High levels of aggressive behavior among young children, including cruelty to others and bullying, are predictive of high levels of aggressive behavior during later periods of the lifespan. The long-term predictive value of assessments of children's early aggression was demonstrated in a longitudinal study conducted in Canada. The study followed White, French-speaking boys from age 6 years to age 15 years (Nagin & Tremblay, 2001). The findings revealed that the average frequency of physical aggression for the total sample was highest when the boys were enrolled in kindergarten. Moreover, the study found that boys with the highest ratings of physical aggression at age 6 were most likely to engage in persistent physical aggression and violence throughout childhood and adolescence. The majority of boys in the study, however, declined in their rates of physical aggression across the years. Similar findings regarding the rates and patterns of children's aggression, especially for boys, also were recently reported for a six-site, multinational study (Broidy et al., 2003).

Not all young children who engage in aggressive bullying and victimization are doomed to a life of violent, criminal behavior. Nonetheless, the predictive value of young children's bullying and victimization to more serious forms of aggression in later childhood and adolescence makes the study of young children's bullying and victimization important. Additionally, findings from studies of young children's bullying and victimization should help experts better understand the etiology and developmental sequence of bullying, which can have implications for developing more successful prevention and intervention programs.

Snyder et al. (2003) found that aggressive bullying was widespread among 5 to 7 year old children who were observed on the playground. In fact, aggressive bullying and victimization (either physical aggression or verbal harassment) was observed once every 3 to 6 min. Not only was aggressive bullying frequent, but it was pervasive. Between 80 and 90% of children on the playground were the recipients of some form of peer aggression. Only about 10% of children were never victimized during the multiple observation periods. Other studies also have demonstrated the common occurrence of bullying among young children. For example, Culp et al. (2003) observed four year-old Head Start children on the playground and found that bullying occurred relatively frequently.

Many studies of bullying distinguish between overt, physical aggression or overt bullying, and relational aggression or relational bullying. Overt bullying refers to physical aggression directed at peers with the intent of causing physical harm to others, or making threats to cause physical harm. Overt bullying includes behaviors such as hitting, kicking, pushing, or threatening to engage in these acts of physical aggression (Crick, Casas, & Mosher, 1997). In contrast, relational bullying refers to aggression directed at individuals with the intent of causing harm to their peer relationships. Relational bullying involves reliance on verbal aggression and includes social exclusion (i.e., purposefully not including a peer in a group activity) rumor spreading, humiliating another child in front of a group of peers, or threatening to engage in acts of relational aggression.

Several studies conducted in the U.S. have identified overt and relational bullying among children as young as four and five years by using either peer- or teacher-nominations or both (Bonica, Arnold, Fisher, Zeljo, & Yershova, 2003; Crick et al., 1997; Crick, Casas, & Ku, 1999; Kochenderfer& Ladd, 1996; McEvoy, Estrem, Rodriguez, & Olson, 2003; McNeilly-Choque, Hart, Robinson, Nelson, & Olsen, 1996). Crick et al., 1997 found that 12% of boys and 3% of girls in their sample were only overtly aggressive; 0% of boys and 26% of girls were only relationally aggressive; and 15% of boys and 7% of girls were both overtly and relationally aggressive.

Two studies conducted in Switzerland also have identified bullying among children as young as five years by using peer and teacher nominations (Alsaker & Valkanover, 2001). Two of the studies conducted in the US included children from low socioeconomic family backgrounds (Bonica et al., 2003; McNeilly-Choque et al., 1996), and of these two studies, only one reported the inclusion of a subsample of African American children ((Bonica et al., 2003). Most of the other studies were of White children of middle class background. Consequently, more research is needed to examine bullying and victimization among young African American children from low-income families.

Most studies linking parenting to children's bullying behaviors have been conducted with school-aged children and adolescents. These studies find that bullying has been related to maternal anger (Curtner-Smith, 2000), maternal depression (Curtner-Smith, 2000), low parent-child involvement (Curtner-Smith, 2000; Flouri & Buchanan, 2003), and harsh forms of discipline (Curtner-Smith, 2000; Olweus, 1993). A study of Irish school-aged children found that bullies reported experiencing more negative emotions about their mothers and fathers than children who were neither bullies nor victims (Connolly & O'Moore, 2003), and a study of British school-aged children found that bullies reported greater emotional distance between themselves and their parents than nonbullies/nonvictims (Bowers, Smith, & Binney, 1992). The latter finding was replicated by a study of Italian school children (Berdondini & Smith, 1996). Along similar lines, Australian adolescents who were classified as bullies also reported more difficulties in communicating with their parents as compared to their nonbully/nonvictim classmates (Rigby, 1994).

The theoretical links between parenting and children's social functioning were delineated by Zhou et al. (2002). First, the researchers contend that empathy is an important inhibitor of aggression because it motivates people to behave in ways that will not be hurtful to others. Second, citing the work of Barnett (1987), Zhou et al. noted that parenting behaviors which foster children's empathy include those that are responsive to the child's emotional needs, allow for emotional expression, and encourage the child's sensitivity to others. Finally, findings from a longitudinal study conducted by Zhou et al. which included elementary school-aged children and their mothers confirmed the theoretical links between parenting, children's empathy and children's social functioning. In particular, Zhou et al. found that high levels of parental warmth and “positive expressiveness” (i.e., the expression of positive emotions in front of children) were related to high levels of children's empathy. In turn, children's empathy was related to fewer teacher-reported child externalizing behaviors. Likewise, the previously mentioned observational study of Head Start Children conducted by Culp et al. (2003) found that mothers who lacked warmth and empathy were more likely to have children who were observed bullying on the playground.

The purpose of our study was to examine (1) the prevalence of overt and relational bullying among a sample of low income, predominantly African American preschool-aged children, and (2) the contribution of mothers' parenting to young children's overt bullying and relational bullying.

METHOD

Participants

The participants were 44 mothers and their four-year old children, M child age = 4.11 years, SD = .64. Nearly 64% of the sample of children was female and 36% of the sample of children was male. All children were enrolled in a local Head Start Program. Head Start is a federally funded preschool program for children of low income families. The vast majority of the children and their mothers (95.5%) were African American. Two children and their mothers were White (4.5%). The percentage of African American children in the sample is similar to the percentage of children enrolled at the local Head Start program where the data were collected. Almost half of the mothers (47.7%) were single, never married; 20.5% of the mothers were single, due to being either divorced or separated, and 2.3% or 1 mother reported being widowed. Mothers who were married for the first time represented 18.2% of the sample; 2 mothers (4.5%) reported being remarried; 3 mothers (6.8%) reported “other” for their marital status.

Mothers' monthly household income before taxes ranged from $0.00 to $3,499.00. Nearly 59% of mothers reported a monthly household income that ranged from $500.00 to $1,499.00. Nearly 16% of the sample of mothers reported a monthly household income that ranged between $1,500.00 and $1,999.00, and another 16% of the sample reported a monthly household income that ranged between $2,000.00 and $2,999.00. Finally, 9% of the mothers reported a monthly household income ranging between $2,500.00 and $3,500.00.

Procedures

A flyer describing the study and a copy of the Agreement to Participate Form were sent home with all children enrolled in Head Start classrooms serving 4-year old children. Mothers who signed and returned the Agreement to Participate Form to their children's teacher were called by a research assistant and an appointment was scheduled to conduct a face-to-face interview. Mothers also were recruited into the study during the morning drop off time at the Head Start school. This recruiting procedure involved setting up a table with a poster describing the study in the hallway outside of the school cafeteria. Many mothers who read the poster signed the Agreement to Participate Form, and then consented to be interviewed at that time, or they scheduled interview to take place later in the week. All interviews took place in a vacant office at the Head Start school unless mothers specifically requested a home visit. Only 5 of the 44 mothers participating in the study preferred being interviewed in their homes rather than at the school.

During the face-to-face interview, mothers completed questionnaires pertaining to their demographics and parenting. For each questionnaire, the research assistant read the directions aloud and asked mothers to answer the sample questions. Mothers were then asked if they preferred to complete the questionnaires on their own, or if they wanted to have the questionnaire read aloud. All mothers preferred to read and complete the questionnaires on their own. Research assistants remained available to assist mothers in completing the questionnaires. Each mother interview took approximately 40 min to complete. Head Start teachers completed assessments on children's bullying. Each child assessment took approximately 5 min to complete.

Measures

Relational Bullying

Teachers completed the Preschool Social Behavior Scale-Teacher (Crick et al., 1997; PSBS-T) on each child whose mother both gave permission for the child to be assessed and who also agreed to participate in the study. The PSBS-T contains a Relational Bullying subscale. This subscale is comprised of 6 items that give a behavioral description of relational bullying. Examples of items describing relational bullying include, “This child tells others not to play with or be a kid's friend,” “When this child gets mad at another kid, this child keeps that kid from playing in a play group,” and “This child tries to get others to dislike a kid.” For each item, teachers indicated the frequency with which the child engaged in that form of relational bullying. Response options include 1 = “Almost never true,” 2 = “Occasionally true,” 3 = “True about 1/2 of the time,” 4 = “Frequently true,” to 5 = “Almost always true.” Scores for Relational Bullying had a possible range of 6 to 30. In the present sample, scores ranged from 6 to 25, M = 10.82; SD = 5.0 Cronbach alpha = .93.

Overt Bullying

The PSBS-T also contains an Overt Bullying subscale. This subscale is comprised of 6 items that give a behavioral description of overt bullying. Examples of items describing overt bullying include, “This child hurts other kids by pinching them,” “When this child wants her way, this child will threaten to physically harm another kid,” and “This child says he or she will beat up other kids.” For each item, teachers indicated the frequency with which the child engaged in that form of overt bullying. Response options for overt bullying were the same as those for relational bullying. Scores for Overt Bullying had a possible range of 6 to 30. In the present sample, scores ranged from 6 to 24, M = 10.68; SD = 5.31, Cronbach alpha = .94.

Mother's Parenting

Mothers completed the Adolescent/Adult Parenting Inventory (AAPI-2), which is a 40-item questionnaire that assesses parenting attitudes, values, and behaviors (Bavolek & Keene, 2001). For each item, participants respond on a Likert scale that ranges from 1 = “Strongly agree” to 5 = “Strongly disagree.” High scores reflect attitudes, values, and behaviors that are typical of optimal parenting whereas low scores reflect attitudes, values, and behaviors that are typical of abusive, neglectful parenting. The AAPI-2 contains five subcales: (a). Empathy, (b) Appropriate Expectations for Children's Behavior, (c) Power vs. Independence, (d) Values Alternatives to Corporal Punishment, and (e) Family Role Reversals.

The Empathy subscale contains 10 items. Sample items include, “Children should keep their feelings to themselves,” “Because I said so! is the only reason parents need to give children,” and “Children who feel secure often grow up expecting too much.” Scores for this subscale have a possible range of 10 to 50. In the present study, scores ranged from 20 to 47, M = 37.64, SD = 5.47, Cronbach alpha = .66.

The Appropriate Expectations for Children's Behavior subscale contains 7 items. Sample items include, “Good children always obey their parents,” “Strong-willed children must be taught to mind their parents,” and “Children should do what they're told to do, when they're told to do it. It's that simple.” Scores have a possible range of 7 to 35. In the present study, scores ranged from 8 to 26, M = 17.84, SD = 3.88, Cronbach alpha = .67.

The Power vs. Independence subscale contains 5 items. Sample items include, “Children need to be allowed freedom to explore their world in safety,” “Children who receive praise will think too much of themselves,” and “Parents who encourage their children to talk only end up listening to complaints.” Possible scores can range from 5 to 25. In the present study, scores ranged from 12 to 24, M = 19.27, SD = 2.78, Cronbach alpha = .41.

The Values Alternatives to Corporal Punishment subscale contains 11 items. Sample items include, “Spanking teaches children right from wrong,” “It's OK to spank as a last resort,” and “Sometimes spanking is the only thing that will work.” Possible scores range from 11 to 55. In the present study, scores ranged from 21 to 52, M = 32.45, SD = 6.6, Cronbach alpha = .71.

The Family Role Reversals subscale contains 7 items. Sample items include, “Parents should be able to confide in their children,” “Children should be aware of ways to comfort their parents after a hard day's work,” and “Children should be responsible for the well being of their parents.” Possible scores range from 7 to 35. In the present study, scores ranged from 12 to 32, M = 22.55, SD = 4.61, Cronbach alpha = .67.

Table I. Frequencies, and Percents of Young Children's Relational Bullying and Overt Bullying
Full size table

RESULTS

Table I shows the results of teachers' reports of children's bullying. Four children (approximately 9% of the sample) were rated by teachers as “frequently to almost always” engaging in relational bullying. Slightly over half of the sample (54%) was rated by teachers as engaging “occasionally to almost half of the time” in relational bullying. Finally, nearly 34% of the sample was perceived by teachers as “almost never” engaging in relational bullying.

Nearly 16% of children were rated by teachers as “frequently to almost always” engaging in overt bullying. Slightly over half of the sample (54%) was rated by teachers as “occasionally to almost half of the time” engaging in overt bullying, and nearly 30% of the sample was perceived by teachers as “almost never” engaging in overt bullying.

A second way to examine the prevalence of relational and overt bullying among this sample was to examine the percentage of boys and girls who were classified into extreme groups of relationally and overtly aggressive children. For each aggression variable (relational bullying, overt bullying), children with scores one standard deviation above the sample mean were considered aggressive and the remaining children were considered nonaggressive. Results revealed that 13% of boys and 0% of girls in the sample were only relationally aggressive; .06% of boys and 10% of girls in the sample were only overtly aggressive; and .07% of boys and nearly 14% of girls were both relationally and overtly aggressive.

To assess gender differences in relational bullying, we conducted an independent t-test comparing boys and girls. No differences between the mean scores for boys and girls were found (for Males, M = 10.69, SD = 4.8; for Females M = 10.89, SD = 5.2; t(42) = −.13, p = .45).

Gender differences in overt bullying were also examined by performing an independent t-test comparing boys and girls. No differences between the mean overt bullying scores for boys and girls were found (for Males, M = 10.69, SD = 4.8; for Females, M = 10.68, SD = 5.66; t(42) = .005, p = .50).

Table II shows the pattern of correlations among relational bullying, overt bullying and parenting. The findings reveal that relational bullying was more related to the parenting variables than overt bullying. In addition, Table III shows that the parenting variable most strongly related to each type of bullying was maternal empathy. Mothers who were high in empathy for their children had children who engaged in less frequent relational bullying and less frequent overt bullying of peers than mothers who were low in empathy for their children. In contrast, mothers who lacked empathy for their children had children who were rated by teachers as engaging in more frequent relational bullying and more frequent overt bullying of peers. Mothers' scores for Appropriate Developmental Expectations and Power vs. Independence were mildly related to children's relational bullying. Mothers who had more appropriate developmental expectations had children who were rated by teachers as engaging in less frequent relational bullying of peers. In contrast, mothers who had less appropriate developmental expectations of their children had children who were rated by teachers as engaging in more frequent relational bullying of peers. Similarly, mothers who were more concerned about exerting power over their children had children who were rated by teachers as engaging in more frequent relational bullying than mothers who were more concerned about fostering independence in their children.

Table II. Correlations Between Young Children's Relational Bullying, Overt Bullying, and Selected Parenting Variables
Full size table

Table II also shows the intercorrelations of the parenting variables. For the most part, mothers' scores for Empathy, Appropriate Developmental Expectations, Power vs. Independence and Values Alternatives to Corporal Punishment are moderately to mildly intercorrelated. For this reason, scores for these variables were standardized and summed to compute a composite variable for mother's parenting. Relational bullying scores were then regressed on the composite variable of parenting. Results revealed that less optimal parenting explained 11% of the variance in teacher reports of children's relational bullying [R2 = .11; β = .34; F = 5.31; p < .05]. Overt bullying scores were not regressed on the composite parenting variable because the only parenting subscale that was significantly related to overt bullying was maternal empathy.

DISCUSSION

Many of the findings from our study are consistent with findings from previous studies of bullying among preschool-aged children and with findings from studies of bullying among older children and adolescents. Teachers who completed assessments on the young children in this study were able to identify children who engaged in occasional to frequent relational bullying and overt bullying. This finding adds to those from other studies that provide evidence for the existence of the two distinct forms of bullying among a population of young preschool-aged children ((Bonica et al., 2003; Crick et al., 1997; 1999; Kochenderfer & Ladd, 1996; McEvoy et al., 2003; McNeilly-Choque et al., 1996).

The percentages of children in our study who were classified in the extreme groups of overt bullies, relational bullies, and overt/relational bullies, however, were not similar to the percentages of children classified into extreme groups in the study conducted by Crick et al. (1997). Crick et al. identified a larger percentage of children in their study who were classified into one of the extreme aggressive groups than were in this study. For both studies, the method of classifying children into extreme groups involved identifying children whose aggression scores (relational bullying and overt bullying) were 1 standard deviation above the sample mean. This method is obviously dependent on the value of the sample mean and will vary from study to study, thus making it difficult to compare across studies.

On the other hand, boys' and girls' mean scores for both relational bullying and overt bullying were somewhat higher in this study than comparable scores for boys and girls who participated in a study conducted by McEvoy et al. (2003). Specifically, McEvoy et al. assessed relational and overt bullying with a sample of 59 White middleclass preschool-aged children by using the same measure that was used in the current study, the PSBS-T. Boys in their study had mean scores of 9.35 for relational bullying and 10.53 for overt bullying. Recall that boys in the current study had mean scores of 10.69 for relational bullying and 10.69 for overt bullying. Girls in the study conducted by McEvoy et al. had mean scores of 9.32 for relational bullying and 6.92 for overt bullying. In contrast, girls in the current study had mean scores of 10.89 for relational bullying and 10.68 for overt bullying. Thus, it appears that children in the present study, especially girls, were somewhat more aggressive than children studied by McEvoy et al. One possible reason for why the relational and overt bullying scores of children in the current study were higher than the relational and overt bullying scores of children who participated in the study conducted by McEvoy et al. may be difference in culture or socioeconomic (SES) levels represented by children in the two studies.

Children who participated in the present study were from predominantly low income families who reside in the Southern region of the US, which is an area where poverty rates are high. In fact approximately 24.4% of children under age 5 years live in poverty in the state in which data for this study were collected, and approximately 19.8% of children under age 5 live in poverty in the county served by the Head Start that participated in this study (U.S. Bureau of Census, 2000). Many children who live in poverty face considerable environmental risks every day that are known to impact child and family development (Duncan & Brooks-Gunn, 2000; Gutman & Eccles, 1999). These risks include a lack of cognitive stimulation in the home, low parental education, high levels of parental stress and depression, harsh parental discipline, increased family conflict, and exposure to crime and/or violence within the neighborhood. Other studies have documented how these risks impact the quality of parenting received by children, which in turn, influences the quality of children's peer interactions (Brody & Flor, 1998; McLoyd, 1990; McLoyd, Jayaratne, Ceballo, & Borquez, 1994).

McNeilly-Choque et al. (1996) examined SES differences in preschool-aged children's overt and relational bullying and found that children from families of lower socioeconomic status engaged in significantly higher teacher-reported overt bullying than children from families of higher socioeconomic status. In contrast, children from families of higher socioeconomic status engaged in significantly higher levels of relational bullying than children from families of lower socioeconomic status. Similar patterns of SES and relational bullying were found in a study by Bonica et al. (2003). That is, children from families of higher SES engaged in more relational bullying than children from families of lower SES. The investigators in this study did not report scores for overt bullying. The current study's finding that children engaged in considerable relational bullying may be an artifact of the sample. Clearly, more research that involves standardizing the instrument among children of different ages, different levels of SES, and different ethnicities is needed.

Most investigations of relational and overt bullying among both preschool-aged children and older children have found that boys tend to engage in more overt bullying whereas girls tend to engage in more relational bullying (Crick et al., 1997; McEvoy et al., 2003; McNeilly-Choque et al., 1996; Rigby, 2002). Alsaker and Valkanover (2001), however, found that more boys in their sample of 344 young children (ages 5 to 7 years) were classified as bullies (65%) than girls, and that boys continued to be more likely than girls to be classified as bullies even when items assessing forms of relational bullying were given the same priority in categorizing bullies. In the present study, no gender differences in children's relational or overt bullying were found.

A few studies have linked aspects of parenting to elementary school children's bullying, and one recent observational study of young children enrolled in Head Start linked low maternal empathy with higher rates of young children's observed bullying (Culp et al., 2003). Likewise, we found that mothers who were low in empathy for their children had children with higher scores for both relational bullying and overt bullying than mothers who were high in empathy. We also found that mothers who had more appropriate developmental expectations of children and who valued children's independence over the need to exert power over children had children with the lowest scores for relational bullying. Although speculative, it could be that mothers who are low in empathy for their children are unable to understand, respond to, and meet their children's emotional needs. In addition, parents who lack understanding of their children's emotional needs may be more likely to engage in parenting behaviors that inhibit the development of children's empathy for others. Parenting behaviors that inhibit children's empathy include harsh forms of discipline (yelling, sarcasm, humiliation, harsh spanking), little inductive reasoning, and low warmth or affection. These parenting behaviors inhibit children's empathy because they fail meet children's emotional needs and they fail to encourage children's emotional sensitivity to others. Moreover, empathy-inhibiting parenting behaviors restrict children's emotional expression. Our findings are consistent with the theory that empathic parenting which relies on age-appropriate expectations for children's behaviors and allows for emotional expression may help to foster children's empathy and lower children's peer aggression.

Several limitations of this study should be noted. First, the small sample size limits the scope of this study. Additional studies of low income minority children and their mothers using larger sample sizes are needed. Second, the measure of children's relational bullying and overt bullying relied on teacher reports. Future investigations would be aided by including observational assessments of children's relational bullying and overt bullying because these forms of peer aggression often occur when there is little adult supervision of children (Craig, Pepler, Connolly, & Henderson, 2001). Finally, future studies would benefit by including a measure of children's empathy. This would allow investigators to examine how children's empathy mediates the influence of maternal empathic parenting on children's relational bullying and overt bullying.