ADOLESCENT ALCOHOL USE AND DATING VIOLENCE PERPETRATION:
THREE STUDIES EXAMINING CONCURRENT AND LONGITUDINAL
RELATIONS ACROSS GRADES 8 THROUGH 12
Heathe Luz McNaughton Reyes
A dissertation submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the
Department of Health Behavior and Health Education
Chapel Hill
2009
Approved by
Vangie A. Foshee, PhD
Susan T. Ennett, PhD
Daniel J. Bauer, PhD
Carolyn T. Halpern, PhD
J. Michael Bowling, PhD
UMI Number: 3387971
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.
UMI 3387971
Copyright 2010 by ProQuest LLC.
All rights reserved. This edition of the work is protected against unauthorized copying under Title 17, United States Code.
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106-1346
Numerous studies suggest a link between alcohol use and adult partner violence, but research on how this relationship unfolds during adolescence is limited. The three studies comprising this dissertation each used a different theoretical lens to guide an empirical examination of the relations between alcohol use and physical dating violence perpetration using data from a longitudinal study spanning grades 8 through 12.
Study one (n=2272) used autoregressive latent curve models to examine several different theoretical models of the linkages between alcohol use and dating violence perpetration over time. Trajectories of alcohol use and dating violence were correlated and this correlation was reduced substantially after adjusting for the effects of common predictors. However, concurrent associations between the two behaviors persisted across nearly all grades. There was no evidence of prospective relations from alcohol use to dating violence or vice-versa.
Study two (n=2311) examined the role of heavy alcohol use in the developmental process of desistance from dating violence perpetration. Growth models were used to test iii the hypotheses that both early and continuing alcohol use would hinder desistance from dating violence during late adolescence. Contrary to expectations, the effects of early alcohol use on dating violence diminished over time. Although the contemporaneous effects of alcohol use on dating violence were significant across most grades, effects weakened during late adolescence and were stronger in the spring than in the fall semesters. Study three (n=2311) examined the hypothesis that increased exposure to violence would strengthen the relationship between heavy alcohol use and dating violence. Growth models were used to examine the main and joint effects of alcohol use and exposure to family, peer, and neighborhood violence on levels of dating violence across grades 8 through 12. Across all grades, the relationship between alcohol use and dating violence was stronger for teens exposed to higher levels of family conflict and friend dating violence. Prevention programs that target risk factors common to both dating violence and alcohol use may reduce involvement in both behaviors. Programs that seek to reduce alcohol-related dating violence should target younger teens and those exposed to family conflict or friend dating violence.
iv
To Rodrigo and Bella Lani
v
ACKNOWLEDGEMENTS
Completion of this dissertation would not have been possible without the unerring love, patience and encouragement of my husband Rodrigo. Thank you for all of the sacrifices you made that allowed me to complete this work.
I am grateful to all of my committee members for their wonderful support and mentorship. Over the past five years each of you have generously opened your doors to me whenever I needed guidance and have provided me with challenges and opportunities that have made me a better scholar and teacher. I feel very fortunate to have been able to work with you.
Support for this dissertation research was provided by an individual National
Research Service Award Pre-doctoral Fellowship awarded by the National Institute on
Alcohol Abuse and Alcoholism (F31AA017015), and by an institutional National
Research Service Award Pre-Doctoral Fellowship awarded by the National Institute on
Child and Human Development (T32 HD07376) to the Carolina Consortium on Human
Development at the Center for Developmental Science. The study that provided the data for this research was funded by the National Institute on Drug Abuse (R01DA16669, S.
T. Ennett, PI) and the Centers for Disease Control and Prevention (R49CCV423114, V.
A. Foshee, PI)
vi
TABLE OF CONTENTS
List of Tables ...................................................................................................................... x
List of Figures .................................................................................................................... xi
Overview ............................................................................................................................. 1
Paper 1: Developmental Associations between Adolescent Alcohol Use and Dating Violence Perpetration ....................................................................................... 5
Abstract ........................................................................................................................... 5
Empirical Studies of the Relationship between Alcohol Use and
Dating Violence .......................................................................................................... 6
Theoretical Models of the Linkages between Alcohol Use and Dating
Violence ...................................................................................................................... 8
Specifying the Relations between Dating Violence and Alcohol Use over Time .................................................................................................................. 11
The Current Study......................................................................................................... 12
Method .......................................................................................................................... 13
Participants ................................................................................................................ 13
Measures ................................................................................................................... 15
Missing Data ............................................................................................................. 19
Analytic Approach .................................................................................................... 20
Results ........................................................................................................................... 24
Univariate Latent Curve Models for Alcohol Use and Dating
Violence .................................................................................................................... 24
Multivariate Growth Model ...................................................................................... 26
Autoregressive Latent Trajectory (ALT) Model ....................................................... 29
Limitations and Future Directions ............................................................................ 34
Conclusions ................................................................................................................... 35 vii Paper 2: The Role of Heavy Alcohol Use in the Developmental Process of
Desistance from Dating Violence Perpetration during Adolescence ................................ 42
Abstract ......................................................................................................................... 42
Introduction ................................................................................................................... 43
Desistance from Antisocial Behavior and Dating Violence ..................................... 44
The Role of Substance Use in Desistance from Dating Violence
Perpetration ............................................................................................................... 46
The Current Study......................................................................................................... 50
Method .......................................................................................................................... 51
Participants ................................................................................................................ 51
Measures ................................................................................................................... 52
Analytic Approach .................................................................................................... 56
Results ........................................................................................................................... 62
Discussion ..................................................................................................................... 66
Limitations and Future Directions ............................................................................ 71
Implications .................................................................................................................. 72
Paper 3: Alcohol Use and Dating Violence Perpetration during
Adolescence: Exposure to Family, Peer and Neighborhood Violence as
Moderators ........................................................................................................................ 79
Abstract ......................................................................................................................... 79
Introduction ................................................................................................................... 80
Moderators of the Relationship between Alcohol Use and Partner
Violence .................................................................................................................... 81
The Current Study......................................................................................................... 83
Method .......................................................................................................................... 85
Participants ................................................................................................................ 85
Measures ................................................................................................................... 86
Analytic Strategy ...................................................................................................... 90
Results ........................................................................................................................... 93 viii Discussion ..................................................................................................................... 96
Limitations and Future Directions ............................................................................ 99
Implications for Prevention ........................................................................................ 100
References ....................................................................................................................... 108
ix
List of Tables
Table 1: Alcohol Use and Dating Violence Perpetration Correlations by
Grade-level and Cohort ..................................................................................................... 37
Table 2. Parameter Estimates (Robust Standard Errors) and Fit Indices for
Unconditional Latent Curve Models of Alcohol Use and Dating Violence ..................... 38
Table 3. Parameter Estimates (Robust Standard Errors) and Fit Indices for
Conditional Multivariate Growth Models of Alcohol Use and Dating
Violence ............................................................................................................................ 39
Table 4. Study Assessment Points by Grade, Cohort and Data Collection
Wave ................................................................................................................................. 74
Table 5. Results for the Unconditional Trajectory Model and for the
Conditional (Launch) Model of the Effects of Early Heavy Alcohol Use on Trajectories of Dating Violence Perpetration .............................................................. 75
Table 6. Results for the Conditional (Snares) Model of the
Contemporaneous Effects of Heavy Alcohol Use on Dating Violence
Perpetration across Grades 8 through 12. ......................................................................... 76
Table 7. Means and Standard Deviations for Dating Violence Perpetration by Grade .......................................................................................................................... 104
Table 8. Results for Models Examining Measures of Violence Exposure as
Moderators of the Effects of Heavy Alcohol Use on Dating Violence across Grades 8 through 12. ............................................................................................ 105
x
List of Figures
Figure 1. Mean Trajectories for Alcohol Use and Dating Violence
Perpetration across Grades 8 through 12. Top: Alcohol Use; Bottom:
Dating Violence ................................................................................................................ 40
Figure 2. Unconditional Multivariate Latent Curve Model of Adolescent
Alcohol Use and Dating Violence Perpetration. ............................................................... 41
Figure 3. Predicted Mean Trajectories of Dating Violence Perpetration across Grades 8 through 12 at Different Levels of Early (Baseline) Heavy
Alcohol Use. ..................................................................................................................... 77
Figure 4. Regression Coefficients and 95% Confidence Intervals for the
Contemporaneous Effects of Heavy Alcohol Use on Dating Violence
Perpetration across Grades 8 through 12. ......................................................................... 78
Figure 5. Parameter Estimates and Standard Errors for the Effects of
Heavy Alcohol Use on Dating Violence Perpetration at Low and High
Levels of Family Violence across Grades 8 through 12. ................................................ 106
Figure 6. Parameter Estimates and Standard Errors for the Effects of
Heavy Alcohol Use on Dating Violence Perpetration at Low and High
Levels of Friend Involvement in Dating Violence Perpetration across
Grades 8 through 12. ....................................................................................................... 107
xi
Overview
Whereas a large body of research has established a consistent and robust link between alcohol use and partner violence perpetration in adulthood (Foran & O’Leary,
2008), little research has examined their relationship during adolescence. Given that patterns of relationship conflict that are established during adolescence may carry over into adulthood (Bouchey & Furman, 2003; Gidycz, Orchowski, King & Rich, 2008;
Magdol, Moffitt, Caspi & Silva, 1998; Smith, White & Holland, 2003), studies that clarify how the relationship between the two behaviors unfolds during adolescence may help to inform primary prevention efforts that reduce partner violence across the lifespan.
To this end, each of the three studies that comprise this dissertation research used a different theoretical lens to guide an empirical examination of the relationship between alcohol use and physical dating violence perpetration using longitudinal data collected from adolescents in grades 8 through 12.
Study one was guided by theoretical models that suggest that: (i) alcohol use will be concurrently related to dating aggression through its effects on cognitive function
(proximal effects model), (ii) alcohol use will be prospectively related to dating violence
(indirect effects model), (iii) dating violence will be prospectively related to alcohol use
(reverse indirect effects model), and (iv) the two behaviors are both manifestations of a general propensity towards deviance driven by common risk factors (common cause model; Foran & Oleary, 2008; Klosterman & Fals-Stewart, 2006; Leonard & Quigley,
1999; White, Brick & Hansell, 1993). An autoregressive latent curve modeling approach
was used to examine each of the pathways implied by these theoretical models in the context of a single analytic framework (Curran & Bollen, 2001; Bollen & Curran, 2004).
Specifically, we simultaneously modeled correlations between latent growth trajectories describing change in alcohol use and dating violence across grades 8 through 12
(correlations implied by the common cause model), as well as concurrent and prospective effects between the repeated measures of each behavior (pathways implied by the proximal and indirect effects models). In addition, we examined the extent to which baseline measures of several risk factors (family conflict, peer aggression, social bonding and emotional distress) that are common to both alcohol use and dating violence accounted for linkages between the two behaviors over time.
Study two was informed by Moffitt’s (1993) theory of antisocial behavior and the work of Hussong, Curran, Moffitt and Caspi (2004) which propose that heavy alcohol use acts as a developmental snare that hinders desistance from dating aggression during late adolescence and young adulthood. Specifically, based on the work of Hussong, et al.
(2004), we hypothesized that higher levels of heavy alcohol use early in adolescence would be associated with higher overall levels of dating violence and decreased deceleration from dating violence perpetration during late adolescence. In addition, we hypothesized that higher levels of heavy alcohol use during assessment points in late adolescence, when the normative pattern is one of desistance from dating violence, would be concurrently associated with higher levels of dating violence perpetration during those time points. To test these hypotheses, study two used a random coefficients growth modeling approach to examine the effects of both early (baseline) and continuing (time-
2
varying) heavy alcohol use on desistance (i.e. deceleration) from dating violence perpetration during late adolescence.
Study three was motivated by empirical evidence and theoretical models that suggest that individual and contextual or situational factors moderate the relationship between alcohol use and dating aggression (Klosterman & Fals-Stewart, 2006). In particular, several theoretical models suggest that the proximal relationship between alcohol intoxication and dating violence may be stronger among individuals who have aggressive propensities and in contexts or situations that facilitate (rather than constrain) aggressive behavior (Chermack & Giancola, 1997; Klosterman & Fals-Stewart, 2006;
Parker, 1995). Based on social cognitive theory (Bandura, 1973; 1977), we reasoned that higher levels of exposure to family, peer and neighborhood violence would, through processes of modeling and reinforcement, each be associated with the development of aggressive perceptual and behavioral tendencies. As such, we expected that the concurrent relationship between heavy alcohol use and dating violence perpetration would be stronger for teens who were exposed to higher as compared to lower levels of family, peer and neighborhood violence. To test the hypotheses implied by this expectation, we examined the main and joint effects of time-varying measures of heavy alcohol use and exposure to family, peer and neighborhood violence on repeated measures of dating violence perpetration across grades 8 through 12.
The data for this dissertation research come from a multi-wave cohort sequential study of adolescent health risk behaviors that spanned middle and high school (National
Institute on Drug Abuse, R01DA16669, S. T. Ennett, PI; Centers for Disease Control and
Prevention, R49CCV423114, V. A. Foshee, PI). Four waves of data were used starting
3
when participants were in the 8th, 9th and 10th grades (wave one) and ending when participants were in the 10th, llth, and 12th grades (wave four). Data were collected at sixmonth time intervals for the first three waves and there was a one-year time interval between waves three and four. Participants were enrolled in two public school systems located in two predominantly rural counties with higher proportions of African
Americans than in the general United States (U.S. Census Bureau, 2001).
4
Paper 1: Developmental Associations between Adolescent Alcohol Use and Dating
Violence Perpetration
Abstract
Although numerous studies suggest alcohol use is associated with adult partner violence, few studies have examined how this relationship unfolds during adolescence.
The current study examined the interrelations between alcohol use and physical dating violence perpetration across grades 8 through 12 using four waves of data from a longitudinal study of adolescent health risk behaviors. Autoregressive latent curve models were used to estimate associations between developmental trajectories of alcohol use and dating violence while also examining time-specific concurrent and bidirectional prospective relations. On average, higher levels of alcohol use were associated with higher levels of dating violence across all grades. Consistent with the common cause model, associations between trajectories of alcohol use and dating violence were reduced substantially after controlling for psychosocial predictors common to both behaviors.
However, consistent with the proximal effects model, significant concurrent associations between the two behaviors persisted across nearly all grades. There was no evidence of prospective relations from prior alcohol use to subsequent dating violence or vice-versa.
Results suggest that primary prevention efforts should target shared determinants of alcohol use and dating violence including family conflict, peer aggression, emotional distress and low social bonding.
Introduction
Whereas a large body of research has documented a consistent and robust link between alcohol use and adult intimate partner violence (for reviews, see Foran &
O’Leary, 2008; Lipsey, Wilson, Cohen & Derzon, 1997; Stith, Smith, Penn, Ward &
Tritt, 2004; Testa, 2004), few studies have examined the relationships between alcohol use and dating violence perpetration during adolescence. Both dating violence and alcohol use become increasingly prevalent during the middle and high school years and can have serious negative consequences for adolescent health and well-being (Ackard,
Eisenberg & Neumark-Sztainer, 2007; Chassin, et. al, 2004; Roberts, Klein & Fisher,
2003; Windle & Windle, 2004). Moreover, patterns of relationship conflict that are established during this period are likely to carry over into adulthood (Bouchey & Furman,
2003; Gidycz, Orchowski, King & Rich, 2008; Magdol, Moffitt, Caspi & Silva, 1998;
Smith, White & Holland, 2003). Therefore, a better understanding of how the relationship between alcohol use and dating violence unfolds during adolescence may inform prevention efforts across the lifespan. To this end, the current study provides an empirical examination of the concurrent and longitudinal relationships between alcohol use and dating violence across middle and late adolescence.
Empirical Studies of the Relationship between Alcohol Use and Dating Violence
There is extensive empirical evidence documenting an association between alcohol use and adult intimate partner violence perpetration (Foran & O’leary, 2008;
Stith, et al., 2004; Lipsey, et al., 1997). For example, in their recent meta-analysis, Foran and O’leary (2008) found a small to moderate effect size for the association between alcohol use/abuse and male-to-female violence and a small effect size for female-to-male
6
violence. However, only five studies, three cross-sectional and two longitudinal, have examined the associations between adolescent alcohol use and dating violence perpetration. Each of the three cross-sectional studies examined the concurrent association of alcohol use with a dichotomous measure of any dating violence perpetration in the past year. Malik, Sorenson, and Anehensel (1997) found that a past year measure of alcohol use frequency was not associated with involvement in physical dating violence perpetration. Similarly, Hird (2000) found that a lifetime measure of any alcohol use was not associated with girls’ involvement in physical, psychological or sexual dating violence perpetration. In contrast, Champion, Long Foley, Sigmon-Smith,
Sutfin, & DuRant (2008) found that past 30-day alcohol use involvement was associated with involvement in date fighting perpetration in the past year (starting a fight with or hitting a boyfriend, girlfriend or date).
Both longitudinal studies of the association between alcohol use and dating violence perpetration examined the relationship between alcohol use among nonperpetrators at time one and onset of perpetration at time two. In the first study, past-30 day frequency of alcohol use was found to predict perpetration onset among girls, but not boys, one year later (Foshee, MacDougall, Linder, & Bangdiwala, 2001). In the second study, a lifetime measure of alcohol use frequency was not found to predict onset of perpetration by either boys or girls six months later (Foshee, Reyes, & Ennett, in press).
Other cross-sectional (O’Keefe, 1997) and longitudinal (Simons, Lin, and Gordon, 1998;
Lavoie, et al., 2001) studies have found that measures of substance use and antisocial behavior that combine indicators of alcohol use with indicators of other types of substance use and/or delinquent behavior are both concurrently and prospectively
7
associated with dating violence perpetration. However, because alcohol use was measured as a composite with other behaviors, it is impossible to determine whether alcohol use was uniquely related to dating violence in those studies.
Overall, the few studies that have examined the relationship between adolescent alcohol use and dating violence have been hampered by limited alcohol use measures which likely do not tap into the kind of heavy or problematic use more likely to be associated with aggression (Foran and O’Leary, 2008; Oleary & Schumacher, 2003).
Furthermore, no studies have examined interrelations between the two behaviors across more than two points in time.
Theoretical Models of the Linkages between Alcohol Use and Dating Violence
Three primary theoretical explanations have been posited to explain the observed relationship between alcohol use and partner violence: (a) the proximal effects model, (b) the indirect effects model, and (c) the common cause or spurious effects model (Foran &
O’Leary, 2008; Klosterman & Fals-Stewart, 2006; Leonard & Quigley, 1999). The proximal effects model posits that alcohol intoxication plays a causal role in increasing risk of dating abuse perpetration through its psychopharmacological effects on cognitive function. Specifically, intoxication can intensify feelings of excitement and curiosity, lead a person to overreact to perceived provocation, and decrease the saliency of cues that aggressive behavior will have negative consequences (i.e., threat inhibition), thereby increasing risk of confrontation and violence (Phil and Hoaken, 2002). This model implies that alcohol use increases risk of dating violence exclusively during the time frame when alcohol is exerting a pharmacological effect.
8
The indirect effects model posits that the causal relationship between alcohol use and dating violence is mediated by other variables such as relationship quality. For example, several researchers have suggested that elevated alcohol use by one or both partners in a dating relationship leads to relationship dissatisfaction and greater frequency of interpersonal conflict and, in turn, to increased risk of dating violence perpetration
(Fagan & Browne, 1994; Fischer, et al., 2005; Quigley & Leonard, 2000; White & Chen,
2000). In contrast to the proximal effects model, which implies that alcohol use and dating violence will be concurrently associated, the indirect effects model implies that the causal influence of alcohol use on dating violence may be studied over a longer timewindow. That is, the indirect effects model suggests that elevated alcohol use during one time period may prospectively predict dating violence perpetration measured at a subsequent time period.
Another version of the indirect effects model suggests that prior aggression, including dating violence, may indirectly lead to subsequent alcohol use (White, Brick, and Hansell, 1993). Mechanisms explaining this relationship (from prior aggression to subsequent alcohol use) include the notions that: (i) involvement in aggression may lead to delinquent peer affiliations and, in turn, to substance use (e.g., Fite, Colder, Lochman
& Wells, 2007) and (ii) involvement in aggression may lead to alcohol use as a means for coping with the negative social and emotional consequences of being abusive (White, et al., 1993).
Regardless of the specific mediating mechanism, indirect effects models imply that elevated alcohol use during one time period may lead to increased dating violence at a subsequent time period and/or vice-versa. Indeed, longitudinal studies of the
9
developmental associations between substance use and non-dating aggression have found some evidence that elevated levels of substance use prospectively predict increased aggression and vice versa, supporting the notion of a reciprocal relationship between the two behaviors (Huang, White, Kosterman, Catalano & Hawkins, 2001; White, Loeber,
Stouthamer-Loeber, & Farrington, 1999).
A third conceptual model that has been posited to explain the link between alcohol use and dating violence is the common cause model. This model suggests that alcohol use and dating violence are linked because they share causal determinants. For example, several risk factors have been found to predict both alcohol use and dating or partner violence among adolescents and young adults including: peer aggression or antisocial behavior (e.g., Andrews, Foster, Capaldi & Hops, 2000; Fite, et al., 2007), emotional distress (e.g., Wolfe, Wekerle, Scott, Straatman, & Grasley, 2004; Tschann,
Flores, Pasch & Van Oss, 2005), and aspects of the family environment including poor parenting practices (e.g., Hotton & Haans, 2004; Lavoie et al., 2001) and family conflict
(e.g., Bray, Adams, Getz, & Baer, 2001; Ehrensaft et al., 2003).
The notion that alcohol use and dating violence share etiological origins is also consistent with several theories of adolescent health risk behavior (e.g., problem behavior theory, general deviance theory, primary socialization theory), that suggest that alcohol use and dating violence perpetration are both manifestations of an underlying propensity towards deviance. These theories identify numerous general causal determinants (e.g., low social bonding, negative family environment) that can lead to involvement in a range of problem behaviors, including substance use and aggression (see Jessor, Donovan &
10
Costa, 1991; Jessor & Jessor, 1977; Oetting & Donnermeyer, 1998; Osgood, Johnson,
O’Malley, & Bachman, 1988).
Specifying the Relations between Dating Violence and Alcohol Use over Time
Taken together, the theoretical models reviewed above suggest there may be any one of a number of pathways linking a set of repeated measures of dating abuse and alcohol use over time. To clarify the nature of these pathways and to help map each pathway onto the modeling framework used in the current study, we classify the theoretical relationships between dating violence and alcohol use into two types: (i) timespecific relations and (ii) time-stable relations. Time-specific relations comprise associations between levels of alcohol use at a particular time point and levels of dating violence at a particular time point. Time-specific relations between repeated measures of alcohol use and dating violence are implied by both the proximal effects and indirect effects models. The proximal effects model suggests that elevated levels of alcohol misuse at a given time-point will be concurrently associated with elevated levels of dating violence perpetration at that same time-point. Indirect effects models suggest that elevated alcohol use at a given time-point may prospectively predict dating violence perpetration at a later time point and/or vice-versa.
In contrast to the time-specific relations suggested by the proximal and indirect effects models, theories that view both alcohol use and dating violence as forms of deviant behavior driven by common causes suggest that there may be an overall timestable association between levels of alcohol use and levels of dating violence over time.
That is, it follows from these theories that overall levels of and changes in ones’ propensity towards deviance will influence levels of involvement in both alcohol use and
11
dating violence over time, resulting in time-stable correlations between the underlying trajectories for both behaviors. These correlations are referred to as “time-stable” because they represent the overall associations between levels of and changes in alcohol use and dating violence across the time period assessed. It also follows from these theories that correlations between trajectories of alcohol use and dating violence would be attenuated once the influence of shared risk factors is accounted for.
The Current Study
The current study used an autoregressive latent trajectory modeling approach to examine both time-specific and time-stable relations between repeated measures of dating violence and alcohol use using data from a multi-wave longitudinal study of adolescent boys and girls that spanned grades 8 through 12. Following from theories that view alcohol use and dating aggression as manifestations of an underlying propensity towards deviant behavior, we examined the correlations between the underlying growth processes governing trajectories of alcohol use and dating violence perpetration over time. In addition, based on the common-cause model, we examined relations between the two behaviors both before and after controlling for baseline psychosocial risk factors that have been identified as contributors to both alcohol use and dating violence including: family conflict, social bonding, emotional distress and peer aggression. Based on the proximal and indirect effects models, we also simultaneously examined concurrent and bidirectional prospective relations between the repeated measures for each behavior.
Finally, because many studies suggest that dating violence perpetration is as prevalent for girls as for boys (Foshee & Reyes, 2009), and that the etiological processes leading to
12
dating violence may differ for boys and girls (Foshee, et al., 2001; Foshee, et al., in press), we tested for sex differences in the pathways relating the two behaviors over time.
Method
Participants
The sample for this study was drawn from a multi-wave cohort sequential examination of adolescent health risk behaviors that spanned middle and high school
(National Institute on Drug Abuse, R01DA16669, S. T. Ennett, PI; Centers for Disease
Control and Prevention, R49CCV423114, V. A. Foshee, PI). Dating violence was assessed beginning when participants were in the 8th, 9th and 10th grades. As such, the current study uses four waves of data starting when participants were in the 8th, 9th and
10th grades (wave one) and ending when participants were in the 10th, llth, and 12th grades
(wave four). Data were collected at six-month time intervals for the first three waves and there was a one-year time interval between waves three and four. Participants were enrolled in two public school systems located in two predominantly rural counties with higher proportions of African Americans than in the general United States (U.S. Census
Bureau, 2001).
At each assessment all enrolled students in the targeted grades who were able to complete the survey in English and who were not in special education programs or out of school due to long-term suspension were eligible for the study. Parents had the opportunity to refuse consent for their child’s participation by returning a written form or by calling a toll-free telephone number. Adolescent assent was obtained from teens whose parents had consented immediately prior to the survey administration. Trained data collectors administered the questionnaires in student classrooms on at least two
13
occasions to reduce the effect of absenteeism on response rates. To maintain confidentiality, teachers remained at their desks while students completed questionnaires and the students placed questionnaires in envelopes before returning them to the data collectors. The Institutional Review Board for the School of Public Health at the
University of North Carolina at Chapel Hill approved the data collection protocols.
At wave one, 6% of parents refused consent, 6% of adolescents declined to participate and 8% were absent on the days when data were collected for a total of 2636 students completing a survey at wave one. The response rate, calculated as the proportion of adolescents who completed a survey out of those eligible for the survey at wave 1 was
79%. For this study, analyses excluded students who; (1) did not report their age or who reported being out of the typical age range of 12-19 for the grades studied (n=50, 2%), (2) did not report their dating status or reported never dating across all of the assessments
(n=247, 9%) or (3) were missing data on the alcohol use or dating violence measures across all waves of the study (n=67, 3%), yielding a sample size of 2272. Almost all students participated in at least two waves of data collection (n=2127, 94%), with 75% participating in 3 or more waves (n=1722).
Approximately half of the sample was male (47%) and the self-reported race/ethnicity distribution was 49% White, 43% Black and 5% other race/ethnicity.
Approximately 29% of participants reported that the highest education attained by either parent was high school or less across all waves of the study. At wave 1, prevalence of any alcohol use in the past three months was 28% and prevalence of any physical dating violence perpetration in the past three months was 18%.
14
Measures
Measures included the two outcomes of interest, alcohol use and dating violence perpetration as well as psychosocial and demographic covariates. The alcohol and dating violence measures were collected at all waves. Measures of the psychosocial covariates
(family conflict, emotional distress, social bonding and peer aggression) were drawn from the baseline assessment to be consistent with the common cause model, which views these variables as precursors to alcohol use and dating violence.
Alcohol use. Alcohol use was measured as a composite of frequency, quantity and heavy use. For all measures, alcoholic beverages were defined as including beer, wine, wine coolers and liquor and a “drink” was defined as a glass of wine, a can of beer, a bottle or can of wine cooler, a shot glass of liquor or a mixed drink. The frequency item assessed the number of days that the adolescent had one or more drinks of alcohol in the past three months with six response categories ranging from 0 days to 20 days or more.
The quantity item assessed how many drinks the adolescent usually consumed on a typical drinking occasion in the past three months with six response categories ranging from less than one drink to five or more drinks. Heavy alcohol use was assessed by five items asking adolescents how many times they had: 3 or 4 drinks in a row, 5 or more drinks in a row, gotten drunk or very high from drinking alcohol, drunk alcohol while alone or been hung over. Each item had five response categories that ranged from 0 to 10 or more times in the past 3 months. The heavy use items were averaged to create a scale and then the frequency, quantity and heavy use measures were standardized and summed to create a composite measure of alcohol use at each wave (average Cronbach’s α =.92).
15
Physical dating violence perpetration. Dating violence perpetration was measured each wave using a short version of the Safe Dates Physical Perpetration Scale (Foshee, et al., 1996). Adolescents were asked, “During the past 3 months, how many times did you do each of the following things to someone you were dating or on a date with? Don’t count it if you did it in self-defense or play.” Six behavioral items were listed: “slapped or scratched them,” “physically twisted their arm or bent back their fingers,” “pushed, grabbed, shoved, or kicked them,” “hit them with your fists or with something else hard,”
“beat them up,” and “assaulted them with a knife or a gun.” Each item had five response categories ranging from 0 to 10 times or more in the past three months. Responses were summed across items to create a physical dating violence perpetration scale measure
(average Cronbach’s alpha=.93).
Psychosocial covariates. We measured peer aggression using six items that assessed how many times in the past three months the respondent had pushed, slapped or kicked someone, physically twisted someone’s arm or bent back their fingers, hit someone with their fist or something else hard, beat someone up or assaulted someone with a knife or gun. Adolescents were specifically asked to exclude acts that they had perpetrated against a date. Scores were averaged across the items to create a composite scale of adolescent physical aggression (Cronbach’s α =.87).
Family conflict was assessed by three items from Bloom’s (1985) self-report measure of family functioning. Adolescents were asked how strongly they agreed or disagreed with the following three items when thinking about their family life in the past three months: we fight a lot in our family, family members sometimes get so angry they throw things and family members sometimes hit each other. Response options ranged
16
from strongly agree (4) to strongly disagree (0). Items were averaged to create a measure of baseline exposure to family conflict (Cronbach’s α =.87).
Emotional distress was measured as a composite of three scales assessing anger, anxiety and depression in the past three months. Anger was assessed by three items drawn from the revised Multiple Affective Adjective Checklist (MAACL-R) that asked adolescents how often they felt mad, angry or furious in the past three months
(Zuckerman & Lubin, 1985). Four response categories ranged from never or almost never to almost always. Anxiety was measured using a shortened version of the Revised
Children’s Manifest Anxiety scale (Reynolds & Richmond, 1979) and depression was measured using three items from the Short Mood and Feelings Questionnaire (Angold,
Costello, & Messer, 1995). Both anxiety and depression were assessed by presenting adolescents with a list of statements describing how they may have felt in the past three months. The statements listed seven symptoms of anxiety (e.g., I felt sick to my stomach) and three symptoms of depression (e.g., I did everything wrong). Each item had five response categories that ranged from strongly disagree to strongly agree. Items were averaged to create a scale score for each construct. Cronbach’s alphas were satisfactory for each of the individual subscales (α =.88 for anger, α =.88 for anxiety, α = .92 for depression) and subscale scores were significantly correlated (p
