Early correlates of preschool aggressive behavior according to type of aggression and measurement

Early correlates of preschool aggressive behavior according to type of aggression and measurement

Journal of Applied Developmental Psychology 27 (2006) 395 – 410 Early correlates of preschool aggressive behavior according to type of aggression and...

189KB Sizes 0 Downloads 42 Views

Journal of Applied Developmental Psychology 27 (2006) 395 – 410

Early correlates of preschool aggressive behavior according to type of aggression and measurement Mariel Juliano, Rebecca Stetson Werner, Kimberly Wright Cassidy ⁎ Department of Psychology, Bryn Mawr College, 101 North Merion Avenue, Bryn Mawr, PA 19010, USA Available online 5 July 2006

Abstract This study investigated both relational and physical aggression in preschool children, explored potential differences in informant reporting and child sex on these subtypes, and examined relationships between types of aggressive behaviors and other types of negative and positive social behaviors. Naturalistic observations of social behavior, teacher ratings, and peer reports were used to assess 67 children's (35 female) behaviors. Evidence was found for the existence of both relational and physical aggressive behavior in preschool children, although observers infrequently identified relational aggression. Categorization of children into a preferred style of aggression differed between informants, with more behavioral observers than teachers characterizing children as displaying only one type of aggression. Results also showed that reports were correlated between informants for physical aggression but not for relational aggression. Sex differences emerged for children's levels of physical aggression but not for relational aggression. Both physical and relational aggression were associated with ratings of problem behaviors in general; however, only physical aggression was negatively related to positive aspects of social functioning. © 2006 Published by Elsevier Inc. Keywords: Physical aggression; Relational aggression; Preschool social functioning; Teacher reports; Behavior observations

1. Introduction Early aggressive behavior has been shown to be a good predictor of later aggressive behavior (see Farrington, 1994; Stattin & Magnusson, 1989; for examples). Recently, there has been a surge of research investigating the different types of aggression, what Little, Jones, Henrich, and Hawley (2003) refer to as the “whats” of aggression. They found evidence for at least two distinct subtypes of aggression, overt, and relational aggression. Overt or physical aggression damages others through physical means; examples of these behaviors include threats or actually engaging in such acts as hitting, kicking, or other means of physically hurting a peer. Relational aggression harms others through damage to their peer relationships and includes acts or threats that socially exclude a peer, such as not allowing a peer to play with them, attempting to get others to dislike a peer, turning their back to someone with whom they are angry, and ignoring the peer (Crick, Casas, & Mosher, 1997). Few studies have investigated the existence of these aggression subtypes in preschool children and an even smaller number have examined the relationships between the aggression subtypes and psychosocial adjustment. The current study examines these questions in a sample of preschool children. ⁎ Corresponding author. E-mail address: [email protected] (K. Wright Cassidy). 0193-3973/$ - see front matter © 2006 Published by Elsevier Inc. doi:10.1016/j.appdev.2006.06.008

396

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

There is evidence that sex differences exist in the prevalence of these subtypes, particularly in school-aged children. Studies suggest that boys show more physical aggression (Hartup, 1974; Xie, Farmer, & Cairns, 2003) and that girls show more relational aggression (see, for example, Archer, Pearson, & Westeman, 1988; Crick & Grotpeter, 1995). In fact, Crick and Grotpeter's (1995) inclusion of relational aggression in measures of aggressive behavior resulted in equal nominations of school-aged boys and girls as aggressive overall, in contrast to previous work that indicated that boys are more aggressive than girls. Some studies, however, have not found sex differences. For example, Tomada and Schneider (1997) found that school-aged Italian boys were both more physically and relationally aggressive than girls. These researchers suggested that cultural differences might account for why their results were not in sync with the results from many other studies. Though there has been increasing realization of the importance of relational aggression as a type of social behavior and recognition of the fact that there may be sex differences in the expression of aggression, much of the previous research has focused on school-aged children. Some recent data suggest that sex differences in type of aggressive behavior may appear as early as preschool. Studies using teacher report to assess preschool aggression found that boys were rated as displaying more physical aggression than girls and girls were rated as displaying more relational aggression than boys (Bonica, Arnold, Fisher, Zeljo, & Yershova, 2003; Crick et al., 1997; Russell, Hart, Robinson, & Olsen, 2003). However, Hart, Nelson, Robinson, and McNeilly-Choque, (1998) found no sex differences in teacher ratings and McEvoy, Estrem, Rodriquez, and Olsen (2003) found that preschool boys, as rated by teachers, peers, and behavioral observers, engaged in more relational aggression than girls. Other studies have measured aggression using observation of behavior in either naturalistic free play situations or semi-structured activity situations and found sex differences (McNeilly-Choque, Hart, Robinson, Nelson, & Olsen, 1996; Ostrov & Keating, 2004; Ostrov, Woods, Jansen, Casas, & Crick, 2004), although Ostrov et al. (2004) found that only the differences in relational aggression reached statistical significance. In these studies, unlike the study by McEvoy et al., relational aggression was more prevalent in girls, whereas physical aggression was more prevalent in boys. Given the small number of studies that have examined aggression subtypes in preschool children, the conflicting findings regarding the existence of sex differences and the variety of methods used to measure aggression, further research studying preschool children using multiple methods of assessment is necessary. While there may be sex differences in the level of the two types of aggressive behavior overall, another important question is whether young children use primarily one kind of aggression or the other. It is important to consider whether differences in preschool children's use of aggressive behavior are similar to the differences found in school-aged children. Presumably most school-aged children possess the social cognitive ability to engage in both relational and physical aggression and have the behavioral inhibition to willfully avoid both physical and relational acts in most situations, or to select one form over the other if they so choose. In contrast, preschool children may not possess the behavioral control to selectively engage in one form of aggressive behavior over another when faced with the situation where an aggressive act is “called for.” Therefore, preschool children may not be differentiated in their expression of aggression and thus may engage in the same amounts of both types of behavior relative to their peers. Thus, another important question is whether preschool children have a characteristic “style” of aggressing. This can help us to get closer to answering whether sex differences in preschool aggression are similar to the differences sometimes observed in older children. Much of what we know about aggression is limited by the wide variety of measures used, the definitions applied, and the informants rating aggression. Issues of informant agreement on measures of aggression have been discussed in the literature. Achenbach, McConaughy, and Howell (1987) suggest that informant agreement is based on the similarity between each rater's role in the child's life and their opportunity to observe the child in the same context. Ladd and Kochenderfer-Ladd (2002) provide evidence that informant agreement is low, particularly when rating younger children, and informant agreement increases with age of the child. Most importantly, they found that a multi-informant measure of aggression was the best measure of relational aggression. Similarly, McNeilly-Choque et al. (1996) found that behavioral observation of aggressive behavior on the playground was weakly related to teacher ratings of relational aggression (r = .10) and only modestly related to teacher ratings of physical aggression (r = .30). In contrast, Ostrov and Keating (2004) found teacher ratings and behavioral observations of both kinds of aggression to be related. Perhaps, as has been suggested by others in the literature, ratings of different informants tap different dimensions of aggression and therefore their ratings might map differently onto social adjustment variables. Thus, it seems to be important to examine the connection between teacher report and behavioral observation.

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

397

Given that different types of aggressive behavior exist, do these forms of aggression lead to the same kinds and degree of psychosocial maladjustment? School-aged children view relational aggression as mean, hostile acts that cause harm and are often enacted in anger (Crick, 1995). Crick and Grotpeter (1995) also found that third through sixth grade children who engaged in relational aggression reported higher levels of loneliness, depression, peer rejection, and negative self-perceptions than other peers. In the few studies that have examined preschool children, psychosocial maladjustment has also been linked with aggression. Crick et al. (1997) found that both physical and relational aggression were predictive of peer rejection as rated by peers and of lower levels of prosocial behavior as rated by teachers. In addition, for boys, relational aggression was positively correlated with peer and teacher ratings for being accepted, though this was not true for girls. Girls who were physically aggressive were less likely to be accepted by peers (as reported by teachers) whereas boys who were more physically aggressive were rated by teachers as being less likely to be accepted only among opposite sex peers. Thus, for these adjustment variables, differences emerged among the psychosocial correlates of aggression for boys and for girls, as well as according to whether the raters of the psychosocial adjustment were teachers or peers. Recently, Ostrov et al. (2004) examined the relationship between observed aggressive behaviors in preschoolers and teacher ratings of adjustment. Preschool children were grouped in triads and observed for 9-min intervals in a conflict-provoking situation. The researchers also obtained measures of psychosocial adjustment from the head teacher of each child. They found a positive relationship between engagement in aggressive behavior in the semistructured play format and psychosocial maladjustment for both boys and girls. The type of psychosocial maladjustment, however, was differentially associated with both sex and type of aggression (Ostrov et al., 2004). For boys, physical aggression was positively related to peer rejection and negatively related to prosocial behavior. Relational aggression was positively associated with exclusion from peers and negatively associated with receiving prosocial behavior (teacher rated). In girls, relational aggression was negatively related to prosocial behavior. No other relationships reached significance. In sum, few prior studies have examined psychosocial correlates of different types of aggression in preschoolers and those that have, found these correlates to differ by sex. Additionally, these studies limited adjustment variables largely to peer acceptance and prosocial behavior. Thus, the current study expands the correlates of aggressive behavior to include broader measures of social competence and measures of other kinds of problem behaviors. We predicted that physical aggression would be positively associated with other problem behaviors and negatively related to global measures of social competence. We expected that relational aggression would be similarly positively associated with measures of problem behaviors; however, we expected that relational aggression might be positively associated with social competence. Given that many forms of relational aggression may require a more sophisticated social understanding, we predicted that the greater social sophistication of the child that uses relational aggression might also be employed by this same child to achieve social competence in other situations. The relationship between psychosocial maladjustment and aggression in boys and girls may differ at different developmental stages. Cillessen and Mayeux (2004), in their longitudinal study of children from the fifth to ninth grade, found evidence that the relationship between social status and type of aggression changes with age. As children become older, relationally aggressive acts become more socially acceptable, whereas physically aggressive acts become less acceptable. Less is known about the relationship between social status and type of aggression in preschool children. There is some evidence that suggests relational aggression may have fewer negative consequences for preschool children than for older children. A recent qualitative review of observational data suggests that preschoolers can use relatively sophisticated forms of relational aggression such as gossiping, secret telling, and rumor spreading (Ostrov et al., 2004); however, research evidence indicates that preschoolers may be unreliable informants for behaviors that are more subtle (Ladd & Mars, 1986). Thus, preschool children may have more difficulty identifying and recalling relationally aggressive acts, and thus, these acts may have less effect on peer outcomes such as social acceptance. Aggression may also have important implications for adjustment when the aggression takes a non-normative form. This has the obvious implication that psychosocial correlates of the different types of aggression will vary by sex. For example, Crick (1997) found that 9–12-year-old boys and girls who used sex non-normative aggressive behavior were at increased risk for social–psychological maladjustment. It is not clear, however, that the same pattern emerges in preschool-aged children. Ostrov and Keating (2004) found that teachers rate some of the children who used nonnormative forms of aggression as “dominant”. For example, they found that dominant preschool girls tended to use physical aggression against their female peers. While evidence suggests that both the type of aggression and the sex of

398

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

the child affect the severity of the adjustment difficulties that aggressive children show, it is again not clear if this relationship is the same in preschool and school-aged children. Given the small number of studies on subtypes of aggression in preschoolers, this study had three main aims: To further document the existence of both kinds of aggression in preschoolers and any sex differences in these subtypes; to examine the question of informant agreement on assessment of aggression; and to examine a wider range of psychosocial adjustment correlates of aggression. The current study, therefore, included assessment of preschool children's aggressive behavior, categorized according to physical and relational aggression, as well as multiple measures of social behavior and social competence. Multiple informants were included: Naturalistic observers recorded aggressive behavior during free play and documented children's social functioning, teachers rated children's behavior and social functioning, and peers rated children according to likeability. Our procedure involved two different ratings of aggression, which differed by informant and context. We predicted that physical aggression would demonstrate higher agreement as this type of aggression is more directly observable and that agreement would be lower for relational aggression, as this behavior is less visible and/or obtrusive. Other researchers have found that there is higher interrater reliability for physical aggression in preschoolers than for relational aggression (McNeilly-Choque et al., 1996). Informants may also have expectations for how children will aggress and this may affect their reports. For example, there is evidence that even children as young as preschool age have sex biases in their expectations for aggressive behavior (Giles & Heyman, 2005). In this study we did not measure peer ratings of aggressive behavior, but we expected that teacher ratings of aggression would show stronger sex differences than actual counts of behavior as measured by observers as they may be more susceptible to sex expectations for behavior. 2. Method 2.1. Participants Eighty-seven of 157 eligible children received parental permission to participate in this study. Ten of these 87 children were not included in the study because they refused to come with the experimenter. Three were not included because the teacher did not complete the rating scales, and 7 were not included because they had an excessive number of absences or had moved during the behavioral observation period. The final sample of 67 (32 male) did not differ from the 20 children who were excluded on age, number of siblings and sex distribution. The mean (SD) age of the final sample of children was 52 (7.73) months, with a range of 37–65 months. Children were from one of two preschools or two daycare centers in the lower middle to upper middle class suburbs of a major Eastern city. The ethnic diversity of the sample was consistent with the demographics from the community. Eighty-four percent of the sample was European American, 6% was African American, 6% was Asian American, 3% was Hispanic, and 1% was Lebanese American. 2.2. Procedure This study was a part of a larger study that investigated preschool children's social behavior, individual differences in children's social cognitive abilities, and measures of social adjustment and competence. For the current study, there were two types of measures used: Measures of aggressive behavior and measures of general social competence, peer popularity, and problem behaviors. 2.2.1. Aggressive behavior 2.2.1.1. Behavior observations of aggressive behavior. Children were observed in unstructured free play on eight different occasions. Each preschool and daycare center that participated in the study had scheduled parts of the day during which children determined the activities they engaged in and their location in the room or playground. In all settings, this was sometimes in the classroom and sometimes outside. Coders were randomly assigned to observe one child at a time. This randomized approach to observing the children enabled sampling of each child's behavior across a number of different free play activities. Given that our intention was to observe the child's behavior during activities that were typical of their behavioral preferences, efforts were not made to control for group versus solitary or dyadic

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

399

play, toy focused versus physical/gross motor play, or pretense versus non-pretense play. One of two coders observed each child for eight sessions of 5-min duration, with each session occurring on a different day. Observers tried to be as unobtrusive as possible, but they remained close enough to the children to be able to clearly hear their conversations. At times when the noise levels were quite high, or a child was particularly soft spoken, observers could be quite close. The observation coding scheme was adapted from Ladd, Price, and Hart's (1988) and Pettit, Bakshi, Dodge, and Coie's (1990) behavioral scoring systems. Physical aggression and relational aggression were differentiated. Physical aggression was defined as harm to others through physical damage or threat of such damage. Examples of physical aggression were hitting, shoving, pushing, and raising fists. The coding scheme included both physical acts and threats to engage in physically aggressive acts. Relational aggression was defined as harming others through threat to damage or damage of peer relationships. Examples of relational aggression were statements that the child would not be someone's friend if they do not do something he/she wanted, turning one's back on someone when angry with another child, and telling other children not to play with another child. The coding scheme included relational aggression acts presented as statements, verbal threats, and through body language and gestures. Instances of aggression were determined according to turns, separated by a response from the peer. For example, when a child shoved a peer, that act constituted one instance of physical aggression. If the peer responded and the aggressing child then shoved again or called the peer a name, that behavior would constitute a second instance of physical aggression. For verbal acts, each statement that expressed a thought was one aggressive act. A pause that exceeded 3 s also signaled that the next behavior would be considered a new instance of aggression. The categories of aggression were mutually exclusive, meaning that they were coded as either relational or physical aggression, but not both. However, both types could occur consecutively, with a child excluding a child verbally, for example, and then shoving that child out of the way, which was scored as one instance of relational aggression and one instance of physical aggression. Each instance of any of the above aggressive behaviors during the 5-min observational period resulted in a score of 1 assigned to that child, and the number of acts was summed for the observation for physical and relational aggression. One of two observers performed all of the coding. The coding was done “live” with each coder performing about 50% of the observations. In order to assess reliability throughout the data collection phase, periodically the two observers would simultaneously code the behaviors of eight different children for 5 min per child and any discrepancies were discussed to ensure ongoing consistency. Cohen's kappa for interrater reliability across all of the reliability sessions was .88 for physical aggression and .75 for relational aggression. 2.2.1.2. Teacher ratings of aggressive behavior. The head teachers in each participating child's classroom (12 total) were asked to complete the Preschool Social Behavior Scale–Teacher Form (PSBS; Crick et al., 1997). The PSBS consists of 23 items, 8 of which assess relational aggression and 8 of which assess physical aggression. Examples of items that assess physical aggression are “This child kicks or hits others” and “This child verbally threatens to hit or beat up other children.” Examples of relational items are “This child tells others not to play with a friend” and “This child tells a friend that s/he won't play with that friend unless s/he does what the child asks.” The remaining seven items on this scale are intended to assess depression and were not included in the current analysis. Teachers rated the physical and relational aggressive acts of each child included in the present study using a 5-point Likert-type scale (1 = never or almost never; 2 = not often; 3 = sometimes; 4 = often; 5 = always or almost always). Thus, the possible range of scores for each child was from 8 to 40 for physical aggression and 8 to 40 for relational aggression. This measure of aggression was included because it offered a comprehensive assessment of each preschoolers' aggressive behavior, as well as providing an opportunity to compare the findings for the present sample to previously reported studies of relational versus physical aggression in preschoolers. Crick et al. (1997) have reported that factor analyses yielded distinct and separate factors for relational and physical aggression as rated by the PSBS. For the current sample, Cronbach's alphas were .94 for relational aggression and .95 for physical aggression, which is consistent with those found by Crick, Casas, and Mosher (.96 and .94, respectively). 2.2.2. Social functioning 2.2.2.1. Teacher rating of social functioning. Teachers were asked to complete the Social Skills Rating System– Preschool Form (SSRS; Gresham & Elliott, 1990) for each participant child. The SSRS assesses positive and prosocial skills; items focus on cooperativeness, assertiveness, and self-control as well as a brief rating of problematic behaviors.

400

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

The Problem Behavior Scale focuses on impulsive, oppositional, anxious or depressive behaviors. Examples of the Problem Behavior Scale items are “has temper tantrums”, “fidgets or moves excessively”, “appears lonely”, and “shows anxiety about being with a group of children”. Thus, the social skills tapped in this teacher measure are mainly behaviors related to compliance with school routine and comfort in school settings. The SSRS has been established as a reliable and valid measure of children's social skills (Gresham & Elliott, 1990) and provides standardized scores as well as global behavior levels. Criterion-related validity (r = .70) for Total Social Skills was satisfactory when compared to the Harter Teacher Rating Scale. Criterion-related validity for Problem Behaviors when compared to the Child Behavior Checklist–Teacher Report Form (r = .81) was adequate as well. The SSRS has good internal consistency (for the current study, Cronbach's alpha = .93; for the standardization sample, alpha = .94) and test–retest reliability (r = .85). The child's raw score on the Total Social Skills Scale and on the Problem Behaviors Scale score were used in the current study. The possible range for the Total Social Skills Scale is 0–60 and is 0–20 for the Problem Behavior Scale. 2.2.2.2. Peer ratings. Sociometric ratings were used in the current study because they have been shown to have high reliability and validity for preschool children (Olson & Lifgren, 1988). Children completed a sociometric rating based on Denham and Holt's (1993) procedures. Children sorted pictures of their classmates into one of three boxes. Each box had a picture of a child of the same sex as the participant: One depicted a happy face with arms outstretched to show “I like the child this much (a lot)”; another presented a child with a neutral expression with arms partly outstretched to indicate “I kind of like the child”; and the third presented a picture of a child with a frowning face and hands very close together, showing “I do not like this child very much at all.” To assure that they understood the task, children participated in a training task in which they sorted pictures of food into the boxes according to how much they liked the food. After the training, the children were shown all of the pictures of the children who received permission to participate in our study and were asked to name them. Only the pictures of the children for whom they knew names were included in each child's sorting task. Each child was rated by a minimum of 3 children (mean number of raters per child = 7.6, SD = 3.2). Each child was given a sociometric rating based upon their picture's placement into the boxes. Ratings of “like a lot” were given a score of 3, ratings of “kind of like” were given a score of 2, and ratings of ”only like a little bit” were given a score of 1. Then the mean of all of the ratings was computed for each child. 2.2.2.3. Observer global ratings. Following each of the eight behavior observation sessions for each child, the observers completed a Global Rating Scale of the child's social competence. Children's behaviors were coded on four dimensions (positive interactions, social awareness, well liked by peers, and social competence) using a five point Likert scale (1 = does not apply; 5 = certainly applies) and a mean was calculated for each of the four dimensions across the eight sessions. The child's score on this measure was the total of the eight Global Ratings completed at the end of each of the 8 Behavior Observation sessions for each dimension. Children's ratings on the four Global Rating dimensions were strongly correlated with one another (r's ranged from .78 to .92, all p's < .001) and therefore were combined into a composite Global Rating of Social Competence. This measure appears to tap more of a holistic sense of social skill. While agreement between the observers was reliable on this measure (Cohen's kappa = .78), it is important to note that the descriptions of social competence that the observers were rating were quite global (e.g., positive interactions, social competence) and are based on each observer's construal of those general terms. 3. Results Each social behavior variable that showed evidence of skewness or kurtosis underwent arcsine or log transformations. After these transformations, all met the requirements for normal distribution. Table 1 shows the means, standard deviations and range of scores for the behavior observation (physical and relational aggression scores), the global rating of social competence, the teacher rating scales (physical and relational aggression), the SSRS scores (teacher rating of total social skills and problem behaviors) and the peer-rated sociometric scores.1

1

T-tests were conducted to assess whether social behavior and social functioning scores varied with ethnicity status. There were no significant differences observed between groups so we collapsed sample across ethnicity groups for the remainder of the analyses.

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

401

Table 1 Mean (and SD) scores for behavior observations and teacher ratings of aggression and social functioning for male and female participants and total sample Informant and measure Observer ratings Aggression (number of instances during 40 min of observation) Physical aggression Relational aggression Global Rating Scale of the Behavior Observation (possible range 1–20) Social competence Teacher Rated Social Behavior Scales Preschool Social Behavior Scale (possible range 8–40) Physical aggression Relational aggression Social Skills Rating System Total Social Skills (0–60) Problem behaviors (0–20) Peer ratings (possible range 1–3) Sociometric (likeability)

Participant sex

Total Sample

Boys

Girls

5.19 (5.74) .71 (1.37)

1.76 (2.50) .94 (1.54)

3.42 (4.67) .82 (1.95)

12.26 (2.91)

13.77 (2.91)

13.05 (2.99)

13.35 (6.06) 15.30 (5.80)

10.85 (5.00) 16.59 (7.32)

12.05 (5.63) 15.98 (6.63)

39.90 (7.23) 5.53 (4.53)

42.35 (8.06) 3.09 (2.98)

41.20 (7.72) 4.23 (3.95)

2.03 (.42)

1.89 (.32)

1.96 (.38)

3.1. Comparisons of incidence of types of aggressive acts 3.1.1. Observations of physical and relational aggression by child sex The incidence of both types of aggression was low during observations of free play. Although the range was considerable, fewer than an average of five instances of aggression of either type occurred during the 40-min observations. Relational aggression was particularly infrequent with 35 children showing no relationally aggressive behavior at all. A 2 (child sex) × 2 (type aggression) mixed model analysis of variance (ANOVA) applied to the aggression frequency scores obtained from the free play observations revealed a main effect of child sex, F(1, 64) = 4.81, p < .05; η2 = .07, and a main effect of type of aggression, F(1, 64) = 30.78, p < .001; η2 = .33. These effects were qualified by a significant interaction involving child sex and type of aggression, F(1, 64) = 11.07, p < .01; η2 = .15. Follow-up t-tests comparing the means, which are presented in the upper portion of Table 1, revealed that boys demonstrated more physical aggression than girls did, t(66) = 3.24, p < .001, but boys and girls did not differ from one another in the observations of relational aggressive acts p > .40. The distribution of boys and girls who displayed only one type of aggression (i.e., showed a characteristic “style”) is another way to explore sex differences in aggression. Previous research suggests that when children engage in only one form of aggression, more girls than boys enact only relational aggression and more boys than girls produce only physical aggression (Crick & Grotpeter, 1995). Thus, in the current study we expected that children who produced high levels of exclusively relational aggression (relative to their peers) would be girls and children who produced high levels of exclusively physical aggression relative to their peers would be boys. To examine this trend, four categories were created: (1) The child's scores showed no aggression above the median; (2) only the child's physical aggression score was above the median; (3) only the child's relational aggression score was above the median; and (4) both of the child's scores–for physical and for relational aggression–were above the median (see Table 2 for the frequencies of these classifications by child sex and type of informant). Note that scores of 1.0 and above were considered above the median for relational aggression. Based on the behavioral observations, 68% (13 out of the 19) of the children who exhibited higher levels of only physical aggression were boys. In contrast, 75% (9 of the 12 children) of the children who exhibited higher levels of only relational aggression were girls. Thus, child sex distribution among the children with higher levels of aggression of only one type differs depending on type of aggression, χ2(1, N = 31) = 5.57, p < .05. 3.1.2. Teacher ratings of physical and relational aggression by child sex Teacher ratings also indicated that aggression was an infrequent form of behavior in the centers. The mean total score ratings for both physical and relational aggression obtained from responses to the PSBS are reported in the lower portion of Table 1. Given that both the relational and physical aggression scales had 8 items on them, the teachers' total

402

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

Table 2 Number of children in each aggressive behavior category by type of informant and child sex Behavior classification

Type informant and child sex Teacher report

No aggression above median Physical aggression only above median Relational aggression only above median Both physical and relational above median

Behavioral observation

Boys

Girls

Boys

Girls

11 4 1 16

20 1 7 7

7 13 3 9

15 6 9 5

rating scores suggest that teachers said that these behaviors occurred on average “not often” on the 5-point scale for each item. A 2 (participant sex) × 2 (type aggression) mixed model ANOVA applied to teachers' ratings of aggression showed a main effect of type of aggression, F(1, 64) = 36.44, p < .001; η2 = .37. Overall, teachers reported lower rates of physically aggressive acts by the children, M = 12.05, SD = 5.63, than relational aggressive acts, M = 15.98, SD = 6.6. This main effect was qualified by a significant participant sex by type of aggression interaction, F(1, 64) = 7.23, p < .01; η2 = .10. Follow-up tests revealed that teachers rated boys as higher in physical aggression than they rated girls t(66) = 2.09, p < .05, but they rated girls and boys as equally high in relational aggression, p > .5 (see means in the middle of section of Table 1). Once again, we examined the distribution of boys and girls who displayed levels of aggression higher than the median in only one type of aggression. Although the frequencies are small, teacher ratings showed the same pattern as those seen for the behavioral observation: 80% (4 out of 5) of the children who exhibited higher levels of physical aggression only were boys, whereas 88% (7 out of 8) of the children who exhibited higher levels of relational aggression only were girls. Small expected frequencies per cell prevented the use of a chi square test on the teacher data (Siegel, 1956). 3.2. Behavior observations and teacher ratings: Informant effects The same pattern of findings was observed in the analyses of variance of the behavior observations and analyses of the teacher ratings, i.e., boys were more aggressive than girls in physical aggression, but not in relational aggression. On the other hand, the analyses of the behavioral observational data showed that incidents of physical aggression occurred more often (M = 3.42; SD = 4.67) than relational aggression (M = 0.3; SD = 1.45) did, whereas teachers' ratings of the frequency of each child's acts of aggression showed higher scores for relational aggression (M = 15.98; SD = 6.63) than for physical aggression (M = 12.05; SD = 5.63). Examination of the relationship between behavioral observations of aggressive acts and teachers' PSBS ratings of children's aggression showed that the two measures were correlated for physical aggression, r(65) = .53, p < .001, but not for relational aggression, r(65) = .04. These two correlations were significantly different, z = 2.75, p < .01. Thus, teachers and observers seemed to agree in their assessment of children's physical aggression, but not relational aggression. In addition, the teacher ratings did not appear to be as differentiated with respect to physical and relational aggression, whereas observers' ratings did. Teacher ratings of physical aggression and teacher ratings of relational aggression were correlated, r(67) = .61, p < .001. In contrast, observers' counts of the number of incidents of physical and relational aggression were not correlated with one another, r(67) = .10, p > .4. These correlations were significantly different, z = 3.05, p < .005. Thus, to the extent that teachers saw a child as exhibiting aggression, they seemed to see the child as aggressing (or not) in both ways. On the other hand, observers who saw a child exhibit higher levels of physical aggression did not necessarily see that child as exhibiting higher levels of relational aggression as well, although this lack of relationship could have been due to the low levels of observed relational aggression. The number of children who were reported to show high levels of aggression (above the median for the sample), but only of one type also varied with informant. For observers, 48% of children were scored for displaying only physical or only relational aggression more frequently than their peers during the behavioral observations. Observers saw the rest of the children as either rarely aggressive (relative to their peers) or as using both types of aggression more

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

403

frequently than their peers. Teacher ratings did not produce as much differentiation. Teachers rated most children (78%) as either exhibiting little of either type of aggression, or more aggression than their peers for both types of aggression. This difference in proportions was significant, z = 3.14, p < .005. Thus, the number of children categorized as showing a characteristic style of aggression varied according to whether teacher ratings or behavioral observations during free play was the source of the information. 3.3. The relationship of aggression and children's social functioning Table 1 shows the means and standard deviations for the various measures of social functioning. The global rating of social competence (created by summing children's scores on the 5-point Likert scale for the four items: Positive interactions, social awareness, well liked by peers, and social competence, averaged across the eight sessions) was the measure of social competence for the observers. The mean suggests that on average observers rated children slightly above the middle of the scale for social competence. The teachers provided a measure of social skills as well as a measure of problem behavior. The means for these measures suggests that the teachers saw the children as possessing a high level of social skills (although not at ceiling) and a low level of problem behaviors. Children provided sociometric ratings for their peers as well. We were interested in whether being relationally and/or physically aggressive was related to other aspects of social behavior such as peer popularity and positive social behavior, as has been observed in older children. Given the issue of effects of type of informant on reported relationships between aggression and social functioning, and the fact that each measure of social competence was likely tapping a somewhat different aspect of social competence, analyses considered each informant separately. Table 3 presents the correlations between the frequency of physical and relational aggression and scores for social functioning. As expected, measures of observed physical aggression were negatively correlated with observers' and teachers' ratings of social competence and positively related to problem behaviors (see left portion of Table 3). Teacher ratings of children's physical aggression were similarly related to teachers' ratings of children's social functioning and ratings of problem behaviors (upper right portion of Table 3). No measures of aggression were related to the sociometric scale. Behavioral observation scores for relational aggression were not related to any measures of social competence (as reported by either teachers or observers), nor were they related to the peer sociometric measures. This is likely due to the fact that there was little variability (floor effect) in the observations of relational aggression. Teacher ratings of relational aggression using the PSBS correlated positively with teacher ratings of problem behaviors on the SSRS. 3.3.1. Aggressive behavior and positive social functioning in girls and boys To further analyze the relationship between aggressive behaviors and social functioning, we conducted a series of hierarchical regressions that predicted social outcomes. To avoid concerns of common method bias, we

Table 3 Correlations between scores for each type of aggressive behavior and social functioning by type of informant Social functioning scores

Type of informant and type of aggressive behavior Behavior observation Physical

Teacher ratings Total social skills–SSRS Problem behaviors–SSRS Peer rating Sociometric Observer global rating Social competence

−.35⁎⁎ .44⁎⁎⁎

Teacher PSBS ratings

Relational .03 .20

Physical

Relational

− .31⁎ .72⁎⁎⁎

−.03 .56⁎⁎⁎

−.07

−.00

− .09

−.03

−.30⁎

−.14

− .16

.10

Note: SSRS = Social Skills Rating System (Teacher Report); GR = Observer Global Rating Scale of Behavior Observation. ⁎p < .05. ⁎⁎p < .01. ⁎⁎⁎p < .001.

404

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

Table 4 Summary of hierarchical regression analyses with physical aggression as measured by observers and child sex predicting teacher ratings of total social skills Dependent variable and predictors

B

SE B

β

R2

Total social skills (teacher rating) Step 1 Physical aggression (observer) Step 2 Physical aggression × Sex

− 2.25

.79

− .35

.12⁎⁎

− .94

1.21

− .28

.13

Change R2

.01

⁎p < .05. ⁎⁎p < .01.

used one type of informant (teacher or observer) to predict social outcomes as reported by the other type of informant. Given that we expected that physical aggression might be related negatively to positive social functioning, but that relational aggression might be related positively to positive social functioning (higher levels of social understanding might be necessary to enact relational aggression and these skills might lead to social competence in other areas), we analyzed relational and physical aggression in separate regression analyses. In addition, we expected that child sex might affect the nature or the strength of the relationship between aggressive behavior and social outcome; therefore, we included an interaction term involving child sex and type of aggression. We conducted the regression analyses with interactions terms even when the zero-order correlations between aggression and social functioning scores were not significant in order to examine whether aggression may predict psychosocial outcomes for one sex and not the other. Due to the low variability in the observers' reports of relational aggression, we did not include them in the regression analyses. Table 4 shows the results of the hierarchical regression analysis predicting teacher ratings of social skills on the SSRS using physical aggression as measured by behavioral observations as the predictor. Observer ratings of physical aggression were entered on Step 1, and a sex by physical aggression (as scored by observers) interaction term was entered on Step 2. As can be seen in Table 4, the observer ratings of physical aggression accounted for a significant portion of the variance in teacher ratings of children's positive social skills. The interaction terms did not offer a unique contribution to the prediction of variance in social skills suggesting that the nature or strength of the relationship did not differ for boys and girls. Next, we examined the prediction of the observers' measure of social competence. Given that we were predicting observers' global ratings, we used teachers' measures of aggression as the predictor. Again, physical and relational aggressions were analyzed in separate regressions because we predicted that the nature of the relationship might differ by aggression type. The results of the hierarchical regression analysis examining observers' global ratings of social competence are shown in Table 5. For these analyses, teacher ratings of physical aggression were entered on Step 1. On Step 2, a sex × physical aggression (as reported by teachers)

Table 5 Summary of hierarchical regression analyses for aggression variables and child sex predicting observers' ratings of social competence Dependent variables and predictors Social competence (observers) Step 1 Physical aggression (teacher) Step 2 Physical aggression × Sex Social competence (observers) Step 1 Relational aggression (teacher) Step 2 Relational aggression × Sex ⁎p < .05. ⁎⁎p < .01.

B

− .19

SE B

2.9

β

R2

−.01

.03

− 1.20

.70

−.25

.07

.79

.51

.21

.01

− .36

.19

−.26

.07

Change R2

.04

.06

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

405

Table 6 Summary of hierarchical regression analyses for aggression variables and child sex predicting sociometric scores Dependent variables and predictors Sociometric Step 1 Physical aggression (observer) Step 2 Physical aggression × Sex Sociometric Step 1 Physical aggression (teacher) Step 2 Relational aggression × Sex Sociometric Step 1 Relational aggression (teacher) Step 2 Relational aggression × Sex

B

SE B

β

R2

− .13

.12

−.43

.01

.01

.06

.39

.02

− .45

.37

−.20

.01

.11

.09

.19

.03

− .01

.07

−.11

.01

.01

.03

.19

.03

Change R2

.01

.02

.02

⁎p < .05. ⁎⁎p < .01.

interaction term was entered. As can be seen in Table 5, teacher ratings of physical aggression did not predict observers' global ratings of social competence. The interaction term also was not significant. The same regression analyses using teacher rating of relational aggression as the predictor and the sex × relational aggression (reported by teachers) interaction term also did not show any predictive relationship between teacher rated relational aggression and social competence as measured by observers (bottom of Table 5). While no form of aggression as measured by either type of observer was related to peer popularity at the zero-order level, it was possible that child sex may have moderated this relationship. For example, because physical aggression is more normative for boys, being physically aggressive may not adversely affect popularity for boys. In contrast, because physical aggression is less common in girls, engaging in this non-normative behavior may affect their popularity. Thus, we regressed sociometric scores on both types of aggression using different informants, including an interaction term for child sex in all equations. As can be seen in Table 6, these interaction terms also did not contribute to the prediction of sociometric scores. 3.3.2. Aggressive behavior and other problem behaviors in girls and boys Table 7 displays the results of a hierarchical regression analysis examining observers' scores of aggression and child sex as predictors of teacher ratings of problem behaviors. Observer ratings of physical aggression were entered on Step 1 and accounted for a significant portion of the variance in teacher ratings of problem behaviors on the SSRS questionnaire. On Step 2, an interaction term between physical aggression (observers' scores) and participant sex was entered. The interaction term did not offer a unique contribution to the prediction of problem behaviors.

Table 7 Summary of hierarchical regression analysis for observed aggression predicting teacher ratings of problem behavior Dependent variable and predictors Problem behaviors (teacher) Step 1 Physical aggression (observer) Step 2 Physical aggression × Sex ⁎p < .05. ⁎⁎p < .01. ⁎⁎⁎p < .001.

B

SE B

β

R2

1.38

1.12

.41

.18⁎⁎⁎

.01

.59

.01

.18

Change R2

.00

406

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

4. Discussion The results of this study provide information regarding the presence of different types of aggression in preschoolers and the correlates of these aggressive behaviors with positive and negative aspects of social functioning. We found evidence for both physical and relational aggression in our preschool sample. However, the incidence of aggressive behavior in the sample was low. Boys engaged in significantly more physical aggression than girls, which is consistent with other studies of preschool children. We did not find any significant sex differences in engagement in relationally aggressive acts, which differs from findings of some studies of preschool children (Bonica et al., 2003; Crick et al., 1997; McNeilly-Choque et al., 1996; Ostrov & Keating, 2004; Ostrov et al., 2004; Russell et al., 2003). Sex differences, however, have not been consistently observed in studies of preschool children. For example, McEvoy et al. (2003) found that boys engaged in higher levels of both relational and physical aggression than girls. In their study, as in the current study, low levels of aggressive behavior were found, and they cautioned that it might be difficult to make direct comparisons with so few occurrences of the behavior. In addition to mean differences in levels of aggression, we looked at children who engaged in higher levels (relative to the peer group) of only one type of aggressive behavior. We expected to find evidence for sex differences at the aggressive extremes. We found support for this idea: Most of the children who engaged in higher levels of exclusively physical aggression relative to their peers were boys, whereas most of the children who engaged in higher levels of exclusively relational aggression relative to their peers were girls. Thus, when the children displayed an aggressive “style,” it tended to be sex consistent. This is similar to the findings in older children when sex differences emerge in children with a characteristic style of aggression (Crick & Grotpeter, 1995). This finding offers support for the idea that at least some preschoolers possess the behavioral control to selectively engage in one type of aggression over another when they choose to aggress. This finding, however, has to be interpreted with caution because few children were aggressing at the extremes, and these extremes were relative to a comparison group that showed very little aggression on the whole. While both teachers and observers seemed to identify differences in the distribution of boys and girls who showed a characteristic style of each type of aggression, the observers identified a greater proportion of children as having a characteristic style of aggression. Teachers were more likely to see children as exhibiting either both types of aggression or no aggression. It is not clear whether the difference between reports from teachers and observers was a function of type of informant or type of measure used. Teachers may be more likely to categorize children as aggressive and then respond affirmatively to any question about that child's aggression. On the other hand, the difference may have been due to the different measures used. The informants differed on level of training in noticing discrete types of aggressive behaviors. Teachers based their ratings on their own interpretations of the aggressive behaviors on the report they were given and the observers were specifically trained to notice particular types of aggression. The teachers and observers also differed in the way that they were asked to count the aggressive behaviors. While the teachers gave reports based on their experience and impressions of their students over a long period of time and across different contexts, the observers based their counts of aggressive behavior on live observations made over a few discrete time periods. The analyses of cross-informant agreement yielded some areas of commonality. As predicted, ratings by teachers and counts by observers were significantly related to one another for physical aggression scores. There was no relationship, however, between the teacher ratings and observer counts of relational aggression. The results of studies that have been conducted on informant agreement for aggression are not all consistent with one another. Some other studies have also found higher levels of agreement for physical aggression than relational aggression (McNeillyChoque et al., 1996). However, Ostrov and Keating (2004) found a significant association between teacher and observer reports of relational aggression, but no significant correlations for overt aggression, which is the opposite of the finding in the current study. In the Ostrov and Keating study and the current study, teacher ratings were assessed using the PSBS. Behavioral observations, however, were not made in the same way. Ostrov and Keating measured physical, relational, and verbal aggression, whereas in the current study, verbal aggression was not directly assessed. Aggressive verbal acts in the current study were categorized as either physically aggressive or relationally aggressive. Acts were defined in this way to make them more consistent with the PSBS, which does not include a measure of verbal aggression. Perhaps the aggressive acts noted by the behavior observers in the Ostrov and Keating study were not categorized in the same way that they were categorized in the current study. Additionally, in the Ostrov and Keating study, children were observed at five different times in 10-min intervals, whereas in the current study, the children were observed at eight

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

407

different times in 5-min intervals. It is possible that longer intervals provided time for the children to acclimate to the observers, and therefore, the observers had a more genuine view of the children's behavior. A longer window of observation may also make it possible detect some of the more subtle aspects of relationships among the children that would then make the relational aggression more obvious. There are several possible reasons for the low levels of informant agreement for relational aggression in the current study. Relational aggression is a relatively low frequency behavior in the population investigated in this study, and thus there were relatively few instances of relational aggression scored by the observers, whereas the teacher was making a more general rating for each child. Teachers could be more sensitive to the interactive style of the child and have a longer window of observation on which to base their ratings. Additionally, the presence of the observers may have prevented some relationally aggressive acts from occurring. It is not clear why this would not also apply to physical aggression, except that relational aggression, as opposed to physical aggression, may be easier to hide in the presence of an observer. Or perhaps children engaging in relational aggression have a greater ability to delay their impulse to aggress until they are no longer under the watchful eye of an adult. An additional issue may be the differential impact of observer biases in counts versus global impressions. The teacher ratings may have been influenced by the teacher's own biases regarding aggression and by the teachers' global impression of the child. Observers may be somewhat protected (although undoubtedly not completely) from personal biases because they are assigned to count observable, discrete behaviors periodically checked with a reliability coder. In support of this idea, it appears that being rated as engaging in one type of aggression by one's teacher makes one more likely to be rated as engaging in the other type of aggression. This was not true for observers. Thus, teachers' ratings may stem from general impressions of children and from their belief that children who aggress in one way likely aggress in another way. However, given the low incidence of aggression, it is also possible that the lack of significant results for observers could be due to low variability in the data. Interestingly, even though observers saw relational aggression as rare, teachers rated it as more frequent than physical aggression. Again, this could be because relational aggression is more difficult to observe, or it could be because teachers are using more global aggression schemas to rate the children. We also predicted that both physical and relational aggression would be positively related to other problem behaviors, but that differences would emerge in the association with social skills/competence. We predicted that physical aggression would be negatively related to social competence, but that relational aggression would be positively related to social competence. We also predicted that these relationships might vary by sex, as relational aggression may be more normative for girls and physical for boys and thus have less of a negative social impact for them. Regression analyses supported the hypothesis for the prediction of problem behaviors. Only observed physical aggression was related to problem behaviors, and this relationship did not differ according to child sex. As predicted, for teachers' ratings of positive social skills, a significant negative relationship with physical aggression was found for observer counts of physical aggression. However, teacher ratings of physical aggression did not predict observer ratings of social competence, and, contrary to our hypothesis, relational aggression as rated by teachers, was also not related to observer ratings of social competence. None of the interaction terms in any of these regressions were significant, suggesting that relationships among the variables did not differ according to sex of the children. Taken together, these results suggest that children who show more physical aggression tend to show higher levels of other kinds of problems behaviors as well, so their behavioral issues are not isolated to aggressive acts. Teachers (but not observers) also see these children as having fewer social skills. We expected that aggressive behavior would have a different relationship with social behavior depending on the child's sex, because of gender differentiated norms of appropriate behavior. There were no such effects in the data suggesting that adults in this context are less affected by child sex in judging the social impact of aggression. Thus, aggression was seen as related to other bad behaviors no matter who was doing it. Interestingly, preschool children's aggressive behavior appears to have a negative impact on teacher ratings of social behavior, but not on peer ratings of likeability. This may be because young children are not yet able to attend to or respond to more subtle aspects of social behavior, such as relational aggression. Also, given children's still developing inhibitory and behavioral control ability, aggressive behavior in general may be more normative and thus does not have the intense social impact that may occur in later years. Other studies, however, have shown that preschool children are sensitive to peer aggression and may negatively factor that into their ratings of peer likeability. Denham, McKinley, Couchoud, and Holt (1990) found that prosocial behavior (which was an aggregate score that was determined by subtracting out an aggression score) was positively related to peer ratings of likeability. Arsenio, Cooperman, and Lover (2000) found that aggressive behavior was negatively associated with peer acceptance. They also found that the

408

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

emotion displayed by the aggressor might be an important factor in peer ratings of likeability. Happiness displayed during aggressive acts by the preschoolers in their study was related to negative ratings of peer likeability, whereas anger displayed during aggressive acts was not related to peer likeability. This study points to the importance of the emotional context of aggressive behavior and how this affects preschoolers' perceptions of behavior. Stormshak, Bierman, Bruschi, Dodge, and Coie (1999) have shown that peer ratings of first grade aggressive children depend on the degree to which aggressive behavior is normative in the classroom context. Stormshak et al. found that aggressive children in classrooms in which the levels of aggressive behavior in general were higher did not evidence the negative effects that aggressive children in classrooms with lower levels of aggressive behavior evidenced. One problem with this argument for the current study is that aggression was relatively infrequent. Given that very little is known about prevalence, onset age, and persistence of the different types of aggression assessed in this study, it is difficult to draw meaningful conclusions about each child's risk for later maladjustment, or for whether they are displaying significantly negative behavior in comparison to their peers. Though the work of Moffitt (1993) suggests that there are meaningful pathways between early aggressive behavior and later delinquent behavior, there is no research that plots the interaction of the types of aggression (relational and physical) investigated in this study with these developmental pathways. Most of the research to date that suggests significant relationships between early preschool aggression and later conduct problems and social maladjustment has only included measures of physical aggression. Crick and Rose (2000) have suggested that relational aggression increases in acceptability as children age (see also Cillessen & Mayeux, 2004). It may be that children who engage in relational aggression do not have the same risks for later maladjustment as children who engage in physical aggression. Additionally, in the current study, only low levels of each type of aggression were found. Thus, these children may have been engaging in a normal amount of aggression and therefore may not be at risk for later maladjustment. Still, there is a clear need for improved measures of aggressive behavior to be developed, which differentiate physical and relational aggression so that the potential differential developmental pathways can be studied. This study provides some information about how aggression is concurrently related to both negative and positive behavior in typically developing preschool children. Further research is needed to help understand if these types of aggression, which appear to be meaningful distinctions in preschool children, are meaningful distinctions across time. However, it is important to keep in mind that the current study was correlational; thus, no assumptions can be made about causal links between aggressive behavior and measures of adjustment. The current study included not only rating scales to measure aggression, but also included naturalistic observations of the child in the classroom environment. This multi-method, multi-informant design was a strength of this study as this design provides a much richer database that contributes to our understanding of this complicated and multifaceted behavior. The current study also had a number of limitations. Unfortunately, different informants and different methods were confounded, where teachers filled out rating scales and observers conducted behavior observations using frequency counts. Future studies should investigate the relationships between teacher and observer ratings of children using the same method, in order to begin to understand the differences that were found in this study between our measures of aggressive behavior. A further limitation of our study was the low level of aggressive behavior found in our sample. This limited our statistical power and ability to closely examine fine distinctions in levels of aggression and, as McEvoy et al. (2003) have indicated, low levels of aggression make it difficult to make generalizations to children who engage in high levels of aggression. Further limitations in generalizability result from our sample, which was fairly homogenous. It also did not specifically include children at high risk for engaging in aggressive behavior. In addition to issues related to measurement of aggression, results of the NICHD Study of Early Child Care suggest that the context in which aggression occurs exerts an interesting and measurable influence upon the behaviors assessed in the current study. Though the current study did not include measures of time spent in childcare or type, quality, and orientation of the child care centers involved, this will be a necessary and important set of variables to include in any future work in this area. It is clear that significant stability exists for early aggressive behavior. Not only does aggressive behavior in young children correlate with concurrent measures of psychosocial adjustment, but early aggressive behavior also significantly predicts later assessment of adjustment and competence (Cairns & Cairns, 1994; Ladd & Burgess, 2001). There is also evidence that young children are already displaying preference for types of aggression enacted and that these types of behaviors have different predictors, correlates, and trajectories. Attention needs to be paid to the context in which aggression is measured, to individual differences such as sex and family history, and to risk factors and protective factors that will have significant impact upon the course of aggression for each child. Our understanding of

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

409

the differential trajectories for children engaging in these diverse types of aggressive behavior will inform our efforts to intervene appropriately when necessary to prevent potential negative outcomes associated with this complicated and pervasive behavior. Acknowledgement This research was supported by a Research Grant to the third author and a Dissertation Fellowship to the second author from the Harry Frank Guggenheim Foundation. We are grateful to the children and childcare centers that participated in this study. We thank the many undergraduate students who helped in the data collection. We also thank Mary Rourke, Leslie Rescorla, Robert Wozniak, and Marc Schulz, as well as Kimberly Wright Cassidy's research group for their numerous contributions to this work. References Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: Implications for crossinformant correlations for situational specificity. Psychological Bulletin, 101, 213−232. Archer, J., Pearson, N., & Westeman, K. E. (1988). Aggressive behavior of children aged 6–11: Sex differences and their magnitude. British Journal of Social Psychology, 20, 31−36. Arsenio, W. F., Cooperman, S., & Lover, A. (2000). Affective predictors of preschoolers' aggression and peer acceptance: Direct and indirect effects. Developmental Psychology, 36(4), 438−448. Bonica, C., Arnold, D. H., Fisher, P. H., Zeljo, A., & Yershova, K. (2003). Relational aggression. Relational victimization and language development in preschoolers. Social Development, 12, 551−562. Cairns, R. B., & Cairns, B. D. (1994). Lifelines and risks: Pathways of youth in our time. Cambridge, UK: Cambridge University Press. Cillessen, A. H. N., & Mayeux, L. (2004). From censure to reinforcement: Developmental changes in the association between aggression and social status. Child Development, 75(1), 147−163. Crick, N. R. (1995). Relational aggression: The role of intent attributions, feelings of distress, and provocation type. Development and Psychopathology, 7, 313−322. Crick, N. R. (1997). Engagement in sex normative versus nonnormative forms of aggression: Links to social–psychological adjustment. Developmental Psychology, 33, 610−617. Crick, N. R., Casas, J. F., & Mosher, M. (1997). Relational and overt aggression in preschool. Developmental Psychology, 33(4), 579−588. Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression, sex, and social–psychological adjustment. Child Development, 66, 710−722. Crick, N. R., & Rose, A. J. (2000). Toward a gender balanced approach to the study of social–emotional development: A look at relational aggression. In P. H. Miller, & E. Scholnick (Eds.), Developmental psychology through the lens of feminist theories (pp. 153−168). Florence: Routledge Press. Denham, S. A., & Holt, R. W. (1993). Preschooler's likeability as cause or consequence of their social behavior. Developmental Psychology, 29, 271−275. Denham, S. A., McKinley, M., Couchoud, E. A., & Holt, R. (1990). Emotional and behavioral predictors of preschool peer ratings. Child Development, 61(4), 1145−1152. Farrington, D. P. (1994). Childhood, adolescent, and adult features of violent males. In L. R. Huesmann (Ed.), Aggressive behavior: Current perspectives (pp. 215−240). New York: Plenum. Giles, J. W., & Heyman, G. D. (2005). Young children's beliefs about the relationship between sex and aggressive behavior. Child Development, 76 (1), 107−121. Gresham, F. M., & Elliott, S. N. (1990). Social skills rating system. Circle Pine, MN: American Guidance Service. Hart, C. H., Nelson, D. A., Robinson, C. C., & McNeilly-Choque, M. K. (1998). Overt and relational aggression in Russian nursery-school-age children: Parenting style and marital linkages. Developmental Psychology, 34(4), 687−697. Hartup, W. W. (1974). Aggression in childhood: Developmental perspectives. American Psychologist, 29, 336−341. Ladd, G. W., & Burgess, K. B. (2001). Do relational risks and protective factors moderate the linkages between childhood aggression and early psychological and school adjustment? Child Development, 72(5), 1579−1601. Ladd, G. W., & Kochenderfer-Ladd, B. (2002). Identifying victims of peer aggression from early to middle childhood: Analysis of cross-informant data for concordance, estimation of relational adjustment, prevalence of victimization and characteristics of identified victims. Psychological Assessment, 14(1), 74−96. Ladd, G. W., & Mars, K. T. (1986). Reliability and validity of preschoolers' perceptions of peer behavior. Journal of Clinical Child Psychology, 15 (1), 16−25. Ladd, G. W., Price, J. M., & Hart, C. H. (1988). Predicting preschoolers' peer status from their playground behaviors. Child Development, 59, 986−992. Little, T. D., Jones, S. M., Henrich, C. C., & Hawley, P. H. (2003). Disentangling the “whys” from the “whats” of aggressive behavior. International Journal of Behavioral Development, 27, 122−133. McEvoy, M. A., Estrem, T. L., & Olsen, M. L. (2003). Assessing relational and physical aggression among preschool children: Intermethod agreement. Topics in Early Childhood Special Education, 23(2), 53−63.

410

M. Juliano et al. / Journal of Applied Developmental Psychology 27 (2006) 395–410

McNeilly-Choque, M. K., Hart, C. H., Robinson, C. C., Nelson, L. J., & Olsen, S. F. (1996). Overt and relational aggression on the playground: Correspondence among different informants. Journal of Research in Childhood Education, 11, 47−67. Moffitt, T. E. (1993). Adolescence-limited and life-cycle-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674−701. Olson, S. L., & Lifgren, K. (1988). Concurrent and longitudinal correlates of preschool peer sociometrics: Comparing rating scale and nomination measures. Journal of Applied Developmental Psychology, 9, 409−420. Ostrov, J. M., & Keating, C. F. (2004). Sex differences in preschool aggression during free play and structured interactions: An observational study. Social Development, 13, 255−277. Ostrov, J. M., Woods, K. E., Jansen, E. A., Casas, J. F., & Crick, N. R. (2004). An observational study of delivered and received aggression, sex and social-psychological adjustment in preschool: “This white crayon doesn't work…”. Early Childhood Research Quarterly, 19, 355−371. Pettit, G. S., Bakshi, A., Dodge, K. A., & Coie, J. D. (1990). The emergence of social dominance in young boys' play groups: Developmental differences and behavioral correlates. Developmental Psychology, 26(6), 1017−1025. Russell, A., Hart, C. H., Robinson, C. C., & Olsen (2003). Children's sociable and aggressive behavior with peers: A comparison of the US and Australia and contributions of temperament and parenting styles. International Journal of Behavioral Development, 27, 74−86. Siegel, S. (1956). Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill. Stattin, H., & Magnusson, D. (1989). The role of early aggressive behavior in the frequency, seriousness, and types of later crime. Journal of Consulting and Clinical Psychology, 57, 710−718. Stormshak, E. A., Bierman, K. L., Bruschi, C., Dodge, K. A., & Coie, J. D. (1999). The relation between behavior problems and peer preference in different classroom contexts. Child Development, 70(1), 169−182. Tomada, G., & Schneider, B. H. (1997). Relational aggression, sex, and peer acceptance: Invariance across culture, stability over time, and concordance among informants. Developmental Psychology, 33, 601−609. Xie, H., Farmer, T. W., & Cairns, B. D. (2003). Different forms of aggression among inner-city African-American children: Sex configurations and school social networks. Journal of School Psychology, 41(5), 355−375.