k;

    whereLis the models maximized likelihood;nis the sample size; andkisthe number of the models parameters. It is recommended that the modelwith the maximum BIC is selected. Different trajectory groups can followdifferentordersof the polynomial: a second-order trajectory is based on aquadratic equation; a first-order trajectory follows a linear equation in

    whichb2is equal to zero; and a zero-order trajectory is indicated by a flatline in which both b1and b2are equal to zero (Nagin, 1999).

    The links between the trajectories of physical aggression and prosocial-ity were analyzed using cross-tabulation and between the trajectories andlater dichotomous problem behaviors using logistic regression analysis. Ineach case, physical aggression and prosociality trajectories were first en-tered, and in the next step an interaction effect between them was added.Specifically planned contrasts were also used in order to better under-stand the interactive effects between physical aggression and prosociality.

    That is, the developmental outcomes of highly aggressive boys who werehigh versus low in prosociality were compared.

    RESULTS

    Descriptive Results

    As shown by mean values in Table 1, there was a general trend for both

    aggression and prosocial behavior to decrease with increasing age. Fur-thermore, the table indicates that the correlations among the aggressionscales over time (r5.22.51) were higher than among the prosocialityscales (r5.16.23). Finally, the correlations between the aggression and

    412 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    11/26

    prosocial scales were negative, but generally low, both concurrently

    (r5 .07 to .18) and across time (r5 .05 to .18).

    Trajectories of Physical Aggression and Prosocial Behavior from Age 6to 12

    On the basis of the maximized BIC (and previous results in the case ofphysical aggression; see Brame et al., 2001), a three-group model wasselected for physical aggression, whereas a two-group model was chosen

    for prosocial behavior as the best-fitting models. Table 2 shows the meangroup assignment probabilities, conditional on assignment by the max-imum likelihood probability rule for each physical aggression group andeach prosociality group. The maximum likelihood probability rule impliesthat individuals on a given trajectory should have a higher mean prob-ability of assignment to this group compared with other groups. Allprobabilities were higher than .80, a value that is generally assumed toimply a good model fit to the data (Nagin, 2005).

    As indicated in Figure 1, the trajectories of physical aggression from age

    6 to 12 (and the estimation of the proportion of the population followingeach trajectory) were as follows: (1) low (35.5%), (2) moderate (46.2%), and(3) high declining (18.3%). The low group followed a trajectory reflecting agradual linear decline from low physical aggression at age 6 to minimal

    TABLE 1

    Descriptives and Pearsons ProductMoment Correlations among Aggression and Prosocial

    Behavior at Ages 6 and 1012

    Variable 1. 2. 3. 4. 5. 6. 7. 8. M SD

    1. Aggression (6) 1.41 1.76

    2. Aggression (10) .37nnn 1.37 1.75

    3. Aggression (11) .34nnn .51nnn 1.18 1.62

    4. Aggression (12) .22nnn .41nnn .48nnn .76 1.24

    5. Prosociality (6) .16nnn .17nnn .17nnn .15nnn 8.00 4.96

    6. Prosociality (10) .18nnn .18nnn .08n .07n .23nnn 6.92 4.89

    7. Prosociality (11) .06 .06 .09nn

    .06 .20nnn

    .22nnn

    6.41 4.658. Prosociality (12) .07n .05 .08n .07n .16nnn .19nnn .20nnn 5.65 4.19

    Note. nnnpo.001; nnpo.01; npo.05.

    Numbers in parentheses indicate participants age in years at time of measurement. Number

    of participants varied from 849 to 1,025 depending on the data available. Scale for Aggression

    from 0 to 6 and for Prosociality from 0 to 20.

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 413

  • 8/13/2019 Kokko_2006_JRA

    12/26

    aggression by age 12. The moderate and high declining groups had quad-ratic trajectories, indicating that there was a slight increase in the rate ofthe decline in their physical aggressiveness as they aged. Figure 2 showsthat the two trajectories of prosocial behavior (and the estimated propor-tions) were as follows: (1) low declining (57.6%) and (2) moderate declin-

    ing (42.4%). Both trajectories declined gradually from age 6 to 12. For thelow group, this trend was captured by a linear trajectory whereas for thehigh group the trend was quadratic, which reflects a small acceleration inthe rate of decline.

    Co-occurence of Physical Aggression and Prosociality

    Cross-tabulation of physical aggression and prosociality indicated thatboys low in aggression were evenly distributed to the low (50.4%) and

    moderate (49.6%) groups of prosociality (Table 3). However, as the level ofaggression increased, the level of prosocial behavior decreased: 62.8% ofboys moderate in physical aggression were also low in prosociality and79.3% of boys who followed high trajectory in physical aggression showeda low level of prosociality. There was a small subgroup of boys (n5 35)who, despite their severe physical aggression throughout middle child-hood, also showed a moderate level of prosocial behavior during the sametime period.

    Links between Trajectories and Problem Behaviors at Age 17

    Academic attainment. Using logistic regression analysis, we firststudied whether the trajectories of physical aggression and prosocial

    TABLE 2

    Mean Group Average Physical Aggression and Prosociality Group Assignment Probabilities,

    Conditional on Assignment by Maximum Probability Rule

    Mean SD

    Trajectory groups

    Physical aggression

    1. Low .83 .14

    2. Moderate .80 .13

    3. High declining .83 .15

    Prosociality

    1. Low declining .83 .152. Moderate declining .81 .15

    414 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    13/26

    behavior in middle childhood were linked to school dropout at age 17.Aggression and prosociality were entered first in the model. The results

    showed that aggression was significantly related to later school dropout,w2(2, N5984)544.35, p5 .000, whereas prosociality was not, w2(1,N5984)50.02, p5 .88. Being on the high declining trajectory ofphysical aggression, compared with the trajectory of low physicalaggression, increased the risk of school dropout by a factor of 5.86 (95%confidence interval [CI], 3.489.86). Also, being on the moderateaggression trajectory increased the likelihood of dropping out of schoolcompared with low aggression by a factor of 2.78 (95% CI, 1.774.39).There was also an increased risk of school dropout among the highly

    aggressive compared with the moderately aggressive (OR52.10; 95% CI,1.393.18). Interaction between the trajectories of physical aggressionand prosociality was then added to the model and was found tobe nonsignificant (Table 4). Similarly, the contrast between highly

    0

    1

    2

    3

    4

    5

    6 10 11 12

    Age

    PhysicalAggression

    Low-actual Moderate-actual High declining-actual

    Low-pred-35.5% Moderate-pred-46.2% High declining-pred-18.3%

    FIGURE1 Trajectories of physical aggression from age 612. The solid lines representactual behavior and the dashed lines represent predicted behaviors (for their respectivecomputation, see Nagin, 1999, p. 3).

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 415

  • 8/13/2019 Kokko_2006_JRA

    14/26

    aggressive children with and without prosociality was not significant(Table 4).

    Physical violence. We again began by testing the effects of physicalaggression and prosocial behavior on physical violence at age 17 and thenadded an interaction to the model. Only aggression was significantlyrelated to physical violence, w2(2, N5 790)525.04, p5 .000; prosocialityhad no significant association w2(1, N5790)50.05, p5 .82. High decliningaggression increased the risk of being physically violent compared with alow level of aggression by a factor of 3.21 (95% CI, 1.975.22), whereas therespective value for moderate aggression was 2.04 (95% CI, 1.412.94).

    01

    23

    456

    78

    910

    1112

    6 10 11 12

    Age

    ProsocialBehavior

    Low declining-actual and low declining-pred-57.6%

    Moderate declining-actual and moderate declining-pred-42.4%

    FIGURE2 Trajectories of prosocial behavior from age 612.

    TABLE 3

    Cross-Tabulation of Trajectories of Physical Aggression and Prosocial Behavior (Ages 612)

    Physical Aggression

    Prosocial Behavior

    1. Low Declining 2. Moderate Declining Total

    1. Low 193 (50.4) 190 (49.6) 383 (100)

    2. Moderate 297 (62.8) 176 (37.2) 473 (100)3. High declining 134 (79.3) 35 (20.7) 169 (100)

    Note.w2(2,N51,025)542.47,po.001.

    Number of participants (%) in each cell indicated.

    416 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    15/26

  • 8/13/2019 Kokko_2006_JRA

    16/26

    DISCUSSION

    The goals of the present study were, first, to test, with developmentaltrajectories between 6 and 12 years of age, to what extent socializationwould both decrease the frequency of physical aggression and increase thefrequency of prosocial behavior; second, to investigate the links betweenthe developmental trajectories of physical aggression and prosocial be-haviors from age 6 to 12 among prospectively followed boys and; third, tostudy whether trajectories of physical aggression and prosocial behavior

    during the elementary school years predicted school dropout and physicalviolence at 17 years of age, with a special focus on those children who werehigh in aggression but differed in terms of prosocial manifestations.

    In line with previous research (Brame et al., 2001; Broidy et al., 2003), wefound that there were different developmental subgroups of physical ag-gression from age 6 to 12. Low and moderate levels of aggressive behaviorshowed stability throughout middle childhood, whereas a high initiallevel of aggression tended to decrease over the childhood years. Broidyet al. (2003) also identified three trajectories of physical aggression in three

    TABLE 5

    Summary of Comparisons between Trajectories for Physical Aggression and Prosocial Be-

    havior and Contrasts between them in Predicting Problem Behaviors at Age 17 (minimumn5790, maximumn5984): Odds Ratios (95% Confidence Intervals)

    Trajectory groups

    Variable

    School dropout Physical violence

    High declining versus low

    physical aggression (PA)

    5.86 (3.489.86)nnn 3.21 (1.975.22)nnn

    High declining versus

    moderate PA

    2.10 (1.393.18)nnn 1.57 (1.012.46)n

    Moderate versus Low PA 2.78 (1.774.39)nnn 2.04 (1.412.94)nnn

    Moderate declining versus

    low prosociality (PRO)

    1.03 (.711.49) 1.04 (.741.45)

    High PA, low declining PRO

    versus high declining PA,

    moderate declining PRO

    1.12 (.492.58) .64 (.251.64)

    Note. nnnpo.001; npo.05.

    violence, all the logistic regression analyses were re-run controlling for the low maternal ed-ucation. The findings remained the same, except for the insignificant changes in the estimates.These results can be obtained from the first author on request.

    418 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    17/26

  • 8/13/2019 Kokko_2006_JRA

    18/26

    moderate prosocial trajectory were also on a low physical aggression tra-jectory. Thus, the cross-tabulation of the prosocial behavior and physical

    aggression trajectories during the elementary school years for males fromlow socioeconomic areas support the notion of physical aggression andprosociality being relatively independent types of behavior, that is, notonly the opposite ends of the same continuum (Feshbach & Feshbach,1986; Krueger et al., 2001; Pulkkinen, 1984). When aggression and proso-cial behavior are considered relatively independent types of behaviors, itis reasonable to assume that the same individual can show different typesof responses in different situations. It has actually been shown that so-called bistrategic adolescents (i.e., who manifest both prosocial and co-

    ercive behaviors) may be quite effective in, for example, their peer contexts(Hawley, 2003). Still, individuals may show some general patterns of be-haviors across aggregated situations and over time.

    The prediction of problem behaviors at 17 years of age using the middlechildhood trajectories of physical aggression and prosocial behavior con-firmed that problems in elementary school predict problems at the end ofhigh school, but our main focus was on the additive and interactive effectsof these two behavioral trajectories. Concerning additive effects, trajecto-ries of procial behavior did not add to the predictive power of physical

    aggression for school dropout or physical violence. Neither did it protectaggressive children against later problem behavior.

    It was shown thatschool dropoutat age 17 was strongly related to beingon a high trajectory of aggression: the highly aggressive boys were aboutsix times more likely to drop out of school compared with the non-ag-gressive boys. Prosociality did not contribute to school problems; neitherdid it have interactive effects with aggression. These findings are con-cordant with results from previous studies that show that early aggressionis linked to subsequent problems in the domain of academic achievement

    (e.g., Bergman & Magnusson, 1997; Brook & Newcomb, 1995; Caspi et al.,1987; Fergusson & Horwood, 1998; Kokko & Pulkkinen, 2000a) and evenschool dropout (e.g., Cairns et al., 1989; Ensminger & Slusarcick, 1992).

    However, they differ from the only other study, which contrastedphysical aggression and prosocial behavior as predictors of schoolachievement. With an Italian sample, Caprara, Barbaranelli, Pastorelli,Bandura, and Zimbardo (2000) observed that it was prosocial behavior atage 89, instead of aggression that positively predicted school achieve-ment at age 1314, even after controlling for the earlier level of school

    achievement. The difference with the results of the latter study could bedue to differences in samples and content of variables. First, whereas theMontreal sample included only boys from families with low education,the Italian sample included both boys and girls from families with various

    420 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    19/26

    educational backgrounds. Second, the present study was focused onschool dropout while the Italian study was focused on variation in school

    grades. Physical violence at age 17, like school dropout, was predictedsolely by earlier physical aggression. Numerous previous studies reportedthe associations between early aggression and later violence and delin-quency (e.g., Coie & Dodge, 1998; Loeber & Hay, 1997; Nagin & Tremblay,2001a; Rutter et al., 1998). The link between physical aggression in child-hood and physical violence in adolescence indicates that there exists astrong, homotypic continuity (Caspi & Roberts, 1999) in this type of overtproblem behavior.

    The fact that prosociality did not protect against the later harmful ef-

    fects of physical aggression might be explained by the age of the presentparticipants. Both school dropout and physical violence were measured atage 17. It has been argued by Arnett (2000) that in western countries, thelate teenage years are part of a distinct life phase, called emerging adult-hood, which is neither adolescence nor adulthood. This period of life ischaracterized by many changes (e.g., graduation from school, movingfrom home), explorations of possible life directions and identities, andrelative independence from normative social roles. It is common thatemergent adults have not yet settled down either to work or family life.

    The late teens of the present study were in this turbulent life phase and it ispossible that individual characteristics, such as prosocial tendencies, didnot play as much of a role in their social behavior as is expected in later lifephases.

    In fact, previous studies show that childhood personality characteris-tics, especially aggression, do not explain long-term unemployment at age27 (Kokko, Bergman, & Pulkkinen, 2003), but do contribute to it 10 yearslater when the participants lives were more independent of the devel-opmental transitions and explorations of early adulthood (Kokko &

    Pulkkinen, 2000a). At age 36, childhood aggression significantly explainedserious difficulties in the domain of work, but among some initiallyaggressive individuals, childhood prosociality protected them againstlong-term unemployment. This might indicate that the protective effectsof prosociality for some outcomes are long term, rather than short term. Itneeds to be seen whether the positive effects of prosocial tendencies areobservable at later ages among these followed Montreal boys.

    Another potential explanation for the lack of compensatory or protec-tive effects of prosocial behavior might be related to methodological is-

    sues: the present sample and the present context of the prosocialassessment. These Montreal boys were originally from low SES family-environment and it is possible that their prosocial behavior did not helpthem because of the contextual factors (i.e., it may not even be valued). In a

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 421

  • 8/13/2019 Kokko_2006_JRA

    20/26

    different context, a positive compensatory or protective effect might befound. Eisenberg et al. (2002) argue that even less stability of prosociality

    can be expected in children growing up in low SES families because of theinstability of the life circumstancesalthough this is not true in the Cap-rara et al. (2001) study with middle-class children. It may also be that thelevel of prosocial behavior achieved by the boys from this low SES samplewas not high enough to protect them against the harmful consequences ofusing physical aggression. This interpretation is supported by the absenceof the high trajectory of prosociality found by Coteet al. (2002). Further-more, teacher assessments of prosocial behavior were used in this study.Teachers see the students in structured settings (i.e., the classroom), which

    may not reflect the levels of prosocial behavior in less structured settings,such as peer groups, within and outside of school. These methodologicalissues need to be taken into account when making any interpretations ofthe present findings for prevention and intervention efforts.

    In addition to the issue of a homogeneous sample drawn from a lowSES background, there are two other issues that might restrict the gen-eralizability of the present results: first, we studied only boys in thepresent study. Although the importance of studying prosociality amongboys has been stressed (Hay, 1994), it would be interesting to examine

    whether the present findings would replicate among girls. It also needs tobe seen whether our findings will apply to participants from families withdifferent cultural backgrounds. Second, our measure of prosocial behaviorwas related to overt manifestations of this characteristic, such as helpingbehavior. In addition to the behavioral (helping) dimension, for instance,Hoffman (2001) conceptualizes prosociality to include cognitive (moralreasoning) and emotional/motivational (empathy, guilt) aspects. It wouldbe important to study whether these aspects of prosociality show similarage-related trends and how they relate to contemporaneous physical ag-

    gression and subsequent social behavior. There are some studies that im-ply that different types of prosociality show different patterns over time(e.g., Eisenberg et al., 1987; Fabes et al., 1999; Hastings et al., 2000; Hay,1994; Jackson & Tisak, 2001; Keltikangas-Jarvinen, 2001), depending alsoon the informant used (Caprara et al., 2001).

    As described by Tremblay (2000), measures of aggressive behavioroften aggregate indices of aggression, antisocial behavior, and conductdisorder. Further measures of direct and indirect aggression as wellas measures of reactive and proactive aggression are sometimes

    combined. Regarding direct physical aggression, the differentiation ofreactive and proactive aggression seems to be crucial. Pulkkinen (1996)and Vitaro, Gendreau, Tremblay, and Oligny (1998) showed thatproactively aggressive children were more likely to manifest antisocial

    422 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    21/26

    behavior later than nonaggressive and reactively aggressive children.We focused on physical aggression in the present study. In the future,

    it would be useful to study the developmental trajectories of differentforms of aggression (reactive and proactive physical aggression, andindirect aggression) and their links to different forms of prosociality,such as helping, moral reasoning, and emotional aspects of prosocialbehavior. Furthermore, it would be important to investigate the comple-mented effects of physical aggression and prosociality on outcome meas-ures other than school dropout and violence for women and men atdifferent ages.

    There were several strengths in the present attempt to understand the

    development of physical aggression and prosocial behavior in middlechildhood and their links to later problem behavior. First of all, we studiedaggression and prosociality at the same time. Several researchers have ur-gently advocated investigating their joint effects. Second, we used a group-based semi-parametric mixture method (Nagin, 1999), which is particularlywell suited for analyzing developmental trajectories of the behaviors inquestion. Using this method, it was possible to identify groups that differedfrom each other in both the level of behavior and the shape of the devel-opmental trend of this behavior. In addition to constant and linear devel-

    opmental trends, non-linear trajectories were found. By examining thedevelopmental trends, we were able to identify children who were low,moderate, or high (in physical aggression only) in aggression and proso-ciality throughout the childhood years. We assume that the classificationbased on developmental trajectories yielded more reliable information aboutthe childs behavioral tendencies than just relying on the classificationsbased on single time-point measures and somewhat arbitrary cut-off points.

    ACKNOWLEDGMENTS

    This research has been supported by research grants from the Academy ofFinland (Nr. 54489 and Nr. 55289), the Fonds pour la Formation deChercheurs et lAide ala Recherche (FCAR-Centre), the Conseil Quebecoisde la Recherche Sociale (CQRS), the Fonds de la Recherche en Sante duQuebec (FRSQ), the Social Sciences and Humanities Research Council ofCanada (SSHRC), the Molson Foundation, the Canadian Institute of Ad-vanced Research, and the US National Institute of Mental Health (RO1

    MH65611-01A2). We thank Danielle Forest for her help with the statisticalanalyses, Helene Beauchesne for coordinating the data collection, MurielRorive for coordinating the data management, and Lyse Desmarais-Gervais for general coordination of the research activities.

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 423

  • 8/13/2019 Kokko_2006_JRA

    22/26

  • 8/13/2019 Kokko_2006_JRA

    23/26

  • 8/13/2019 Kokko_2006_JRA

    24/26

    Hay, D. F. (1994). Prosocial development.Journal of Child Psychology and Psychiatry,35, 2971.Hay, D. F., Castle, J., Davies, L., Demetriou, H., & Stimson, C. A. (1999). Prosocial action in

    very early childhood.Journal of Child Psychology and Psychiatry,40, 905916.Hoffman, M. L. (2001). Toward a comprehensive empathy-based theory of prosocial moral

    development. In A. C. Bohart, & D. J. Stipek (Eds.), Constructive & destructive behavior:Implications for family, school, & society(pp. 6186). Washington, D.C.: American Psycho-logical Association.

    Howes, C., & Phillipsen, L. (1998). Continuity in childrens relations with peers. Social De-velopment,7, 340349.

    Hughes, C., & Dunn, J. (2000). Hedonism or empathy?: Hard-to-manage childrens moralawareness and links with cognitive and maternal characteristics. British Journal of Devel-opmental Psychology,18, 227245.

    Jackson, M., & Tisak, M. S. (2001). Is prosocial behaviour a good thing? Developmental

    changes in childrens evaluations of helping, sharing, cooperation, and comforting. BritishJournal of Developmental Psychology,19, 349367.

    Jones, B. L., Nagin, D. S., & Roeder, K. (2001). A SAS procedure based on mixture models forestimating developmental trajectories.Sociological Methods and Research,29, 374393.

    Keltikangas-Jarvinen, L. (2001). Aggressive behaviour and social problem-solving strategies:A review of the findings of a seven-year follow-up from childhood to late adolescence.Criminal Behaviour and Mental Health,11, 236250.

    Kohlberg, L., Levine, C., & Hewer, A. (1983).Moral stages: A current formulation and response tocritics. Basel, Switzerland: Karger.

    Kokko, K., Bergman, L. R., & Pulkkinen, L. (2003). Child personality characteristics andselection into long-term unemployment in Finnish and Swedish longitudinal samples.

    International Journal of Behavioral Development,27, 134144.Kokko, K., & Pulkkinen, L. (2000a). Aggression in childhood and long-term unemployment in

    adulthood: A cycle of maladaptation and some protective factors. Developmental Psychol-ogy,36, 463472.

    Kokko, K., & Pulkkinen, L. (2000b). Breaking the cycle of maladaptation. Prevention andTreatment, 3, Article 36. Available on the World Wide Web: http://journals.apa.org/prevention/volume3/pre0030036r.html.

    Krueger, R. F., Hicks, B. M., & McGue, M. (2001). Altruism and antisocial behavior: Inde-pendent tendencies, unique personality correlates, distinct etiologies. Psychological Sci-ence,12, 397402.

    Loeber, R., & Hay, D. (1997). Key issues in the development of aggression and violence fromchildhood to early adulthood.Annual Review of Psychology,48, 371410.

    Nagin, D. S. (1999). Analyzing developmental trajectories: A semi-parametric, group-basedapproach.Psychological Methods,4, 139177.

    Nagin, D. S. (2005). Group-based modeling of development. Cambridge, MA: Harvard UniversityPress.

    Nagin, D. S., & Tremblay, R. E. (1999). Trajectories of boys physical aggression, opposition,and hyperactivity on the path to physically violent and nonviolent juvenile delinquency.Child Development,70, 11811196.

    Nagin, D. S., & Tremblay, R. E. (2001a). Analyzing developmental trajectories of distinct butrelated behaviors: A group-based method.Psychological Methods,6, 1834.

    Nagin, D. S., & Tremblay, R. E. (2001b). Parental and early childhood predictors of persistentphysical aggression in boys from kindergarten to high school. Archives of General Psychi-atry,58, 389394.

    Nangle, D. W., & Foster, S. L. (1992). The effects of a positive behavioral context on the socialimpact of aggressive behavior. Journal of Abnormal Child Psychology,20, 543553.

    426 KOKKO, TREMBLAY, LACOURSE, NAGIN, AND VITARO

  • 8/13/2019 Kokko_2006_JRA

    25/26

    Pitkanen, L. (1969).A descriptive model of aggression and nonaggression with applications to chil-drens behaviour (Jyvaskyla Studies in Education, Psychology and Social Research, Nr. 19).

    Jyvaskyla, Finland: JyvaskylaUniversity Printing House.Pulkkinen, L. (1984). Discussion: The inhibition and control of aggression.Aggressive Behavior,

    10, 221225.Pulkkinen, L. (1996). Proactive and reactive aggression in early adolescence as precursors to

    anti- and prosocial behavior in young adults.Aggressive Behavior,22, 241257.Pulkkinen, L., & Tremblay, R. E. (1992). Patterns of boys social adjustment in two cultures and

    at different ages: A longitudinal perspective. International Journal of Behavioral Develop-ment,15, 527553.

    Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American Journal ofOrthopsychiatry,57, 316331.

    Rutter, M., Giller, H., & Hagell, A. (1998). Antisocial behavior by young people. Cambridge, UK:

    Cambridge University Press.Scott, S., Knapp, M., Henderson, J., & Maughan, B. (2001). Financial cost of social exclusion:

    Follow up study of antisocial children into adulthood. British Medical Journal, 323,191194.

    Serbin, L. A., Cooperman, J. M., Peters, P. L., Lehoux, P. M., Stack, D. M., & Schwartzman, A. E.(1998). Intergenerational transfer of psychosocial risk in women with childhood historiesof aggression, withdrawal, or aggression and withdrawal. Developmental Psychology, 34,12461262.

    Shiner, R. L. (1998). How shall we speak of childrens personalities in middle childhood?A preliminary taxonomy. Psychological Bulletin, 124, 308332.

    Shiner, R. L., & Caspi, A. (2003). Personality differences in childhood and adolescence:

    Measurement, development, and consequences.Journal of Child Psychology and Psychiatry,44, 232.

    Tremblay, R. E. (1991). Commentary: Aggression, prosocial behavior, and gender: Threemagic words but no magic wand. In D. J. Pepler, & K. H. Rubin (Eds.),The development andtreatment of childhood aggression(pp. 7178). Hillsdale, NJ: Lawrence Erlbaum.

    Tremblay, R. E. (2000). The development of aggressive behaviour during childhood: Whathave we learned in the past century? International Journal of Behavioral Development, 24,129141.

    Tremblay, R. E. (2003). Why socialization fails?: The case of chronic physical aggression. In B.B. Lahey, T. E. Moffitt, & A. Caspi (Eds.),Causes of conduct disorder and juvenile delinquency(pp. 182224). New York: Guilford Press.

    Tremblay, R. E., Loeber, R., Gagnon, C., Charlebois, P., Larivee, S., & LeBlanc, M. (1991).Disruptive boys with stable and unstable high fighting behavior patterns during juniorelementary school.Journal of Abnormal Child Psychology,19, 285300.

    Tremblay, R. E., & Nagin, D. S. (2005). The developmental origins of physical aggression inhumans. In R. E. Tremblay, W. W. Hartup, & J. Archer (Eds.), Developmental origins ofaggression(pp. 83106). New York: Guilford Press.

    Tremblay, R. E., Vitaro, F., Gagnon, C., Piche, C., & Royer, N. (1992). A prosocial scale for thePreschool Behaviour Questionnaire: Concurrent and predictive correlates. International

    Journal of Behavioral Development,15, 227245.Vaillancourt, T. (2005). Indirect aggression among humans: Social construct or evolutionary

    adaptation? In R. E. Tremblay, W. W. Hartup, & J. Archer (Eds.), Developmental origins ofaggression(pp. 158177). New York: Guilford Publications Inc.

    Vitaro, F., Gendreau, P., Tremblay, R. E., & Oligny, P. (1998). Reactive and proactive aggressiondifferentially predict later conduct problems. Journal of Child Psychology and Psychiatry,39, 377385.

    TRAJECTORIES OF PROSOCIAL BEHAVIOR AND AGGRESSION 427

  • 8/13/2019 Kokko_2006_JRA

    26/26