Citation
Comparison of Early Pubertal Timing Effects on Aggression and Delinquency between Urban Minority Girls and Boys: Demonstrating Effects and Identifying Pathways

Material Information

Title:
Comparison of Early Pubertal Timing Effects on Aggression and Delinquency between Urban Minority Girls and Boys: Demonstrating Effects and Identifying Pathways
Creator:
BENSENHAVER, SARAH L. ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Adolescence ( jstor )
Adolescents ( jstor )
Adulthood ( jstor )
African Americans ( jstor )
Delinquency ( jstor )
Delinquent behavior ( jstor )
Early development ( jstor )
Hispanics ( jstor )
Human aggression ( jstor )
Young offenders ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Sarah L. Bensenhaver. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/31/2007
Resource Identifier:
436098774 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

COMPARISON OF EARLY PUBERTAL TIMING EFFECTS ON AGGRESSION AND DELINQUENCY BETWEEN URBAN MINORITY GIRLS AND BOYS: DEMONSTRATING EFFECTS AND IDENTIFYING PATHWAYS By SARAH L. BENSENHAVER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2005

PAGE 2

ii ACKNOWLEDGMENTS There are several people who have played supportive roles for me during my years as a graduate student. I would like to take this opportunity to thank these people for their guidance and encouragement through this challenging moment of my life. First, I would like to express my gratitude to my advisor, Dr. Julia Graber, without whom I could not be pursuing my interests in adolescent development. She is a tremendous academic who has served as a wonderful role model for me in all facets of my professional development. Her expertise and guidance have provided me with the experience and confidence to further pursue academic research in the field of adolescent behavior. Special thanks go to my committee members, Dr. Alan Agresti and Dr. Lise Youngblade. With their help, I have learned the importance of thoroughness in conducting and reporting statistical analyses, and I truly feel confident in my ability to do so. I would also like to offer thanks to my fellow graduate students in the developmental psychology department for continually offering unconditional support, friendship, and good-times. Without the support of these friends, the stressors of graduate school and of life would surely have overtaken me. Finally, I am extremely thankful to my family. The support and encouragement that my mother, brother, and sister have given me have been invaluable. I thank my mother for showing me, by example, that I can achieve any goal I set for myself through hard work, patience, and perseverance. It is my familyÂ’s unconditional and unwavering support, love, and encouragement that have sustained me and helped me to succeed.

PAGE 3

iii TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................................ ii LIST OF TABLES...................................................... iv LIST OF FIGURES...................................................... v ABSTRACT........................................................... vi INTRODUCTION ....................................................... 1 LITERATURE REVIEW................................................. 3 Pubertal Timing ..................................................... 3 Aggression in Adolescence............................................. 5 Variations in Aggression by Gender, R ace/Ethnicity, and Demographic Factors .... 6 The Present Study................................................... 16 METHODS ........................................................... 19 Research Design.................................................... 19 Participants......................................................... 19 Procedure.......................................................... 20 Measures.......................................................... 21 Analysis Plan....................................................... 24 RESULTS............................................................ 28 Descriptive Analysis................................................. 28 Association of Pubertal Timing with Aggression and Delinquency Outcomes .... 30 DISCUSSION......................................................... 50 Limitations......................................................... 57 Implications........................................................ 58 REFERENCES........................................................ 60 BIOGRAPHICAL SKETCH.............................................. 64

PAGE 4

iv LIST OF TABLES Table page 1Demographic data by pubertal timing in the 6th grade ..................... 28 2Means and standard deviations by pubertal timing for each grade ............. 29 3Intercorrelations among independent and dependent measures ............... 31 4 F ratios and effect sizes for univariate main effects of grade, gender, race/ethnicity, and pubertal timing on aggression, nonviolent delinquency and violent delinquency ................................................. 32 5Moderating effects of friend delinquency on the relation between pubertal timing and aggression............................................... 43 6M oderating effects of friend delinquency on the relation between pubertal timing and nonviolent delinquency ..................................... 44 7Moderating effects of friend delinquency on the relation between pubertal timing and violent delinquency ........................................ 45

PAGE 5

v LIST OF FIGURES Figure page 1Mediational model testing friend delinquency as the mediating variable ....... 27 2Moderator model testing friend delinquency as the moderating variable ....... 27 3Grade X timing X race interaction on nonviolent delinquency ............... 35 4Grade X timing X race/ethnicity X gender interaction on violent delinquency... 38 5Moderating effect of friend delinquency on reports of aggression in the 6th grade............................................................ 47 6Moderating effect of friend delinquenc y on reports of nonviolent delinquency.. 48 7Moderating effect of friend delinquency on reports of violent delinquency ..... 49

PAGE 6

vi ABSTRACT Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science COMPARISON OF EARLY PUBERTAL TIMING EFFECTS ON AGGRESSION AND DELINQUENCY BETWEEN URBAN MINORITY GIRLS AND BOYS: DEMONSTRATING EFFECTS AND IDENTIFYING PATHWAYS By Sarah L. Bensenhaver May 2005 Chair: Julia A. Graber Major Department: Psychology Early pubertal maturation has been linked to higher rates of aggressive and delinquent behaviors among white, middle class females. Less is known about this relationship for males or minorities. The association between early pubertal maturation and aggressive or delinquent behaviors for both male and female African Americans and Latinos ( n = 816) was examined as well as different pathways between early maturation and these negative outcomes longitudinally across 6th, 7th, and 8th grades. Early maturation was associated with higher rates of aggression, nonviolent delinquency, and violent delinquency for both males and females regardless of race or ethnicity at all time points. The impact of early maturation was more deleterious for Latinos compared to African Americans. Associating with deviant peers mediated the relationship between early maturation and aggression and violent delinquency in the 6th grade. Moderating

PAGE 7

vii effects of deviant peers were found in all grades. These results provide valuable information regarding at-risk groups and inform future intervention efforts.

PAGE 8

1 INTRODUCTION The entry into adolescence marks the beginning of a phase of an individualÂ’s life where rapid biological, psychological, and so cial changes occur (e.g., Graber, BrooksGunn, & Petersen, 1996; Simmons & Blyth, 1987). Along with these changes, rates of aggression and delinquency increase substantially for both males and females. Understanding the mechanisms behind this increase in aggression and delinquency in adolescence is important because these adolescents are at an increased risk for school drop-out, peer rejection, and future incar ceration (Coie & Dodge, 1998). Various models relating puberty to adjustment have been proposed to address this increase in the occurrence of disorder and problem behaviors upon entry into adolescence, one of which examines the timing of pubertal maturation. Timing models suggest that it is not the level of pubertal maturation that is critical to the development of positive or negative outcomes, rather it is the timing of pubertal maturation relative to peers (Graber, BrooksGunn, & Archibald, in press). Specifically, negative adjustment outcomes are associated with either early or late maturation compared to most other same age adolescents. Until just recently, most research examining the relationship between pubertal timing and negative adjustment has focused on non-Hispanic, white females from middle class families. Therefore, less is known about the impact that early or late maturation has on racial or ethnic minorities or males. The current study proposes to examine the relationship between early pubertal maturation and rates of aggression and delinquency in a sample of racially and ethnically diverse, early adolescent males and females. The

PAGE 9

2 period of early adolescence, specifically 6th, 7th, and 8th grades, encompasses the time in which the increases in aggressive and delinquent behaviors first begin to emerge and start to approach their peak levels. The current study has chosen to focus on this time period to attempt to examine the transition from childhood rates of aggressive and delinquent behaviors to the higher rates associated with adolescence. Specifically, the first goal of the current study is to examine how increases in aggression and delinquency relate to early pubertal maturation. In addition, different pathways relating early maturation to aggressive and delinquent outcomes will also be examined, specifically association with deviant peers.

PAGE 10

3 LITERATURE REVIEW Pubertal Timing As indicated, pubertal timing refers to when an individual begins puberty in relationship to their peers, that is, being either earlier, on-time, or later than oneÂ’s peers. Two different timing models have been proposed to illustrate the connection between pubertal timing and adjustment problems. They are the deviancy model and the stage termination model (Brooks-Gunn, Petersen, & Eichorn, 1985). The deviancy model states that being different from the majority of same-age peers in the timing of puberty places an adolescent at risk for negative adjustment outcomes. Given that, on average, girls tend to show external signs of puberty about two years earlier than boys, the two groups of adolescents who would be at the extremes of the pubertal timing continuum are the early maturing females, because they begin earlier than any other group, and the late-maturing males, because they begin later than any other group. Therefore, according to the deviancy hypothesis, it would be these two groups of adolescents who are most at risk for adjustment difficulties. An alternate model that has been proposed to explain who will experience negative outcomes associated with their pubertal timing is the stage termination model. This model proposes that those adolescents who begin puberty earlier than their sameage peers are most at risk for negative adjustment outcomes. The reasoning behind this model is that those adolescents who physically appear older will be exposed to more adult situations at an earlier age than their on-time peers and may not have developed the

PAGE 11

4 skills necessary to cope effectively with these situations. Accordingly, early maturing girls would display the highest rates of adjustment difficulties, followed by early maturing males, in comparison with their on-time or late-maturing peers. Huddleston and Ge (2003) have proposed an adaptation to these original models of the effects of pubertal timing. The basic concepts underlying the deviancy model and stage termination model remain the same; however, Huddleston and Ge introduce the importance of understanding who comprises the reference group of peers against which an adolescent evaluates their perception of the timing of their puberty. The original models assume that same-age adolescents of both genders are included in the adolescentÂ’s reference group. If this is the case then the predictions of negative outcomes stated previously still hold. The proposed variations on the previous models state that perhaps perception of pubertal timing should be considered within gender. It is plausible that boys may only compare their level of physical development to other same-age boys and are not as affected by when same-age girls begin to develop, and vice versa. These are known as variant 2 of each of the respectiv e models. If this is the case, then for variant 2 of the deviance model both early and late-maturing males would display higher rates of adjustment difficulties, compared to on-time maturing males. This same pattern would apply to females as well. Additionally, for variant 2 of the stage-termination model, early maturing males and females would show high rates of adjustment difficulties compared to their on-time and late-maturing same-age peers. The following sections will discuss the association between pubertal timing and aggressive or delinquent behaviors. We begin with an overview of the development of aggression in adolescence.

PAGE 12

5 Aggression in Adolescence The study of the development of aggressive behaviors from childhood through adolescence and adulthood has long been a topic of interest among developmental psychopathologists. One reason for this interest is the association between aggression, delinquency, and future adult criminal behavior. While these terms differ in the type and severity of behavior displayed, they are also logically and empirically related (Farrington, 2004). This can be attributed to similarities in the definitions of these terms as well as the fact that many individuals who engage in one of these behaviors often engage in multiple externalizing behaviors. Therefore, it follows that risk factors, as well as successful interventions that apply to one of these behaviors are likely to also apply to the others. As is the case with many psychological constructs, debate regarding a formal definition of aggression or delinquency is ongoing and could easily encompass an entire paper. Given that the focus of the current study is not to debate definitional issues, there will be no attempt to thoroughly discuss all controversies regarding the definitions of aggression and delinquency, however for practical reasons this issue must be addressed (see Coie & Dodge, 1998 for a more thorough review of definitions). For the current study, aggression is defined as behavior that is intended to and has the potential to harm another person (Coie & Dodge). Delinquency refers to acts prohibited by criminal law, including theft, burglary, robbery, violence, and vandalism (Farrington, 2004). Researchers that study aggression and other externalizing behaviors have differentiated between different degrees of seve rity and types of behaviors to capture the range of subclinical symptoms as well as disorders related to externalizing behaviors. At the more severe end of the spectrum is conduct disorder. Two different periods of onset have been identified regarding conduct disorder, each of which is associated with different prognoses. Early-onset conduct diso rder (childhood) tends to persist throughout

PAGE 13

6 adolescence and into adulthood and is predictive of future adult criminal behavior, whereas late-onset (adolescence-limited) conduct disorder is generally more limited to the adolescent time period and may be less predictive of adult criminal outcomes (Lahey et al., 1998; Moffitt, 1993). However, prevalence rates of conduct disorder are low. It has been well established in the literature that overall rates of aggressive behavior are higher in males than females with an increase in rates occurring for both genders during adolescence (Coie & Dodge, 1998). This includes increased rates of delinquent behaviors such as assault, r obbery, shoplifting, and drug and alcohol use (Moffitt, Caspi, Rutter, & Silva, 2001). This phenomenon has been labeled adolescencelimited conduct disorder or aggression which broadly refers to aggressive behaviors that first appear during adolescence and which are limited to a shorter time span than those associated with conduct disorder (Moffitt et al.). Interestingly, more recent reports regarding adolescence-limited aggression indicate that those behaviors may persist longer into adulthood than originally thought. These individuals continue to drink heavily, use drugs, get into fights, and commit criminal acts (Farrington, 2004; Pajer, 1998). Rarely do aggressive or delinquent behaviors first appear in adulthood for either gender (Moffitt et al., 2001). Since the majority of individuals who participate in aggressive or delinquent behaviors first do so in adolescence, it is important to understand the mechanisms behind the initiation of these behaviors in adolescence and whether these differ between males and females. Variations in Aggression by Gender, Race/Ethnicity, and Demographic Factors Early Adolescence While late adolescence and early adulthood represent the time when aggressive and delinquent behaviors are most damaging, early adolescence is when the initial increases in these behaviors begin. When examining rates of serious violent offenses

PAGE 14

7 (e.g., assault, robbery, rape), recent studies have found that up to about age 11, occurrence of these behaviors are virtually zero; however, rates of these behaviors double by age 14 (Coie & Dodge, 1998). Moffitt and colleagues (2001) reported the mean age of onset of self-reported aggressive or delinquent behaviors to be 13.3 years for males in the Dunedin Longitudinal Study of Development from early childhood through young adulthood. Given these data, it follows that the period of early adolescence represents the time when it is most critical to evaluate the mechanisms that underlie the initiation of aggressive and delinquent behaviors. It is at this point that any interventions to prevent the onset of these behaviors would be most effective. Gender Differences Due to the fact that rates of aggressive behaviors at all ages are higher in males than in females, the majority of research examining aggression has focused on males and consequently is biased toward forms of aggression that are more common in boys, such as overt aggression (Crick, 1997). Overt aggression encompasses those aggressive behaviors that harm others through physical damage or through the threat of such damage, such as direct acts of physical or verbal assault. Examining other forms of aggression has recently been emphasized to determine if the low rates of aggressive behaviors reported for females could be attributed to biases in the measurement of aggression. Specifically, relational forms of aggression such as rumor spreading, and purposely destroying someoneÂ’s reputation, ma y be more typical manifestations of aggression in females (Coie & Dodge, 1998). Moffitt and colleagues (2001) evaluated the gender disparity in rates of violent and nonvi olent aggressive and delinquent behaviors over the first two decades of their participantsÂ’ lives and found that males were consistently higher in rates of aggression and delinquency at all points in time with only a

PAGE 15

8 few exceptions. The key exceptions were that males and females were remarkably similar in their use of illicit drugs and their involvement in domestic violence. However, it is noteworthy that the gender difference was the greatest in regards to violent behavior and violent crimes. At the same time, girls who engage in more traditionally “male” types of aggression and delinquency, as indicated by conduct disorder rather than relational aggression, are those who go on to have more difficulties in adulthood (Pajer, 1998). Given the differences in rates and types of aggression displayed by males and females, it is important to investigate the developmental pathways to aggressive behavior in both females and males to look for similarities and differences. Moffitt et al. (2001) has conducted the most thorough and current investigation into this area at this time. Importantly, they found that males were not more vulnerable to risk factors than females. For example, family risk factors, such as harsh, inconsistent parenting, living in a single parent household, and low SES, were important predictors of aggressive and delinquent behavior but since they affected males and females similarly, they therefore do not contribute to an explanation of the gender difference. It is beyond the scope of the present investigation to consider all risk factors for aggression and delinquency and whether they vary for males and females. However, family context will be considered in connection with differential pathways for aggression and delinquency based on pubertal timing (see later section). Nor will the proposed study attempt to examine all of the risks previously mentioned, as this is one of the first examinations of aggression and pubertal timing in an ethnically diverse sample. Regarding the influence that peers play in the initiation of aggressive and delinquent behaviors, Moffitt et al. (2001) found that male peer groups were more conducive to the development of aggressive and delinquent behaviors. This association

PAGE 16

9 with peers also accounted for a large part of the gender difference in these behaviors. However, Moffitt et al. (2001) did not l ook at association with delinquent peers specifically, only association with male peers. This is an area that requires further investigation to determine if association with deviant peers operates in the same way for both genders. Current timing research. Initial investigations of the association between pubertal timing and adjustment difficulties seem to indicate that this association varies by gender. The pattern of effects for females resembles what would be expected according to the stage termination model, with early maturing females showing the most signs of adjustment difficulties in comparison to their on-time and late-maturing female peers. For example, in the Oregon Adolescent Depression Project (OADP), two studies have examined the association between early maturation and a wide range of psychological disorders and psychosocial symptoms and found that in general early pubertal maturation in females was associated with higher lifetime prevalence rates of disorders and subclinical symptoms, even into young adulthood, compared to on-time maturers (Graber, Lewinsohn, Seeley, & Brooks-Gunn, 1997; Graber, Seeley, Brooks-Gunn, & Lewinsohn, 2004). For example, early maturing females had significantly higher rates of Disruptive Behavior Disorders compared to on-time or late maturers during the highschool years (Graber et al., 1997). Additionally, in young adulthood, early maturing females had higher rates of elevated Antisocial Personality traits in comparison to ontime maturers (Graber et al., 2004). Importantly, finding increased rates of externalizing disorders and symptoms in early maturing girls does not provide enough information to either support or reject any of the timing models since they all predict negative outcomes associated with early maturation in females. In the Dunedin study of conduct disorder,

PAGE 17

10 delinquency, and aggression, it was found that early maturing females were engaging in higher rates of delinquent activity than their peers (Moffitt et al., 2001). No studies as yet have found that late maturation among females was a risk for aggression or delinquency. In contrast to females, the pattern of results for studies that have examined the relationship between pubertal timing and adjustment difficulties in males support variant two of the deviance model, with both early and late-maturing males showing signs of adjustment difficulties in comparison to their on-time peers. Most of the pubertal timing research that has been done with males has focused on substance use, such as alcohol, tobacco, and marijuana, and also aggression and delinquency (for exception, see Graber et al., 2004). Studies have indicated that late maturation in boys is associated with higher rates of both violent and nonviolent delinquent acts, as well as higher rates of substance use in late adolescence and young adulthood, and delinquent or criminal behavior in adulthood. Early maturation in males is associated with negative outcomes as well, including more advanced drinking habits by mid-adolescence (Andersson & Magnusson, 1990) as well as higher rates of both violent and nonviolent delinquent behavior (CotaRobles, Neiss, & Rowe, 2002; Williams & Dunlop, 1999). In the OADP (Graber et al., 2004), early maturing males did not have higher rates of disorders or symptoms than their on-time or late peers; however, they did have higher rates of tobacco use in mid adolescence and by young adulthood, 40% were daily smokers. Huddleston and Ge (2003) report that while some research has found early maturation in males to be associated with adjustment and health in adulthood, there are some inconsistent findings. For instance, early maturing males have been found to have higher rates of deviance, to be more likely to use tobacco and drugs, to engage in sexual activity, and to have higher levels of alcohol consumption. Thus, although late

PAGE 18

11 maturation is associated with elevated conduct problems and substance abuse in late adolescence and young adulthood, these youth are not distinguishable from on-time maturers in early adolescence. In contrast, early maturing males may have elevated problems in comparison to other youth during early adolescence when aggressive and delinquent behaviors begin to increase more dramatically. However, less research has been conducted regarding pubertal timing effects on adjustment outcomes in males compared to females; therefore, there is a n eed for further research with males to clarify this relationship. Racial/Ethnic Differences Research on rates of aggressive and de linquent behaviors has shown racial and ethnic disparities. For example, compared to Whites, African American males do not show a drop off in violent aggressive behavior between the ages of 22 and 30, and nearly twice as many African American males continue to engage in violent and delinquent behaviors into adulthood than White males (Coie & Dodge, 1998). In general, Latinos show higher rates of aggressive behaviors compared to Whites but lower rates compared to African Americans although rates may vary by immigration status or specific ethnic origin (e.g., Mexican American, Puerto Rican.) . However, differences between races and ethnicities in rates of these behaviors are not especially informative on their own. Accordingly, the majority of research that has examined rates of aggression and delinquency by race or ethnicity has focused on differential exposure to contextual factors that are associated with higher rates of these behaviors, such as poverty, exposure to abuse, and single parent households, as well as the impact of daily stressors associated with minority status, to attempt to account for any disparities between races.

PAGE 19

12 Minority ethnic and racial groups in the United States are exposed to higher levels of daily stress associated with discrimination and prejudice compared to their White peers and this has been associated with poorer mental health outcomes and adjustment difficulties (for review, see Contrada et al., 2000). Additionally, many minority groups within the United States are systematically excluded from critical resources and power (Garcia Coll & Garrido, 2000). This differential exposure to risky contexts and stress is a key contributor to the higher rates of aggr essive and delinquent behaviors among African Americans and Latinos compared to Whites. Numerous studies have illustrated the differential effect of risk factors and protective factors on the development of aggr essive or delinquent behavior for different racial or ethnic groups. Some of these factors are poverty compared to economic affluence and parenting behaviors. It is beyond the scope of the present investigation to account for or investigate each of these. However, several investigations have found that risk and protective factors associated with the development of aggressive and delinquent behaviors do not necessarily hold across all races and ethnicities and therefore, it is imperative to evaluate the impact of exposure to different risk or protective factors on the development of externalizing behaviors across different ethnic and racial groups. Current timing research. Most of the research that has been conducted regarding the association between pubertal timing and aggressive behavior has only examined this relationship among non-Hispanic white adolescents. As indicated previously, there are higher rates of aggressive and delinquent behaviors among minority groups, including African Americans and Latinos, compared to Whites. Additionally, there is some evidence to indicate that different risk factors and protective factors associated with aggression and delinquenc y may have different effects by race or

PAGE 20

13 ethnicity. The few studies that have examined the effects of early pubertal maturation with different racial or ethnic groups do support the idea that early maturation may result in different outcomes for different groups and encourage further investigation into this area. For example, pubertal maturation typically begins earlier on average for African American girls compared to their White counterparts (Herman-Giddens et al., 1997). This means that early maturing African American females will be maturing even before their early maturing White counterparts; hence, early maturing African American girls are likely to be at the most extreme end of the continuum of pubertal maturation. This would lead one to hypothesize that early matu ring African American girls would have the highest rates of aggressive and delinquent behaviors associated with their pubertal timing. As yet it is unclear if a similar race difference in the timing of puberty exists among boys (Reiter & Lee, 2001). African American males may also mature earlier than their peers but there are limited data to assess this issue. If this is the case, early maturing African American boys might have higher rates of these behaviors than early maturing boys of other racial or ethnic backgrounds. To date, few investigations have examined whether timing effects were comparable across racial or ethnic groups. In a study of girlsÂ’ substance use based on the ADD Health data set, early maturation in general was associated with higher rates of substance use for all races. Research on the association between substance use and early pubertal maturation among African American girls shows lower prevalence rates compared to early maturing White girls despite the fact that the early maturing African American girls did mature earlier as a group than the early maturing White girls (Lanza & Collins, 2002). This finding provides direct evidence against the hypothesis that timing effects do not differ across race. In a study of boys, Cota-Robles and colleagues (2002)

PAGE 21

14 found that, in general, early maturing males had higher rates of violent and nonviolent delinquent behaviors regardless of race when compared to on-time maturing males. Interestingly, the findings differed across et hnic and racial groups as well as depending on if the delinquent behaviors were violen t or not. For nonviolent delinquent behaviors, White and Mexican American boys had significantly higher rates than African American boys. For violent delinquent behaviors, Me xican American and African American boys had significantly higher rates than White boys. In a study of African American males, Ge, Brody, Conger, Simons, and Murry (2002) found early pubertal maturation to be associated with higher rates of externalizing behaviors. It is important to note however, that across these studies, early maturers consistently have higher rates of adjustment difficulties compared to their on-time peers. Again, these findings support the association between early maturation and negative outcomes; however, they also suggest that the severity of the negative outcomes associated with early maturation may vary across race or ethnicity. Rates of Aggression in Context Research has examined the impact that living in an urban compared to a rural setting can have on the development of aggressive and delinquent behaviors. Generally, the consensus is that living in crowded, urban, inner city environments is associated with higher exposure to violent crimes and higher rates of participation in violent, aggressive, and delinquent behaviors (Coie & Dodge, 1998; Farrington, 2004). Within urban areas, high-crime neighborhoods have higher rates of aggressive and delinquent adolescent boys than the low-crime neighborhoods. It is important to note that there is a higher prevalence of African Americans and Latinos who live in urban areas than Whites. Additionally, many of the urban, inner city neighborhoods are characterized by high rates

PAGE 22

15 of poverty, disorganization, and deteriorati on. However, the direction of the effect between these deprived neighborhoods and the higher rates of aggressive and delinquent behaviors has yet to be firmly established. The conditions of the neighborhoods could be causing the increased rates of aggressive behaviors or it could be that aggressive and delinquent individuals tend to live in deprived areas (Farrington, 2004). Regardless, it has been established that urban neighborhoods have the highest rates of violent, aggressive, and delinquent behaviors and consequently are important locations for research attempting to understand the factors involved in the development of these constructs. Summary In summary, there are numerous demographic factors that are associated with higher rates of aggression and delinquency including but not limited to living in a single parent household, racial or ethnic minority st atus, and living in an urban environment. Given the risk associated with these demographic factors, future research that attempts to look at the development of aggressive and delinquent behaviors in adolescence should consider these variables either as main variables of interest, or at the least, as control variables. The present investigation focuses on young, minority adolescents in a large metropolitan area in order to examine the role of a risk factor for aggression and delinquency that has been previously under-studied in this population, that is, pubertal timing. Psychosocial Factors One social factor that has received a good deal of attention in regards to the development of aggressive and delinquent behaviors in adolescence is associating with aggressive or delinquent peers. It has been well established in the literature that associating with delinquent peers is an important predictor of initiation of and continued

PAGE 23

16 participation in aggressive and delinquent behaviors in adolescence (Farrington, 2004). Studies that have attempted to examine the direction of this association support the idea that associating with delinquent friends infl uences an adolescentÂ’s own aggressive and delinquent behaviors as well as the reverse, that more delinquent adolescents tend to associate with delinquent friends (Farrington, 2004). While many studies have found that associating with deviant peers is linked to aggression and delinquency, the role that early pubertal timing plays in that relationship is still unclear. Of the studies that have examined this relationship, most show that early maturing adolescents report more friends who engage in delinquent activity than their ontime peers. Stattin and Magnusson (1990) found that early maturing girls were associating with older peers. It is assumed however, that these older peers were engaging in more problem behaviors and delinquency. These findings seem to suggest that the types of peers that an adolescent associates with are central to the relationship between early maturation and negative adjustment outcomes. Less is known about the role that associating with deviant peers may play in the relationship between these problem behaviors and pubertal timing in males or females and males from non-white backgrounds. The Present Study The current study examined some of the issues that remain unclear regarding early pubertal maturation and aggressive or delinquent behaviors. The sample consisted of African American and Latino males and females drawn from a large metropolitan community. These participants were tested longitudinally, once a year for 3 years beginning in 6th grade. The goals of this study were, first, to determine if early pubertal timing confers the same risk for aggressive or delinquent behavior among these different

PAGE 24

17 groups; and second, to examine different pathways between early maturation and aggressive or delinquent behaviors, specifically associating with deviant peers. The present investigation compared those who mature early with all other youth; I chose to focus on early maturation for two reasons. First, there is no evidence to support an association between late maturation and aggression or delinquency in females. Second, although late-maturing males have been shown to have higher rates of aggression and delinquency than their on-time peers, these negative outcomes associated with late-maturing males do not emerge until older ages. Given the current study’s focus on the early adolescent time period, examining late maturation is beyond the scope of the study. This investigation was drawn from a larger evaluation study of a drug and violence prevention program. As such, a prior investigation has examined gender differences in aggression and delinquency in the larger study (Nichols, Graber, BrooksGunn, & Botvin, 2004). The present investigation used a subsample of the study for whom information on pubertal timing was available and focuses specifically on the role of pubertal timing in the development of aggression and delinquency. Variations of aggression and delinquency by gender and race/ethnicity in this subsample were assessed for descriptive purposes to provide background for the analysis of pubertal timing. Control variables in all analyses included school type (public vs. parochial) and family structure (two parent household vs. other). The following research questions were addressed in the present study: •Is early pubertal timing associated with aggressive or delinquent behaviors over 6th, 7th, and 8th grades? •Does the association between early pubertal timing and aggressive or delinquent behaviors differ by gender over 6th, 7th, and 8th grades?

PAGE 25

18 •Does associating with deviant peers mediate or moderate the relationship between early pubertal maturation and aggression or delinquency over 6th, 7th, and 8th grades? •Does the effect of associating with deviant peers, on the relationship between early pubertal maturation and aggression or delinquency, differ by gender over 6th, 7th, and 8th grades? •Does the association between early pubertal timing and aggressive or delinquent behaviors differ by race/ethnicity over 6th, 7th, and 8th grades? •Does the effect of associating with deviant peers, on the relationship between early pubertal maturation and aggression or delinquency, differ by race/ethnicity over 6th, 7th, and 8th grades? •Are there any interactions between gender, race/ethnicity, and timing in regards to aggression and delinquency over 6th, 7th, and 8th grades? Early maturers were expected to report significantly higher rates of aggressive and delinquent behaviors, compared to on-time and late maturers. Early maturing females were expected to have higher rates of aggressive and delinquent behaviors compared to early maturing males. Individuals who reported associating with deviant peers were expected to report higher rates of aggressive and delinquent behaviors compared to those who did not associate with deviant peers. Associating with deviant peers was expected to either moderate or mediate the relationship between early pubertal maturation and aggressive or delinquent behaviors.

PAGE 26

19 METHODS Research Design The current study is part of a larger randomized clinical trial designed to expand and test the effectiveness of an already proven drug prevention strategy on violent and aggressive behavior. A total of 42 public and parochial middle schools in New York City participated in the evaluation study. All schools participated in baseline data collection activities with their 6th grade classes, prior to the intervention, and annual surveys in 7th and 8th grades; half the schools received prevention programming for 3 years. Participants The current study uses data from the larger study collected at baseline (6th grade), the 1-year follow-up (7th grade), and the 2-year follow-up (8th grade). Only participants assigned to the control condition at baseline ( N = 2663) were used in order to avoid contamination with potential intervention effects. Students were surveyed via two separate booklets across a 2-day time period (see description below). Those students who were absent for either of the data collection days were only able to complete the first booklet. Therefore 23% ( n = 519) of students who did not complete the second booklet at baseline were dropped from the current study. Additionally, the question assessing pubertal timing was included in two later surveys so only those youth who participated over time could be included in the present investigation. Of those students, individuals who were inconsistent in their reports of their pubertal timing were dropped from the current study to enhance the validity of the self-report pubertal timing measure (see

PAGE 27

20 following description) resulting in a group of 1173 students. Thirty percent ( n = 357) of this group indicated a racial or ethnic affilia tion other than African American or Latino. These other racial or ethnic groups were dropped from the sample because the group sizes were too small for meaningful analysis, leaving a final sample of 816. Of the remaining participants the average level of missingness was 0.5%. Maximum–likelihood procedures to impute the remaining missing data points were not conducted due to the extremely low number of missing data points. Tests of sample bias compared to those dropped from the analyses are reported in a later section. The mean age for the sample included in the current analyses was 11.66 years ( SD = .49) in the 6th grade with a range of 9.64 to 14.00. The sample included slightly more girls than boys, with 57% of the students being female. The majority of students were African American (66%) with the remainder of the sample consisting of Latinos (34%). Over half of the students came from two-parent households (56%). Youth were enrolled in public (87%) and parochial (13%) schools. Procedure A passive consent procedure approved by Weill Cornell Medical College’s IRB was used to inform parents about the nature of the study and to provide them with an opportunity to disallow their child’s participation. A consent form describing the focus of the larger study and the self-report survey was distributed in the schools for students to take home to their parents, as well as mailed directly to students’ homes. Students whose parents indicated they did not want them to participate in the self-report survey did not complete any of the data collection activities. The survey was divided into two separate booklets and data collection was conducted on two separate days during regular 40-minute class periods. A multi-ethnic

PAGE 28

21 team of three to five data collectors administered the questionnaire following a standardized protocol similar to those used in previous research (e.g., Botvin, Schinke, Epstein, & Diaz, 1994). Steps were taken to ensure the quality of self-report data. Identification codes rather than names were used to emphasize the confidential nature of the questionnaire and students were assured about the confidentiality of their responses. Carbon monoxide (CO) breath samples were al so collected at 6th, 7th, and 8th grade to enhance the validity of self-report data utilizing a variant of the bogus pipeline procedure developed by Evans and his colleagues (Evans, Hansen, & Mittlemark, 1977). While this measure was used to increase the validity of questions pertaining to cigarette smoking, studies have shown bogus pipeline procedures can also increase the validity of reporting on other problem behaviors (Tourangeau, Smith, & Rasinski, 1997). Measures Demographic Data Data concerning the characteristics of the participants were collected using standard survey items concerning gender, age, household structure, and race/ethnicity in 6th grade. A single dichotomous variable was created to capture the type of household structure where 1 indicates living in a tw o-parent household. Students from single and no-parent households were coded as 0. Anothe r single dichotomous variable was created to capture participantsÂ’ race or ethnic affiliation where 1 represented participants who were Latino and 0 represented participants who were Black. A single dichotomous variable was created to indicate if the adolescent was attending a public or a parochial school (where 1 represents participants who attended public school and 0 for parochial school students).

PAGE 29

22 Early Pubertal timing Timing was assessed by a single subjective item based on the question “Do you think your development was any earlier, later, or about the same as other “same gender” youth your age?” Response options included 1 ( Early ), 2 ( On-time ), and 3 ( Late ). This item was included in two later surveys and only those participants who were consistent in their responses to this item across both surveys were included in the present analyses. Due to the focus of the present investigation on the effects of early maturation only, a single dichotomous variable was created where 1 represents only participants who reported maturing early. All other responses were coded as 0. Aggression (Past Month) The same ten items were used to assess general aggression in the 6th grade ( = .92), 7th grade ( = .94), and 8th grade ( = .94). These items were drawn from the aggression scale of the Youth Self-Report (YSR, Achenback & Edelbrock, 1986). Students were asked how many times in the past month they had engaged in 10 incidents of aggressive behavior. Items included “Yelled at someone (you were mad at),” “Told someone off,” “Pushed or shoved someone on purpose,” and “Hit someone.” Response categories were on a 5-point scale. Response options included 1 ( Never ), 2 ( Once ), 3 ( 2-3 times ), 4 ( 4-5 times ), and 5 ( More than 5 times ). Items were averaged with higher scores indicating greater aggression. The response opti ons were changed from the 3-point scale in the YSR to the 5-point scale in this study to be consistent with the other measures in the survey. Delinquency (Past Year) Students were asked how many times in the past year they had engaged in 10 incidents of delinquent behavior. The same 10 items were used in the 6th grade ( = .85),

PAGE 30

23 7th grade ( = .88), and 8th grade ( = .90). These items were taken from a violence/delinquency scale developed by Elliot, Huizinga, and Menard (1989). Two separate subscales were created to distinguish violent delinquency (6 items; 6th grade = .79, 7th grade = .83, and 8th grade = .85) and nonviolent delinquency (4 items; 6th grade = .67, 7th grade = .77, and 8th grade = .80). Items for violent delinquency included “Thrown objects such as rocks or bottles at cars or people,” “Hit someone with the intention of hurting them,” and “Taken part in a fight where a group of your friends were against another group.” Items for nonviolent delinquency included “Purposefully damaged or destroyed property or things th at did not belong to you,” “Taken something from a store when the clerk wasn’t looking,” and “Taken something worth less than $50.” Response categories were on a 5-point scale. Response options included 1 ( Never ), 2 ( Once ), 3 ( 2-3 times ), 4 ( 4-5 times ), and 5 ( More than 5 times ). Within each subscale, items were averaged with higher scores indicating greater violent or nonviolent delinquency. Friend Delinquency (Past Year) Students were asked to indicate how many of their friends had engaged in seven incidents of delinquent behavior in the past year. The same seven items were used to assess friend delinquency in the 6th grade ( = .87), 7th grade ( = .91), and 8th grade ( = .92). These were taken from a violence/delinquency scale developed by Elliot et al. (1989). Items for friend delinquency included “Hit or threatened to hit someone without any real reason,” “Beat someone or fought someone physically if they were provoked (other than just playing around),” “Ruined or damaged something on purpose that wasn’t theirs,” and “Stolen something worth less than $50.” Response categories were on a 5-point scale. Response options included 1 ( None ), 2 ( Less than half ), 3 ( About half ),

PAGE 31

24 4 ( More than half ), and 5 ( All or almost all ). Items were rescored onto a scale of 0-4 and then summed to create a continuous measure where higher scores indicate associating with more friends who engage in a higher number of delinquent behaviors for each time point. To test for differences between students in the current study and those who were dropped from the analyses, t-tests were run on all continuous variables and chi-squared analyses were run on all background variables from the 6th grade data. Students who were dropped from the current analyses were more likely to be boys (2(1, N = 2266) = 15.17, p < .001; 52% vs 43%); and more likely to be Latino (2(1, N = 2285) = 10.21, p = .001; 41% vs 34%). They were also less likely to come from two-parent household (2(1, N = 2204) = 6.64, p = .011; 51% vs 57%); and more likely to attend public school (2(1, N = 2285) = 32.29, p < .001; 94% vs 87%) compared to the students included in the present investigation. The two groups did not differ on reported rates of associating with delinquent peers. There were also no significant differences between the participants in the current study and those that were dropped from analyses regarding reported rates of the frequency of aggressive behavior in th e past month or regarding reported rates of nonviolent delinquency; however, those student s who were dropped from analyses did report higher rates of violent delinquency compared to students included in the current study ( M = 1.50 vs M = 1.41, respectively). Thus, the sample used in the current study had equivalent or attenuated baseline rates of all outcome variables as the cross-sectional sample of participants at 6th grade. The result of this is that any sample bias that may be introduced in the reduced sample would only bias against finding significant effects. Analysis Plan Descriptive analyses were conducted to show the mean ratings of aggression, violent delinquency, and nonviolent delinquenc y for the two different ethnic/racial

PAGE 32

25 groups, genders, and the two pubertal timing groups. Preliminary analyses were conducted to determine whether multivariate or univariate analyses were appropriate for these data. A repeated measures doubly multivariate analysis of covariance (MANCOVA) controlling for school type and family structure was determined to be inappropriate for this data set due to violations of a variety of assumptions associated with MANCOVAs in general. These violations included lack of homogeneity of the variance-covariance matrices across design cells as indicated by a significant BoxÂ’s M test ( p < .001) as well as singularity among the dependent variables as indicated by the size of the correlations with one another. Despite the high correlations among the dependent variables in the study, it is still of theoretical and practical importance to evaluate each of the dependent variables separately due to the real world implications associated with each one. Therefore univariate analyses were conducted to evaluate the data. The association between early pubertal tim ing and aggressive behaviors, violent delinquent behaviors, or nonviolent delinque nt behaviors over time was examined via three 2 (timing) x 2 (gender) x 2 (race/ethnicity) x 3 (grade) repeated measures ANCOVAs, controlling for family structure and school type. Grade was the within subject variable (6th, 7th, & 8th) while gender, race/ethnicity, and timing were the between subjects variables. These analyses also provided information regarding any differences between males and females, or African Americans and Latinos in the association between early pubertal timing and aggressive behaviors, violent delinquent behaviors, or nonviolent delinquent behaviors over time and therefore addressed Research Questions 1, 2, 5, and 7. Mediating effects were examined using a series of regression analyses as set forth by Baron and Kenny (1986) to assess linkages of the mediation models. As shown in

PAGE 33

26 Figure 1, the mediation model involves two paths that predict the outcome variable: the direct impact of the independent variable which is pubertal timing (path c ) and the impact of the mediator which is friend delinquency (path b ). There is also a path from pubertal timing (IV) to friend delinquency (mediator–path a ). Friend delinquency functions as a mediator if a linear regression analysis is significant for path a , if a linear regression analysis is significant for path c , and when paths a and b are entered simultaneously, a previously significant relation between pubert al timing and aggressive behavior is no longer significant. This model was also tested for the violent and nonviolent delinquency outcome variables as well. Gender and race/ethnicity were included in all analyses to determine any differences between males and females or African Americans and Latinos regarding the relationship between pubertal timing and aggressive or delinquent behaviors. These analyses answered, in part, Research Questions 3, 4, and 6. Moderating effects were examined using hierarchical regression analyses to test friend delinquency as the moderator variable (controlling for school type and family structure). As seen in Figure 2, the moderator model has three paths that feed into the outcome variable of aggressive behavior: the impact of pubertal timing as a predictor (path a ), the impact of friend delinquency as a moderator (path b ), and the interaction or product of the two (path c ). The moderator hypothesis is supported if the interaction (path c ) is significant. As was the case with the mediator model, this model was also tested for the violent and nonviolent delinquency outcome variables. Gender and race/ethnicity were included in all analyses to determine any differences between males and females or African Americans and Latinos regarding the relationship between pubertal timing and aggressive or delinquent behaviors. Thes e analyses answered, in part, Research Questions 3, 4, and 6.

PAGE 34

27 Figure 1.Mediational model testing friend delinquency as the mediating variable Figure 2.Moderator model testing friend delinquency as the moderating variable

PAGE 35

28 RESULTS Descriptive Analysis Sample demographics are presented in Table 1 within pubertal timing group, as well as for the entire sample. The only notable difference is the higher percentage of African Americans that are early maturers compared to the percentage of African Americans within the total sample. Given previous research that has shown that African American females and potentially males tend to mature earlier on average compared to their white or Latino peers, this discrepancy is not unusual (see Graber, 2003, for a review). Table 1.Demographic data by pubertal timing in the 6th grade Early maturersOn-time/late maturersTotal sample Age, M (SD) 11.69 (.53)11.65 (.48)11.66 (.49) Gender, % female595657 Race, % African American736366 School type, % public888787 Family structure, % two parents575656 General aggressive behaviors were quite prevalent, with almost the entire sample reporting engaging in some form of aggressive behavior within the past month during 6th grade (94%), 7th grade (97%), and 8th grade (96%). Reported rates of violent and nonviolent delinquency were also fairly high in the 6th grade, with 56% of the sample having engaged in one or more acts of violent delinquency at least once in the past year and 53% having engaged in one or more acts of nonviolent delinquency at least once in the past year. However, reported rates of violent delinquency increased more dramatically over the next two years than rates of nonviolent delinquency. Seventy-two

PAGE 36

29 percent of the sample reported having engaged in one or more acts of violent delinquency at least once in the past year in 7th grade, increasing slightly to 73% in the 8th grade. In comparison, 64% of the sample reported having engaged in one or more acts of nonviolent delinquency at least once in the past year in 7th grade, decreasing slightly to 63% in the 8th grade. Table 2 shows the average rates of aggression, violent delinquency, and nonviolent delinquency within pubertal timing group and for the total sample across all three grades. Overall, average reported rates of nonviolent and violent delinquency are relatively low with most adolescents reporting engaging in violent and nonviolent delinquent acts once in the past year. Table 2.Means and standard deviations by pubertal timing for each grade 6th Grade7th Grade8th Grade M(SD)M(SD)M(SD) AggressionEarly maturers2.37 (1.09)3.10 (1.19)3.23 (1.21) On-time/late maturers2.18 (.99)2.86 (1.20)2.95 (1.23) Total sample2.23 (1.02)2.92 (1.20)3.03 (1.23) Violent delinquencyEarly maturers1.50 (.72)1.72 (.83)1.84 (.92) On-time/late maturers1.37 (.59)1.61 (.75)1.70 (.88) Total sample1.41 (.63)1.64 (.77)1.74 (.89) Nonviolent delinquencyEarly maturers1.34 (.54)1.62 (.74)1.68 (.87) On-time/late maturers1.30 (.47)1.54 (.72)1.59 (.81) Total sample1.31 (.49)1.56 (.73)1.61 (.82) Average rates of aggressive behaviors were higher with most adolescents reporting engaging in between 1 and 3 aggressive behaviors in the past month. While average rates of all three outcomes increase over time, there is a disproportionately larger increase that occurs between 6th grade and 7th grade compared to between 7th grade and 8th grade. Additionally, early maturers report higher average rates of aggression, nonviolent delinquency, and violent delinquency than their on-time or late-maturing counterparts in each grade. This is most conspicuous in regards to average rates of aggression in the past month. In 7th and 8th grade, early maturers are reporting having

PAGE 37

30 engaged in between 2 and 5 acts of aggressive behavior in the past month while their ontime or late-maturing counterparts are reporting having in engaged in between 1 and 3 aggressive behaviors in the past month. Correlations between the variables in this study at all three time points are presented in Table 3. As can be seen, the three indicators of aggression and delinquency have strong intercorrelations at each time of assessment and moderate to strong associations over time. Association of Pubertal Timing with Aggression and Delinquency Outcomes Research Questions 1, 2, and 5. Is early pubertal timing associated with aggressive or delinquent behaviors over 6th, 7t h, and 8th grades and does this association differ by gender or race/ethnicity? To evaluate the relationship between early pubertal maturation and aggression, nonviolent delinquency and violent delinquency over time, three 2 (timing) X 2 (gender) X 2 (race) X 3 (grade) repeated measures analyses of covariance (ANCOVAs) were conducted, controlling for whether or not the adolescent comes from a two parent household (family structure) and whether they go to a parochial or public school (school type). Grade was the within subjects variable (6th, 7th, and 8th) while gender, race/ethnicity, and pubertal timing were the between subjects variables. This analysis also provided information regarding any differences between males and females, or African Americans and Latinos in the association between early pubertal timing and aggressive or delinquent behaviors over time. A summary of the F ratios and effect sizes for each of the following three ANCOVAs is provided in Table 4.

PAGE 38

31Table 3.Intercorrelations among independent and dependent measures Variable12345678910111213141516171819 1School type---.002.011-.060-.095*.024.036.062.039-.089. 035.063.087.011-.116**.064.074.096*.092* 2Gender---.020.066-.027.069.047.153**.123**-.00 6.053-.022.089.119**.015.072.018.106*.087 3Pubertal timing---.087.003.089.081 .035.085-.026. 040.091*.050.064-.001.054.098*.046.072 4Race--.118**-.057-.141**-.084-.096*.150**-.027-. 129**-.048-.058.120**-.041-.135**-.041-.086 5Family structure 6th---.040-.076-.097*-.068.733**.004 0.030.014.646**.027-.036-.014-.020 6Delinquent friends 6th--.588** .528**.572**-.055.351**.306**.312**.368**-.044.314**.247**.283**.313** 7Aggression 6th-.616**.704**-.109*.361**.546**.400**.478**-.046.296**.428**.370**.405** 8Nonviolent delinquency 6th--.699**-.059.291**.313**.439**.389**-.020.247**.236**.378**.349** 9Violent delinquency 6th---.088.324**.411**.389**.516**-.041.322**.327**.365**.421** 10Family structure 7th---.049-.045.002-.046.742**.016-.038-.012-.068 11Delinquent friends 7th--.558**.555**.593**-.064.433**.344**.390**.375** 12Aggression 7th--.564**.668**-.007.377**.599**.424**.477** 13Nonviolent delinquency 7th--.676**.019.398**.375**.541**.442** 14Violent delinquency 7th---.021.471**.470**.486**.581** 15Family structure 8th--.001-.007.013-.059 16Delinquent friends 8th--.555**.576**.594** 17Aggression 8th--.562**.650** 18Nonviolent delinquency 8th--.736** 19Violent delinquency 8th -* Correlation is significant at the .01 level ** Correlation is significant at the .001 level

PAGE 39

32 Table 4. F ratios and effect sizes for univariate main effects of grade, gender, race/ethnicity, and pubertal timing on aggression, nonviolent delinquency and violent delinquency Grade (G) Gender (Gr) Race/Ethnicity (RE) Pubertal Timing (PT) A.Main effects F (2, 1550)2 F (1, 775)2 F (1, 775)2 F (1, 775)2Aggression17.14***.02.17.009.77**.016.10*.01 Nonviolent delinquency3.36*.0015.10***.02.46.003.74.01 Violent delinquency5.26**.0112.85***.021.85.007.33**.01 B. Two-way interaction effects G x GrG x REG x PT F (2, 1550)2 F (2, 1550)2 F (2, 1550)2Aggression1.61.001.01.001.00.00 Non-violent Delinquency.28.001.51.001.80.00 Violent Delinquency1.17.00.97.00.41.00 Gr x REGr x PTRE x PT F (1, 775)2 F (1, 775)2 F (1, 775)2Aggression3.12.00.94.00.51.00 Non-violent Delinquency2.69.00.52.002.27.00 Violent Delinquency.54.00.51.002.02.00 C. Three-way interaction effects G x Gr x REG x Gr x PT F (2, 1550)2 F (2, 1550)2Aggression.13.00.56.00 Non-violent Delinquency1.45.00.88.00 Violent Delinquency.35.001.70.00 G x RE x PTGr x RE x PT F (2, 1550)2 F (1, 775)2Aggression2.11.00.15.00 Non-violent Delinquency3.05*.00.01.00 Violent Delinquency1.32.00.22.00 D. Four-way interaction effects G x Gr x RE x PT F (2, 1550)2Aggression1.44.00 Non-violent Delinquency1.96.00 Violent Delinquency3.12*.00 Note: * p < .05, ** p < .01, *** p < .001.

PAGE 40

33 Aggression The analysis of aggression revealed a significant main effect of grade F (2, 1550) = 17.14, p < .001, a significant main effect for pubertal timing, F (1, 775) = 6.10, p = .014, and a significant main effect of race F (1, 775) = 9.77, p = .002. There were no significant interactions among grade, pubertal timing, gender, or race and no main effect for gender. Three follow-up independent samples t-tests were conducted between pubertal timing and aggression in the 6th, 7th, and 8th grades. Early maturers consistently reported significantly higher levels of aggression in the past month compared to on-time or late maturers in the 6th, 7th, and 8th grades (Table 2). Follow-up paired samples t-tests revealed that there were significant increases in aggression between 6th, 7th, and 8th grades (Table 2). Three follow-up independent samples t-tests between race/ethnicity and aggression revealed that African Americans consistently reported significantly higher levels of aggression in the past month compared to Latinos in the 6th grades( M = 2.34, M = 2.03), 7th ( M = 3.03, M = 2.70), and 8th ( M = 3.14, M = 2.79). These results support the hypothesis that early maturers report higher levels of aggression than on-time or late maturers across time; however, analyses did not find that the association between pubertal timing and aggression differed by gender or race/ethnicity. Additionally, overall reported le vels of aggression increased significantly at each time point, which was expected give n previous research. Finally, significantly higher mean reports of aggression were found among African American adolescents compared to Latinos.

PAGE 41

34 Nonviolent Delinquency To evaluate the relationship between early pubertal maturation and nonviolent delinquency over time, the same repeated measures ANCOVA, 2 (timing) X 2 (gender) X 2 (race) X 3 (grade), was conducted, again, controlling for family structure and school type. This analysis revealed a significant main effect of grade F (2, 1550) = 3.36, p = .035 and a significant main effect of gender F (1, 775) = 15.10, p < .001, as well as a significant three-way interaction between grade, pubertal timing, and race/ethnicity F (2, 1550) = 3.05, p = .047. Beginning with the grade and gender main effects, follow-up paired samples t-tests revealed that there were significant increases in nonviolent delinquency between 6th grades ( M = 1.31), 7th ( M = 1.56), and 8th ( M = 1.62). Three follow-up independent samples t-tests between gender and nonviolent delinquency revealed that males consistently reported significantly higher levels of nonviolent delinquency in the past year compared to females in the 6th grades ( M = 1.39, M = 1.24), 7th ( M = 1.63, M = 1.50) and 8th grades( M = 1.72, M = 1.54). In addition, the significant three-way interaction indicated that the relationship between pubertal timing and race/ethnicity regarding nonviolent delinquency differed by grade (Figure 3). Three follow-up univari ate ANCOVAs, 2 (timing) X 2 (gender) X 2 (race), were conducted to evaluate how reports of nonviolent delinquency changed over the 6th, 7th, and 8th grades by group, controlli ng for family structure and school type. There were neither significant main effects of pubertal timing or race/ethnicity in the 6th grade and 8th grade, nor was there a significant interaction between the two in either of those grades. However, there was a significant interaction between pubertal timing and race/ethnicity in the 7th grade F (1, 791) = 5.83, p = .016. Two follow-up independent samples t-tests were conducted. The first examined the association between pubertal

PAGE 42

35 1.2 1.3 1.4 1.5 1.6 1.7 1.8 6th Grade7th Grade8th GradeMean reports of non-violent delinquency Early AA On-time/Late AA Early Latinos On-time/Late LatinosFigure 3.Grade X timing X race interaction on nonviolent delinquency timing and nonviolent delinquency in the 7th grade separate by race/ethnicity. This revealed that there were no significant diffe rences in reports of nonviolent delinquency between early maturing and on-time or late-maturing African American adolescents in the 7th grade; however, early maturing Latinos reported significantly higher rates of nonviolent delinquency compared to on-time or late-maturing Latinos ( M = 1.71, M = 1.46). The second t-test examined the association between race/ethnicity and nonviolent delinquency in the 7th grade with the output split by pubertal timing. There were no significant differences between early maturing African Americans and Latinos regarding reported rates of nonviolent delinquency in the 7th grade; however, on-time or late-maturing African Americans reporte d significantly higher rates of nonviolent delinquency compared to on-time or late-maturing Latinos ( M = 1.58, M = 1.46). In summary, overall reports of nonviolent delinquency significantly increased each year and males reported significantly higher levels of nonviolent delinquency than

PAGE 43

36 females. In addition, the follow-up analyses of the three-way interaction between grade, pubertal timing, and race/ethnicity on nonviolent delinquency revealed that there were no difference between these groups in the 6th grade or 8th grade. In contrast, 7th grade proved to be particularly troublesome for early maturing Latinos, who evidenced a sharp increase in reports of nonviolent delinquency, especially in comparison to on-time or late-maturing Latinos, whose rates remained relatively low compared to other groups. Violent Delinquency The same repeated measures ANCOVA was also used to evaluate the relationship between early pubertal maturation and violent delinquency over time: a 2 (timing) X 2 (gender) X 2 (race) X 3 (grade) repeated measures ANCOVA, controlling for family structure and school type. This analysis revealed a significant four-way interaction between grade, gender, race/ethnicity, and pubertal timing F (2, 1544) = 3.12, p = .044 as well as significant main effects of grade F (2, 1544) = 5.26, p = .005, pubertal timing F (1, 772) = 7.33, p = .007, and gender F (1, 772) = 12.85, p < .001. The direction of the main effects was as expected based on follo w-up t-tests. Reports of violent delinquency increased significantly between 6th ( M = 1.41), 7th ( M = 1.64) and 8th grade ( M = 1.74). Males reported significantly higher rates of violent delinquency than females in 6th ( M = 1.49, M = 1.34), 7th ( M = 1.74, M = 1.56), and 8th grade ( M = 1.82, M = 1.67). Early maturers reported significantly higher rates of violent delinquency compared to ontime or late maturers in 6th ( M = 1.50, M = 1.37), 7th ( M = 1.72, M = 1.61), and 8th grade ( M = 1.84, M = 1.70). So, overall rates of violent delinquency increased significantly between 6th, 7th, and 8th grade, males reported higher rates of violent delinquency than females, and early maturers reported higher rates than on-time/late maturers.

PAGE 44

37 In order to evaluate the relationships between the variables in more detail at each point in time, three 2 (timing) X 2 (gender) X 2 (race) analyses of covariance (ANCOVAs) were conducted with violent delinquency in the 6th, 7th, and 8th grade as the outcomes of interest, controlling for family structure and school type (Figure 4). In the 6th grade, significant main effects of gender F (1, 802) = 13.26, p < .001, race/ethnicity F (1, 802) = 4.46, p = .035, and pubertal timing F (1, 802) = 4.91, p = .027 were found for reports of violent delinque ncy; however, there were no significant interactions. These main effects revealed that males reported significantly higher rates of violent delinquency than females in the 6th grade ( M = 1.49, M = 1.34), African Americans reported significantly higher rates than Latinos ( M = 1.45, M = 1.32), and early maturers reported significantly higher rates than on-time or late maturers ( M = 1.50, M = 1.37). Results from the 7th grade analysis differed slightly from the 6th grade in that there were only significant main effects of gender F (1, 789) = 13.84, p < .001 and pubertal timing F (1, 789) = 6.27, p = .013. Males continued to report significantly higher rates of violent delinquency in the 7th grade compared to females ( M = 1.74, M = 1.56) and early maturers reported significantly higher rates compared to on-time or late maturers ( M = 1.72, M = 1.61). Significant differences between racial/ethnic groups were not found in the 7th grade, nor were there any significant interactions. Lastly, the analysis of violent delinquency in the 8th grade revealed a significant main effect of pubertal timing F (1, 785) = 4.05, p = .045, where early maturers continued to report higher rates of violent delinquency compared to their on-time or late-maturing counterparts ( M = 1.84, M = 1.70). However, there were no significant main effects of gender or race/ethnicity, nor were there any significant interactions.

PAGE 45

38 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 6th Grade 7th Grade 8th GradeMean reports of violent delinquency Early AA On-time/Late AA Early Latinos On-time/Late Latinos 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 6th Grade 7th Grade 8th GradeMean reports of violent delinquency Early AA On-time/Late AA Early Latinas On-time/Late LatinasA B Figure 4. Grade X timing X race/ethnicity X gender interaction on violent delinquency. A) Rates of violent delinquency among females. B) Rates of violent delinquency among males.

PAGE 46

39 In summary, in the 6th grade, statistical differences were observed regarding reported rates of violent delinquency between genders, with males reporting higher rates than females; between racial/ethnic groups , with African Americans reporting higher rates than Latinos; and between pubertal timing groups, with early maturers reporting higher rates than on-time or late maturers. By 7th grade, only differences between genders and pubertal timing group were observed, both in the same direction as 6th grade. By 8th grade, only pubertal timing remained significant, again in the same direction as was observed in both 6th and 7t h grade. The lack of differences between racial and ethnic groups in the 7th and 8th grades can be attributed to the sharp increase in reported rates of violent delinquency among early maturing Latino males and females that occurred in the 7th grade. On-time or late-maturing Latino males also showed an increase in reported rates of violent delinquency in the 7th grade; however, they still reported the lowest rates among all males in the 7th grade. In contrast, on-time or late-maturing Latina females did not show the same high increase in reported rates of violent delinquency in either the 7th or 8th grades. Additionally, by 8th grade, all females, with the exception of on-time or late maturing Latinas, reported rates of violent delinquency comparable to the rates reported by males. Most notably, the early maturing Latina females, by 8th grade, reported the highest rates of violent delinquency among all other groups of adolescents except for early maturing African American males. Overall, it appears that while early maturation incurred a specific risk for increased rates of violent delinquency for all groups at all time points, this risk seems particularly influential for Latino males and females in 7th grade whereas not being an early maturer appeared protective against increases in this behavior for Latina females in 8th grade.

PAGE 47

40 The Role of Deviant Peers Research Questions 3, 4, and 6. Does associating with deviant peers mediate or moderate the relationship between early pubertal maturation and aggression or delinquency over 6th, 7th, and 8th grades and does this differ by gender or race/ethnicity? Initially, both the mediating and moderating effects of associating with deviant peers on the association between pubertal timing and the aggression and delinquency outcomes were evaluated by gender or race/ethnicity to determine any differences by group; however, none of these analyses reveal ed significant group differences. Hence, the following discussion will only encompass overall mediating and moderating effects of associating with delinquent friends on aggr ession or delinquency over the 6th, 7th, and 8th grades. Mediating effects of associating with delinquent peers were examined via a series of regression analyses as set forth by Ba ron and Kenny (1986). Figure 1 illustrates the two paths that predict the outcome variables: the direct impact of the independent variable pubertal timing (path c ) and the impact of the mediator friend delinquency (path b ). There is also a path from pubertal timing (IV) to friend delinquency (mediator), path a . Recall that in order for friend delinquency to function as a mediator, regression analyses should be significant for path a , and significant for path c , and when paths a and b are entered simultaneously, a previously significant association between pubertal timing and the outcome of interest is no longer significant. To control for effects of extraneous variables, family structure and school type were used as covariates. They were entered in the first step of all hierarchical regression analyses. In testing friend delinquency as a mediator, analyses for path a were run yielding a significant association between pubertal timing and friend delinquency in the 6th grade

PAGE 48

41 F (3, 808) = 2.84, p = .037, but not in the 7th grade or 8th grade. Therefore, early maturers reported associating with differentially higher rates of delinquent friends in the 6th grade only. Given this, further analyses testing the mediation model were conducted using friend delinquency in the 6th grade as the mediator on all outcomes at all time points. These analyses determined if early maturers were exhibiting higher rates of externalizing behaviors across all time points due to differentially greater association with deviant peers early on compared to their on-time or late maturing counterparts. Analyses for path c were run yielding a significant association between pubertal timing and aggression in the 6th grade [ F (3, 812) = 3.82, p = .010], 7th grade [ F (3, 797) = 3.71, p = .011], and 8th grade [ F (3, 796) = 3.34, p = .019]. In addition, there was a significant association between pubertal timing and violent delinquency, F (3, 811) = 3.68, p = .012, in the 6th grade only. Timing was not predictive of nonviolent delinquency at any point in time, nor was it predictive of violent delinquency in the 7th grade or 8th grade. This left only four outcomes for which the mediation model could continue to be tested, aggression in the 6th, 7th, and 8th grad e, and violent delinquency in the 6th grade. For all four of these outcomes, pubertal timing and 6th grade friend delinquency were entered simultaneously into the regression equations. In these analyses, only 6th grade friend delinquency predicted aggression in the 6th and 8th grade, as well as violent delinquency in the 6th grade. For all three of these outcomes the previously significant associations between pubertal timing and aggression and violent delinquency were nonsignificant. Therefore, associating with differentially higher rates of delinquent friends in the 6th grade fully mediated the relationship between pubertal timing and aggression in the 6th and 8th grades and violent delinquency in the 6th grade.

PAGE 49

42 Additionally, the association between pubertal timing and aggression in the 7th grade was partially mediated by friend delinquency in the 6th grade. These results support the idea that early maturers reported higher rates of aggression across all time points due to differentially greater association with deviant peers early on compared to their on-time or late maturing counterparts. Friend delinquenc y in the 6th grade also explained higher rates of violent delinquency among early maturers compared to on-time or late maturers in the 6th grade only. To test the moderating role of associating with delinquent friends in the link between pubertal timing and aggression, nonviolent delinquency, and violent delinquency in the 6th, 7th, and 8th grades, nine separate hierarchical multiple regression procedures were used to test for quadratic moderating effects as recommend by Baron and Kenny (1986). Quadratic effects would indi cate that high levels of reported friend delinquency would have a stronger influence on the relationship between pubertal timing and the outcome of interest than low levels of reported friend delinquency. Centered variables (i.e., mean deviation scores) were used to reduce multicollinearity between the interaction term and the main effects when testing for moderator effects as recommended by Aiken and West (1991). In each of these analyses, the control variables (family structure and school type) were entered at Step 1, the main effects (pubertal timing and friend delinquency) were entered at Step 2, th e interaction term (pubertal timing X friend delinquency) was entered at Step 3, friend delinquency squared was entered at Step 4, and finally the interaction between pubertal timing and friend delinquency squared was entered at Step 5. Quadratic moderation is supported if the final interaction between pubertal timing and friend delinquency squared significantly predicts the outcome of interest. As shown in Tables 5, 6, and 7, si gnificant quadratic moderating effects were

PAGE 50

43 43Table 5.Moderating effects of friend delinquency on the relation between pubertal timing and aggression 6th Grade7th Grade8th Grade BtR2 inc. BtR2 inc. BtR2 inc. Step 1 School type.07.02.79.01.14.041.32.01.18.051.71.01 Family Structure-.12-.06-2.04*-.02-.01-.32-.02-.01-.29 Step 2 Pubertal Timing.19.082.50*.34***.19.071.97*.31***.19.071.94.30*** Friend Delinquency.15.7716.83***.14.7617.98***.14.7518.65*** Step 3 Timing X Friend delinquency.04.091.73.00-.01-.03-.81.00-.01-.03-.77.00 Step 4 Friend Delinquency2.00-.22-4.63***.02***.00-.28-6.58***.04***-.01-.29-7.27***.05*** Step 5 Timing X Friend delinquency2.00-.14-2.53*.01*.00.00-.04.00.00-.03-.54.00 Model R2.38.36.36 Model F69.45***62.17***62.16***df7, 8017, 7867, 786 Note. B , , t, R 2 , F, and df reflect values from the final regression equation; inc. = increment. * p < .05. ** p < .01. *** p < .001.

PAGE 51

44Table 6.M oderating effects of friend delinquency on the relation between pubertal timing and nonviolent delinquency 6th Grade7th Grade8th Grade BtR2 inc. BtR2 inc. BtR2 inc. Step 1 School type.06.041.46.01**.15.072.36*.01.14.061.99*.01* Family Structure-.07-.07-2.30*.05.031.12.04.02.78 Step 2 Pubertal Timing.04.03.91.27***.00.00.04.30***.12.061.75.32*** Friend Delinquency.05.5611.50***.07.6314.44***.07.5513.49*** Step 3 Timing X Friend delinquency.01.02.44.01*-.01-.04-.80.00.01.02.46.01 Step 4 Friend Delinquency2.00-.01-.21.00.00-.11-2.56*.00*.00.03.67.00 Step 5 Timing X Friend delinquency2.00-.13-2.23*.01*.00.071.37.01.00-.12-2.53*.00* Model R2 .30.32.34 Model F 47.73***51.66***58.41*** df 7, 7987, 7877, 786 Note . B , , t, R 2 , F, and df reflect values from the final regression equation; inc. = increment. * p < .05. ** p < .01. *** p < .001.

PAGE 52

45Table 7.Moderating effects of friend delinquency on the relation between pubertal timing and violent delinquency 6th Grade7th Grade8th Grade BtR2 inc. BtR2 inc. BtR2 inc. Step 1 School type.03.02.58.01-.02-.01-.24.00.13.051.73.01* Family Structure-.06-.04-1.53-.03-.02-.65-.10-.06-2.00* Step 2 Pubertal Timing.17.123.60*** .32** *-.02-.01-.35.35***.22.113.08**.35*** Friend Delinquency.07.5912.42***.09.7217.16***.08.6415.91*** Step 3 Timing X Friend delinquency.05.203.83***.00.00-.01-.34.00.02.061.39.00 Step 4 Friend Delinquency2.00.01.12.00.00-.18-4.33***.02***.00-.07-1.85.00 Step 5 Timing X Friend delinquency2.00-.27-4.64***.02**.00.101.98*.00*.00-.15-3.32***.01*** Model R2 .35.37.37 Model F 61.14***65.22***65.48*** df 7, 8007, 7857, 783 Note. B , , t, R 2 , F, and df reflect values from the final regression equation; inc. = increment. * p < .05. ** p < .01. *** p < .001.

PAGE 53

46 found for aggression, nonviolent delinquency, and violent delinquency in the 6th grade; violent delinquency only in the 7th grade; and both violent and nonviolent delinquency in the 8th grade. To aid in interpreting these complicated regression analyses, friend delinquency was divided into 3 equal parts, consisting of low friend delinquency (friends engaged in between 0 and 9 delinquent behaviors in the past year), moderate friend delinquency (friends engaged in between 10 and 19 delinquent behaviors), and high friend delinquency (friends engaged in betw een 20 and 28 delinquent behaviors). Graphs depicting the average rates of outcomes that are moderated by friend delinquency are presented in Figures 5 for aggression, 6a and 6b for nonviolent delinquency, and 7a, 7b, and 7c for violent delinquency for adolescents who report low friend delinquency and those who report high friend delinquency by pubertal timing. As the graphs clearly demonstrate, associating with friends who e ngage in low rates of delinquent behaviors does not appear to affect differentially on-time or late maturers compared to early maturers. In contrast, associating with friends who engage in high rates of delinquent behaviors appears to have a more deleterious impact for on-time or late maturing adolescents compared to their early maturing peers regarding reported rates of aggression, nonviolent, and violent delinquency in the 6th grade, and nonviolent and violent delinquency in the 8th grade. Regarding violent delinquency in the 7th grade, it appears that early maturers who associate with friends who engage in high rates of delinquent behaviors report slightly higher rates of violent delinquency than on-time or late maturing adolescents. In sum, associating with delinquent friends moderated the relationship between pubertal timing and aggression, nonviolent, and violent delinquency in the 6th grade and nonviolent and violent delinquency in the 8th grade. When youth had low rates of friend

PAGE 54

47 1 1.5 2 2.5 3 3.5 4 4.5 On-time/LateEarlyOn-time/LateEarly Low friend delinquencyHigh friend delinquency 6th Grade Meandelinquency, they also engaged in low rate s of aggression and delinquent behaviors regardless of pubertal timing; however, associating with friends who engage in high rates of delinquent behaviors was more deleterious for on-time or late maturers compared to early maturers. A different pattern emerged in 7th grade regarding violent delinquency, where associating with friends who engage in high rates of delinquent behaviors appeared to impact early maturers more negatively than on-time or late maturers. Figure 5.Moderating effect of friend delinque ncy on reports of aggression in the 6th grade.

PAGE 55

48 0.5 1 1.5 2 2.5 3 Ontime/Late EarlyOntime/Late Early Low friend delinquencyHigh friend delinquency Mean 0.5 1 1.5 2 2.5 3 3.5 Ontime/Late EarlyOntime/Late Early Low friend delinquencyHigh friend delinquency Mean A B Figure 6.Moderating effect of friend deli nquency on reports of nonviolent delinquency. A) Moderating effect in the 6th grade. B) Moderating effect in the 8th grade.

PAGE 56

49 0.5 1 1.5 2 2.5 3 3.5 Ontime/Late EarlyOntime/Late Early Low friend delinquencyHigh friend delinquency Mean 0.5 1 1.5 2 2.5 3 3.5 Ontime/Late EarlyOntime/Late Early Low friend delinquencyHigh friend delinquency Mean 0.5 1 1.5 2 2.5 3 3.5 Ontime/Late EarlyOntime/Late Early Low friend delinquencyHigh friend delinquency MeanA B C Figure 7.Moderating effect of friend deli nquency on reports of violent delinquency. A) Moderating effect in the 6th grade. B) Moderating effect in the 7th grade. C) Moderating effect in the 8th grade.

PAGE 57

50 DISCUSSION Previous research has most often focused on the association between early maturation and negative outcomes for white, middle class females. The current study examined this association more fully by including both males and females in the analyses, as well as examining associations among an urban sample of African American and Latino adolescents. The first aim of the current study was to examine the effects of early pubertal maturation on reports of aggression and delinquency over the 6th, 7th, and 8th grades, as well as evaluate if these effects differed by gender or race/ethnicity. As hypothesized, early maturers reported significantly higher rates of aggression, violent delinquency, and nonviolent delinquency compared to on-time or late maturers at all points in time regardless of gender or race/ethnicity. Overall rates of aggression were fairly high in this sample, a finding that has been shown in previous samples of urban, minority adolescents (Cotton et al., 1994; Farrell, Kung, White, & Valois, 2000). Rates of violent and nonviolent delinquency were also high, although not at the same rate of frequency as aggression, which is also not unusual given previous research (Coie & Dodge, 1998; Farrington, 2004). Rates of these outcomes also increased predictably across the 6th, 7th, and 8th grades. However, it is important to note that while early maturation is associated with higher rates of these externalizing behaviors in general; the severity of negative outcomes associated with early maturation varied across race/ethnicity. For instance, early maturation appeared to have a stronger negative influence for Latinos than for African

PAGE 58

51 Americans in regards to nonviolent delinquency in the 7th grade and violent delinquency in both the 7th and 8th grades. This does not necessarily mean that Latinos reported higher rates of these outcomes compared to African American adolescents; in fact, among on-time or late maturers, African Americans consistently report higher rates of aggression, violent and nonviolent delinquency. What this does mean is that for Latinos there is a greater disparity between early maturers compared to on-time or late maturers in reports of nonviolent and violent delinquency than was found comparing early maturing African American adolescents with on-time or late maturing African American adolescents. Specifically, the transition from 6th to 7th grade was particularly problematic for early maturing Latinos, evidencing sharp increases in both nonviolent and violent delinquency. Additionally, the transition from 7th to 8th grade remained problematic for early maturing Latina females in regards to violent delinquency, as their reported rates continued to increase to levels higher than all other groups with the only exception being early maturing African American males. Contrary to their early maturing counterparts, on-time or late maturing Latina females consistently reported the lowest rates of violent delinquency. The same holds true for on-time or late maturing Latino males only not to quite the same degree. This is, to a great extent, most illustrative of why early maturation is associated with more severe outcomes for Latinos compared to African Americans. These results expand upon the research of Cota-Robles and colleagues (2002) who also found higher rates of violent a nd nonviolent delinquent behaviors associated with early maturation in a sample of White, Mexican American, and African American males. However, contrary to Cota-Robles et al., the current study found that the impact of early pubertal maturation did vary by race as described previously. There are a few

PAGE 59

52 explanations for this difference. First, the current study assessed rates of aggression, nonviolent, and violent delinquency longitudinally over the 6th, 7th, and 8th grades while Cota-Robles et al. evaluated a cross-sectional sample of adolescents at one time point. Hence, the current study was able to capture age related changes over time while CotaRobles et al. did not evaluate age differe nces. Additionally, the current study focused on both males and females from urban communities roughly between the ages of 11 and 14 in 6th grade, while Cota-Robles et al. used a nationally representative all male sample that ranged in age from 11 to 17. Finally, Cota-Robles et al. used a sample of Mexican Americans specifically, whereas the current study did not differentiate between different ethnicities within the Latino population. All of these factors may explain why this study found interactions between pubertal timing and race in regards to violent and nonviolent delinquency while they did not. While the negative effects of early maturation for White females is well documented in the literature for both internalizing and externalizing symptoms and behaviors (Graber et al., 1997, 2004), the current study extended these findings by showing that early maturation is not only a risk factor for females regarding externalizing behaviors such as aggression and delinquency, but that it is also a risk factor for males as well. Additionally, the effects of early matu ration did not seem to differ significantly by gender, with the only exception being Latina females in the 8th grade reporting significantly higher rates of violent delinquency compared to on-time or late maturing Latina females while Latino males did not show this same pattern. This provides modest evidence against the deviance model of pubertal timing. Recall that the deviance model attempts to illustrate the connection between pubertal timing and adjustment problems by stating that being different from the majority of same-age peers in the timing of puberty

PAGE 60

53 places an adolescent at risk for negative adjustment outcomes (Brooks-Gunn et al., 1985). Therefore, only early maturing females and late maturing males should show adjustment problems related to pubertal timing because they are the two groups that are the most deviant from their same-age peers. The current study found virtually the same level of externalizing behaviors among early maturing males and females. Unfortunately, given that the current study only examined the negative effects associated with early maturation, our data can neither support nor reject the stage termination model or variant 2 of either the deviance model or the stage termination model because all three of these models predict negative adjustment outcomes for both early maturing males and females. It would be beneficial for future research to examine if late pubertal maturation is associated with higher rates of aggression or delinquency specifically for an ethnically diverse group of males and females during mid to late adolescence when problems for late maturers seem to emerge. The second aim of the current study was to move beyond the demonstration of main effects and begin to investigate some of the pathways underlying the association between early pubertal maturation, aggression, and delinquency in adolescence. This has been a topic of interest in the literature regarding the association of negative outcomes with early maturing white females. Other studies have found that associating with older male peers is related to negative outcomes for early maturing white females compared to other girls (Stattin & Magnusson, 1990). It was assumed in previous research that these older male peers were also probably deviant although the deviancy of these older peers was not explicitly evaluated. Therefore, while previous research has found moderating effects of associating with older male peers on negative adjustment in early maturing females, association with deviant peers has not been explicitly examined for moderating

PAGE 61

54 or mediating effects. The current study expanded upon previous findings by examining the mediating and moderating effects of associating with deviant peers in a large, racially and ethnically diverse, urban sample of both males and females. Mediating effects of associating with higher rates of deviant friends in the 6th grade were found for aggression at all points in time and violent delinquency in the 6th grade only. Interestingly, timing was not predictive of friend delinquency in either the 7th or 8th grade. Hence, 6th grade was the only time point during which early maturation was associated with differentially higher rates of delinquent friends compared to on-time or late maturers. This is congruent with previous research that examined a sample of 1012-year-old African Americans from disa dvantaged neighborhoods and found that early maturers were significantly more likely to affiliate with deviant peers (Ge et al., 2002). This age range roughly corresponds with the 6th grade sample studied here. No other research has examined the association between pubertal timing and friend delinquency in an older sample for these constructs. Associating with higher rates of delinquent friends in the 6th grade explained the relationship between early pubertal timing and higher reports of aggression in the 6th, 7th, and 8th grade, as well as higher reports of violent delinquency in the 6th grade. This lends support to the idea that one of the pathways between early maturation and negative outcomes in adolescence is differential association with deviant peers early on. This appears to place adolescents on a negative trajectory early on, which results in higher rates of problematic outcomes at later points in time compared to their on-time or late maturing counterparts. Friend delinquency was also evaluated to determine any potential moderating effects between early pubertal maturation and increased rates of aggression, violent, and

PAGE 62

55 nonviolent delinquency in the 6th, 7th, and 8th grades. Results of these analyses found that associating with delinquent friends moderates the relationship between pubertal timing and aggression in the 6th grade, nonviolent delinquency in the 6th and 8th grades, and violent delinquency in the 6th, 7th, and 8th grades such that youth who associate with friends who engage in low rates of delinquent behaviors also tend to engage in low rates of these behaviors regardless of pubertal timing; however, associating with friends who engage in high rates of delinquent behavior appears to be more deleterious for ontime or late maturers compared to early maturers. Examination of the changes in reported rates of the preceding outcomes at different levels of friend delinquency seem to show that on-time or late maturers are more negatively affected as friend delinquency increases than are early maturers. One explanation for the previous finding may be that since this is a study of early adolescence (6th, 7th, and 8th graders) perh aps there is something different about ontime or late maturers who would associate with friends who engage in high rates of delinquent behavior compared to early maturers within the same grade. For example, ontime or late maturing 6th graders who report high levels of friend delinquency may be fundamentally different from their peers. It is hypothesized that one of the reasons that early maturation is associated with negative adjustment outcomes is due to the fact that early maturing adolescents physically appear older than they actually are, which may lead to association with older deviant peers and exposure to adult-like situations at an earlier age. Hence, these individuals may be at risk simply because of their outward appearance. In contrast, the mechanisms underlying why on-time or late maturing adolescents would be associating with high rates of delinquent peers may be indicative of larger problems. It would be beneficial fo r future studies to address this issue by

PAGE 63

56 evaluating differences between early maturing and on-time or late maturing adolescents who associate with friends who engage in hi gh rates of delinquent behaviors to determine if those on-time or late maturers may have more environmental risk factors associated with negative adjustment compared to the early maturers. It is important to note that association with delinquent friends also moderated the relationship between early pubertal maturation and violent delinquency in the 7th grade; however, this showed a different pattern compared to all of the other significant moderation effects. Adolescents who associate with friends who engage in low or moderate rates of delinquent behaviors in the 7th grade engage in similar rates of violent delinquency regardless of pubertal timing. It is only when adolescents associate with friends who engage in high rates of deli nquent behaviors that a difference appears between pubertal timing groups. However, the opposite pattern emerges where associating with delinquent peers appears to be more deleterious for early maturers compared to on-time or late maturers. It is interesting that this anomaly appears in the 7th grade for violent delinquency, considering that the moderating effect of friend delinquency is not atypical in the 6th or 8th grade for violent delinquency compared to the other outcomes of interest. Given that the transition from 6th to 7th grade is indicative of the largest increase in rates of all of the externalizing behaviors examined in this study, perhaps this anomaly is indicative of larger issues that impact the development of delinquent behaviors at this point in time. The transition from 6th to 7th grade may be a sensitive period in adolescent development regarding the initiation of aggressive and delinquent behaviors. Future research should attempt to elucidate more fully the factors that contribute to the development of these externalizing behaviors at this point in time. Additionally, these results suggest that future interventions attempting to reduce the rates

PAGE 64

57 of these problematic behaviors should ideally begin before the transition into 7th grade takes place. It should be noted that all youth in the current study were enrolled in middle schools and did not change schools between 6th and 7th grades. Limitations A limitation of this study is a focus on adolescents from urban communities only. Future studies would need to be conducted to determine if any results found in the present investigation apply in suburban or rural contexts. Additionally, the measures included in the present study were obtained via a self-report survey format. While steps were taken to increase the validity of the reports of aggressive and deviant behaviors, the current study is still limited to the adolescentÂ’s self-report of these behaviors. While the focus on an ethnic minority sample is a strength of the current study due to the lack of literature regarding the effects of pubertal timing for races and ethnicities other than white adolescents, the current study is also limited by the lack of a white comparison group. While scarce, previous literature seems to indicate that there may be differences between whites and minority adolescents regarding the severity of the impact of pubertal timing on initiation of substance use (Lanza & Collins, 2002). Also, less is know about white adolescents from urban environments. It would be beneficial for future research to include a white comparison group along with minority groups to clarify differential effects of pubertal timing. Finally, recent research that has examined aggression in adolescence has distinguished between overt forms of aggression (e.g., fighting) and relational forms of aggression (e.g., rumor spreading; Crick, 1996; 1997; Crick & Grotpeter, 1995). We are unable to do this in the current study because the larger prevention study did not include information regarding rates of relational aggression.

PAGE 65

58 Implications The results of this study provide useful information regarding the identification of at risk groups and more specific information regarding why they are at risk. Early maturation is a risk factor for the development of aggressive and delinquent behaviors among both male and female African Americans and Latinos, particularly for Latinos. Additionally, both mediating and moderating eff ects of associating with delinquent peers were evaluated to provide the most complete picture of the role of delinquent peers in the association between early maturation and aggressive or delinquent behaviors. The results of these analyses provided information about how associating with deviant peers early on explains the relationship between early maturation and aggression over time and early maturation and violent delinquency in the 6th grade. They also allow us to determine how the impact of associating with delinquent peers might differ between early and ontime or late maturers as well. Early maturation places individuals on a negative trajectory early on resulting in higher rates of problematic behaviors in later years and this is at least partially attributable to higher rates of association with deviant peers in the 6th grade. This information can be applied to intervention efforts in order to prevent poor outcomes more effectively. By targeting an intervention towards early maturers specifically, that focuses on developing nondelinquent peer friendships, it may be possible to attenuate some of the negative outcomes associated with early maturation in later years. The practical significance of even a slight reduction in rates of aggression, violent, and nonviolent delinquency among adolescents would prove to be of great value socially. This is especially obvious regarding both violent and nonviolent delinquency. Reducing the occurrence of property destruction, theft, and physical assault not only benefits society at

PAGE 66

59 large, but also may reduce the probability that the adolescent will continue to engage in these behaviors or escalate in severity of delinquency. Additionally, the moderating effects of friend delinquency point to benef its of offsetting engagement with delinquent peers for on-time and late maturing youth as well. Regarding potential interventions, associating with delinquent peers is not the only factor that may be attributing to higher rates of negative outcomes among early maturers. In line with the stage termination model of pubertal timing, it has been hypothesized that early maturers may be at an increased risk for negative adjustment because they are exposed to more adult-like situations at an earlier age by virtue of their more mature physical appearance; however, they may not have developed the skills necessary to effectively cope with the cha llenges of these situations (Brooks-Gunn et al., 1985). It would be beneficial for future research to examine various social-cognitive skills such as the decision-making abilities, assertiveness, or communication skills of early maturing adolescents and their peers to determine if there are any differences between groups. This could also inform future intervention strategies for adolescents who are at-risk based on early pubertal maturation.

PAGE 67

60 REFERENCES Achenbach, T. M., & Edelbrock, C. S. (1986). TeacherÂ’s report form . Burlington, VT. Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions . Thousand Oaks, CA: Sage. Andersson, T., & Magnusson, D. (1990). Biological maturation in adolescence and the development of drinking habits and alcohol abuse among young males: A prospective longitudinal study. Journal of Youth and Adolescence, 19, 33-41. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1181. Botvin, G. J., Schinke, S. P., Epstein, J. A., & Diaz, T. (1994). The effectiveness of culturally-focused and generic skills training approaches to alcohol and drug abuse prevention among minority youth. Psychology of Addictive Behaviors, 8, 116-127. Brooks-Gunn, J., Petersen, A. C., & Eichorn, D. (1985). The study of maturational timing effects in adolescence. Journal of Youth and Adolescence, 14 (3), 149-161. Coie, J. D., & Dodge, K. A. (1998). Aggression and antisocial behavior. In W. Damon & N. Eisenberg (Eds.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (5th ed., pp. 779-862). New York: John Wiley & Sons. Contrada, R. J., Ashmore, R. D., Gary, M. L., Coups, E., Egeth, J. D., Sewell, A., Ewell, K., Goyal, T. M., & Chasse, V. (2000). Ethnicity-related sources of stress and their effects on well-being. Current Directions in Psychological Science, 9, 136-139. Cota-Robles, S., Neiss, M., & Rowe, D.C. (2002). The role of puberty in violent and nonviolent delinquency among Anglo Ameri can, Mexican American, and African American boys. Journal of Adolescent Research, 17 (4), 364-367. Cotton, N. U., Resnick, J., Browne, D. C., Martin, S. L., McCarraher, & Woods, J. (1994). Aggression and fighting behavior among African-American adolescents: Individual and family factors. American Journal of Public Health, 84 (4), 618-622.

PAGE 68

61 Crick, N. R. (1996). The role of overt a ggression, relational aggression, and prosocial behavior in the prediction of children’s future social adjustment. Child Development, 67 , 2317-2327. Crick, N. R. (1997). Engagement in gender normative versus nonnormative forms of aggression: Links to social-psychological adjustment. Developmental Psychology, 33 , 610-617. Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression, gender, and socialpsychological adjustment. Child Development, 66 , 710-722. Elliot, D., Huizinga, D., & Menard, S. (1989). Multiple problem youth: Delinquency, substance use, and mental health problems . New York: Springer-Verlag. Evans, R. I., Hansen, W. B., & Mittlemark, M. B. (1977). Increasing the validity of selfreports of smoking behavior in children. Journal of Applied Psychology, 62, 521-523. Farrell, A. D., Kung, E. M., White, K. S., & Valois, R. F. (2000). The structure of selfreported aggression, drug use and delinquent behaviors during early adolescence. Journal of Clinical Child Psychology, 29 (2), 282-292. Farrington, D. P. (2004). Conduct disorder, a ggression, and delinquency. In R. M. Lerner & L. Steinberg (Eds.), Handbook of adolescent psychology (2nd ed., pp. 627-664). New York: John Wiley & Sons. Garcia Coll, C., & Garrido, M. (2000). Minorities in the United States: Sociocultural context for mental health and developmental psychopathology. In A. J. Sameroff, M. Lewis, & S. M. Miller (Eds.), Handbook of developmental psychopathology (2nd ed., pp. 177-195). New York: Plenum. Ge, X., Brody, G. H., Conger, R. D., Simons, R. L., & Murry, V. M. (2002). Contextual amplification of pubertal transition effects on deviant peer affiliation and externalizing behavior among African American children. Developmental Psychology, 38 (1), 42-54. Graber, J. A. (2003). Puberty in context. In C. Hayward (Ed.), Gender differences at puberty (pp. 307-325). New York: Cambridge University Press. Graber, J. A., Brooks-Gunn, J., & Archibald, A. B. (in press). Links between puberty and externalizing and internalizing behaviors in girls: Moving from demonstrating effects to identifying pathways. In D. M. Stoff & E. J. Susman (Eds.), Developmental psychobiology of aggression . New York: Cambridge University Press.

PAGE 69

62 Graber, J. A., Brooks-Gunn, J., & Petersen, A. C. (1996). Transitions through adolescence: Interpersonal domains and context . Mahwah, NJ: Lawrence Erlbaum & Associates. Graber, J. A., Lewinsohn, P. M., Seeley, J. R., & Brooks-Gunn, J. (1997). Is psychopathology associated with the timing of pubertal development? Journal of the American Academy of Child and Adolescent Psychiatry, 36 (12), 1768-1776. Graber, J. A., Seeley, J. R., Brooks-Gunn, J., & Lewinsohn, P. M. (2004). Is pubertal timing associated with psychopathology in young adulthood? Journal of the American Academy of Child and Adolescent psychiatry, 43 (6), 718-726. Herman-Giddens, M. E., Slora, E. J., Wasserman, R. C., Bourdony, C. J., Bhapkar, M. V., Koch, G. G., & Hasemeier, C. M. (1997). Secondary sexual characteristics and menses in young girls seen in office practice: A study of pediatric research in office settings network. Pediatrics, 99, 505-512. Huddleston, J., & Ge, X. (2003). Boys at puberty: Psychosocial implications. In C. Hayward (Ed.), Gender differences at puberty (pp. 113-134). New York: Cambridge University Press. Lahey, B. B., Loeber, R., Quay, H. C., Applegate, B., Shaffer, D., Waldman, I., Hart, E. L., McBurnett, K., Frick, P. J., Jensen, P. S., Dulcan, M. K., Canino, G., & Bird, H. R. (1998). Validity of DSM-IV subtypes of conduct disorder based on age of onset. Journal of the American Academy of Child & Adolescent Psychiatry, 37, 435-442. Lanza, S. T., & Collins, L. M. (2002). Pubertal timing and the onset of substance use in females during early adolescence. Prevention Science, 3 (1), 69-82. Moffitt, T. E. (1993). Adolescence-limited and life-course persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P.A. (2001). Sex differences in antisocial behavior: Conduct disorder, delinquency, and violence in the Dunedin Longitudinal Study . New York: Cambridge University Press. Nichols, T. R., Graber, J. A., Brooks-Gunn, J., & Botvin, G. J. (2004). Gender differences in aggression and delinque ncy among middle school students from 6th to 7th grade. Manuscript submitted for publication. Pajer, K. A. (1998). What happens to “bad” girls? A review of the adult outcomes of antisocial adolescent girls. The American Journal of Psychiatry, 155 , 862-870. Reiter, E. O., & Lee, P. A. (2001). Have the onset and tempo of puberty changed? Archives of Pediatric and Adolescent Medicine, 155 , 988-989.

PAGE 70

63 Simmons, R. G., & Blyth, D. A. (1987). Moving into adolescence: The impact of pubertal change and school context . New York: Aldine. Stattin, H., & Magnusson, D. (1990). Paths through life: Vol. 2. Pubertal maturation in female development . Hillsdale, NJ: Erlbaum. Tourangeau, R., Smith, T. W., & Rasinski, K. A. (1997). Motivation to report sensitive behaviors on surveys: Evidence from a bogus pipeline experiment. Journal of Applied Social Psychology, 27 (3), 209-222. Williams, J. M., & Dunlop, L. C. (1999). Pubertal timing and self-reported delinquency among male adolescents. Journal of Adolescence, 22, 157-171.

PAGE 71

64 BIOGRAPHICAL SKETCH Sarah Lynne Bensenhaver was born on September 8, 1979, in Fairfax, Virginia, to Karen Bensenhaver Barrett-Perry. She has two younger siblings, her brother Jonathan and her sister Elyse. While her father was never a parental figure in her life, SarahÂ’s mother was a diligent and driven woman who always provided her children with not only the basic material needs but also a very warm and supportive home. Sarah learned the values of independence, patience, and persistence during childhood and adolescence by watching her mother successfully overcome adversity. She also learned the value and importance of family. After graduating high school, Sarah went on to Lord Fairfax Community College where she earned an Associate of Science degree. She then transferred to Virginia Polytechnic Institute and State University where she earned a Bachelor of Science in psychology. While at Virginia Tech, Sarah worked with elementary age students enhancing literacy and math skills, which contributed, in part, to her desire to study human development. She applied and was accepted into the developmental psychology graduate program at the University of Florida, where she is currently developing her skills as an academician and researcher. Her research interests include examining the psychological, biological, and social factors that impact the development of healthy/adaptive behaviors in adolescence and minimize maladaptive/pathological outcomes. She has received her masterÂ’s degree and is well on her way to completing all

PAGE 72

65 the remaining requirements for the program. She is scheduled to receive her Ph.D. in developmental psychology in 2007.