<%BANNER%>

Neighborhood Context and Alcohol Use among Urban, Low-income, Multi-ethnic, Young Adolescents

Permanent Link: http://ufdc.ufl.edu/UFE0024200/00001

Material Information

Title: Neighborhood Context and Alcohol Use among Urban, Low-income, Multi-ethnic, Young Adolescents
Physical Description: 1 online resource (214 p.)
Language: english
Creator: Tobler, Amy
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: adolescents, alcohol, community, context, minority, neighborhood, urban
Health Education and Behavior -- Dissertations, Academic -- UF
Genre: Health and Human Performance thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This study defined the alcohol-related neighborhood context of a large sample of urban, racial/ethnic minority, young adolescents and examined how this context related to alcohol use. Data were part of a longitudinal, group-randomized controlled trial of an alcohol preventive intervention, which included 42 Chicago community areas. Neighborhood measures included (1) number of alcohol outlets per capita per community area; (2) alcohol purchase attempt rate by pseudo-underage youth; (3) average number of alcohol advertisements per school per community; and (4) a Census 2000-based area deprivation index. Students, parents, and community leaders were also surveyed and provided data on alcohol behaviors, perceived neighborhood problems and support of alcohol policies, and neighborhood strength and preventive action, respectively. Multilevel latent class analysis, mixed effects regression, and multilevel structural equation modeling were used to (1) identify the number and characteristics of heterogeneous latent alcohol-related neighborhood risk classes, (2) determine how membership in these risk classes influenced trajectories of alcohol use, and (3) explore how alcohol-related neighborhood context directly influenced alcohol use and was mediated by protective home and family management practices. Five heterogeneous classes of alcohol-related neighborhood risk were identified. Among this sample of low-income urban youth, there were no neighborhoods defined as truly low-risk (i.e., high social capital and low exposure and access to alcohol). None of the neighborhood risk classes were significantly associated with the trajectories of alcohol use/intentions relative to the other classes. When considering each neighborhood construct separately, neighborhood strength was negatively, and exposure to alcohol advertisements positively, associated with alcohol use. Neighborhood strength and commercial alcohol access were associated with home alcohol access and protective family management practices. Home alcohol access had a positive association with alcohol use. Home alcohol access may partially mediate the relation between neighborhood strength and alcohol use. Findings suggest inner-city parents respond to environmental risk, such that as neighborhood risk increases, so also do protective home and family management practices. Parent engagement in restricting alcohol access and improving family management practices may be key to preventive efforts to reduce alcohol use among inner-city, adolescent youth.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amy Tobler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Weiler, Robert M.
Local: Co-adviser: Komro, Kelli Ann.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024200:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024200/00001

Material Information

Title: Neighborhood Context and Alcohol Use among Urban, Low-income, Multi-ethnic, Young Adolescents
Physical Description: 1 online resource (214 p.)
Language: english
Creator: Tobler, Amy
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: adolescents, alcohol, community, context, minority, neighborhood, urban
Health Education and Behavior -- Dissertations, Academic -- UF
Genre: Health and Human Performance thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This study defined the alcohol-related neighborhood context of a large sample of urban, racial/ethnic minority, young adolescents and examined how this context related to alcohol use. Data were part of a longitudinal, group-randomized controlled trial of an alcohol preventive intervention, which included 42 Chicago community areas. Neighborhood measures included (1) number of alcohol outlets per capita per community area; (2) alcohol purchase attempt rate by pseudo-underage youth; (3) average number of alcohol advertisements per school per community; and (4) a Census 2000-based area deprivation index. Students, parents, and community leaders were also surveyed and provided data on alcohol behaviors, perceived neighborhood problems and support of alcohol policies, and neighborhood strength and preventive action, respectively. Multilevel latent class analysis, mixed effects regression, and multilevel structural equation modeling were used to (1) identify the number and characteristics of heterogeneous latent alcohol-related neighborhood risk classes, (2) determine how membership in these risk classes influenced trajectories of alcohol use, and (3) explore how alcohol-related neighborhood context directly influenced alcohol use and was mediated by protective home and family management practices. Five heterogeneous classes of alcohol-related neighborhood risk were identified. Among this sample of low-income urban youth, there were no neighborhoods defined as truly low-risk (i.e., high social capital and low exposure and access to alcohol). None of the neighborhood risk classes were significantly associated with the trajectories of alcohol use/intentions relative to the other classes. When considering each neighborhood construct separately, neighborhood strength was negatively, and exposure to alcohol advertisements positively, associated with alcohol use. Neighborhood strength and commercial alcohol access were associated with home alcohol access and protective family management practices. Home alcohol access had a positive association with alcohol use. Home alcohol access may partially mediate the relation between neighborhood strength and alcohol use. Findings suggest inner-city parents respond to environmental risk, such that as neighborhood risk increases, so also do protective home and family management practices. Parent engagement in restricting alcohol access and improving family management practices may be key to preventive efforts to reduce alcohol use among inner-city, adolescent youth.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amy Tobler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Weiler, Robert M.
Local: Co-adviser: Komro, Kelli Ann.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024200:00001


This item has the following downloads:


Full Text

PAGE 1

NEIGHBORHOOD CONTEXT AND ALCOHOL USE AMONG URBAN, LOW-INCOME, MULTI-ETHNIC, YOUNG ADOLESCENTS By AMY L. TOBLER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Amy L. Tobler 2

PAGE 3

ACKNOWLEDGEMENTS This study was funded by grants from the National Institute on Alcohol Abuse and Alcoholism and National Center on Minority He alth and Health Disparities (R01 AA013458; R01 AA016549), awarded to Dr. Kelli A. Komro. I thank Karen Alfano, MBA, for survey design and management of data collection, Kian Farbak hsh, M.S., for database design and management, and Cheryl Perry, Ph.D., for her overall contri butions to the PNC st udy. I also gratefully acknowledge the participation of students, pa rents and community leaders in the Project Northland Chicago trial. I express many thanks to my doctoral comm ittee, Drs. Robert Weiler, Kelli Komro, Mildred Maldonado-Molina, Steven Pokorny and Dennis Thombs, for their guidance and support throughout my course of study and completion of this work. It has been a pleasure to work with, and learn from, them. I appreciat e their dedication to making me a better scientist and providing me the freedom and encouragement to tackle ne w projects and statistical methods. I am better personally and professionally for having worked with them. I am grateful to Drs. Kelli Komro and Alex ander Wagenaar for hiring and embedding me in an environment rich with ideas and determina tion to use the best science to make a difference in the lives of individuals and co mmunities at-large. They both insp ired me to pursue this course and I am deeply thankful for their support and encouragement. It has been a tremendous opportunity to learn from the best, in such a healthy, productive e nvironment. They are outstanding people and I am thankful to know them. I look forward to many years of continued work and collaboration! I am thankful to my parents, who instilled in me a desire for excellence in every pursuit and an understanding of the im portance of higher education. Th eir support and encouragement have helped me get this far, and will continue to sustain me as I move forward. I am grateful for 3

PAGE 4

the foundation they provided and hope that I ha ve, and will continue to make them proud. I am thankful for my little brother and sister, Braden and Greer, who have inspired me to do what I can to make life a little easier for adolescent youth. I admire them both for who they are and the trials they have overcome in their lives. I than k my twin sister, Ashley; I am grateful each day that from the beginning we have been able to make our journey through life together. She has pushed me to do and be better and is always a great source of support. I treasure her friendship and will always be thankful that she did not go to FSU. I must acknowledge the great blessings I have received from my Heavenly Father, who is helping me do more with my life than I could ha ve ever done alone. I have been blessed with many opportunities to learn and grow and I am tha nkful for each. My hope is that I can use what I have been given to make a difference in the world around me. Lastly, I must acknowledge my best friend, a nd husband, Jeff. He has been my greatest source of love, support and encouragement. He was always willing to listen to me talk about my triumphs and troubles, even when he did not understand them. He kne w I was having trouble with my statistical models and that was enough. I appreciate his sense of humor and love for a good time, always keeping me grounded and focused on what is really important. I am thankful that he loves me for (or in spite of) my ambition and does not expect anything less from me. I share this accomplishment with him and look forw ard to all that we w ill accomplish together in the years to come. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGEMENTS.............................................................................................................3 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES.......................................................................................................................10 ABSTRACT...................................................................................................................................11 CHAPTER 1 BACKGROUND AND SIGNIFICANCE..............................................................................13 Prevalence of Alcohol Use Among Youth in the United States.............................................13 Prevalence of Alcohol Use among R acial/Ethnic and Gender Subgroups......................14 Prevalence of Alcohol Use among Ethnic Mi nority Youth in Chicago, Illinois.............15 Prevalence of Alcohol Use among the Project Northland Chicago Sample...................15 Consequences of Adolescent Alcohol Use.............................................................................16 Immediate Consequences................................................................................................16 Distant Consequences......................................................................................................19 Etiology of Adolescent Alcohol Use......................................................................................21 Proximal, Distal and Ultimate Risk and Protective Factors............................................22 Limitations of Current Knowledge..................................................................................26 Theoretical Foundation......................................................................................................... ..29 Theory of Triadic Influence.............................................................................................30 Wagenaar and Perrys Mode l of Drinking Behavior.......................................................31 Summary of Theore tical Framework...............................................................................31 Innovation of the Study..........................................................................................................32 Study Goal, Aims and Research Questions............................................................................33 Multi-ethnic, Urban Youths Exposur e to Patterns of Alcohol-related Neighborhood Characteristics......................................................................................33 Effects of Alcohol-related Neighborhood Context on the Tr ajectories of Alcohol Use and Intentions among Young Adolescents...........................................................34 Relationships between Neighborhood Contex t, Family Management Practices and Alcohol Use among Urban, Mult i-ethnic, Young Adolescents...................................35 Summary.................................................................................................................................35 2 METHODS...................................................................................................................... .......41 Overview and Research Questions.........................................................................................41 Project Northland Chicago.....................................................................................................41 Study Design................................................................................................................... 41 Intervention......................................................................................................................42 Data Collection................................................................................................................42 Students....................................................................................................................43 5

PAGE 6

Parents......................................................................................................................44 Community leaders..................................................................................................44 Alcohol outlets.........................................................................................................45 Alcohol purchase attempts.......................................................................................45 Alcohol advertisements............................................................................................46 Census 2000.............................................................................................................47 Participants.............................................................................................................................47 Measures.................................................................................................................................47 Neighborhood Context....................................................................................................48 Protective factors......................................................................................................48 Risk factors...............................................................................................................50 Ultimate Outcomes...................................................................................................51 Alcohol use...............................................................................................................51 Alcohol use intentions..............................................................................................52 Tendency to use alcohol...........................................................................................52 Hypothesized Mediators..................................................................................................52 Home alcohol access................................................................................................53 Parental monitoring and communication.................................................................53 Alcohol-specific communication.............................................................................53 Hypothesized Moderator.................................................................................................54 Analytical Approaches.......................................................................................................... ..54 Multilevel Latent Class Analysis.....................................................................................55 Measurement model........................................................................................................56 Model Selection...............................................................................................................57 General Linear Mixed Effects Regression......................................................................59 Multilevel Structural Equation Modeling........................................................................60 Missing Data....................................................................................................................64 3 MULTI-ETHNIC, URBAN YOUTHS EX POSURE TO PATTERNS OF ALCOHOLRELATED NEIGHBORHOOD CHARACTERISTICS.......................................................77 Executive Summary................................................................................................................77 Background......................................................................................................................77 Methods...........................................................................................................................78 Results........................................................................................................................ .....79 Conclusions.....................................................................................................................79 4 EFFECTS OF ALCOHOL-RELATED NEIGHBORHOOD CONTEXT ON THE TRAJECTORIES OF ALCOHOL US E AND INTENTION S AMONG YOUNG ADOLESCENTS.................................................................................................................... 81 Abstract....................................................................................................................... ............81 Background......................................................................................................................81 Methods...........................................................................................................................81 Results........................................................................................................................ .....82 Conclusions.....................................................................................................................82 Key Words...................................................................................................................... .82 6

PAGE 7

Introduction................................................................................................................... ..........82 Methods..................................................................................................................................86 Design..............................................................................................................................86 Data Collection................................................................................................................86 Measures..........................................................................................................................87 Neighborhood risk....................................................................................................87 Alcohol use and alcohol use intentions....................................................................88 Covariates.................................................................................................................88 Analytical Strategy..........................................................................................................89 Missing Data....................................................................................................................90 Results.....................................................................................................................................91 Discussion...............................................................................................................................92 5 RELATIONSHIPS BETWEEN NEIGHBORHOOD CONTEXT, FAMILY MANAGEMENT PRACTICES AND ALCO HOL USE AMONG URBAN, MULTIETHNIC, YOUNG ADOLESCENTS....................................................................................99 Abstract....................................................................................................................... ............99 Background......................................................................................................................99 Methods...........................................................................................................................99 Results........................................................................................................................ .....99 Conclusions...................................................................................................................100 Key Words.....................................................................................................................1 00 Introduction................................................................................................................... ........100 Methods................................................................................................................................104 Design............................................................................................................................104 Data Collection..............................................................................................................105 Students..................................................................................................................105 Parents....................................................................................................................105 Community leaders................................................................................................106 Neighborhood characteristics.................................................................................106 Measures........................................................................................................................106 Alcohol-related neighborhood context...................................................................106 Home and family management practices...............................................................109 Alcohol use.............................................................................................................110 Analytical Strategy........................................................................................................110 Missing Data..................................................................................................................113 Results...................................................................................................................................113 Measuremen t Models....................................................................................................113 Alcohol-related neighborhood context...................................................................113 Home and family management practices...............................................................114 Alcohol use.............................................................................................................114 Structural Model............................................................................................................114 Discussion.............................................................................................................................116 6 DISCUSSION................................................................................................................... ....124 7

PAGE 8

APPENDIX A 2002 STUDENT SURVEY..................................................................................................132 B 2002 PARENT SURVEY.....................................................................................................148 C 2002 COMMUNITY LEADER SURVEY..........................................................................156 D ALCOHOL PURCHASE ATTEMPT PROTOCOL............................................................165 E ALCOHOL ADVERTISEMENT AS SESSMENT DATA COLLECTION AND CODING PROTOCOL.........................................................................................................176 LIST OF REFERENCES.............................................................................................................192 BIOGRAPHICAL SKETCH.......................................................................................................214 8

PAGE 9

LIST OF TABLES Table page 1-1 Prevalence of alcohol use among 8th, 10th and 12th grade youth in the United States, 2007....................................................................................................................................371-2 Prevalence of alcohol use among 9th to 12th grade youth in Chicago, Illinois, 2007.........381-3 Prevalence of alcohol us e among youth participating in Project Northland Chicago, 2002-2005..........................................................................................................................392-1 Summary of research questions, st udy design and analytic approaches............................662-2 Comparison of 66 study schools with average for Chicago Public Schools......................672-3 Data used from Proj ect Northland Chicago.......................................................................682-4 Characteristics of students who completed 1 or more PNC surveys.................................692-5 Racial/Ethnic distribution (%) among PNC study communities from Census 2000.........702-6 Descriptive statistics for measures of deprivation from Census 2000...............................712-7 Descriptive statistics for ne ighborhood context measures, 2002.......................................722-8 Frequencies of alcohol use and intentions items T1 T4 (2002-2005)..............................732-9 Frequencies of home and family management items T3 (2004)........................................754-1 Descriptive statistics for va riables included in each model...............................................964-2 Time-invariant predictors at age 12 of tr ajectories of alcohol use and alcohol use intentions from age 12 to 14..............................................................................................974-3 Results from bivariate analyses of individual neighbor hood risk/protective items at age 12 as predictors of the tr ajectories of alcohol use and alcohol use intentions from age 12 to 14 while controlling for ba seline levels of intentions/use..................................985-1 Standardized, geomin-rotated factor lo adings and fit statistics for measurement models..............................................................................................................................121 9

PAGE 10

LIST OF FIGURES Figure page 1-1 Theoretical framework of direct and indirect rela tionships between neighborhood context, family management practices and early adolescent alcohol use and intentions............................................................................................................................405-1 Hypothesized structural model........................................................................................1225-2 Structural model depicting standard ized paths among alc ohol-related neighborhood context, home and family management pr actices, and early adolescent alcohol use. (Nonsignificant paths are indicated with dashed line.)....................................................123 10

PAGE 11

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NEIGHBORHOOD CONTEXT AND ALCOHOL USE AMONG URBAN, LOW-INCOME, MULTI-ETHNIC, YOUNG ADOLESCENTS By Amy L. Tobler May 2009 Chair: Robert Weiler Cochair: Kelli A. Komro Major: Health and Human Performance This study defined the alcohol-related neighbor hood context of a larg e sample of urban, racial/ethnic minority, young adoles cents and examined how this co ntext related to alcohol use. Data were part of a longitudinal, group-random ized controlled trial of an alcohol preventive intervention, which included 42 Chicago commu nity areas. Neighborhood measures included (1) number of alcohol outlets per capita per commun ity area; (2) alcohol pu rchase attempt rate by pseudo-underage youth; (3) average number of alcohol advertisements per school per community; and (4) a Census 2000-based area de privation index. Students, parents, and community leaders were also surveyed and pr ovided data on alcohol behaviors, perceived neighborhood problems and support of alcohol policies, and neighbor hood strength and preventive action, respectively. Multilevel latent class analysis, mixed effects regression, and multilevel structural equation modeling were used to (1) identify the number and characteristics of heterogeneous latent alcohol-related ne ighborhood risk classes, (2) determine how membership in these risk classes influenced tr ajectories of alcohol use, and (3) explore how alcohol-related neighborhood context directly influenced alcohol use and was mediated by protective home and family management practices. Five heterogeneous cla sses of alcohol-related 11

PAGE 12

neighborhood risk were identified. Among this samp le of low-income urban youth, there were no neighborhoods defined as truly lo w-risk (i.e., high social capit al and low exposure and access to alcohol). None of the neighborhood risk classes were significantly associated with the trajectories of alcohol use/inte ntions relative to the other cl asses. When considering each neighborhood construct separately, neighborhood strength was negatively, and exposure to alcohol advertisements positively, associated with alcohol use. Neighborhood strength and commercial alcohol access were associated with home alcohol access and protective family management practices. Home alcohol access ha d a positive associati on with alcohol use. Home alcohol access may partially mediate the rela tion between neighborhood strength and alcohol use. Findings suggest inner-city pa rents respond to environmental risk, such that as neighborhood risk increases, so also do protec tive home and family management practices. Parent engagement in restricting alcohol access and improving family management practices may be key to preventive efforts to reduce alcohol use among inner-city, adolescent youth. 12

PAGE 13

CHAPTER 1 BACKGROUND AND SIGNIFICANCE Alcohol use among youth contributes to a num ber of deleterious health and social outcomes, including traffic crashes, increased risk for disease, risky sexual behavior, homicides, suicides, crime, unintentional injury (Bor owsky, Ireland, & Resnick, 2001; Dunn, Bartee, & Perko, 2003; Greenfeld, 1998; Gyimah-Brempon g, 2001; National Highway Traffic Safety Administration, 2005; National Institute on Alcohol Abuse and Alcoholism, 2000; Smith, Branas, & Miller, 1999; Sorenson & Berk, 2001), and is the third leading actual cause of death (i.e., nongenetic, modifiable fact or contributing to death; McGinnis & Foege, 1993) in the United States (Mokdad, Marks, Stroup, & Gerber ding, 2004). Recent research has shown that exposure to alcohol during adolescence can have detrimental effects on brain development, intellectual capabilities, and increases the likelihood for later addiction (Brown, Tapert, Granholm, & Delis, 2000; Monti et al., 2005). Furthe r, a number of studies provide evidence for increased risk for these problems and earlier age at alcohol initiation (D eWit, Adlaf, Offord, & Ogborne, 2000; Ellickson, Tucker, & Klein, 2003 ; Guo et al., 2002; McGue, Iacono, Legrand, Malone, & Elkins, 2001; Stueve & O'Donnell, 2005; Warner & White, 2003). Notwithstanding risks, alcohol remains one of the most widely us ed substances during earlyand late-adolescence (Johnston, O'Malley, Bachman, & Schulenberg, 2008). Prevalence of Alcohol Use among Youth in the United States Despite slight declines in recent years, alcohol clearly remains the most frequently used drug among youth in the United States (Johnston et al., 2008; Table 1-1). In 2007, 39% of 8thgraders used alcohol in their life time, 32% used alcohol in the past year and 16% used in the past month (Johnston et al., 2008; Johns ton et al., 2008). Heavy, problem atic use is also prevalent during early adolescence; 18% of 8th-graders have been drunk in their lifetime, 13% have been 13

PAGE 14

drunk in the past year, and 6% have been drunk in the past month (Johnston et al., 2008). Further, 10% of 8th-graders reported heavy episodic useha ving had five or more drinks in a row in the previous two weeks (Johnston et al., 2008). These numbers increase dramatically as youth progress through adolescence, with 72% of youth in 12th grade reporting lifetime alcohol use, 44% reporting current use, 46% having been drunk in the past year, 29% having been drunk in the past month, and 26% reporting h eavy episodic use (Johnston et al., 2008). Prevalence of Alcohol Use among Racial/Ethnic and Gender Subgroups Important variations in alcohol use ex ist among racial/ethnic and gender subgroups (Table 1-1; Johnston et al., 2008). For example, Hispanic youth in 8th grade present the highest prevalence of lifetime, annual and past month alcohol use and having been drunk relative to White and African American youth. This pattern persists until the end of high school, when White youth present the highest prevalence in all use categorie s. The prevalence for African American youth across all cate gories is consistently the lowest among the racial/ethnic subgroups. While historically early onset and the prevalence of alc ohol use has been higher among boys than girls, recent research has shown that the gender gap has decreased and may, in fact, be non-existent, or in some cases reversed. As s hown in Table 1-1, the prevalence of lifetime, annual and past month use and having ever been drunk is consistently higher among adolescent girls than boys (Johnston et al., 2008), with few exceptions, until their senior year of high school. For 12th-grade youth, more frequent, problematic al cohol use (e.g., past month alcohol use and having been drunk) is consistently highe r among boys than girls. In 2007, 47% of 12th-grade boys had used alcohol in the past month, compared to 41% of girls. Additionally, by the end of high school, boys have a higher prevalence of havi ng been drunk than girls, such that 47% of 14

PAGE 15

12th-grade boys had been drunk in the past year, compared to 45% of girls; and 32% of boys had been drunk in the past month, compared to 26% of girls (J ohnston et al., 2008). Prevalence of Alcohol Use among Ethnic Minority Youth in Chicago, Illinois Alcohol use among ethnic minority youth in Ch icago, Illinois is similar to that among youth nationwide (Table 1-2). Data from the 2007 Youth Risk Behavior Surveillance Survey (YRBSS; Centers for Disease Control and Prevention, 2008) indicate that 71% of 9th through 12th grade youth in Chicago have used alcohol in their lifetime (72% of girls and 71% of boys), 39% have used alcohol in the past month (40% of girls, 37% of boys), and 20% report heavy episodic alcohol use (21% of girls and 19% of boys ). Sixty six percent of African American and 80% of Hispanic youth in Chicago have used alcohol in their lif etime, 30% and 47% of African American and Hispanic youth have used alcoho l in the past month, re spectively, and 12% of African American and 27% of Hispanic youth report heavy episodic al cohol use. The YRBSS reported prevalence for White yout h was not available in 2007. Prevalence of Alcohol Use among th e Project Northland Chicago Sample The present study used data from Project Northland Chicago (PNC), a randomized controlled trial of a comprehens ive alcohol-preventive interven tion for multi-ethnic urban youth (Komro et al., 2004; Komro et al., 2008). A cohor t of young adolescents participated in the study, which began when they were in the 6th grade and concluded at the end of their 8th grade year. (See Chapter 2 for further details.) Table 1-3 presents the prevalence of alcohol use among youth who participated in the study. For this co hort of youth, alcohol use is higher among boys in 6th grade for nearly all categ ories of alcohol use (past month use at the end of 6th grade provides the exception). However, by the time these youth have reached 7th grade, alcohol use among girls meets or exceeds that of boys for all categories of alcohol use. These patterns are similar to those found among youth nationwide (Johns ton et al., 2008), as are the patterns across 15

PAGE 16

the racial/ethnic subgroups in the study. Hispanic youth presen t the highest prev alence across all use categories and all grade years. White yout h have the lowest prevalence across all use categories until 8th grade, when they exceed African Am erican youth in all use categories. Consequences of Adolescent Alcohol Use The extant literature substa ntiates many immediate consequences of alcohol use among underage youth that include, but are not limited to, increased risk for: criminal activity, alcoholrelated traffic crashes, fatalities, and risky dr iving behavior, unintenti onal injury, risky sexual behavior, delinquency, and depression. Additiona lly, alcohol use during adolescence has also been shown to contribute to a number of more distant consequences, such as: diminished neurocognitive functioning, various cancers, cardiovascular disease, increased risk for alcohol abuse and dependence, other illicit drug use, and alcohol -related traffic crashes. Immediate Consequences A number of studies from th e fields of sociology and cr iminology have found a positive correlation between alcohol and crime (Blount, Silverman, Selle rs, & Seese, 1994; Bromley & Nelson, 2002; Greenfeld, 1998; Gyimah-Brempong, 2001; National Institute on Alcohol Abuse and Alcoholism, 2000; Parker, 1993; Parker, 1995 ; Parker & Cartmill, 1998; Roizon, 1997; Scribner, Cohen, Kaplan, & Allen, 1999; Stitt & Giacopassi, 1992; Valdez Kaplan, Curtis, & Yin, 1995; Vanoers & Garretsen, 1993; Zha ng, Welte, Wieczorek, & Messner, 2000). In an analysis of national, longitudinal data on the prevalence of al cohol involvement in crime, Greenfeld (1998) found that approximately 3 million violent crimes occur each year in which the victim perceived the offender to have been drin king at the time of the offense. Additionally, two-thirds of victims who suffered violence from an intimate partner reported that alcohol had been a factor. Further, among violent offenders, 41% of probationers, 41% of those in local jails, 38% of those in state prisons, and 20% of those in federal prisons were estimated to have been 16

PAGE 17

drinking when the crime was committed. More re cently, Miller and coll eagues (2006) estimated that 57% of violent crimes (e.g. rape, robbery, a ssault, murder), 67% of crimes against children (e.g. physical and sexual abuse), and 58% of property crimes (e.g. larceny, burglary, motor vehicle theft) were attributable to alcohol and other drug use. Moreover, the authors estimate that alcohol-attributable crimes have an economic cost of $84 billion, more than double the estimated $38 billion in costs due to other dr ug-attributable crimes (Miller, Levy, Cohen, & Cox, 2006). Alcohol-related traffic crashes, fatalities a nd risky driving behavi or are other serious immediate consequences of alcohol use by youth (National Highway Traffic Safety Administration, 2006; Zakrajsek & Shope, 2006). Car crashes are the leading cause of death for teenagers, and nearly one-quarter of youth in fa tal traffic crashes have been drinking (National Highway Traffic Safety Administration, 2006). Further, 30-day prevalence data from the YRBSS indicated nearly 30% of youth in 9th through 12th grade nationwide had ridden in a car driven by someone who had been drinking, a nd 10% had driven a car after they had been drinking alcohol (Eaton et al., 2005). These intoxicated driver s are not only a danger to themselves, but a considerable danger to others, as nearly half of the people wh o die in crashes involving an underage drinking driver are peop le other than the driver (Nat ional Highway Traffic Safety Administration, 2005). Additionally, alcohol is a leadi ng contributor to injury-related deaths. Suicide, homicide, assault, drowning, and recreational injury involve alcohol in one quarter to three quarters of cases (Smith et al., 1999). In a meta-analysis of 331 medical examiner studies published between 1975 and 1995 reporting non-traffic injury fatalities, Smith, Branas and Miller (1999) found that alcohol was involved in 27% of poisonings, 90% of deaths due to hypothermia, 21% of gunshot 17

PAGE 18

fatalities, 42% of deaths due to burns or fire, 63% of fatal falls, and approximately 50% of drowning fatalities, homicides, and suicides. Vi ctims of burns and hypothermia had the highest intoxication levels, followed by drowning, falls motor vehicle acciden ts, homicides, and suicides. The authors noted that fewer than 25% of ar ticles reviewed report ed gender and agespecific rates; however, their findi ngs clearly illustrate that alcohol is an important factor in many fatal injuries, age notwithstanding. Alcohol has also been found to influen ce risky sexual behavior among youth (Cook & Clark, 2005; Dunn et al., 2003; Eaton et al., 2005; Guo et al., 2002; Halper n-Felsher, Millstein, & Ellen, 1996; Poulin & Graham, 2001; Santelli, Brener, Lowry, Bhatt, & Zabin, 1998; Santelli, Robin, Brener, & Lowry, 2001; Shrier, Emans, Woods, & DuRant, 1997; Stueve & O'Donnell, 2005). Data from the 2007 YRBSS show that among the 35% of sexually active youth nationwide, nearly one-quarter dr ank alcohol before their last sexual intercourse (Centers for Disease Control and Prevention, 2008). Such beha vior increases the odds of having unprotected sex (i.e. failure to use a condom), multiple partners, pregnancy and contracting sexually transmitted diseases (Cook & Clark, 2005; Dunn et al., 2003; Guo et al., 2002; Stueve & O'Donnell, 2005). Further, nearly three quarters of date rape situations involve individuals who have been drinking (Mohler-Kuo, Do wdall, Koss, & Wechsler, 2004). Additionally, alcohol use has been found to be significantly associated with delinquency. Findings from the 2003 National Survey on Drug Use and Health (NSDUH) i ndicate that as the level of alcohol use in the past year among youth aged 12 to 17 increased, so also did the percentage of youth who had (1) got ten into a fight at work or sc hool; (2) participated in a groupagainst-group fight; (3) a ttacked someone with the intention to seriously hurt them; (4) stole, or tried to steal, something worth more than $50; (5) sold illegal drugs; and (6) carried a handgun 18

PAGE 19

(Substance Abuse and Mental Health Services Administration, 2005). Si milarly, in a study of 2,078 youth aged 14 to 16 years, Best and coll eagues (Best, Manning, Gossop, Gross, & Strang, 2006) found that excessive drinking was positivel y associated with frequent truancy and involvement in deli nquent behavior. Moreover, youth who use alcohol are more likely to be depressed (Best et al., 2006; Windle & Davies, 1999; Zullig, Valois, Huebner, Oeltmann, & Drane, 2001), have reduced life satisfaction (Zullig et al., 2001) and experience academic failure (Grunbaum et al., 2004; Renna, 2007). In a longitudinal study of 975 high school sophomores and juniors, Windle and Davies (1999) found that approximately one -quarter of youth identified as heavy drinking also met the criterion for depression. Further, in a cross-sectional survey of 5,032 high school students, Zullig and colleagues (2001) found that age of alcohol initiation (i.e., younger th an 13 years), regular alcohol use and binge drinking am ong youth were associated with si gnificant reductions in life satisfaction. Lastly, in an analysis of data from the National Longit udinal Survey of Youth (NLSY), Renna (2007) estimated that heavy episodic alcohol use may reduce the probability of graduating high school on-time by as much as 5.2% for girls and 14.5% for boys. Distant Consequences In addition to immediate consequences, alcohol use during adolescence has been associated with a number of more distant hea lth and social consequences. First, underage alcohol use represents a considerable expense to American society. One estimate of the societal cost of underage alcohol use in the United States is $53 billion annually, attributed to loss of young lives, lost productivity, and health care costs (Pacific Institute of Research and Evaluation, 1999). Moreover, in an estimate of alcohol a nd other drug-related cr ime alone, Miller and colleagues (Miller et al., 2006) report that the bill for alcohol and other drug crimes exceeded $205 billion in 1999, attributed to tangible medical, mental health, property loss, future earnings, 19

PAGE 20

public services, adjudication, and sanctioning costs, as well as the value of pain, suffering and lost quality of life. Second, in recent years, alcohol has been shown to have effects on adolescents neurocognitive development and functioning. Using a sample c ontaining 33 alcohol-dependent adolescents matched with 24 adolescents with no history of alcohol or drug problems, Brown and colleagues (2000) found that alcohol-depende nt adolescents showed significantly poorer performance on verbal and nonverbal retention, visuospatial functioning, and retrieval of verbal and nonverbal information. Additionally, Monti and collegues (2005) reported that drinking during adolescence produces: permanent change s in brain physiology; reduced neuronal plasticity; neurocognitive disadvantages; and may make the brain more susceptible for later alcohol dependence. Third, alcohol use can increase the risk fo r several cancers, incl uding cancers of the mouth, esophagus, stomach, pancreas, colon and breast (Bagnardi, Biangiardo, La Vecchia, & Corrao, 2001; Corrao, Bagnardi, Zambon, & La V ecchia, 2004; Kune & Vitetta, 1992; La Vecchia & Negri, 1989; Seitz & Poschl, 1997). Add itionally, alcohol may increase the risk for cardiovascular disease (Bryant, Schulenberg, O'Malley, Bachman, & Johnston, 2003; Corrao et al., 2004; Corrao, Rubbiati, Bagnardi, Zambon, & Poikolainen, 2000; Rehm, Gmel, Sempos, & Trevisan, 2003) and stroke (Corra o et al., 2004; English et al., 1995). Lastly, alcohol use during early adolescence has been associated with increased risk for alcohol and other drug abuse and dependence in late adolescence and into adulthood. According to the Gateway Hypothesis (Kande l & Jessor, 2002), alc ohol serves as a gateway to use of marijuana and other illicit dr ugs (Jackson, Sher, Cooper, & Wood, 2002; Kandel & Jessor, 2002; Kandel, Yamaguchi, & Chen, 1992; Willner, 2001; Wilson, Battistich, Syme, & Boyce, 2002). 20

PAGE 21

Youth who use alcohol have been found to be at increased risk to initiate smoking (Best et al., 2006; Jackson et al., 2002) and use marijuana and other illicit drugs (Best et al., 2006; Kandel et al., 1992; Willner, 2001; Wilson et al., 2002) relative to those who have not used alcohol during adolescence. Additionally, severa l studies have shown that adolescents who initiate alcohol use early during adolescence are at increased risk for subsequent use, abuse and dependence into adulthood (Grant et al., 2006; Guo, Collins, Hi ll, & Hawkins, 2000; Hawkins, Catalano, & Miller, 1992; Hingson, Heeren, & Winter, 2006). One cross-sectional study of 43,093 adults found that relative to respondents who began drin king at age 21 years or older, those who began drinking before age 14 were almo st twice as likely to experience alcohol dependence within 10 years of alcohol initiatio n (Hingson et al., 2006). Those who dr ink earlier in life are also more likely to report driving after drinking (H ingson, Heeren, Levenson, Jamanka, & Voas, 2002; Hingson, Heeren, Winter, & Wechsler, 2003), bei ng in an alcohol-relate d traffic crash in adulthood (Hingson et al., 2002; Hingson, Heeren, Winter et al ., 2003), and are 3 to 4 times more likely to have been in a fight after drinking (Hingson, Heeren, & Zakocs, 2001). Etiology of Adolescent Alcohol Use Given the high rates and considerable conse quences associated with alcohol use among adolescents, a substantial body of scientific literature has been devoted to understanding factors associated with initiation of alcohol use am ong adolescents. The primary theoretical foundation for the present study, the Theory of Triadic Infl uence (TTI), was used to organize the findings. The TTI (Flay & Petraitis, 1994) is a relatively new meta-theory that in corporates individualand environmental-level constructs from other social-behavioral theories [e.g. Health Belief Model (Becker, 1974; Janz & Bekcer, 1984), Protection Motivation Theory (Rogers, 1983), Theory of Reasoned Action (Fishbein & Ajzen, 1975) and Theory of Planned Behavior (Ajzen, 1985, 1988)] into a comprehensive model for understa nding health behaviors (Flay & Petraitis, 21

PAGE 22

1994). The model organizes proximal, distal, and u ltimate factors by three st reams of influence: intrapersonal, social and attitudinal. Proximal, Distal and Ultimate Risk and Protective Factors Proximal factors postulated to influence al cohol use among youth include experiences (i.e. alcohol use expectancies, so cial reinforcements), intentions, and cognitions (i.e. beliefs about subjective norms, att itudes and self-efficacy regard ing alcohol use, perceived accessibility). There is a large, fairly consistent body of literature substantiating the effects of these proximal factors on alcohol use among adol escent youth. For example, attitudes favorable to alcohol use (Barnow, Schultz, Lucht, Ulric h, & Freyberger, 2004; Bo t, Engles, & Knibbe, 2005; Darkes, Greenbaum, & Goldman, 2004; Hawk ins et al., 1992; Hipwell et al., 2005; Scheier & Botvin, 1997), poor resistance self-effi cacy (Hawkins et al., 1992; Scheier, Botvin, Diaz, & Griffin, 1999; U.S. Department of Health and Human Services, 20 00), and beliefs that alcohol use is normative and positive (Aas & Klepp, 1992; Hawkins et al., 1992; U.S. Department of Health and Human Services, 2000) have all been found to be positively associated with alcohol use among adolescents. It is important to note, that most research that has included minority youth has been cross-sectional and has focused on these more proximal influences on behavior. According to the TTI, distal factors are theo rized to cause the proxi mal factors, having an indirect effect on adolescent alcohol use, as well as affecting the behavior directly. A variety of family and peer factors have been found to infl uence adolescent alcohol use, both directly and indirectly through the proximal factors noted previously. Alcohol use among family members (Ary, Tildesley, Hops, & Andrews, 1993; Di elman, Butchart, & Shope, 1993; Jackson, Henriksen, & Dickinson, 1999; Yu, 2003), accessibility of alcohol in the home (Jackson et al., 1999; Komro, Maldonado-Molina, Tobler, Bonds, & Muller, 2007), peer use (Callas, Flynn, & 22

PAGE 23

Worden, 2004; Curran, Stice, & Chassin, 1997; D'Amico & McCarthy, 2006; Geckova & van Dijk, 2001; Henry, Slater, & Oetting, 2005; Loveland-Cherry, Leech, Laetz, & Dielman, 1996; Marsden et al., 2005), and family functioning, namely parental monitoring (Alvarez, Martin, Vergeles, & Martin, 2003; Boraws ki, Ievers-Landis, Lovegreen, & Trapl, 2003; Clark, Thatcher, & Maisto, 2005; Cleveland, Gibbons, Gerrard, Pmery, & Brody, 2005; McArdle et al., 2002; van der Vorst, Engels, Meeus, Dekovic, & Van Leeuwe, 2005), parent/child communication (Kelly, Comello, & Hunn, 2002; Wills, Gibbons, Gerrard, Murry, & Brody, 2003), relationship satisfaction (Ledoux, Miller, C hoquet, & Plant, 2002; Nelson, Patience, & MacDonald, 1999; Wills et al., 2003), and supervision (Aizer, 2004; Coley, Morris, & Hernandez, 2004; Richardson, Radziszewska, Dent, & Flay, 1993), have all been shown to be significantly associated with alcohol use among youth, such that alcohol use among family members, increased accessibility of alcohol in the home and increased peer use are associated with increased levels of adolescent alcohol use. Protec tive family functioning characteristics, such as increased parental monitoring, parent/child communication and rela tionship satisfaction, typically show an inverse relatio nship with alcohol use among youth. According to the TTI, ultimate factors also ha ve direct and indirect effects on behavior, with indirect effects o ccurring through the distal factors; however, the direct effects for the ultimate, macro-level factors are postulated to be smaller than effects observed for factors more proximal to the adolescent. They represent th e broad environment in which youth are embedded and are encompassed by a number of factors that include, but are not limited to, exposure and access to alcohol in the community, extant soci al capital and the broader cultural norms for alcohol use among youth. Such ultimate factor s were the focus of the present study. 23

PAGE 24

Both the number of alcohol outlets in a community, or alcohol outlet density, and the number of alcohol advertisements within commun ities provide measures of exposure to alcohol outside the home. Such exposure has been found to be disproportionately located in urban, lowincome, minority communities (Hackbarth et al., 2001; Pollack, Cubbin, Ahn, & Winkleby, 2005; Treno, Alaniz, & Gruenewald, 2000). For example, a study of 82 neighborhoods in four northern/central California citie s found that the most economica lly disadvantaged neighborhoods had three times the alcohol outlet density than th at of the least depriv ed neighborhoods (Pollack et al., 2005). Further, a study of outdoor alcohol and tobacco advertising in Chicago communities found that low-income and minority communities had significantly more alcohol and tobacco advertisements than communities with White or no-racial-majority communitiesat an approximately 3:1 ratio (Hackbarth et al., 2001). Such apparent targe ting of low-income, ethnic minority populations is not without its consequences. Numerous studies have found significant relations between alcohol outlet dens ities and outdoor alcohol advertisements and a number of deleterious health a nd social outcomes, such as increased alcohol consumption and intentions to drink (Collins, Ellickson, McCa ffrey, & Hambarsoomians, 2007; Ellickson, Collins, Hambarsoomians, & McCaffrey, 2005; Fleming, Thorson, & Atkin, 2004; Pasch, Komro, Perry, Hearst, & Farbakhsh, 2007; Scribner, Cohen, & Fisher, 2000; Scribner et al., 2007; Snyder, Milici, Slater, Sun, & Strizhakova, 2006; Stac y, Zogg, Unger, & Dent, 2004), violence (Gorman, Labouvie, Speer, & Subaiya, 1998; Gorman, Speer Labouvie, & Subaiya, 1998; Scribner et al., 1999; Speer, Gorman, Labouvie, & Ontkush, 1998), and traffic crashes (Scribner, Mackinnon, & Dwyer, 1994). Mkel and colleagues (Mkel Osterberg, & Sulkunen, 1981; Mkel 2002; Scribner et al., 1994) provide a compelling exam ple, where a 46% increase in liquor-licensed restaurants and a 22% increase in retail monopoly st ores in Finland were associated with a 46% 24

PAGE 25

increase in the volume of alcohol consumed, a 63% increase in the frequency of alcohol consumption and a 20% increase in heavy alco hol consumption among individuals aged 15 to 69. Further, despite existing laws that make it illegal for youth under the age of 21 to purchase alcohol in the United States, underage youth can, and do, purchase it. Studies indicate that underage buyers are able to purchase alcohol without showing age identification in 47-97% of attempts (Forster et al., 1994; Grube, 1997; Paschall, Grube, Black, Flewelling et al., 2007; Preusser & Williams, 1992). Such co mmercial access to alcohol may contribute to the number of underage youth who use and abuse alcohol (Johnston, O'Malley, Bachman, & Schulenberg, 2007) and represents a considerable risk to healthy adolescent development. Another ultimate-level constr uct hypothesized to influence alcohol use and alcohol use intentions among adolescents is extant social capital. Cole man (1994, 2007) describes social capital as a confluence of entities having two common characteristics (1) they consist of a social structure and (2) they facilitate individual and/or collective action within the structure. Further defined, it describes engagement social trust, and reciproc ity and help among neighbors and community organizations which contributes to available tangible a nd intangible resources (Putnam, 1993; Weitzman & Che n, 2005). Commonly used measures of social capital include neighborhood problems (real and perceived), community activis m and strength, and available social (e.g., peer networks, efforts/programs offered by not-for-profit organizations) and economic resources (Szreter & Woolcock, 2004). Research has shown an association between such measures of social capital and a number of health-related outcome s, including self-rated health and health behavior s (Poortinga, 2006), smoking (Siahpus h et al., 2006), and mortality (Kawachi, Kenneday, Lochner, & Prothrow-Sti th, 1997). Weitzman and Chen (Weitzman & 25

PAGE 26

Chen, 2005) offer hypotheses as to how social cap ital influences alcohol use and intentions among youth, suggesting that social capital promot es healthy attachment to the community and norms favorable to abstaining from drinking. Lastly, alcohol enjoys a soci al sanction unlike any other drug (National Research Council & Institute of Medicine, 2004). Youth are bombard ed with media messages depicting alcohol use as normative and positive (Chris tensen, Henriksen, & Roberts, 2000). Such messages are often reinforced in the communities in which youth resi de, and unfortunately, in their homes as well. Alcohol is readily accessible both commercia lly and socially (Jones-Webb et al., 1997; Wagenaar et al., 1993; Wagenaar et al., 1996) and enforcement of the minimum legal drinking age is modest at best (Wagenaar & Wolfson, 19 94). Moreover, 13% to 35% of adolescent youth receive alcohol directly from their parent(s) and 11% to 31% take alcohol from their home for their drinking occasions (Ary et al., 1993; Foley, Altman, Durant, & Wolfson, 2004; Harrison, Fulkerson, & Park, 2000; Jackson et al., 1999; Maisey & Davies, 2003; Marsden et al., 2005; Rossow, Pape, & Storvoll, 2005; Smart, Adlaf, & Walsh, 1996; Williams & Mulhall, 2005). Clearly, youth are receiving detrimental messages about alcohol (i.e. that it is normative and positive) from society as a whole, and from thei r families as well, when the message that should be sent is, No for children and teens, moderati on for adults, and excessive drinking is taboo for all (Califano, 2007). Limitations of Current Knowledge While the literature on the proximal and distal domains that influence adolescent alcohol use is prolific, the ultimate, neighborhood-level influences on alcohol use among adolescents have more often been assumed than empirically examined (Britt, Carlin, Toomey, & Wagenaar, 2005; Duncan, Duncan, & Strycker, 2002; Gibbons et al., 2004; National Institute on Alcohol Abuse and Alcoholism, 2004; Roski et al., 1997 ; Toumbourou et al., 2007; Wagenaar, Toomey, 26

PAGE 27

& Lenk, 2004). While this gap in th e literature is unfortunate, it is certainly understandable, as the relations between such ultimate-level f actors and individual behavior are complex, interrelated and require a substan tial body of multilevel da ta whose collection is often restricted by funding and other temporal cons iderations. However, empirically examining these relations remains important as neighborhood context may be a key predictor of substance use for adolescents. Understanding the unique context in which youth are embedded may lead to the development of more efficacious preventive interventions and policy initiatives to reduce alcohol use among underage youth. Additionally, relatively few longitudinal stud ies have been conducted among racial/ethnic minority youth. This is a critical ga p in the literature, as census da ta show that the population in the United States is quickly moving toward a maj ority-minority society. At the time of Census 2000 (U.S. Census Bureau, 2000), three states already had more than 50% minority populations (Hobbs & Stoops, 2002; U.S. Census Bureau, 2003). In 2005, the U.S. Census Bureau estimated that nearly half of children un der age 5 in the United States were racial/ethnic minorities (U.S. Census Bureau, 2007). Further, Af rican American youth drink alcohol in lower quantities and less frequen tly than most other racial groups (Substance Abuse and Mental Health Services Administration, 2006); yet, they suffer disproportionately from the physical and social consequences of alcohol use (N ational Institute on Alcohol A buse and Alcoholism, 2000). At present, they are the only racial/ethnic group experiencing this paradox (Caetano, Clark, & Tam, 1998; National Institute on Alcohol Abuse and Alcoholism, 2001). The explanations for this paradox are limited; however, Wallace (1999) suggests that the disproporti onate distribution of alcohol-related problems are related to the racialized nature of American society which lends to racial/ethnic differences in i ndicators of socioeconomic status and exposure to contextual, 27

PAGE 28

ultimate-level risk factors. These demographic and social trends, and limited explanation of disproportionate alcohol-related problems, clearly elucidate the importa nce of understanding the etiology of alcohol use among such growing, at-r isk, segments of the United States population, and suggests that alcohol use among these youth may be the result of not only individual characteristics, but also the interaction of envi ronment and cultural contexts (Godette, Headen, & Ford, 2006). Further, of studies examining alcohol us e among ethnic minority youth, most are crosssectional and have focused on more proxima l influences on behavior. One important longitudinal study by Griffin and colleagues (2000) found that their set of proximal risk and protective factors was less pr edictive of alcohol use among African American and Hispanic youth compared to White youth. They concluded that ultimate-level factors, such as environment and cultural context, may be more important among minority youth, as many minority youth face a myriad of challenges related to their urban en vironments compared to their rural and suburban counterparts (Griffin et al., 2000). Cl early, more research is needed to determine which proximal, distal and ultimate factors play an important role in drinking behaviors among minority youth, as they may differ from those among White youth (the referent group for much of extant scientific theory). Moreover, there is a paucity of literature examining African American and Hispanic youth residing in an urban context. African Am ericans and Hispanics ar e disproportionately residents of metropolitan cities (U.S. Census Bureau, 2000) and ecological research has shown that many social problems (e.g., crime, delinqu ency, drug use, public disorder, and school dropout) are significantly cluste red in neighborhoods of con centrated poverty, racial heterogeneity and family instability (Coulton, Korbin, Su, & Chow, 1995; Duncan et al., 2002; 28

PAGE 29

Sampson, 1992). As such, urban minority youth are disproportionately at risk for alcohol and other drug use, which has been attributed to a nu mber of factors, such as neighborhood disorder, a sense of hopelessness, psychological distress, increased opportunity for drug use, weaker economic conditions and fewer neighborhood re sources (Arkes, 2007; Crum, Lillie-Blanton, & Anthony, 1996; Duncan et al., 2002; Elliott et al., 1996; Hill & Angel, 2005; Karvonen & Rimpela, 1997; Wilson, Syme, Boyce, Battistich, & Selvin, 2005). Thus, examination of the impact of the unique urban environment in which many minority youth reside is warranted. Lastly, much of the extant literature describing the neighborhood contexts, and subsequent maladaptive behavior s among youth, has relied on either census data (Allison et al., 1999; Chuang, Ennett, Bauman, & Foshee, 2005; El liott et al., 1996; Galea, Ahern, Tracy, & Vlahov, 2007) or self-report measures (Crum et al., 1996; Gibbons et al., 2004; Hill & Angel, 2005). While informative, such studies fail to capture the multiple dimensions inherent in available social capital, focu sing on socioeconomic indicators or neighborhood disorder or dysfunction alone. A more multifaceted approach to describing available social capital among inner-city communities may lend to improved understanding of the context in which many minority youth reside, as well as disparities in health and well-being among urban, minority populations (Fiscella & Williams, 2004). Theoretical Foundation The present study was guided primarily by th e Theory of Triadic Influence (Flay & Petraitis, 1994) and supplemented with Wagenaar and Perrys Model of Drinking Behavior (Wagenaar & Perry, 1994). These two theories recognize that adolescents are embedded within an environment that influences, and is influen ced by, individual behavior. Additionally, these theories assert that a variety of factors at multiple levels of influence (i.e. proximal to ultimate) directly and indirectly influence behavior. Further, both theo ries were developed in the context 29

PAGE 30

of adolescent substance use and provided the theoretical founda tion for the PNC intervention and evaluation. As such, these theories represent a useful framework for examining how more distal, neighborhood-level factors affect adolescent drinking behavior, and if parenting practices may mediate these effects. Theory of Triadic Influence The Theory of Triadic Influence (Flay & Petr aitis, 1994) is a relati vely new meta-theory that postulates that health-related behaviors are affected by three streams of influence environmental (i.e. the macro-environment, or th e broad social and cultura l environment), social (i.e. the micro-environment, or ones more imme diate social context), and intrapersonal (i.e. ones biology or personality). Within each of these streams are five tiers of influence, ultimate, distal and proximal. The top tier repr esents ultimate causes of behaviorfactors in the environment believed to be the root causes of behavior. The second tier, the social-personal nexus, is where ultimate causes inte ract to provide personally releva nt, but still general, social relationships, knowledge and values, and sense of se lfand social competence. In the third tier, termed expectancy-value, the social-personal ne xus becomes more specific to a particular behavior. All streams flow into the cognitive (fourth) tiersocial normative beliefs, attitudes and self-efficacy. Lastly, the social cognitions on the fourth tier determine the decision/intention to act in a certain way in a particular situation. While the streams of influence are conceptualized as flowing vertical ly down the tiers of influence, this theory recognizes potential interstream pathways. This theory is intended to account for factors th at have both direct and indirect effects on behavior and provides a comprehensive approach to understanding the etiology of health-related behaviors. 30

PAGE 31

Wagenaar and Perrys Model of Drinking Behavior A model of drinking behavior developed by Wagenaar and Perry (1994) comprehensively considers the antecedents of alcohol use. Th e authors consider a number of factors at the ultimate, distal and proximal levels. Specifical ly, broader socio-environmental conditions (i.e. public policy, social/institutional structures, ma rket mechanisms, availability, and social controls) are hypothesized to dir ectly influence drinking behavior in addition to indirectly influencing behavior via effects on social interactions (e.g. in formal social controls, significant others, parents, peers, coworkers) and alcohol cognitions and perceptions. This model highlights the centrality of social interac tion and the importance of broade r socio-environmental conditions. Summary of Theoretical Framework Figure 1-1 presents the theoretical framew ork for the present study, based on the TTI and Wagenaar and Perrys model of drinking beha vior. Neighborhood risk was characterized by exposure and access to alcohol and extant soci al capital, defined by perceived neighborhood problems, community preventive efforts, neighb orhood strength and deprivation. Also, home and family management was characterized by home alcohol access, alcohol-specific communication, parent/child communication, and parental mon itoring. It was hypothesized that the ultimatelevel, environmental context (i .e., neighborhood risk) would have both direct and indirect effects on alcohol use, but that the direct effect would be partially mediated by more proximal levels of influence (e.g., home and family management). Neighborhood risk was conceptualiz ed as having a direct effect on alcohol use, such that higher levels of neighborhood risk would be associat ed with more alcohol use. Moreover, it was hypothesized that the effect of neighborhood risk on alcohol would be partially mediated through effects on home and family management practices such that higher leve ls of neighborhood risk 31

PAGE 32

would be associated with increa sed protective home and family management practices, which, in turn, would be associated with reduced alcohol use. While the literature describing the relati ons between neighborhood risk and home and family management practices is scant and inc onclusive, several studies have suggested that protective home and family management practices increase in neighborhoods of greater risk and social disadvantage (Beyers, Bates, Pe ttit, & Dodge, 2003; Brook, Nomura, & Cohen, 1989; Rankin & Quane, 2002; Tobler, Komro, & Maldonado-Molina, 2007). For example, Rankin and Quane (2002) found that concentrated disadvant age, residential stab ility and neighborhood collective efficacy were positively associated with parental monitoring, parental involvement and family rules, with the effects of collective e fficacy on parental monitoring reaching statistical significance. Further, qualitativ e research has also suggested such relationshipsincreased neighborhood protection and parent al monitoring strategies in low-income neighborhoods have diluted the deleterious effects of neighborhood risk on child development (Jarrett, 1997). Innovation of the Study The present study sought to address current limitations in th e scientific literature by examining ultimate, neighborhood-level contextual f actors that influence a dolescent alcohol use, particularly for racial/ethnic mi nority youth, using multilevel data that had been collected as part of a large, randomized preventive interv ention conducted among urban, low-income, multiethnic adolescents. Specifically, this study de scribed the contexts of urban, low-income, minority youth and identified both the direct effects of such cont exts on alcohol use and alcohol use intentions and indirect effects, mediated by distal-level home and family management practices. This study included (1) longitudinal data from a large sample of urban, poor, minority youth (n=5,812); (2) risk and protective factor s measured at different levels (neighborhood, family, peer and individual) and from different sources (observations, leaders, archival, parents, 32

PAGE 33

youth); and (3) use of advanced, st ate-of-the-science sta tistical methods to examine the effects of ultimate-level neighborhood factors on alcohol use among adolescents. Specific neighborhood contextual factors under study included: alcohol promotion and accessibility; perceived neighborhood problems; neighborhood strength; preventive action by communities, law enforcement and community organizations; and indicators of poverty and deprivation from Census 2000 (U.S. Census Bureau, 2000). These ne ighborhood measures were conceptualized as representing exposure and access to al cohol and extant social capital. Study Goal, Aims and Research Questions The overall goal of the present study was to understand how ultimate-level, neighborhood context influences young adolescent drinking behaviors and intentions direc tly and indirectly through distal-level, home and family manage ment practices. This goal was accomplished by conducting three distinct investig ations through secondary analysis procedures using data from Project Northland Chicago, with the respective findings presented in three papers. Each investigation built upon the one previous and le nt to completion of the overall goal. Multi-ethnic, Urban Youths Exposure to Patterns of Alcohol-related Neighborhood Characteristics Much of the extant literature describi ng the neighborhood contexts of youth, and often subsequent maladaptive behaviors, has relied on either census da ta (Allison et al., 1999; Chuang et al., 2005; Elliott et al., 1996; Galea et al., 2007 ) or self-report measures (Crum et al., 1996; Gibbons et al., 2004; Hill & Angel, 2005). This i nvestigation extended the scientific knowledge by providing a unique description of urban communitie s using several measures of social capital, as well as direct assessments of exposure and access to alcohol, all of which have been substantiated in the scientific literature as risk factors for licit and illicit drug use and other maladaptive behavioral outcomes among youth (C ollins et al., 2007; Ellickson et al., 2005; 33

PAGE 34

Fleming et al., 2004; Gorman, Labou vie et al., 1998; Gorman, Speer et al., 1998; Kawachi et al., 1997; Mkel et al., 1981; Mke l 2002; Pasch et al., 2007; Poortinga, 2006; Scribner et al., 1999; Scribner et al., 2000; Scri bner et al., 1994; Scribner et al., 2007; Siahpush et al., 2006; Snyder et al., 2006; Speer et al., 1998; Stacy et al., 2004; Weitzman & Chen, 2005). Specifically, the research questions addressed were (1) How many latent classes are necessary to describe the heterogeneity in neighborhood risk among multi-ethnic young adolescent residents of urban communities? and (2) What are the characteristics and proportions of adolescents residing in the heterogeneous latent neighborhood ri sk classes? The research ques tions, and statis tical approach used to address them, have not been addresse d previously in the literature. As such, no a priori hypotheses were made about the number or ch aracteristics of the heterogeneous neighborhood risk classes. Effects of Alcohol-related Neighborhood Context on the Trajectories of Alcohol Use and Intentions among Young Adolescents Little is known about the etiology of alcohol use among Af rican American and Hispanic youth residing in inner cities. Li kewise, our knowledge of how th eir unique context influences alcohol use and alcohol use intentions is limited. This investigation addressed these gaps in the literature by examining the primary research que stion: How does neighborhood context directly influence the trajectories of alcohol use and intentions during early adolescence? The heterogeneous classes of neighborhood risk defined in Paper 1 were used to predict alcohol use and alcohol use intentions over time. We hypot hesized that neighborhood risk latent class membership would significantly pred ict the trajectories of alcohol use and intentions during early adolescence, such that residence in more risky communities would be accompanied by more advanced trajectories of alcohol use and intentions compared to those in lower risk communities. This is consistent with litera ture suggesting that neighborhood ri sk is significantly associated 34

PAGE 35

with a number of deleterious health and soci al outcomes, including alcohol use (Arkes, 2007; Coulton et al., 1995; Crum et al., 1996; Duncan et al., 2002; Elliott et al., 1996; Hill & Angel, 2005; Karvonen & Rimpela, 1997; Sampson, 1992; Wilson et al., 2005). Relationships between Neighborhood Context, Family Management Practices and Alcohol Use among Urban, Multi-ethnic, Young Adolescents Several studies have suggested small or non-significant direct effects of neighborhood contextual factors on adolescen t alcohol use (Duncan et al., 2002; Fulkerson, Pasch, Perry, & Komro, In Press; Lambert, Brown, Phillips, & Ialongo, 2004). It may be that home and family management practices fully or partially mediate the influence of risky environments. This investigation extended the analyses in Pape r 2 and our knowledge of the mechanisms of observed effects by addressing two research questions (1) How does neighborhood context influence home and family management practices (e.g., home alcohol access and protective family management practices)? and (2) Do family management practices mediate the effects of neighborhood risk on early adolescent alcohol use and intentions? Specif ic hypotheses were (1) neighborhood risk would be significantly asso ciated with home and family management practices, such that residence in more risky communities would be accompanied by higher levels of protective home and family management practices (Beyers et al., 2003; Brook et al., 1989; Rankin & Quane, 2002; Tobler et al., 2007); and (2) the effects of neighborhood risk on early alcohol use during early adolesce nce would be partially mediated by protective home and family management practices (Beyer s et al., 2003; Jarrett, 1997). Summary Alcohol use among adolescents continues to be a great concern, give n high rates of use and extensive immediate and long-term conseque nces. While the proximal and distal factors influencing adolescent alcohol us e have been widely studied, the role of more ultimate-level 35

PAGE 36

factors, such as neighborhood context, have often b een assumed than substant iated scientifically. Additionally, relatively few longitudinal stud ies have been conducted among ethnic minority youth and, of those, most focused on proximal influe nces on behavior. This is a critical gap in the literature, as minority youth represent a growing segment of the United States population, they suffer disproportionately from the consequences of alcohol use, and more frequently reside in communities that place them at greater ri sk for maladaptive behavior. Thus, neighborhood contextual factors may be a key predictor of alcohol use, especially among racial/ethnic minority adolescents. Therefore, unde rstanding the magnitude and processes of neighborhood-level effects is of substantial importa nce. Completion of the three inve stigations in this study (1) provided a detailed description of the contex ts in which a large sample of urban, young adolescents resided, (2) quantifie d the direct effects of these neighborhood contexts on alcohol use and alcohol use intentions over time among these high-risk youth, and (3) elucidated the role parents play in buffering the effects of thes e risky contexts on alcohol use. This knowledge may help scientists, practitioners, and policy-make rs alike as they study the conditions that may increase the risk for alcohol use, refine scientific theory, prior itize resources and public policies, and develop more successful preventive interventions. 36

PAGE 37

Table 1-1. Prevalence of alcohol use among 8th, 10th and 12th grade youth in the United States, 2007. Used alcohol Been drunk Grade: 8th10th12th8th10th 12thLifetime: Male 38.4 59.9 71.8 16.8 40.1 55.0 Female 39.3 63.8 72.2 18.8 42.4 55.0 White 37.3 63.0 74.7 17.9 44.6 60.1 African American 37.7 53.2 62.6 14.0 27.5 37.9 Hispanic 48.0 64.4 72.7 23.9 39.9 53.5 Total 38.9 61.7 72.2 17.9 41.2 55.1 Annual: Male 30.9 54.4 66.2 11.7 34.1 46.7 Female 32.7 58.2 66.1 13.5 34.7 45.1 White 31.6 58.3 69.6 13.4 38.3 52.5 African American 27.4 44.0 54.7 8.2 18.5 27.3 Hispanic 39.8 58.0 64.7 16.6 31.4 40.8 Total 31.8 56.3 66.4 12.6 34.4 46.1 30-Day: Male 15.6 33.4 47.1 5.3 18.9 31.7 Female 16.0 33.3 41.4 5.6 17.4 25.7 White 15.6 35.9 49.3 5.9 21.3 33.7 African American 12.3 21.7 28.7 3.7 8.3 14.6 Hispanic 23.0 34.8 41.4 7.4 15.0 24.0 Total 15.9 33.4 44.4 5.5 18.1 28.7 37

PAGE 38

Table 1-2. Prevalence of alcohol use among 9th to 12th grade youth in Chicago, Illinois, 2007. Ever drank Past month use Heavy episodic use Overall 71.4 38.9 20.0 Girls 71.7 40.4 20.6 Boys 71.0 37.3 19.2 African American 66.0 29.8 11.5 Hispanics 79.5 47.3 26.7 38

PAGE 39

Table 1-3. Prevalence of alcohol use among youth participating in Project Northland Chicago, 2002-2005. Past year Past month Past week Heavy episodic Drunk, lifetime Beginning 6th grade Overall 17.2 6.8 2.9 3.9 5.2 Girls 14.9 6.2 2.3 2.9 3.7 Boys 19.3 7.3 3.4 4.9 6.7 Whites 17.2 4.4 2.0 2.2 3.8 African American 15.8 6.8 3.2 4.9 5.6 Hispanics 18.6 7.4 2.7 3.3 5.3 End of 6th grade Overall 22.8 9.6 4.6 4.9 8.0 Girls 21.6 9.7 4.1 4.6 6.4 Boys 23.9 9.5 5.1 5.1 9.4 Whites 18.0 8.3 3.6 2.7 6.8 African American 22.8 8.7 3.9 4.6 7.5 Hispanics 26.3 11.4 5.9 6.1 8.6 End of 7th grade Overall 33.1 15.6 6.9 6.0 10.9 Girls 34.1 17.0 6.9 5.9 10.8 Boys 32.1 14.3 6.9 6.1 10.9 Whites 28.6 11.9 4.6 3.6 10.2 African American 31.2 14.3 6.2 5.6 9.7 Hispanics 38.9 19.7 9.0 8.0 12.9 End of 8th grade Overall 40.4 21.0 10.7 8.6 17.2 Girls 41.9 21.9 11.1 8.1 17.1 Boys 39.7 20.2 10.3 9.0 17.3 Whites 44.0 22.7 13.1 9.6 18.4 African American 38.2 17.6 8.4 7.1 16.9 Hispanics 45.9 27.0 14.1 11.8 19.3 39

PAGE 40

Figure 1-1. Theoretical framewor k of direct and indirect re lationships between neighborhood context, family management practices and early adolescent alcohol use and intentions. 40

PAGE 41

CHAPTER 2 METHODS Overview and Research Questions This study sought to understand how neighbor hood context influences alcohol use among ethnic minority youth residing in urban comm unities. This goal was achieved by conducting three distinct investigations designed to answ er specific research que stions, with each study building on the one previous. An outline of the re search questions for each investigation and the primary analytical approach used for each is presented in Table 2-1. Project Northland Chicago Study Design Project Northland Chicago (PNC) was a randomized controlled trial of schools and surrounding community areas in the city of Chic ago, Illinois (Komro et al., 2004; Komro et al., 2008). From a list of all public schools in Chi cago, schools were selected for recruitment that included grades 5 through 8, had relatively low mob ility rates (less than 25%), and were larger schools (30 or more students per grade). Magnet schools were not included, as they were less likely to be neighborhood based and the intervention included a neighborhood component. Sixty-six schools agreed to pa rticipate and signed a formal Cooperative Agreement form indicating their commitment to the project for three years. After the 66 schools were recruited, they were collapsed into study units to achieve an average of 200 students per study unit. Study un its were defined by combining geographically close schools within city-defined community area s, corresponding to Chicagos census tracts. Study units were matched on ethni city, poverty, mobility, and readi ng and math test scores. Units were randomized into intervention (n = 10 units and 30 schools) or control (n = 12 units and 36 schools) conditions. Before student baseline surveys were implemented, five schools 41

PAGE 42

dropped out of the study due to ti me constraints (4 from the control condition and 1 from the intervention condition). PNC focu sed on a cohort of students who were in the sixth grade in the 2002-2003 school year. Students con tinued in the intervention and evaluation (e.g. repeated annual surveys of students) activ ities through the 2004-2005 school y ear, when the students were completing the 8th grade. Schools that participated in th e study were located throughout Chicago and had similar demographic characteristics to students throughout the Chic ago school district (Table 2-2). Intervention The goals of the PNC intervention were to change personal, social and environmental factors that support alcohol use among young adoles cents (Komro et al., 2004). The intervention was implemented consecutively from 6th through 8th grades and each year of the intervention involved school, family and community components. PNC included three years of (1) peer-led classroom curricula to 10 sessions per year ; (2) parental involve ment and education home-based sessions per year, plus other educational, schoo l, and community involvement activities; (3) peer leadership and youth-planne d community service projects; and (4) community organizing and environmental neighb orhood change (Komro et al., 2004). At the end of the intervention period, there we re no differences in alcohol use, intentions, norms or outcome beliefs between the intervention and control conditions (Komro et al., 2008). Thus, data from both the control and interventi on conditions were used for the present study. Analyses controlled for treatment group membership as appropriate. Data Collection As part of the PNC trial, data were co llected at the neighborhood, school, family, peer and individual levels and included (1) surveys of cohort students, their parents, and community leaders; (2) alcohol purchase attempts by pseudo-underage buye rs; (3) assessment of outdoor 42

PAGE 43

alcohol advertisements within 1500 feet of each study school; and (4) archival data from city records. Table 2-3 summarizes collected data th at was used for the present study by level, data source and time. Students Surveys were administered to students at the beginning of 6th grade (2002, T1), the end of 6th grade (2003, T2), the end of 7th grade (2004, T3) and the end of 8th grade (2005, T4; See Appendix A for a copy of the baseline student surv ey). All students enro lled in the specified grade each year (T1, N = 4,680; T2, N = 4,511; T3, N = 4,062; T4, N = 4,002) were eligible to complete the surveys. Sixty-one schools and 4,2 59 students (91% response rate) participated in the baseline survey, 59 schools and 4,240 students (94% response rate) participated in the T2 survey, 60 schools and 3,778 students (93% response rate) participated in the T3 survey, and 59 schools and 3,802 students (95% respons e rate) participated in the T4 survey. The cohort followup rate was 89% from T1 to T2, 67% from T1 to T3, and 61% from T1 to T4. Loss to follow-up mostly occurred due to two school s closing and students leaving th e study schools. In total, 5,812 students completed at least one of the four surveys; of those, 2,373 (41%) completed all four, 808 (14%) completed three, 1,534 (26%) completed two, and 1,097 (19%) completed one survey. Table 2-4 presents the characteristic s of students who completed one or more study surveys. Survey implementation was conducted at each participating school using a team of three trained survey interviewers. Surveys were r ead aloud to the class and students followed along and filled out their survey as it was read aloud. A survey team would return to the schools 1-2 times to implement surveys with students who were absent. Parent cons ent and student assent procedures were approved by the University of Minnesota Institutional Review Board for the Protection of Human Subj ects and the Chicago Public Schools Law Department. A Certificate 43

PAGE 44

of Confidentiality was obtained from the U.S. Department of Health and Human Services to further protect the confidentiali ty of the student responses. Se condary data analyses were approved by the University of Florida Institutional Review Board. Parents Parents of the student cohort were survey ed using a paper-penc il questionnaire at T1 (2002; Appendix B). The sample was obtained fr om parent address and telephone lists provided by the Chicago school district. During implemen tation of the school-based student survey, the parent questionnaires were distri buted to students, and they were asked to deliver the closed packet to their primary caregiver. The parent su rvey packet included a pa rent survey in English and Spanish; a cover letter consent form; a return addressed, stamped enve lope; and a participant payment form. Parents were given $25 after the co mpleted survey was returned. Students were given a $5 SubwayTM gift certificate for delivering the packet to their parents. After two weeks, teachers handed out a second copy of the pack et to students whose parents had not yet responded. Teachers were also asked to periodica lly remind students about the parent survey. A total of 3,250 parents we re surveyed at T1 (N = 4,643 eligible, 70% response rate). Community leaders A telephone survey of leaders in each of the study schools and neighborhoods was conducted at T1 (2002; Appendix C). Leaders included lo cal school council chairs and members, religious leaders, managers of park and recreation centers, nei ghborhood beat officers, neighborhood beat facilitators (citizen volunteers who work with beat officers), and managers/leaders of neighborhood organizations. Numerous school and city directories and phone books were used to identify the neighborh ood leaders within close proximity to each study school. The survey was administer ed by trained research staff us ing standardized protocols at 44

PAGE 45

the Telephone Survey Center at the Universi ty of Minnesota. Three hundred forty-four community leaders were surveyed (N = 491 targeted, 70% response rate). Alcohol outlets A listing of all alcohol outlets, and their addr esses, for the city of Chicago was obtained from the Chicago Licensing Department at T1 (2002). A count of the off-sale alcohol outlets (i.e., retailer where alcohol is sold for consumption off the li censed premises) per study unit was obtained by mapping each outlet address to a de tailed street map of PNC study units using ArcView GIS software (ESRI, 2005). Alcohol purchase attempts Propensity for underage access to alcohol from off-sale, commercial sources was tested directly using a standa rdized protocol (Appendix D; Perry et al., 1993; Wagenaar, Murray, Wolfson, Forster, & Finnegan, 1994; Wagenaar, Toomey, & Erickson, 2005). At T1 (2002), women who were at least 21 years old and who were judged by a panel to be younger appearing (e.g. 20 years old or younger) attempted to purchase alcoholic beverages at outlets without age identification. Buyers were matched to establ ishments based on their race/ethnicity and the dominant race/ethnicity of the community area in which an establishment was located to ensure that buyers would be similar in appearance to the typical client ele. Each buyer was accompanied by a driver and scout, a staff person who made obse rvations about the establishment exteriors. Buyers and scouts were trai ned by project staff to follo w a consistent protocol. The number of alcohol establishments pe r community ranged from 1 to 59 (mean = 13.38). Among communities with fewer than 20 off-sa le establishments, attempts were made at all establishments; among communities with mo re than 20 establishments, up to 20 were randomly selected for the purchase attempts. Two purchase attempts were conducted at each establishment, with attempts o ccurring, on average, 3 weeks apar t (N = 326 establishments, N = 45

PAGE 46

653 attempts). Buyers entered the establishmen t alone, with the scout remaining in the car parked out-of-sight of establishment employees. Buyers selected a si xor twelve-pack of Budweiser, Coors, or Miller Lite beer and went to the cash re gister with the shortest line. If clerks asked how old the buyers were, they answ ered honestly. If asked for age identification, they indicated that they did not have their age identific ation with them. If refused alcohol sale, buyers quietly left the establishment without argu ment. Buyers were inst ructed not to consume any of the purchased alcohol. Following each pu rchase attempt, buyers and scouts completed forms describing characteris tics of purchase attempts and establishments. Alcohol advertisements All alcohol advertisements w ithin 1,500 feet of PNC study schools were examined at T2 (2003; Appendix E). All advertisements were documented using digital cameras and the exact location of each print ad, or ad cluster, was documented to the nearest hundred-thousandth of a degree using a Global Positioning Systems Device (e.g. Garmin, eTrex, Venture). Other documented details included the location of the ad (i.e. bus stand/bench, billboard, liquor store, grocery or convenience store, bar, other), number of ads at the location, and a brief description of the ad. The protocol and data collection had b een piloted previously using two Chicago public schools as pilot sites. Detailed street maps with a 1,500 foot radius clearly demarcated around each study school site were created using ArcView GIS software (ESRI, 2005) Data collection staff took a wide angle photo of each GPS location where ads were found, and then a close-up of each individual sign/ad or ad cluster. Close up photographs were used to code the content of each ad. All billboards and bus stops were documented regardless of content. However, storefronts and restaurants/bars were only photographed if the ad s posted contained alcohol-related content. 46

PAGE 47

Census 2000 Seventeen Census 2000 (U.S. Census Bureau, 2000) measures were obtained for each PNC community which corresponded to the Chica go census tracts, includ ing: total population, percentage of the population with less than 9 ye ars and with 12 or more years of education, unemployment rate, occupational composition, median family income, income disparity, median home value, median gross rent, poverty threshold, single-parent household rate, percentage of households without a motor vehicle, telephon e and/or complete plumbing, and household crowding. Participants The sample for the present study include d 5,655 of 5,812 youth residing in the 42 PNC study communities who completed at least one study survey, after removing participants who had moved between intervention conditions during the study were excluded from analyses. These students were predominantly African American or Hispanic (43% and 29%, respectively), had an equal gender distribution (50% boys), spoke English in th eir homes (74%), and were low income (72% receiving free, or reduced price lu nch). Less than half of the students (47%) lived in two-parent households. The communities in which the participants resided showed considerable variability across race/ethnicity a nd measures of deprivation (Tables 2-5 and 2-6, respectively). Measures The present study analyzed baseline (T1, 2002) measures from the parent and community leader surveys, direct assessments of th e alcohol-specific neighborhood environment and secondary data from community archives. Data from all four (T1-T4, 2002-2005) PNC student surveys were analyzed as well, providing data on parental home and family management and alcohol use. 47

PAGE 48

Neighborhood Context The 42 study communities were characterized by nine baseline measures of risk and protective factors from the parent and commun ity leader surveys, Census 2000, and direct assessments of the alcohol-specific neighbor hood environment. Table 2-7 presents the descriptive statistics for each of the measures. Protective factors Neighborhood strength Five items from the community leader survey were used to create a scale of neighborhood st rength: How would you rate the neighborhood in terms of having a strong community identity?; leve l of community resources?; participation level of residents in lo cal activities?; level of influen ce local residents or community groups have on decisions about local polic ies?; and efforts of residents in addressing the prevention of alcohol use among teenagers? (Cronbachs al pha: 0.70, range 5-25). Response options were 1 = low, 3 = medium, and 5 = high, with a higher score on this sc ale indicating greater neighborhood strength. Neighborhood and police preventive action A scale assessing neighborhood and police preventive action was created usi ng nine items from the community leader survey : How would you rate police involvement in the preventi on of alcohol use among teenagers in the neighborhood?; How would you characterize rela tionships between your local beat officers and the neighborhood residents surro unding the schools?; If teenag ers were hanging out on the block, how likely is it that residents in the neighborhood would do something about it?; If a store was selling alcohol to teen agers, how likely is it that residents in the neighborhood would call the police?; If the police were called on a loud party invol ving young people, how likely is it that they would check to see if there was underage drin king?; How likely is it that a group from the neighborhood would work to reduce the amount of alcohol adve rtisements?; How 48

PAGE 49

likely is it that if a business served or sold al cohol to minors, the business would be cited by an enforcement agency?; How likely is it that if an adult provide d alcohol to minors, the adult would be cited or ticketed by police?; and How likely is it that a minor who was in possession of alcohol would be cited or ticketed by police? (C ronbachs alpha: 0.89, range 9-45). Response options were in the form of a 5-option Likert scale ranging from very little involvement/not at all good/not at all likely to a great deal of involvement/very good/ very likely. A higher score on this scale indicated more neighborhood and policy preventive action. Organizational preventive efforts Eight items, with yes/no responses (1 = yes, 5 = no), from the community leader survey were used to create a scale assessing organizational preventive efforts: Has your organization work ed to promote alcohol-free activities for youth?; increase or promot e police enforcement against underage drinking?; reduce public drunkenness?; promot e participation in a neighborhood watch or block club?; restrict alcohol advertisements such as on billboards or storefront s?; reduce the number of businesses that sell or serve alcohol to underage youth?; prom ote participation in an effort to establish a dry precinct?; and change a policy in your organization related to alcohol use? (Cronbachs alpha: 0.79, range 8-40). A hi gher score on this scale indicated more organizational preventive efforts. Community action A community action scale was crea ted using four items from the parent survey that included 5-option Likert scale respons es ranging from Would not do something about it to Definitely would do something about it: If teenagers were hanging out on your block drinking alcohol, how likely is it that you or some of your neighbors would do something about it?; If a store on your block wa s selling alcohol to teen agers, how likely is it that you or some of your neighbor s would call the police?; If there was a loud party involving 49

PAGE 50

young people going on in a house on your block, how likely is it that you or some of your neighbors would do something about it?; and If there was a liquor st ore that had alcohol advertisements visible from outside the store, ho w likely is it that you or some of your neighbors would try to reduce the amount of alcohol adve rtisements (ads)? (Cronbachs alpha: 0.76, range 4-20). A higher score on th is scale indicated greater parental community action. Risk factors Perceived neighborhood problems A scale assessing perceived neighborhood problems was created using seven items from the parent surv ey: Below is a list of urban problems. Please check how much of a problem each of the following is on the block where you live: drug dealing?; unsupervised youth?; people drinking alcohol on the street?; too many stores that sell alcohol?; lack of superv ised activities for youth? ; too many alcohol advertisements (ads)?; and poor police re sponse? (Cronbachs alpha: 0.93, range 7-35). Response options were 1= not a problem, 3 = a minor problem, and 5 = a serious problem. A higher score on this scale indicated greater perceived neighborhood problems. Alcohol advertisements A count of the number of alc ohol advertisements within 1500 feet of each PNC study school was obtained dire ctly in 2003. A measure of the average number of alcohol advertisements around schools within each community area was created by dividing the total number of alcohol advertisements surrounding schools within each community area by the total number of PNC study sc hools in each community area. Off-sale alcohol outlet density The mean number of off-sa le alcohol outlets per 1,000 population per community area was obtained by divi ding the mean number of off-sale alcohol outlets per community area obtained from the Chicago Licensing Department by the total population for each community area, obtained from Census 2000. 50

PAGE 51

Alcohol purchase attempt rate A measure of the commercial accessibility of alcohol to underage youth was obtained by dividing the number of successful purchase attempts by the total number of attempts for each study community area. Area deprivation An area deprivation index was created for each study community following procedures described by Singh (2003). Seventeen Census 2000 indicators for each study community were used and included: educat ional distribution (perce ntage of the population with less than 9 years and with 12 or more years of education) unemployment rate, occupational composition, median family income, income disp arity, median home value, median gross rent, median monthly mortgage, home ownership rate, family poverty rate, population below 150% of the poverty threshold, single-parent household ra te, percentage of house holds without a motor vehicle, telephone, and/or complete plumbing, a nd household crowding. Factor score coefficients from Singh (2003) were used to weight the 17 indicators comprising the index. The scale was then standardized by setting the mean and st andard deviation to 100 and 20, respectively (Cronbachs alpha: 0.87; range 45.6-152.6). Ultimate Outcomes The main outcomes of interest for the presen t study were student al cohol use and alcohol use intentions. Items assessing these outcomes we re drawn from the Monitoring the Future study (Johnston, OMalley, Bachman, & Schulenberg, 2005) and have been used extensively in previous intervention studies (Komro et al., 2004; Komro et al., 2001; Perry et al., 2003; Perry et al., 2002; Perry et al., 1996; Wagenaar, Zobec k, Williams, & Hingson, 1995). Table 2-8 presents the descriptive statistics for these m easures across all study time-points. Alcohol use Alcohol use was assessed longitudinally with five items: During the last 12 months, on how many occasions, or times, have you had alcoholic beverages to drink?; During the last 30 51

PAGE 52

days, on how many occasions, or times, have you had alcoholic beverages to drink?; During the last 7 days, on how many occasions, or times, have you had alcoholic beverages to drink?; Think back over the last two weeks, how many times have you had five or more alcoholic drinks in a row?; and Have you ever gotten really drunk from drinking alcoholic beverages, so you fell down or got sick?. Response options for th e past year, past month and past week items included: occasions, -2 occasions, -5 occasions, -9 occasions, -19 occasions, -39 occasions, and or more occasions. Response options for the heavy episodic use and having ever been drunk items included: never, once, twice, three to five times, six to nine times, and ten or more times. Alcohol use intentions Alcohol use intentions were assessed longitudinally with three items: Would you drink alcohol if you best friend offered it to you?; D o you think you will be drinking alcohol. .in the next month?; and when a senior in high sc hool?. Response options included yes, not sure, and no. Tendency to use alcohol For Paper #2 (Chapter 4), the alcohol use a nd intentions items were combined into one, nine-item scale describing the tendency to us e alcohol. The scale ranged from 9 to 45, with a higher score on this scale indica ting greater alcohol use/inten tions (Cronbachs alpha: 0.85-0.88). Hypothesized Mediators Home alcohol access, parental monito ring/communication, and alcohol-specific communication were hypothesized mediators of in terest. Data from the T3, 2004 student survey were used in order to establish clear tem poral precedence for observed relationships. The descriptive statistics for these data are presented in Table 2-9. 52

PAGE 53

Home alcohol access Three items from the student survey asse ssed the accessibility of alcohol from their homes and parents. Two items measured the ease with which students coul d obtain alcohol from their parents and homes: How hard would it be for you to obtain alcohol from your parent or guardian? and How hard would it be for you to take it from your home?. Response options included hard, in-between, and easy. One item required students to identify the sources of their last alcoholic beverage: If you have ever ha d an alcoholic drink, think back to the last time you drank. How did you obtain the al cohol?. Your parent or guard ian gave it to you and You took it from home were the two resp onse options included in this study. Parental monitoring and communication Students responded to five items assessing th eir parental monitoring and communication: How often do/does you/your pa rent or guardian ask you about what you are doing in school?; praise you when you do a good job?; eat dinner with a parent or guardian?; ask you where you are going or who you will be with?; and have a conversation with you that lasts 10 minutes or more?. Response op tions included: never, hardly ever, sometimes, a lot, and all the time. Alcohol-specific communication Four items from the student survey assessed alcohol-specific communication: How often does your parent or guardian talk with you a bout problems drinki ng alcohol can cause young people?; family rules against young pe ople drinking alcohol?; what would happen if you were caught drinking alcohol?; and how ads and commercials are used to get you to buy things?. Response options included: never, hardly ever, sometimes, a lot, and all the time. 53

PAGE 54

Hypothesized Moderator A baseline measure of family composition was se lected for inclusion as a covariate in the analyses for the second investigation (Paper #2, Chapter 4), as it is th eoretically related to alcohol use among youth (Bergmark & Andersson, 1999; Bjarnason et al ., 2003; Foxcroft & Lowe, 1991; Miller, 1997) and had a disproporti onate distribution among the neighborhood risk classes identified in Paper #1 (Chapter 3). Fa mily composition was dichotomized such that mother and father together (53%) was compared to other (mother and father equally, at separate homes, mother mostly, father mos tly, grandparent, other relative, foster parents, other). Analytical Approaches Fitzmaurice, Laird and Ware (2004) describe m ultilevel data as that for which there is a hierarchical or clustered structure. Namely, units of observation are nested within larger units of assignment or organization or within individual s over time. Such observations are correlated, a violation of assumptions for standa rd statistical methods. A nave an alysis that fails to consider such dependence of observations introduces a downward bias into the standard errors of the estimates, inflating the Type I error rate beyond the nominal level (Twisk, 2006). First developed in the context of educational research (Gol dstein, 1987; Goldstein & Cuttance, 1988; Nuttall, Goldstein, Prosser, & Rasbash, 1989), multilevel analysis addresses this shortcoming by considering the dependency of observations, thus improving the precision of estimates (Twisk, 2006). As a group-randomized trial, data from PN C is multilevel in nature. The unit of assignment to a treatment condition was a group in stead of an individua l; however, the unit of observation was individuals within the groups. Thus the nesting of the un it of observation in the unit of randomization resulted in a multilevel data structure (Murray, 1998). In the present study, 54

PAGE 55

community served as the grouping unit. Such a specification more approp riately reflects the relationships examined here, as the primary intere st is the effect of co mmunity-level, contextual variables on individual behavior. Therefore, all analyses considered the dependency of observations among students within thei r respective neighborhood, or community. Three primary statistical methods were used to address the proposed research questions (1) multilevel latent class analysis, (2) general linear mixed effects regression, and (3) multilevel structural equation modeling. The analyses utiliz ed personand variable -centered approaches, where the variable-centered a pproach assumes the data come from a homogeneous population and describes the average behavior of the sa mple and a person-centered approach allows identification of heterogeneous s ubgroups within the sample and examination of effects by these subgroups (Muthn & Muthn, 2000). The general la tent variable framework in Mplus (Muthn & Muthn, 2004) was used for Papers 1 and 3 (Cha pter 3 and 5, respectively), which allows for models containing continuous and/or categorical latent variables. Additionally, this framework has the capacity for multilevel modeling, where individual-level (withi n) and cluster-level (between) variation can be esti mated (Muthn & Muthn, 2004). An alyses for Paper 2 (Chapter 4) were conducted using the general linear regression framew ork in SAS version 9.1 (SAS Institute, 2004). Multilevel Latent Class Analysis Multilevel latent class analysis was used in Paper 1 (Chapter 3) to identify the heterogeneity in social capital and exposure and access to alcohol among the communities in which the sample resided. Latent class analysis (L CA) is a person-centered analysis strategy that determines the smallest number of latent cl asses describing the association among a set of observed categorical variables (Muthn & Muth n, 2000). This was an optimal approach for describing the different contex ts in which these youth were embedded, as it acknowledges that 55

PAGE 56

there may be disproportionate risk across ne ighborhoods within urban areas and uses item response probabilities among individuals in th e sample to identify underlying heterogeneity, rather than covaria tion among variables. Measurement model The estimated proportion of unique response pa tterns is expressed as a function of two parameters (1) latent class probabilities ( ), which describe the pr oportion of individuals expected in each latent class and (2) conditional item-response probabilities ( ), which describe the probability of a particular response to an obs erved, manifest variable, conditional on latent class membership. The probability of a particular response pattern is given by, where cnkcjci C c c nkjiuuuP|,|2,|1, 1 ,2,1,... ),...,(c represents the probabili ty of being in latent class c and ,and are the probabilities of responses i, j and k to items 1 through n, respectively, conditional on me mbership in latent class c (Lanza, Flaherty, & Collins, 2003). Maximum likelihood estimates of these parameters are usually estimated using the EM algorithm (Dempster & Rubin, 1977; Goodman, 1974), which iteratively converges on the maximum of the likelihood function through a lternating between the E-step, or expectation step, and the M-step, or maximization step (Lanza et al ., 2003; Muthn & Muthn, 2004). ci |1,cj |2,cnk |,Parameters estimated in LCA are similar to f actor loadings in factor analysis, whereby the parameters describe the rela tionship between the manifest vari ables and the latent variable. A probability near 0 or 1 reflects a strong relationship between the variable and the latent construct, such that given a latent class, there is a lo w or high probability predicting how an individual would respond to an item, respect ively. Alternatively, a probabil ity near 0.5 for a dichotomous indicator suggests that given a latent class, ther e is no greater ability th an chance to predict how an individual would respond to an item (Lanza et al., 2003). 56

PAGE 57

Multilevel LCA is a fairly recent extension of traditional LCA that allows the measurement model to be fit while accounti ng for non-independence of observations due to cluster sampling (Asparouhov & Muthn, 2006; Muthn & Muthn, 2004). The present study specified students as being nested within comm unities. As such, the item response probabilities for each individual student were modeled at le vel one and the item res ponse probabilities for each community modeled at level two, along with the variation in parameters between individuals within communities. There are two assumptions to any latent class mo del. First, all individuals in a latent class are assumed to have the same conditional response probabilities for the items. Second, there is conditional independence given the latent class, namely the indi cators are independent of one another (Lanza et al., 2003). Model Selection Class enumeration proceeded through two step s. First, competing models (i.e., models with 1through n-class solutions) were estimated and compared to determine the number of classes necessary to represent th e heterogeneity in responses to the manifest variables among the sample. Selection of the appropr iate number of classes was ai ded by model selection indices, such as the Bayesian Information Criterion (B IC; Raftery, 1986) and Akaike Information Criterion (AIC; Akaike, 1973), which provide a re lative measure of model fit. They are calculated based on the li kelihood ratio statistic ( yobs obs G exp log22), the number of parameters estimated, and, in the case of BIC, sample size. Simulation studies have suggested that when comparing model selection indices, th e BIC is superior when determining the number of classes (Collins, Fidler, Wugalter, & Long, 1993; Hagenaars & McCutcheon, 2002; Lanza et 57

PAGE 58

al., 2003; Nylund, Asparouhov, & Muthn, 2007); t hus, the solution with the lowest BIC is typically preferred. An alternative index, the Lo, Mendell, Rubi n likelihood ratio test (LMR; Lo, Mendell, & Rubin, 2001), may also be helpful during cla ss enumeration. The LMR test allows for comparison of nested latent class mo dels, where the fit of a model with k classes is compared to its neighboring model with k -1 classes. Its estimate is two ti mes the loglikelihood difference of the competing models and yields a p -value that can be used to dete rmine if there is a statistically significant improvement in fit with the inclusion of an additional class. While its application has been limited, it may be a helpful, secondary tool for determining the most appropriate solution (Nylund, Asparouhov, & Muthn, 2007). Unfortunately, such tools for model selection do not always agree on the best-fitting model (Kuha, 2004). Thus, the inte rpretation of the classes must be considered throughout the model selection process and be founded upon substa ntive theory (Lanza et al., 2003; Muthn et al., 2002), here the Theory of Triadic Influe nce (Flay & Petraitis, 1994) and Wagenaar and Perrys Model of Drinking Behavi or (Wagenaar & Perry, 1994). Second, the conditional item-response probabilities ( ) were examined to ascertain the meaning of each class. The strength of the parameters lends to assignment of substantive labels for each latent class. The class membership probabilities ( ) allow the comparison of the prevalence of each neighborhood class. For a well identified model, the parameters suggest distinct interpretable labels for each class (Lanza et al., 2003). For the present study, the parameters indicated the proba bility of being above the mean of a neighborhood risk or protective item conditional on ne ighborhood class membership. For instance, LCA estimates the probability of being above the mean of a ne ighborhood risk item given membership in the 58

PAGE 59

highest risk class. This probability is expected to be high when compared to the probability of being above the mean of a neighbor hood risk item given membership in a lower risk class. General Linear Mixed Effects Regression General linear mixed effect regression was us ed in Paper 2 (Chapter 4) to examine the predictive utility of the latent alcohol-related neighborhood risk classes identified in Paper 1 (Chapter 3) on the traject ories of alcohol use and intentions. Th is analysis integrated personand variable-centered approa ches, allowing the person-centered, latent, alcohol-related neighborhood risk class membership to be incorporated into a longitudinal, multilevel regression model, where the mean of the outcome variable is modeled as a function of one or more independent variables with hierarchical or clustered data (Fitzmaurice et al., 2004). This type of mixed model incorporates fixed and random effects, where fi xed effects represent the mean response modeled as a combination of characteristics that are a ssumed to be homogeneous across individuals and random effects represent effects that are unique to the individual and represent the variability around the mean response (Fitzmaurice et al., 2004) For longitudinal data, incorporating random effects allows the covariation among repeated m easures to be expressed as a function of time. Here, the mixed model approach allowed for es timation of the variabil ity between communities and the variability within individuals over time. Ignoring the positiv e correlations among responses within these units, or levels, may poten tially introduce bias into the estimated effects and increase the likelihood of Type I error (Murray, 1998). Using mixed effects regression to fit a growth model involves several steps, as outlined by Muller and Fetterman (Muller & Fetterman, 2002). First, it is necessary to determine the appropriate covariance structure to account for repeated indi vidual observations over time. Previous work with these data (K omro et al., 2007) indicated that an unstructured error structure is most appropriate. This covariance structure is the least restrictive and assumes that all 59

PAGE 60

correlations between observations are different within subjec ts (Twisk, 2006). Second, the maximum model considered is specified and co-lin earity between predictors evaluated. Next, the unconditional model for the outcome variable is examined. This model describes the underlying, unadjusted trajectory of the outcome. Lastly, bi variate relationships be tween predictors and outcome variables are examined and significant predictors ( p < .10) are retained in the final model. Raudenbush and Bryk (Raudenbush & Bryk, 2002) describe three advantages to mixed effects regression over traditional linear regression (1) improved estimation of individual effects, (2) ability to model cross-level effects, and (3) ability to pa rtition variance and covariance components. In addition, using mixed effects regr ession to estimate a growth curve model allows inclusion of all data for a participant, even if there is missing data at one or more time point. These advantages considered, mixed effects regr ession was an appropriate method for examining how alcohol-related neighborhood context predicted the growth in alcoho l use and intentions over time. Multilevel Structural Equation Modeling Paper 3 (Chapter 5) used multilevel structural equation modeling to examine the direct and indirect relationships be tween alcohol-related neighborhood context, home and family management, and alcohol use. This approach was particularly advantageous for examining these relations, as it analyzes relations between late nt variables without ra ndom error (Bollen, 1989) and, in Mplus, allows direct and indirect e ffects to be estimated simultaneously (Muthn & Muthn, 2004). Structural equation modeling (S EM) is a technique used to test causal models where the independent and dependent variab les are latent (Vogt, 2005). Each SEM consists of two parts (1) a measurement model, where observed indicators are related to latent constructs (e.g., 60

PAGE 61

confirmatory factor analysis); and (2) a stru ctural model, which specifies hypothesized causal relationships among the latent variables through a series of linear, logist ic or probit regression equations, for continuous, categ orical and ordinal factors, respectively (Muthn & Muthn, 2004). Analyses generally proceed through a sequen ce of five processes (1) model specification, (2) model identification, (3) model estimation, (4) model testing, and (5) model modification (Schumacker & Lomax, 2004). First, model specification involves determining all variables, relations and parameters of interest. As SEM is a method for model-testi ng, rather than model development, these determinations should be founded upon sound scientific theory and the extant scientific literature (Klem, 2000). Failure to identify important constructs, or inclusion of extraneous constructs, contributes to specification erro r, whereby the model being tested is not consistent with the true population model (Schumacker & Lomax, 2004). Such a model is said to be misspecified, and produces estimates that are systematically bias ed from their true valu es. Thus, consideration of all potential variables related to the model is th e critical, and often hardest, first step to SEM. Second, model identification involves dete rmining whether a unique set of parameter estimates can be found given the sample covari ance matrix and the popula tion covariance matrix implied by the tested model (Schumacker & Lomax, 2004). Schumacker (2004) describes three levels of identification, where a model is (1) underidentified, such that there is not enough information in the covariance matrix to uniquely determine one or more parameters, (2) just identified, where there is just enough information in the covarian ce matrix for all parameters to be uniquely determined, or (3) overidentified, such that there is more than one way to estimate the parameters because there is more than e nough information in the covariance matrix. Three methods can be used to address these identifi cation issues (1) parame ter constraints may be 61

PAGE 62

imposed on the measurement model; either one indicat or for each latent variable can be fixed to 1, or the variance of each latent variable must be fixed to 1; (2) the parameters can be estimated using maximum likelihood estimation, rather than ordinary least square s; and (3) estimation could begin with the most parsimonious model, with additional parameters included only after the simplest model is identified. Once the first two processes have been satisfied, the implied theoretical model is then estimated. Several methods for estimating the parameters in the model ar e available, including unweighted, weighted or ordina ry least squares, generalized least squares, and maximum likelihood (Schumacker & Lomax, 2004). Maximum likelihood (ML) estimation is the method most frequently used in SEM, and has been found to be robust to violation of normality (Klem, 2000); however it is not feasible with more complex models involving more than 3 or 4 latent variables, as it requires integration across ma ny different matrices (Muthn & Muthn, 2004). The present study utilized the minimum variance weighted least squares estimator, which uses pairwise deletion to handle missing data among categorical and/or ordinal variables (Muthn & Muthn, 2004). The fourth process in SEM is model testing. Once parameter estimates have been obtained, it is necessary to determine how well the data fit the model (Schumacker & Lomax, 2004). Several indices may be used for this asse ssment (1) comparative fit index (CFI), (2) Tucker-Lewis fit index (TLI), (3) root mean square error of approxi mation (RMSEA) and (4) standardized root mean square residual (SRMSR ). The CFI and TLI describe the improvement in fit of the tested model compared with that of a null model assuming zero covariance among the variables (Kline, 2005). A value greater than 0.90 indicates reasonably good model fit (Hu & Bentler, 1999). The RMSEA is a parsimony-adjusted index, where a value < 0.05 indicates close 62

PAGE 63

approximate fit, values between 0.05 and 0.08 suggest reasonable fit, and values > 0.10 suggest poor model fit (Kline, 2005). The last index, the SRMSR, is a measure of the mean residual correlation, where values < 0.10 are considered adequate (Kline, 2005). Further, parameter values should make sense, being in the expected direction and within the range of plausible values (Schumacker & Lomax, 2004). Lastly, model modification ensues when the strength of the theoretical model is questioned. Several procedures can be used to further examine model specification, including (1) consideration of the statistical significance of estimated parametersnon-significant parameters could be removed from the model if deemed theo retically irrelevant, or constrained to zero in subsequent models; (2) examination of fitted residualsthe differences in the observed covariance matrix and the model covariance matrix should be small and not vary in magnitude from one variable to another; (3) examinati on of the squared multiple correlations for each observed variable; and (4) use of expected parameter change, Lagrange multiplier, and Wald statistics to evaluate the effect of freeing parameters (Schumacker & Lomax, 2004). Recent methodological developments have extended SEM to accommodate multilevel data. In the context of the present study, the SE M consisted of three measurement models and the structural model relating them. The measuremen t model components included (1) an exploratory factor analysis (EFA) to determine the appropr iate factor structure for the alcohol-related neighborhood context items; (2) an EFA of the home alcohol access, parental monitoring/communication, and alcohol-specific communication items; and (3) a CFA to verify the factor structure of the alc ohol use items. Community membersh ip was specified as a nested random effect to account for the dependenc y of observations among youth within each community. 63

PAGE 64

Garson (2007) describes several advantages of SEM relative to multiple regression methods, including more flexible assumptions, a reduction in measurem ent error related to having multiple indicators per latent variable, and the ability to (1) test models overall rather than coefficients individually, (2) test models with multiple dependents, (3) model mediating variables, (4) model error terms, (5) test co efficients across multiple between-subjects groups, and (6) handle difficult data. Given the complexity of the data and relationships investigated in the present study, and the recent methodological a dvancements in Mplus and other statistical software, multilevel structural equati on modeling was a valid technique. Missing Data Seventy-two percent of the cohort students co mpleted three or four surveys, while 28% completed one or two. Students who completed th ree to four surveys were more likely to be White ( 2 (5) = 107.417, p<0.001) and live with both parents ( 2 (1) =37.887, p<0.001), compared to those who only completed one or two surveys. There were no significant differences in alcohol use between those who completed three or four surveys and those completing one or two. To handle missing data, the respective an alyses used either maximum likelihood estimation or pairwise deletion. The multilevel LC A (Paper 1, Chapter 3) in Mplus and general linear mixed effects regression models in SAS (Pap er 2, Chapter 4) used ML estimation, where estimated parameters represent the parameter values for which the probability of the observed data takes its maximum (Agresti, 2007; Muthn & Muthn, 2004). ML estimation is one of two recommended strategies for ha ndling missing data that provid es robust parameter estimates (Schafer & Graham, 2002). The mu ltilevel SEM (Paper 3, Chapter 5) used pairwise deletion to handle missing data (Muthn & Muthn, 2004). Estimates are based on the polychoric correlations for all pairwise present data, wher e only missing values on the two variables under 64

PAGE 65

consideration are ignored, not the entire case. ML estimation is preferable to pairwise deletion, which may introduce bias into the estimates of effect and produce nonpositi ve definite matrices. However, these threats are reduced among larger samples (Marsh, 1998) and is the best available option when estimating complex models with more than 3 or 4 latent variables, as was the case here. 65

PAGE 66

Table 2-1. Summary of re search questions, study desi gn and analytic approaches. Research question(s) Study design Analytical approach How many latent classes are necessary to describe the heterogeneity in neighborhood risk among young adolescent residents of urban, multi-ethnic communities? Cross-sectional Multilevel latent class analysis Paper #1 What are the prevalence and characteristics of the heterogeneous latent neighborhood risk classes? Cross-sectional Multilevel latent class analysis Paper #2 How does neighborhood context influence the trajectories of alcohol use and intentions during early adolescence? Longitudinal Mixed Effects Regression How does neighborhood context influence home and family management practices (e.g., home alcohol access, parental monitoring, parent/child communication, alcohol-specific communication)? Longitudinal Multilevel structural equation modeling Paper #3 Do home and family management practices mediate the effects of neighborhood risk on alcohol use during early adolescence? Longitudinal Multilevel structural equation modeling 66

PAGE 67

67 Table 2-2. Comparison of 66 study schools w ith average for Chi cago Public Schools. Intervention Control Chicago Public Schools Frequency/Percent # Units 10 12 NA # Schools 28 32 330 # Communities 20 22 77 # Students 1,755 2,285 420,322 % White 13.9 11.4 9.1 % African American 46.8 40.4 49.7 % Hispanic 21.8 34.0 37.6 % Mixed/Other 17.5 14.1 12.7 % Low income 65.7 72.8 85.2 % Truant 2.1 1.7 3.6 % Math at/above norms 48.3 48.8 43.0 % Reading at/above norms 42.3 42.5 42.9

PAGE 68

Table 2-3. Data used from Project Northland Chicago. Construct Data source # of items T1 T2 T3 T4 NEIGHBORHOOD Deprivation Census 2000 17 X Number of alcohol outlets per community area Chicago licensing department 1 X Commercial alcohol acc ess Alcohol purchase attempts 1 X Alcohol ads with 1500 feet of each school Observation 1 X Neighborhood strength Lead er survey 5 X Neighborhood and police prevention ac tion Leader survey 9 X Neighborhood problems Parent survey 7 X HOME & FAMILY Home alcohol access Student survey 4 X Parental monitoring Student survey 1 X Parent/child communication Student survey 3 X Alcohol-specific communication Student survey 4 X INDIVIDUAL Alcohol use Student survey 5 X X X X Alcohol use intentions Student survey 4 X X X X 68

PAGE 69

Table 2-4. Characteristics of students who completed 1 or more PNC surveys. Boys Girls Total All 2942 2868 5812 Race Percent Percent Percent Asian (N = 305) 5.3 5.2 5.3 African American (N = 2506) 41.6 44.7 43.1 Hispanic (N = 1657) 28.8 28.2 28.5 Native American (N = 58) 1.3 0.7 1.0 White (N = 730) 13.4 11.7 12.6 Mixed/Other (N = 552) 9.6 9.4 9.5 Family composition Living with both parents (N = 2719) 49.2 44.6 46.9 Living with both parent s, but at separate homes/Parent & Step-parent (N = 642) 11.1 11.0 11.1 Living with mother mostly (N = 1792) 28.1 33.8 30.9 Living with father mostly (N = 157) 3.1 2.2 2.7 Living with grandparent or other relative (N = 422) 7.2 7.3 7.3 Other (N = 67) 1.2 1.1 1.2 Free or reduced price lunch Yes (N = 4178) 68.5 75.5 72.0 No (N = 1628) 31.5 24.5 28.0 Language at home English (N = 4267) 73.2 73.9 73.6 Spanish (N = 1070) 18.5 18.4 18.4 Other (N = 464) 8.3 7.7 8.0 Note: Numbers may not add up to 5812 due to missing values. 69

PAGE 70

70 Table 2-5. Racial/Ethnic distribution (% ) among PNC study communities from Census 2000. Race/Ethnicity Community N Asian African American Hispanic Native American White Other Armour Square 66 19.70 33.33 28.79 0.00 9.09 9.09 Ashburn 209 0.96 65.55 19.14 2.39 3.83 8.13 Austin 193 0.00 86.01 5.18 4.15 0.00 4.66 Avalon Park 79 0.00 92.41 0.00 1.27 0.00 6.33 Beverly 115 0.00 44.35 1.74 2.61 33.04 18.26 Bridgeport 133 60.15 2.26 18.80 0.00 9.02 9.77 Clearing 88 1.16 2.33 39.53 6.98 41.86 8.14 Douglas 77 2.60 89.61 0.00 1.30 0.00 6.49 East Garfield Park 46 0.00 89.13 2.17 4.35 2.17 2.17 East Side 64 4.69 0.00 76.56 0.00 12.50 6.25 Englewood 72 2.78 90.28 2.78 1.39 0.00 2.78 Fuller Park 36 0.00 91.67 5.56 2.78 0.00 0.00 Gage Park 86 0.00 0.00 94.19 0.00 3.49 2.33 Garfield Ridge 63 1.61 6.45 30.65 1.61 50.00 9.68 Grand Boulevard 48 0.00 93.75 0.00 2.08 2.08 2.08 Hegewisch 29 0.00 24.14 37.93 0.00 17.24 20.69 Humboldt Park 89 0.00 7.87 83.15 1.12 1.12 6.74 Hyde Park 48 0.00 83.33 4.17 8.33 0.00 4.17 Jefferson Park 38 7.89 13.16 23.68 0.00 50.00 5.26 Kenwood 116 0.86 90.52 0.00 1.72 0.00 6.90 Lake View 118 0.85 13.56 63.56 2.54 14.41 5.08 Lincoln Park 38 2.63 21.05 63.16 0.00 2.63 10.53 Logan Square 139 0.00 7.91 82.73 2.88 2.88 3.60 Lower West Side 83 0.00 0.00 85.54 6.02 3.61 4.82 Morgan Park 72 0.00 45.83 0.00 1.39 37.50 15.28 Mount Greenwood 59 0.00 15.79 5.26 1.75 66.67 10.53 Near North Side 70 0.00 91.30 2.90 0.00 0.00 5.80 Near West Side 204 0.01 77.45 10.29 2.94 0.00 7.35 New City 89 0.00 29.21 21.35 4.49 32.58 12.36 North Center 73 8.22 9.59 31.51 0.00 35.62 15.07 North Lawndale 181 0.56 92.22 0.56 2.78 0.00 3.89 North Park 111 36.94 1.80 27.93 0.90 22.52 9.91 Norwood Park 39 0.00 5.26 15.79 0.00 71.05 7.89 Portage Park 280 2.51 0.72 45.52 1.43 42.29 7.53 Roseland 75 0.00 85.33 0.00 1.33 0.00 13.33 South Lawndale 199 0.50 0.50 95.98 0.50 1.01 1.51 Washington Heights 38 0.00 94.74 0.00 2.63 0.00 2.63 West Elsdon 142 0.70 0.00 84.51 1.41 9.15 4.23 West Garfield Park 181 0.55 90.61 1.10 3.31 0.55 3.87 West Pullman 36 2.78 72.22 2.78 5.56 0.00 16.67 West Ridge 212 24.17 6.16 26.54 0.95 34.12 8.06 Woodlawn 125 0.00 85.60 1.60 2.40 0.00 10.40

PAGE 71

Table 2-6. Descriptive statistics for m easures of deprivation from Census 2000. Measure Mean Standard deviation Minimum Maximum Population aged > 25 y with < 9 y of education, % 0.58 0.56 0.01 2.50 Population aged > 25 y with a high school diploma, % 24.18 8.21 5.71 37.88 Employed persons aged > 16 y in white-collar occupations, % 3.79 4.12 0.23 20.90 Median family income, $ 47,232.46 21,619.11 14,476.40 133,832.80 Income disparity (Aggregate) 1,104,490.08 1,137,047.25 140,358.82 7,349,900.00 Median home value, $ 5,163,971.09 4,132,734.99 165,600.00 16,903,700.00 Median monthly rent, $ 580.66 145.47 224.00 953.30 Median monthly mortgage, $ 1228.35 441.84 773.00 3005.80 Owner-occupied housing units, % 4.94 2.93 0.26 11.85 Civilian labor force population aged > 16 y unemployed, % 12.51 8.10 2.93 33.53 Families below poverty level, % 2.46 2.82 0.03 11.74 Population below 150% of the poverty threshol d, % 31.86 18.04 6.11 69.25 Single-parent households with children aged < 18 y, % 18.29 13.24 2.68 52.13 Households without a motor vehicle, % 3.87 3.79 0.07 18.05 Households without a telephone, % 0.98 1.25 0.00 5.26 Occupied housing units without complete plumbing, % 0.31 0.39 0.00 1.56 Households with more than 1 person per room, % 1.34 1.46 0.03 6.42 71

PAGE 72

Table 2-7. Descriptive statistics for neighborhood context measures, 2002. Variable Mean SD Min Max Protective factors Neighborhood strengtha 16.45 2.17 5.00 25.00 Neighborhood & police preventive actionb 31.70 4.14 9.00 45.00 Organizational preventive effortsc 23.89 4.66 8.00 40.00 Community actiond 15.32 3.55 4.00 20.00 Risk factors Perceived neighborhood problemse 20.01 9.93 7.00 35.00 Mean number of alcohol ads per school per community 17.11 21.61 0.00 74.00 Mean number of off-sale alcohol outlets per 1, 000 population 0.22 0.16 0.07 0.70 Alcohol purchase attempt success rate 0.35 0.20 0.00 0.72 Area deprivation indexf 95.58 18.01 45.60 152.60aA higher score on this scale indi cates more neighborhood strength. bA higher score on this scale indicates more neighborhood and police preventive action. cA higher score on this scale indicates more organizational preventive efforts. dA higher score on this scale indicates more community action. eA higher score on this scale indi cates more neighborhood problems. fA higher score on this scale indi cates more area deprivation. 72

PAGE 73

Table 2-8. Frequencies of alcohol use and intentions items T1 T4 (2002-2005) Frequency (%) T1 (2002) N = 4,259 T2 (2003) N = 4,240 T3 (2004) N = 3,778 T4 (2005) N = 3,804 Alcohol use Past year 0 occasions 81.72 76.16 66.19 58.32 1-2 occasions 12.51 15.76 19.40 21.00 3-5 occasions 3.65 4.80 8.08 10.04 6-9 occasions 0.94 1.80 3.11 4.79 10-19 occasions 0.68 0.87 1.65 3.11 20-39 occasions 0.16 0.28 0.61 1.42 40 or more occasions 0.33 0.33 0.96 1.32 Past month 0 occasions 92.23 89.47 83.32 77.88 1-2 occasions 5.93 7.83 11.58 14.50 3-5 occasions 1.08 1.66 2.92 4.67 6-9 occasions 0.35 0.47 0.96 1.32 10-19 occasions 0.19 0.31 0.40 0.90 20-39 occasions 0.05 0.05 0.24 0.32 40 or more occasions 0.16 0.21 0.58 0.42 Past week 0 occasions 96.26 94.57 92.33 88.10 1-2 occasions 2.80 4.23 5.76 8.98 3-5 occasions 0.56 0.66 0.96 1.69 6-9 occasions 0.16 0.14 0.27 0.55 10-19 occasions 0.07 0.14 0.35 0.21 20-39 occasions 0.02 0.07 0.08 0.05 40 or more occasions 0.12 0.19 0.27 0.42 Ever drunk Never 94.47 91.53 88.66 82.14 Once 4.21 5.89 6.56 9.95 Twice 0.80 1.56 2.52 4.12 3-5 times 0.26 0.66 1.38 2.43 6-9 times 0.14 0.12 0.42 0.50 10 or more times 0.12 0.24 0.45 0.87 Heavy episodic use Never 95.15 94.24 93.13 90.54 Once 3.20 3.76 3.85 5.11 Twice 1.08 1.25 1.64 2.34 3-5 times 0.31 0.38 0.88 1.19 6-9 times 0.09 0.07 0.16 0.40 10 or more times 0.16 0.31 0.34 0.42 73

PAGE 74

Table 2-8. Continued Frequency (%) T1 (2002) N = 4,259 T2 (2003) N = 4,240 T3 (2004) N = 3,778 T4 (2005) N = 3,804 Alcohol use intentions Drink if best friend offered Yes 1.91 3.54 7.39 12.82 Not sure 12.02 17.37 24.75 27.24 No 86.07 79.08 67.85 59.94 Drink next month Yes 1.41 2.62 5.65 11.37 Not sure 8.51 12.85 19.18 22.49 No 90.08 84.52 75.17 66.14 Drink in high school Yes 7.39 10.75 17.16 24.35 Not sure 29.12 35.19 37.93 37.06 No 63.49 54.05 44.91 38.59 Drink when and adult Yes 20.43 24.67 33.47 38.09 Not sure 42.70 44.33 41.97 39.81 No 36.87 31.00 24.56 22.10 74

PAGE 75

Table 2-9. Frequencie s of home and family management items T3 (2004). Frequency (%) N = 3,778 Home alcohol access Easy to get alcohol from parent Easy 7.05 In-between 12.09 Hard 80.86 Easy to get alcohol from home Easy 17.55 In-between 18.59 Hard 63.85 Last time drank, received alcohol from parent No 89.09 Yes 10.91 Last time drank, took alcohol from home No 96.72 Yes 3.28 Parent monitoring/communication Parent ask about school Never 4.11 Hardly ever 7.11 Sometimes 28.34 A lot 24.52 All the time 35.91 Parent praise when do a good job Never 5.89 Hardly ever 9.13 Sometimes 26.57 A lot 21.61 All the time 36.79 Eat dinner with parent Never 4.99 Hardly ever 8.94 Sometimes 20.72 A lot 21.70 All the time 43.65 Parent ask who with Never 2.58 Hardly ever 3.51 Sometimes 12.54 A lot 18.12 All the time 63.26 75

PAGE 76

Table 2-9. Continued. Frequency (%) N = 3,778 Parent/child conversations Never 8.25 Hardly ever 13.90 Sometimes 34.96 A lot 22.74 All the time 20.15 Alcohol-specific communication Parent talk about problems drinking alcohol can cause Never 17.63 Hardly ever 18.85 Sometimes 27.32 A lot 17.89 All the time 18.32 Parent talk about family ru les against drinking alcohol No 87.96 Yes 12.04 Parent talk about conseque nces of drinking alcohol Never 21.73 Hardly ever 18.91 Sometimes 25.01 A lot 17.33 All the time 17.01 Parent talk about influen ce of ads and commercials Never 33.25 Hardly ever 16.51 Sometimes 27.07 A lot 12.60 All the time 10.57 76

PAGE 77

CHAPTER 3 MULTI-ETHNIC, URBAN YOUTHS EXPO SURE TO PATTERNS OF ALCOHOLRELATED NEIGHBORHOOD CHARACTERISTICS1 Executive Summary Background Research has shown the importance of soci al capital (Coleman, 2007) and exposure and access to alcohol in preventing and/or promoti ng a number of negative outcomes (Collins et al., 2007; Gorman, Labouvie et al., 1998; Kawachi et al., 1997; Pasch et al ., 2007; Poortinga, 2006; Scribner et al., 1999; Scribner et al., 1994; Scribner et al., 2007; Siahpush et al., 2006). Youth residing in urban neighborhoods may be at dispropor tionate risk, as their extant social capital may be less and their exposure and access to alcohol more than their suburban and rural counterparts (Coleman, 2007; Forrest & Kearns, 2001; Hackbarth et al., 2001; Pollack et al., 2005; Treno et al., 2000). However, this disparat e risk may not only be present across urban, suburban and rural areas, but also across nei ghborhoods within urban areas. Further, risk disparities may exist acro ss race/ethnicity, as racial/ethnic minority youth are disproportionately residents of metropolitan cities (U.S. Census Bureau, 2000), where many social problems (e.g., crime, delinquency, drug use, public disorder, and school dropout) are si gnificantly clustered (Coulton et al., 1995; Duncan et al., 2002; Sampson, 1992). The present study provides a description of the urban communities in which a large sample of racial/ethnic minority youth reside, usi ng measures of social capital and exposure and access to alcohol obtained from Census 2000, repo rts by parents and community leaders within each study community, or assessed directly. Speci fically, the research questions are (1) How many latent classes are necessary to describe heterogeneity in neighborhood characteristics of 1 For the full text of this work, see Tobler, A.L., Komr o, K.A. & Maldonado-Molina, M.M. (In Press). Multi-ethnic urban youths exposure to patterns of alcohol-related neighborhood characteristics. Journal of Community Health 77

PAGE 78

urban communities in which multi-ethnic early adolescents reside? and (2) What are the characteristics and proportions of adolescents residing in the heteroge neous latent neighborhood classes? Methods Data included baseline measures from a grouprandomized trial of an alcohol preventive intervention for multi-ethnic urban youth (Pro ject Northland Chicago; Komro et al., 2008), encompassing 42 of 77 city-defined Chicago community areas. Th e sample in this secondary data analysis included 4,215 youth who comple ted school-based surveys when beginning 6th grade. Students were predominan tly African American or Hispan ic (42% and 30%, respectively), had an equal gender distribution (50% male), lived in the United States for all their life (86%), spoke English in their homes (72%), lived in tw o-parent households (53% ) and were low income (70%). Neighborhood measures included (1) mean number of alcohol outlets per 1,000 population per community area; (2 ) alcohol purchase attempt rate by pseudo-underage youth; (3) average number of alcohol advertisements with in 1500 feet of each scho ol per community; and (4) a Census 2000-based deprivation index. Pa rents and community leaders provided data on perceived neighborhood problems and parental prevention actions, a nd neighborhood strength and preventive action by communities, law enforcement, and community organizations, respectively. Measures were dichotomized at their means to improve interpretability of results and identification of how the community characteristics differed. Multilevel latent class analysis (LCA; Asparouhov & Muthn 2006) was used to identify the number and characteristics of heterogene ous latent neighborhood classes among students residing in study communities using the nine dichotomous indicators. Analyses were conducted at the individual (student) level (given the in dividual-level data provided by the parent), 78

PAGE 79

adjusting for the correlation of responses among students within each community. Mplus 4.21 was used for all analyses (Muthn & Muthn, 2007). Results Five classes best described the heter ogeneity among the sample (1) Low social capital/low exposure/high access to alcohol (19.8% ), (2) Low social capital/low exposure/low access to alcohol (24.5%), (3) Moderate social capital/low exposure/ high access to alcohol (30.0%), (4) Moderate social capital/moderate exposure/low access to alcohol (20.1%), and (5) High social capital/moderate exposure/high access to alcohol (5.6%). We found considerable segregation across the neighborhood classes, where Hispanic youth we re the clear majority in the Low social capital/low exposure/low access to alcohol and Moderate social capital/moderate exposure/low access to alcohol clas ses. African American youth we re the clear majority for the other classes: Low social capital/low exposure/high access to alcohol, Moderate social capital/low exposure/high access to alcohol, an d High social capital/moderate exposure/high access to alcohol. The majority of both African Americans and Hi spanics resided in communities characterized by low social capita l (58.4% and 51.5%, respective ly). Only 10% of the African American and 3% of the Hispanic youth in this sample resided in communities characterized by high social capital. The majority of Af rican American youth resided in communities characterized by high access to alcohol (79.8%); wh ereas, the majority of Hispanic youth resided in communities characterized by low access to alcohol (19.9%). Conclusions Results suggest there is s ubstantive heterogeneity among this seemingly homogeneous urban population. Youth at greatest risk may be those in communities characterized by low social capital and high access to alcohol (Class 1) and moderate social capital and high access to alcohol (Class 3). Further, findings highlight the socioeconomic di sadvantage of these inner-city 79

PAGE 80

communities and the resource disparity across the racial/ethnic groups. Understanding the nuances in urban neighborhood contexts provides the foundation for understanding subsequent effects on family functioning and problem behaviors, may appropriately inform theory, suggest targets for intervention, and help prioritize limited resources (Dun can et al., 2002; Griffin et al., 2000). 80

PAGE 81

CHAPTER 4 EFFECTS OF ALCOHOL-RELATED NEIGHBORHOOD CONTEXT ON THE TRAJECTORIES OF ALCOHOL US E AND INTENTION S AMONG YOUNG ADOLESCENTS Abstract Background African American and Hispanic youth disproportionately reside in inner cities. As a result, they are at greater risk for maladaptive social and behavioral out comes, including alcohol and other drug use. Little is known about the e tiology of alcohol use among this growing, highrisk segment of the population. Likewise, our knowledge of how their uni que context influences their alcohol use and alcohol use intentions is limited. This study extends the scientific knowledge about these relationships by examin ing the direct effect s of alcohol-related neighborhood contextual risk on the trajectories of alcohol use beha viors and alcohol use intentions among a large samp le of urban, racial/eth nic minority, young adolescents. Methods Participants included 4,215 youth residing in 42 community areas in Chicago, Illinois who completed surveys at the beginning of 6th grade (2002), end of 6th grade (2003), end of 7th grade (2004) and end of 8th grade (2005). Particip ants responded to nine items assessing their frequency of alcohol use and intentions to use alcohol. Neighborhood risk was characterized using five, mutually exclusive, alcohol-related neighborhood risk cl asses describing extant social capital and exposure and access to al cohol. Participants were assigned to one of the five classes using maximum rule assignment. General linear mixed effects regression was used to assess the change in alcohol use and alcohol use intentions over time beginning at gr ade six to grade eight attributable to alcohol-related neighborhood risk. 81

PAGE 82

Results None of the neighborhood risk classes were sign ificantly associated w ith the trajectories of alcohol use/intentions relative to the other classes for eith er African Americans or Hispanics. Among the individual neighborhood ri sk/protective items comprising the composite indicator, organizational preventive efforts and percei ved neighborhood problems were significantly associated with reduced alcohol use/intentions over time for African American and Hispanic youth, respectively. However, the direction of the perceived neighborhood problems effect was opposite to that hypothesized. None of the effect s for the other risk/protective items reached statistical significance. Conclusions The lack of significant effect s highlights the need for exam ining the role of family management and functioning in mediating/m oderating the deleterious effects of risky neighborhood contexts. Key Words Adolescents, Communities, Context, Alcohol, Social Capital Introduction Despite slight declines in recent years, alcohol remains the most frequently used drug among youth in the United States throughout adolescence. Among 8th-grade adolescents in particular, 39% have used alcohol in their lifetime, 32% have used alcohol in the past year, and 16% have used in the past month (Johnston et al., 2008). Heavy, probl ematic use is also prevalent; 18% of 8th -grade students have been drunk in their lifetime, 13% have been drunk in the past year, and 6% have been drunk in the past month (Johnston et al., 2008). Further, 10% report heavy episodic usehaving had five or more drinks in a row in the previous two weeks (Johnston et al., 2008). Such use is not with out considerable cons equence, as it contributes to 82

PAGE 83

traffic crashes, increased risk for disease, risky sexual behavior, homicides, suicides, crime, and unintentional injury (B orowsky et al., 2001; Dunn et al., 2003; Greenfeld, 1998; GyimahBrempong, 2001; National Highway Traffic Safety Administration, 2005; National Institute on Alcohol Abuse and Alcoholism, 2000; Smith et al., 1999; Sorenson & Berk, 2001). Additionally, exposure to alcohol in adoles cence can have detrimental effects on cognitive growth and functioning and increases the likelihood for late r addiction (Brown et al., 2000; Monti et al., 2005). Given the prevalence and consequences of alcohol use among adolescent youth, a substantial body of literature has been devoted to understanding factors associated with this behavior. However, to date, most has focused on more proximal influences (e.g., resistance selfefficacy, attitudes and norms favorable to use, peer use), whereas the influences of more distal, contextual factors (e.g., social capital, community disorganization, alcohol control policies) have more often been assumed than quantified empiri cally (Britt et al., 2005; Duncan et al., 2002; Gibbons et al., 2004; National Inst itute on Alcohol Abuse and Al coholism, 2004; Roski et al., 1997; Toumbourou et al., 2007; Wage naar et al., 2004). Additionally, relatively few longitudinal studies have been conducted among racial/ethnic minority youth residing in urban communities. This is a critical gap in the literature, as census data indicate that the United States is quickly moving toward a majority-minority society (Hobbs & Stoops, 2002; U.S. Census Bureau, 2003) and African American and Hispanic youth di sproportionately reside in urban cities (U.S. Census Bureau, 2000). These youth are at increased risk for a number of maladaptive social and behavioral outcomes, including alco hol and other drug use, attribut able to a number of factors, including neighborhood disorder, a sense of hope lessness, psychological distress, increased opportunity for drug use, weaker economic conditions and fewer neighborhood resources 83

PAGE 84

(Arkes, 2007; Crum et al., 1996; Duncan et al ., 2002; Elliott et al., 1996; Hill & Angel, 2005; Karvonen & Rimpela, 1997; Wilson et al., 2005). Fu rther, African American youth drink alcohol in lower quantities and less frequently than mo st other racial/ethnic groups (Substance Abuse and Mental Health Services Admi nistration, 2006); yet, they su ffer disproportionately from the physical and social consequences of use (National Institute on Alcohol Abuse and Alcoholism, 2000). Current demographic and social trends and chas ms in the scientific literature elucidate the importance of understanding the etiology of alco hol use among such growing, at-risk segments of the United States population. Alcohol us e among racial/ethnic minority youth residing in urban communities may be the result of not only more proximal, individual characteristics, but also the interaction of their un ique, distal environmental and cultural contexts (Godette et al., 2006). The present study extends the scientific know ledge about the etio logy of alcohol use among young adolescents residing in urban commun ities by examining longitudinally the direct effects of alcohol-related nei ghborhood risks on alcohol use beha viors and intentions over time. Specifically, neighborhood risk was described by ex tant social capital and exposure and access to alcohol. Research has shown associations between social capital (Coleman, 1994, 2007) and exposure and access to alcohol and a number of negative outcomes. For example, social capital has been associated with se lf-rated health and health be haviors (Poortinga, 2006), smoking (Siahpush et al., 2006), and mortality (Kawachi et al., 1997). Additionally, alcohol outlets and advertisements are disproportionately located in urban, low-income, minority communities (Hackbarth et al., 2001; Pollack et al., 2005; Treno et al., 2000) and associated with alcohol consumption and intentions to drink (Collins et al., 2007; Ellickson et al., 2005; Fleming et al., 84

PAGE 85

2004; Pasch et al., 2007; Scribner et al., 2000; Scribner et al., 2007; Snyder et al., 2006; Stacy et al., 2004), violence (Gorman, Labouvi e et al., 1998; Gorman, Speer et al., 1998; Scri bner et al., 1999; Speer et al., 1998), and tra ffic crashes (Scribner et al., 1994). Moreover, despite existing laws that make it illegal for youth under the age of 21 to purchase alcohol in the United States, underage youth can, and do, purchase it. Studies indicate that underage buyers are able to purchase alcohol without showing age identification in 47-97% of attempts (Forster et al., 1994; Grube, 1997; Paschall, Grube, Black, Flewelling et al., 2007; Preusser & Williams, 1992). Such commercial access to alcohol may contribute to the large percent of underage youth who use and abuse alcohol (Johnston et al., 2008). Consider ed together, these c ontextual factors are considerable risks to the h ealthy development of urban mi nority youth and warrant further scientific investigation. This study provides a distinctive description of the direct effect s of alcohol-related neighborhood risk on alcohol use among young adolescen ts residing in an urban context by using a multi-dimensional latent class indicator of alcohol-related nei ghborhood risk (Chapter 3) to predict the trajectories of alcohol use and alcohol use intentio ns. Much of the extant literature describing the neighborhood context, and subse quent maladaptive behaviors, among youth has relied on either census data (Allison et al., 1999; Chuang et al., 2005; Elliott et al., 1996; Galea et al., 2007) or self-report measures (Crum et al., 1996; Gibbons et al., 2004; Hill & Angel, 2005), whereas context in this study was defined by fi ve heterogeneous latent classes of neighborhood risk. These classes were identified previously usin g nine indicators of so cial capital and exposure and access to alcohol obtained from Census 2000, reports by parents and community-leaders within each study community, or assessed directly. 85

PAGE 86

Methods Design Data were part of a longitudinal group -randomized controlled trial of an alcohol preventive intervention for multi-ethnic urban youth [Project Northland Chicago (PNC); see Komro et al., 2004; 2008 for a complete description of the projects resear ch design, participant recruitment, intervention components, and outcomes], which included 42 of 77 city-defined Chicago community areas as part of the study. The sample included 4,215 youth residing in the 42 study communities who were present and complete d school-based surveys at the beginning of 6th grade (2002). The students were predominantly African American or Hispanic (42% and 30%, respectively), had an equal ge nder distribution (50% male), lived in the United States for all their life (86%), spok e English in their homes (72%), liv ed in two-parent households (53%) and were low income (70% receiving free, or re duced price lunch). In terms of demographic characteristics, participating students were si milar to students enrolled throughout the Chicago Public School (CPS) system, where 50% and 38% of youth were African Am erican or Hispanic, respectively, and 85% were lo w income. Also, the study schools were similar to schools throughout CPS with respect to truancy (1.9% PNC, 3.6% CPS) and the percentages of students at, or above, the norms for math (48.6% PNC, 43% CPS) and reading ( 42.4% PNC, 42.9% CPS). Data Collection Student surveys were administered in st udy schools during the fa ll of 2002, spring of 2003, spring of 2004 and spring of 2005, when the students were in the 6th, 7th and 8th grades. Surveys were administered by trained university -based research teams using standardized protocols. Prior to survey administration, pa rents and students were given the opportunity to refuse participation. Response rates were between 91% and 96% each year (students who completed a survey/student enrolle d in the relevant grade in th e study schools each year). Data 86

PAGE 87

collection protocols were approved by the Univer sity of Minnesota Inst itutional Review Board, with secondary data analyses approved by the Univ ersity of Florida Institutional Review Board. Measures Neighborhood risk Students were classified into one of five, mutually exclusive, latent neighborhood risk classes identified with a multileve l latent class analysis of nine indicators of neighborhood risk and protective factors (1) ne ighborhood strength; (2 ) neighborhood and police preventive action; (3) organizational preventive efforts; (4) community action; (5) perceived neighborhood problems; (6) mean number of alcohol adver tisements within 1500 of each public school per community; (7) mean number of off-sale alc ohol outlets per 1,000 popul ation; (8) alcohol purchase attempt success rate; and (9) a Census 2000-based area deprivation index (Chapter 3). Six of these measures were conc eptualized as measures of soci al capital (Coleman, 1994, 2007): neighborhood strength, neighborhood and police preventive action, organizational preventive efforts, community action, perceived neighborho od problems, and area deprivation. The mean number of alcohol advertisements and off-sale alcohol outlet density provided measures of exposure to alcohol, whereas the al cohol purchase attempt success rate provided a direct measure of commercial accessibility of alcohol. The five neighborhood risk classes and the propo rtion of the sample within each included (1) low social capital/low exposure/high access to alcohol (19.8%), (2) low social capital/low exposure/low access to alcohol (24.5%), (3) mode rate social capital/low exposure/high access to alcohol (30.0%), (4) moderate social capital/m oderate exposure/low access to alcohol (20.1%), and (5) high social capital/moderate exposure/hi gh access to alcohol (5.6%). High, Moderate and Low labels were assigned using the number of items in each category (i.e., social capital, exposure to alcohol, access to al cohol) that had high probabilitie s of being above the mean for 87

PAGE 88

each latent class. Membership in the respective latent neighborhood risk class was represented with a categorical variable with five levels. Alcohol use and alcohol use intentions The tendency to use alcohol was assessed longi tudinally with a nine item scale measuring alcohol use/intentions (1) During the last 12 months, on how many occasions, or times, have you had alcoholic beverages to drink?; (2) Dur ing the last 30 days, on how many occasions, or times, have you had alcoholic beverages to drin k?; (3) During the la st 7 days, on how many occasions, or times, have you had alcoholic beverage s to drink?; (4) Think back over the last 2 weeks, on how many times have you had five or mo re alcoholic drinks in a row?; (5) Have you ever become really drunk from drinking alcoholic beverages so you fell down or became sick?; (6) Would you drink alcohol if your best friend offered it to you?; (7) Do you think you will be drinking alcohol in the next month?; (8) Do you think you will be drinking alcohol when a senior in high school?; and (9) Do you thi nk you will be drinking alcohol when you are an adult?(Cronbachs alpha: 0.85-0. 88; Range: 9-45). A higher score on this scale indicated greater alcohol use/intentions. Covariates A baseline measure of family composition was se lected for inclusion as a covariate in the analyses, as it is theoretically related to al cohol use among youth (Bergmark & Andersson, 1999; Bjarnason et al., 2003; Foxcroft & Lowe, 1991; Miller, 1997) and ha d a disproportionate distribution among the neighborhood risk classes. Family composition was dichotomized such that mother and father together was compared to other (mother and father equally, at separate homes, mother mostly, father mostly, grandparent, other relative, foster parents, other). Race/ethnicity was considered for inclusion as a covariate; however, the 88

PAGE 89

race/ethnicities were dispropor tionately distributed across th e neighborhood risk classes (See Chapter 3). Therefore, each race/ethnicity was modeled separately. Analytical Strategy Using the neighborhood classes identified and described previously among this sample (Chapter 3), each individuals class member ship was determined using maximum rule assignment, where the class assigned reflects that for which their posterior probability is the highest. Nagin (1999; 2005; 2001) su ggested that an average poster ior probability for each class of 0.80 indicates adequate assignment. Here, th e average posterior probability for each class ranged from 0.96 to 1.0. To assess the change in alcohol use/intentions over time attributable to membership in the respective neighborhood risk classes, we used general linear mixed modeling (i.e., multilevel growth curve analysis). The al cohol use/intentions scale used in the present study had an approximate normal distribution (skewness = 2.33; kurtosis = 7.34) which did not significantly benefit from log or root-based transformations. Therefore, analyses we re conducted using PROC MIXED (SAS version 9.1), a procedure in SAS de signed to handle dependent variables that are Gaussian and which does not apply list-wise de letion. Based on diagnostics outlined by Muller and Fetterman (2002) and descri bed previously (Komro et al., 2007), we specified an unstructured error structure and linear functional form for the dependent variable. Community was specified as a nested random effect to account for the dependency of observations among students within each study community. Additionally, given the disproportionate racial/ethnic distribution among the neighborhood risk classes (Table 4-1; Chapter 3), analyses were conducted separately for the African American and Hispanic youth. The alcohol use/intentions s cale was regressed on time, baseline alcohol use/intentions, family composition, and alcohol-related nei ghborhood risk class. Table 4-1 presents the 89

PAGE 90

descriptive statistics for all variables included in the models by race/ethnicity. A partial test approach was used, where estimates do not depe nd on the order of variab les in the model and represent the impact of each predictor while cont rolling for all others. As there was no clear, low/no risk referent class among the five (e.g ., high social capital/low exposure/low access to alcohol), all possible contrasts of the neighborhood risk classes were examined. Missing Data White, Asian, Native American, and youth fr om Other race/ethnicities were excluded from analyses because their sample sizes did not provide sufficient statistical power to detect significant effects when modeled independently. The analysis samp les for the African American and Hispanic youth included 98.1% (n = 1,710) and 97.8% (n = 1,213) of the eligible samples, respectively, as a resu lt of missing values among at least one of the independent variables at baseline. There were no statisti cally significant differences between the eligible and analysis samples across the neighborhood risk classes for either African American or Hispanic. The eligible and analysis samples did differ signif icantly for both African Americans and Hispanics in that youth reporting less alcohol use/intentions were more likely to be included in the analysis sample. The eligible and analysis samples did not differ significantly across family composition for either race/ethnicity. Seventy-two percent of the cohort students comp leted three or four surveys, while 28% completed one or two. All available follow-up data were used in the analysis. Students who completed three or four surveys differed significantly from those who completed one or two in that older students were less likely to have completed three or four surveys. Students who completed three or four surveys did not differ significantly from those who completed only one or two w ith respect to neighborhood risk class, race/ethnicity, gender, family composition, and al cohol use/intentions. 90

PAGE 91

Results Table 4-2 presents the results from the ge neral linear mixed model regressions for each racial/ethnic group. We examined all possible contrasts of the neighborhood risk classes. However, the only effects that were near margin al significance were for contrasts with the low social capital/low exposure/low access to alcohol cla ss (Class 2). Therefore, all results presented in Table 2 are relative to this class. All analyses controlled for baseline values of the outcome and family composition. Neighborhood risk class membership at age 12 was not significantly associated with the trajectories of alcohol use/intentions from ag e 12 to 14 when controlling for baseline alcohol use/intentions and family composition for either the African American or Hispanic samples. Two contrasts among the Hispanic sample a pproached marginal significance (i.e., p < 0.10) compared to the low social capital/low exposure/ low access to alcohol class, Hispanic youth in the moderate social capital/low exposure/hi gh access to alcohol and moderate social capital/moderate exposure/low access to alcohol classes were less likely to report alcohol use/intentions over time, albeit not statistically significant ( p = 0.1349 and p = 0.1247, respectively). Given the lack of significant results using the composite neighborhood risk class indicator as a predictor of alcohol use/intentions over time, we examined the predictive utility of the individual neighborhood risk/protective items that defined th e classes. Table 4-3 presents the results of these post hoc bivariate analyses, where alcohol use/intentions over time was regressed on the individual risk /protective item, time and base line alcohol use/intentions. For African American youth, organiza tional preventive efforts were significantly associated with reduced trajectories of alcohol use/inte ntions (Standardized Slope: -0.0452, p = 0.0224). Among Hispanics, perceived neighborhood problems exhibi ted a significant inverse relationship with 91

PAGE 92

alcohol use/intentions over time, such that increases in perceived neighborhood problems were associated with reduced trajectories of alc ohol use/intentions (Standardized Slope: -0.0602, p = 0.0072). None of the other risk/p rotective items were significan tly associated with alcohol use/intentions over time for either racial/ethnic group. Discussion The present study examined the direct eff ects of alcohol-related neighborhood contexts (i.e., risk) on the trajectories of alcohol us e/intentions among a larg e sample of urban, racial/ethnic minority youth. None of the neighborhood risk classe s were significantly associated with the trajectories of alcohol use/intentions relative to the ot her classes for either African Americans or Hispanics. Among the Hispanic youth, two contrasts approached marginal significancerelative to the lo w social capital/low exposure/lo w access to alcohol class (Class 2), Hispanic youth in the moderate social cap ital/low exposure/high access to alcohol and moderate social capital/moderate exposure/low access to alcohol classe s were less likely to report alcohol use/intentions over time. However, these effects were not statistically significant. Among the individual neighborhood ri sk/protective items comprising the composite indicator, organizational preventive efforts and percei ved neighborhood problems were significantly associated with reduced alcohol use/intentions over time for African American and Hispanic youth, respectively. However, the direction of the perceived neighborhood problems effect was opposite to that hypothesizedincreases in perc eived neighborhood problems were associated with reduced alcohol use/intentions over time. N one of the effects for th e other risk/protective items were statistically significant. Results are consistent with previous res earch on the role of neighborhood context in alcohol and other drug beha vior (Duncan et al., 2002; Lambert et al., 2004) and suggest a need to examine potential mediators of the relationships examined here. For example, Lambert and 92

PAGE 93

colleagues (2004) found that among their sample of urban, African American adolescents, the effects of neighborhood characteris tics on substance use were entirely mediated by drug beliefs for females and partially mediated by drug beliefs for males. Fulk erson and colleagues (In Press) also reported no significant direct effects of a contextual measur e, informal social control, on alcohol use, alcohol use intenti ons, and other violent and delin quent behaviors in their crosssectional analyses among the same sample used in the present study. Rather, the authors found that the effects of informal social control on high risk beha viors were entirely mediated by parental monitoring. Certainly, more proximal, individual-level factors, such as attitudes and norms regarding substance use, warrant investiga tion as mediators, as do family-level factors, such as parental monitoring, parent/c hild communication, and the like. All of these youth were at considerable risk given their urban context, as none of the youth resided in low/no risk neighborhood (e.g., high social capital/low exposure/low access to alcohol). Research has shown that urban envi ronments are at increased risk for crime, delinquency, drug use, public disorder, school dropout, and exposure and commercial access to alcohol relative to their suburba n and rural counterparts (Bernstein et al., 2007; Coulton et al., 1995; Duncan et al., 2002; Gruenewald et al ., 2002; Kwate et al., 2007; LaVeist & Wallace, 2000; Sampson, 1992). Thus, the lack of effects here given the risks associat ed with residence in urban environments, suggest the need for me diation analyses to determine how family functioning and/or interpersonal characteristics may buffer the effects of neighborhood risk on alcohol use/intentions. A mediati on analysis will be the focus of future work and may provide a more complete understanding of the etiology of alcohol use/intentions among urban, minority, adolescents. 93

PAGE 94

This study had several limitations. First, there was no clear, low/no ri sk, referent class among the five (e.g., high social capital/low exposure/low access to alcohol). We examined all possible contrasts and the selection of Class 2, low social capital/low exposure/low access to alcohol, was somewhat arbitrary, as it provided the only near marginally significant effects. Second, we examined the effects of neighborhood context at one point in time (age 12) as a potential predictor of the trajector ies of alcohol use and alcohol us e intentions from age 12 to 14. The saliency of context in shaping alcohol use among youth may vary throughout adolescence (Szapocznik & Coatsworth, 1999). Accordingly, future research should examine associations between context and drinking behaviors and inte ntions of youth as they evolve and develop across time. Our data precluded such an examina tion. Third, the sample for this study were lowincome, racial/ethnic minority, young adolescents re siding in Chicago, Illi nois. More studies are needed to examine the relationships presente d here both during midto late-adolescence and among youth residing in other metro politan cities as well as rura l and suburban areas. Effects may not be consistent across developmental pe riods or across differing economic and cultural contexts. Lastly, measures of neighborhood risk and protection used do not represent the universe of neighborhood or social capital desc riptors. Future resear ch should examine the influence of neighborhood contexts on alcohol use and alcohol use inte ntions while including additional community measures, such as crime rates, political activism, public policies and measures of social structure. Limitations notwithstanding, this study contributes to a sparse literature describing both the etiology of alcohol use among urban, raci al/ethnic minority youth and the effects of neighborhood context on alcohol us e and alcohol use intentions over time. Moreover, this study used a distinct alcohol-related neighborhood risk indicator created by us ing data from Census 94

PAGE 95

95 2000, self-report measures from parents and co mmunity leaders and direct environmental assessments (Chapter 3). This is a particular strength, as much of the literature describing the influence of neighborhood context on drug use and other deleterious health and social outcomes has relied solely on census data (Allison et al., 1999; Chuang et al., 2005; Elliott et al., 1996; Galea et al., 2007) or self-report measures (Crum et al., 1996; Gibbons et al., 2004; Hill & Angel, 2005). The lack of significant effects highl ights the need for examining the role family management and functioning in mediating/m oderating the deleterious effects of risky neighborhood contexts.

PAGE 96

Table 4-1. Descriptive statistics for variables included in each model Full sample (n=4215) African Americans (n=1770) Hispanics (n=1265) Variable Mean SD Prevalence/ % Yes Mean SD Prevalence/ % Yes Mean SD Prevalence/ % Yes Neighborhood risk class Class 1: Low social capital/Low exposure/High access to alcohol 19.8 41.1 2.34 Class 2: Low social capital/Low exposure/Low access to alcohol 24.5 17.3 49.2 Class 3: Moderate social capital/Low exposure/High access to alcohol 30.0 29.0 14.2 Class 4: Moderate social capital/Moderate exposure/Low access to alcohol 20.1 2.9 30.9 Class 5: High social capital/Moderate exposure/High access to alcohol 5.6 9.8 3.4 Covariates at baseline Family composition (mother and father together) 52.7 33.5 70.3 Alcohol use & alc ohol use intentionsa 11.05 2.88 10.96 2.98 11.16 2.75 Outcome variable Alcohol use & alcohol use intentionsa 6th grade 11.72 3.49 11.49 3.29 12.00 3.63 7th grade 12.80 4.32 12.54 4.19 13.34 4.64 8th grade 13.96 5.39 13.36 4.68 14.61 5.71 -a Minimum = 9.00, Maximum = 45.00; A higher score on this scale indicate d greater alcohol use/intentions. 96

PAGE 97

Table 4-2. Time-invariant predictors at age 12 of trajectories of alcohol use and alcohol use intentions from age 12 to 14 Standardized slope SE t-value P-value African Americans Intercept 0.0372 0.0698 Time 0.1993 0.0136 14.70 <.0001 Neighborhood risk class Class 1: Low social capital/Low exposure/High access to alcohol -0.0031 0.0815 -0.04 0.9695 Class 2: Low social capital/Low exposure/Low access to alcohola -Class 3: Moderate social capital/Low exposure/High access to alcohol -0.0062 0.0807 -0.08 0.9391 Class 4: Moderate social capital/Moderate exposure/Low access to alcohol -0.1146 0.1222 -0.94 0.8521 Class 5: High social capital/Moderate exposure/High access to alcohol 0.0252 0.1351 0.19 0.3488 Covariates Baseline alcohol use & alcohol us e intentions 0.4988 0.0178 27.99 <.0001 Family composition (mother & father together) -0.0263 0.0164 -1.61 0.1078 Hispanics Intercept 0.0688 0.0455 1.51 0.1410 Time 0.2295 0.0136 16.90 <.0001 Neighborhood risk class Class 1: Low social capital/Low exposure/High access to alcohol -0.0558 0.1381 -0.40 0.6864 Class 2: Low social capital/Low exposure/Low access to alcohola -Class 3: Moderate social capital/Low exposure/High access to alcohol -0.1085 0.0725 -1.50 0.1349 Class 4: Moderate social capital/Moderate exposure/Low access to alcohol -0.1022 0.0666 -1.54 0.1247 Class 5: High social capital/Moderate exposure/High access to alcohol -0.0428 0.1168 -0.37 0.7143 Covariates Baseline alcohol use & alcohol us e intentions 0.4395 0.0180 24.43 <.0001 Family composition (mother & father together) -0.0225 0.0180 -1.25 0.2118a Class 2, Low social capital/Low exposure and access to alcohol, was the referent class. 97

PAGE 98

Table 4-3. Results from bivariate analyses of individual neighborhood risk /protective items at age 12 as predictors of the tra jectories of alcohol use and alcohol use intentions from age 12 to 14 while controlling for baseline levels of intentions/use. African Americans Hispanics Standardized slope SE P-value Standardized slope SE P-value Protective factors Neighborhood strength -0.0235 0.0220 0.2844 -0.0068 0.0236 0.7739 Neighborhood & police preventive action 0.0093 0.0214 0.6648 -0.0131 0.0252 0.6035 Organizational preventive efforts -0.0452 0.0198 0.0224 -0.0032 0.0236 0.8908 Community action 0.0143 0.1900 0.4518 -0.0017 0.0220 0.9379 Risk factors Perceived neighborhood problems -0.0164 0.0195 0.3996 -0.0602 0.0224 0.0072 Mean number of alcohol advertisements -0.0092 0.0255 0.7186 0.0031 0.0303 0.9174 Mean number of off-sale alcohol outlets per 1,000 pop. 0.0427 0.0302 0.1571 -0.0367 0.0268 0.1700 Alcohol purchase attempt success rate 0.0070 0.0252 0.7818 -0.0071 0.0230 0.7561 Area deprivation index 0.0205 0.0245 0.4022 0.0199 0.0267 0.4557 98

PAGE 99

CHAPTER 5 RELATIONSHIPS BETWEEN NEIGHBORHOOD CONTEXT, FAMILY MANAGEMENT PRACTICES AND ALCOHOL USE AMON G URBAN, MULTIETHNIC, YOUNG ADOLESCENTS Abstract Background African American and Hispanic youth are di sproportionately reside nts of inner-city communities. Little is known about how their uni que context influences early onset of alcohol use. We examined relations between alcohol-re lated neighborhood context, protective home and family management practices, and alcohol use among urban, racial/ethnic minority, adolescents. Methods Data were part of a longitudinal study of multi-ethnic urban youth (Project Northland Chicago). The sample comprised 5,655 youth w ho were primarily low SES (72%), African American (43%) and Hispanic (29%). Particip ants completed surveys in 2002-2005 (ages 11-14 years). Items assessed alcohol use, accessibility of alcohol at home and parental family management practices. Neighborhood context meas ures at baseline included (1) alcohol outlet density; (2) commercial alcohol acce ssibility; (3) alcohol advertis ement exposure; (4) perceived neighborhood strength, reported by parents and community leaders; and (5) area deprivation. Structural equation modeling was used to assess direct and indirect re lations between alcoholrelated neighborhood context at baseline, home alc ohol access and family management practices in 7th grade, and alcohol use in 8th grade. Results Neighborhood strength was negatively, and exposure to alcohol advertisements positively, associated with alcohol use ( = -0.082, p < 0.05 and = 0.049, p < 0.05, respectively). Neighborhood strength and commerci al alcohol access were associated with home 99

PAGE 100

alcohol access ( = 0.063, p < 0.05 and = -0.124, p < 0.001, respectively) and protective family management practices ( = -0.070, p < 0.01 and = 0.073, p < 0.001, respectively). Home alcohol access had a positive asso ciation with alcohol use ( = 0.399, p < 0.001), while the association between protective family management practices and alcohol use was not significant when home alcohol access was considered. Tests for indirect effects suggest that home alcohol access may partially mediate the relation betw een neighborhood strength and alcohol use, although this indirect effect was marginally significant ( = 0.025, p = 0.062). Conclusions Neighborhood context had significant direct and indirect associ ations with alcohol use. Results suggest inner-city parents respond to e nvironmental risk, such that as neighborhood risk increases, so also do protective home and family management practices. Parent engagement in restricting alcohol access and improving family ma nagement practices may be key to preventive efforts to reduce alcohol use am ong inner-city, adol escent youth. Key Words Adolescents, Communities, Family, Context, Alcohol Introduction Alcohol remains the drug of choice among youth in the United States. Among 8th-grade adolescents in particular, 39% have used alcohol in their lifetime, 32% have used alcohol in the past year, and 16% have used in the past m onth (Johnston et al., 2008). Heavy, problematic use is also prevalent; 18% of 8th -grade students have been drunk in their lifetime, 13% have been drunk in the past year, and 6% have been drunk in the past month (Johnston et al., 2008). Further, 10% report heavy episodic usehaving had five or more dr inks in a row in the previous two weeks (Johnston et al., 2008). Such alcohol use has been associated with a number of deleterious health and social problems, including alcohol abuse and dependence, alcohol-related 100

PAGE 101

violence and injuries, drinking and driving, trua ncy, traffic crashes, risky sexual behavior, and other drug use throughout adolescence and into adulthood (Grant, Stinson, & Harford, 2001; Gruber, DiClemente, Anderson, & Lodico, 1996; Hingson et al., 2002; Hingson, Heeren, Winter et al., 2003; Hingson et al., 2001; Hingson, Heeren, Zakocs, Wint er, & Wechsler, 2003; Hingson et al., 2006). Additionally, exposure to alcohol in adolescence can have detrimental effects on cognitive growth and functioning and increases the likelihood for later addiction (Brown et al., 2000). Given the prevalence and consequences of alcohol use among youth, a substantial body of literature describing the etiology of this problem atic behavior has amassed. However, to date, most of these studies have focused on individual-, peerand family-level influences (Britt et al., 2005; Duncan et al., 2002; Toumbourou et al., 2007) and few longit udinal studies have examined the etiology of alcohol use among racial/ethni c minority youth residing in urban communities. This is a critical gap in the literature, as census data indicate that the United States is quickly moving toward a majority-minority society (Hobbs & Stoops, 2002; U.S. Census Bureau, 2003) and African American and Hispanic youth di sproportionately reside in urban cities (U.S. Census Bureau, 2000). These youth are at increased risk for a number of maladaptive social and behavioral outcomes, including alcohol use, rela ted to their unique environments (Arkes, 2007; Duncan et al., 2002; Hill & Angel, 2005; Wilson et al., 2005). Further, African American youth drink alcohol in lower quantitie s and less frequently than mo st other racial/ethnic groups (Substance Abuse and Mental Health Services Administration, 2006); yet, they suffer disproportionately from the physical and social consequences of use (National Institute on Alcohol Abuse and Alcoholism, 2000). 101

PAGE 102

These demographic and social trends and chasms in the scientific literature elucidate the importance of understanding the etiology of alco hol use among such growing, at-risk segments of the United States population. Further, they suggest that alcohol use among racial/ethnic minority youth residing in urban communities may be the result of not only proximal, individual characteristics, but also the interaction of th eir unique, community and family environments (Godette et al., 2006). For example, several neig hborhood characteristics ha ve been associated with alcohol use among youth, including alcohol ou tlet density (Pollack et al., 2005; Scribner et al., 2007; Treno et al., 2000), exposure to alcohol advertisements (Collins, Wileyto, Murphy, & Munafo, 2007; Pasch et al., 2007; Snyder et al., 2006), commercial alcohol accessibility (Forster et al., 1994; Grube, 1997; Paschall, Grube, Black, & Ringwalt, 2007; Sobel, 1982), and neighborhood strength and deprivation (Boardman & Saint-Onge, 2005; Scheier, Botvin, & Miller, 2000). The distribution of these characteristics has been shown to be disproportionate across urban, suburban and rural communities (Coleman, 2007; Forrest & Kearns, 2001; Pollack et al., 2005; Treno et al., 2000). A variety of home and family management factors have also been found to influence adolescent alcohol use, including accessibility of alcohol in the home (Jackson et al., 1999; Komro et al ., 2007), parental monitoring (Alvarez et al., 2003; Borawski et al., 2003; Cleveland et al., 2005), parent/child co mmunication (Kelly et al., 2002; Wills et al., 2003), relationship satisfaction (Ledoux et al., 20 02; Nelson et al., 1999; Wills et al., 2003), and supervision (Aizer, 2004; Coley et al., 2004; Richards on et al., 1993). However, what remains unclear is how these neighborhood and family charact eristics in urban settings relate to each other and to alcohol use. Some studies suggest th at parents respond to risk in their environments, exhibiting higher levels of protective family management practices (Beyers et al., 2003; Chuang et al., 2005) and mediating the effects of risky neighborhood environments on alcohol use 102

PAGE 103

(Beyers et al., 2003; Chuang et al., 2005). Th ese findings are consis tent with Beck and Lockharts model of parental involvement in adolescent dri nking (Beck & Lockhart, 1992), as well as more comprehensive theories of adoles cent health behaviors (F lay & Petraitis, 1994; Wagenaar & Perry, 1994). These theories suggest that as parental awar eness of the alcoholrelated risks in their environment increase, so also do their protective home and family management practices (e.g., availability of alcohol in the home, parental monitoring, communication). However, other studies suggest that neighborhood risk is compounded by lower levels of protective home and family management practices, leading to higher levels of use and other maladaptive behaviors (Ingoldsby & Shaw, 2002; Rankin & Quane, 2002). The present study extends the scientific knowledge about the etiology of alcohol use among racial/ethnic minority, young adolescents residing in urban communities by examining longitudinally the direct and i ndirect relations between alc ohol-related neighborhood context, home and family management practices, and alc ohol use. The hypothesized structural model was founded upon substantive theory [i.e., theory of triadic influence (Flay & Petraitis, 1994) and Wagenaar and Perry's model of drinking beha vior (Wagenaar & Perry, 1994)] and previous research (Figure 5-1). We hypothe sized that each of the alcohol -related neighborhoo d contextual constructs would show direct, positive associations w ith alcohol use while neighborhood strength would have a direct, negative association (Boa rdman & Saint-Onge, 2005; Pasch et al., 2007; Paschall, Grube, Black, & Ringwalt, 2007; Scribn er et al., 2007). Additionally, correlations among each of these constructs were expected Home alcohol access and protective family management practices were hypothesized to have direct effects on alc ohol use (positive and negative, respectively), as well as correlate with each other (A izer, 2004; Cleveland et al., 2005; Jackson et al., 1999; Komro et al., 2007; Swahn & Hammig, 2000). We hypot hesized that there 103

PAGE 104

would be more complicated associations betw een neighborhood strengths and risks, in that parents may respond to high risk environments by increasing protective factors within the home (Dent, Grube, & Biglan, 2005; Hawkins et al ., 1992; Ingoldsby & Shaw, 2002; Rankin & Quane, 2002). Methods Design Data were part of a longitudinal, group -randomized controlled trial of an alcohol preventive intervention for multi-ethnic urban youth [Project Northland Chicago (PNC); see Komro et al., 2008 for a complete description of the projects research design, participant recruitment, intervention components, and outcomes], which included 42 of 77 city-defined Chicago community areas as part of the study. The sample included 5,655 youth residing in the 42 study communities who completed at least on e study survey when they were in the 6th, 7th, or 8th grade. The students were predominantly Af rican American or Hispanic (43% and 29%, respectively), had an equal gender distribution ( 50% boys), spoke English in their homes (74%), and were low income (72% receiving free, or re duced price lunch). Less than half of the students (47%) lived in two-pare nt households. In terms of demographic characteristics, participating students were similar to student s enrolled throughout the Chicago Public School (CPS) system, where 50% and 38% of youth were African American or Hispanic, respectively, and 85% were low income. Also, the study school s were similar to schoo ls throughout CPS with respect to truancy (1.9% PNC, 3.6% CPS) and the percentages of students at, or above, the norms for math (48.6% PNC, 43% CPS) and reading (42.4% PNC, 42.9% CPS). At the end of the intervention period, there were no statistically significant differen ces in alcohol use, intentions, norms or attitudes between the inte rvention and control conditions (Komro et al., 104

PAGE 105

2008). Thus, data from both the control and inte rvention conditions were used for the present study. Data Collection Students Student surveys were administered in st udy schools during the fa ll of 2002, spring of 2003, spring of 2004 and spring of 2005, when the students were in the 6th, 7th and 8th grades. Data from the 2002, 2004, and 2005 surveys were used for the present study. All students enrolled in the appropriate grade each year we re eligible to participate. Surveys were administered by trained university-b ased research teams using standardized protocols. Prior to survey administration, parents a nd students were given the opportuni ty to refuse participation. Response rates were between 91% and 96% each year (students who comple ted a survey/student enrolled in the relevant grade in the study sc hools each year). Data collection protocols were approved by the University of Minnesota Instit utional Review Board, with secondary data analyses approved by the University of Florida Institutional Review Board. Parents Parents of the students were surveyed in fall, 2002 (n = 3,250; 70% response rate). Hardcopy surveys were given to students, and they were asked to deliver the packet to their primary caregiver (Komro et al., 2008). Parents were given $25 after the completed survey was returned. Students were given a $5 gift certificate for de livering the packet to their parents. Parents completing the surveys (n = 3,250) were predomin antly married (54%), ha d one to three children living in their home (70%) and had, at the leas t, graduated from high school (78%). Parents responded to seven items that assessed pe rceived neighborhood problems. Students whose parents did not complete the parent surv ey were not excluded from the study. 105

PAGE 106

Community leaders A telephone survey of community leaders in each community was conducted in 2002 (n = 344, 70% response rate). Community leaders included school council members, religious leaders, managers of recreation centers, neighborhood beat officers, neighborhood beat facilitators, and managers/lead ers of neighborhood organizations The survey instrument was based on others administered in similar resear ch projects (Komro et al., 1999; Wagenaar & Streff, 1990) and contained fourteen items assessing neighborhood strength and neighborhood and police preventive action. Neighborhood characteristics Data describing alcohol-re lated neighborhood charac teristics included (1) mean number of off-sale alcohol outlets per community area, obtained from the Chicago Licensing Department in 2002; (2) commercial alcohol accessibility, tested directly in 2002 by pseudo-underage youth (Komro et al., 2008); and (3) average number of alcohol advertisements within 1500 feet of each school per community, assessed in spring, 2003 (Pasch, Komro, Perry, Hearst, & Farbakhsh, In Press; Pasch et al., 2007). Census 2000 data for each community were also retrieved. Measures Alcohol-related neighborhood context Neighborhood strength. Five community leader survey items were used in a scale of neighborhood strength: How would you rate the neig hborhood in terms of having a strong community identity?; level of community resources?; pa rticipation level of residents in local activities?; level of influence local residents or community groups have on decisions about local policies?; and efforts of residents in addressing the prevention of alcohol use among teenagers? (Cronbachs alpha: 0.70; Range: 5-25). Response options were 1 106

PAGE 107

= low, 3 = medium, and 5 = high, with a higher score on this sc ale indicating greater neighborhood strength. Neighborhood and police preventive action. Nine community leader survey items were used in a scale of neighborhood and police pr eventive action: How would you rate police involvement in prevention of alcohol use am ong teenagers in the neighborhood?; How would you characterize relationships between loca l beat officers and neighborhood residents surrounding schools?; If teenager s were hanging out on the bloc k, how likely is it that residents in the neighborhood would do something about it?; If a store wa s selling alcohol to teenagers, how likely is it that residents in the neighborhood woul d call the police?; If police were called on a loud party involv ing young people, how likel y is it that they would check to see if there was underage drinking? ; How likely is it that a gr oup from the neighborhood would work to reduce the amount of alcohol advertisem ents?; How likely is it that if a business served or sold alcohol to minors, the business would be cited by an enforcement agency?; How likely is it that if an adult provided alcohol to minors, the adult would be cited or ticketed by police?; and How likely is it that a minor w ho was in possession of alcohol would be cited or ticketed by police? (Cronbachs al pha: 0.89, Range 9-45). Response options were in the form of a 5-option Likert scale ranging from very little involvement/not at all good/not at all likely to a great deal of involvement/very good/very likel y. A higher score on this scale indicated more neighborhood and police preventive action. Perceived neighborhood problems. A perceived neighborhood problems scale was created using seven items from the parent surv ey: Below is a list of urban problems. Please check how much of a problem each of the following is on the block where you live: drug dealing?; unsupervised youth?; people drinking alcohol on the street?; too many 107

PAGE 108

stores that sell alcohol?; lack of superv ised activities for youth? ; too many alcohol advertisements?; and poor police respons e? (Cronbachs alpha: 0.93, Range 7-35). Response options were 1 = not a problem, 3 = a minor problem, and 5 = a serious problem. A higher score on this scale indicat ed greater perceived neighborhood problems. Exposure to alcohol advertisements. The number of alcohol advertisements within 1500 feet of each study school was obtained in 2003 (Pasch et al., In Pres s; Pasch et al., 2007). The location of each ad was documented using a Global Positioning System. Street maps with a 1,500 foot radius around each school were create d using ArcView GIS. The average number of alcohol advertisements around sc hools within each community ar ea was obtained by dividing the total number of alcohol advertisements surr ounding schools within each community area by the total number of schools in each community area. Off-sale alcohol outlet density. The mean number of off-sale alcohol outlets per 1,000 population per community area was obtained by dividing the mean number of off-sale alcohol outlets per community area by the total population for each community area. Commercial accessibility of alcohol. Commercial accessibility of alcohol to underage youth was assessed using a standardi zed protocol (Komro et al., 2008). Women who were judged by a panel to be younger appearing (i.e. 20 years old or younger) attempted to purchase alcoholic beverages without age identific ation. Two purchase attempts were conducted at each randomly selected off-sale alcohol outlet (n = 326 outlets n = 652 attempts). The purchase attempt success rate was obtained by dividing the number of su ccessful purchase attempts by the total number of attempts for each community area. Area deprivation. An area deprivation index was created following procedures described by Singh (2003). Se venteen Census 2000 indicators were used: educational 108

PAGE 109

distribution (percentage of populatio n with less than 9 years and 12 or more years of education), unemployment rate, occupational co mposition, median family income, income disparity, median home value, median gross rent, median monthl y mortgage, home ownership rate, family poverty rate, population below 150% of poverty threshold, single-parent household rate, percentage of households without a motor vehicle, telephone, and/or complete plumbing, and household crowding. Factor score coefficien ts from Singh (2003) were used to weight the indicators. The scale was standardized, setting the mean and standard deviation to 100 and 20, respectively (Cronbachs alpha: 0.87; a higher score on this scale indicated greater area deprivation). Home and family management practices Home alcohol access. Three items from the student surv ey assessed the accessibility of alcohol from their homes and parents. Two items measured the ease with which students could obtain alcohol from their parent s and homes: How hard would it be for you to obtain alcohol from your parent or guardian? and How hard would it be for you to take it from your home?. Response options included hard, in-between, and easy. One item required students to identify the sources of their la st alcoholic beverage: If you have ever had an alcoholic drink, think back to the last time you drank. How di d you obtain the alcohol? Your parent or guardian gave it to you and You took it from home were the two response options included in this study. Parental monitoring/communication. Students responded to five items assessing their parental monitoring and communication: How often do/does you/your parent or guardian ask you about what you are doing in school ?; praise you when you do a good job?; eat dinner with a parent or guardian?; ask you where you are going or who you will be with?; and have a conversation with you that lasts 10 minutes or more?. Response options included: never, hardly ever, sometimes, a lot, and all the time. 109

PAGE 110

Alcohol-specific communication. Four items from the stude nt survey assessed alcoholspecific communication: How often does your pa rent or guardian talk with you about problems drinking alcohol can cause young peopl e?; family rules against young people drinking alcohol?; what would happen if you were caught drinking alcohol?; and Does your parent or guardian talk to you about how ad s and commercials are used to get you to buy things?. Response options includ ed: never, hardly ever, s ometimes, a lot, and all the time. Alcohol use Students responded to five items from the Monitoring the Future study (Johnston et al., 2008) that assessed alcohol use: During the la st 12 months, on how many occasions, or times, have you had alcoholic beverages to drink?; D uring the last 30 days, on how many occasions, or times, have you had alcoholic beverages to drink?; During the la st 7 days, on how many occasions, or times, have you had alcoholic bevera ges to drink?; Think back over the last 2 weeks, on how many times have you had five or more alcoholic drinks in a row?; and Have you ever become really drunk from drinking al coholic beverages so you fell down or became sick?. Response options for the past year, pa st month and past week items included: occasions, -2 occasions, -5 occasions, -9 occasions, -19 occasions, 0-39 occasions, and or more occasions. Response options for the heavy episodic use and having ever been drunk items included: never, once, twice, three to five times, six to nine times, and ten or more times. Analytical Strategy Structural equation modeli ng (SEM) in Mplus (version 5. 2; Muthn & Muthn, 2007) was used to assess the direct a nd indirect relations between al cohol-related neighborhood context at baseline (6th grade), home alcohol access and family management practices in 7th grade, and 110

PAGE 111

alcohol use in 8th grade. The SEM framework was partic ularly advantageous for the present study, as it analyzes relationshi ps between latent variables without random error (Bollen, 1989) and, in Mplus, allows direct and indirect e ffects to be estimated simultaneously (Muthn & Muthn, 2004). Analyses proceeded through two phases. First, measurement models were evaluated to determine the relationships between the observe d variables and underlying latent constructs. Multilevel exploratory factor anal yses (EFA) were conducted to de termine the appropriate factor structure for the home and family management and alcohol-related ne ighborhood context items. EFA, rather than confirmatory factor analysis (CFA), was used for these items because we did not have a priori hypotheses about the underlying factor structure for these data. A CFA was conducted for the alcohol use items, as we hypot hesized all would load on a single, Alcohol Use, factor. Three measurement models were f it, determining the factor structure for the alcohol-related neighborhood context, home and fam ily management practices, and alcohol use items separately. Community membership was specifi ed as a nested random effect to account for the dependency of observations among students within each community for each measurement model. All available data from each appropriate time point (6th, 7th or 8th grade) were used, with sample sizes ranging from 3801-4170, with 2.1% 0% and 0.03% missing data, respectively. Minimum variance weighted least squares (WLSMV ) was used for parameter estimation and an oblique, geomin factor rotation was specified. Fit of the measurement models was a ssessed with four goodness-of-fit indices: comparative fit index (CFI), Tucker-Lewis f it index (TLI), root mean square error of approximation (RMSEA) and standa rdized root mean square residual (SRMSR). The CFI and TLI describe the improvement in fit of the test ed model compared with that of a null model 111

PAGE 112

assuming zero covariance among the variables (Klin e, 2005). A value greater than 0.90 indicates reasonably good model fit (Hu & Bentler, 1999). The RMSEA is a parsimony-adjusted index, where a value < 0.05 indicates close approximate fit, values between 0.05 and 0.08 suggest reasonable fit, and values > 0.10 suggest poor model fit (Kline, 2005). The last index, the SRMSR, is a measure of the mean residual corr elation, where values < 0.10 are considered adequate (Kline, 2005). The second analysis phase tested structur al models specifying hypothesized causal relations among the identif ied constructs. The structural mo del was built in st ages, where the relations were modeled between (1) home and family management and alcohol use; (2) alcoholrelated neighborhood context a nd home and family manage ment; (3) alcohol-related neighborhood context and alcohol use; and (4) alcohol-related neighborhood context, home and family management, and alcohol use. Paths that were not statistically si gnificant and/or whose inclusion did not improve the fit of the mode l were excluded in each stage. Model fit was assessed with the CFI, TLI, and RMSEA. Multilevel analyses were conducted for the first three model building stages; however, the final model was estimated at the individual-level, given insufficient statistical power to estimate the most complex model at the community-level. The final model retained only statistically significant paths identified from the first three multilevel models. All regression paths were estimated while controlling for treatment group assignment. Direct effects on alcohol use in 8th grade were estimated while controlling for baseline levels of use. Indirect effects were calcu lated as the product of the regre ssion coefficients describing the effect of the independent variable on the hypothe sized mediator and the hypothesized mediator on the outcome. Sobels method (Sobel, 1982) was used for calculation of the standard errors of the indirect effects (Muthn & Muthn, 2004). 112

PAGE 113

Missing Data WLSMV estimation with categoric al and/or ordinal variables in Mplus uses pairwise deletion to handle missing data (Muthn & Muth n, 2004). Estimates are based on the polychoric correlations for all pairwise present data, wher e only missing values on the two variables under consideration are ignored, not the entire case. While maximum likelihood (ML) estimation is optimal for handling missing data (Schafer & Gr aham, 2002), it is not com putationally feasible when estimating more complex models with several latent variables (Muthn & Muthn, 2004), as was the case here. Seventy-two percent of the cohort students completed three or four surveys, while 28% completed one or two. Students who completed three to four surveys were more likely to be White ( 2 (5) =107.417, p < 0.001) and live with both parents ( 2 (1) = 37.887, p < 0.001), compared to those who only completed one or two surveys. There were no significant differences in alcohol use between those who completed three or four surveys and those completing one or two. Results Measurement Models Three measurement models were fit to dete rmine the factor structure for the alcoholrelated neighborhood context, home and family ma nagement practices, and alcohol use items. Table 5-1 shows the standardized, geomin-rotated loadings and the fit statistics for each model. The identified factor structures were consistent across all study time-points. Alcohol-related neighborhood context One factor, Neighborhood Strength, best described the covari ation among the items reported by community leaders and pare nts when the students were in 6th grade (Perceived Neighborhood Strength, Neighborhood and Polic e Preventive Action, and Perceived Neighborhood Problems; CFI = 1.000, TLI = 1.000, RMSEA < 0.01, SRMSR < 0.01). The other 113

PAGE 114

four alcohol-related neighborhood contextual items (exposure to alcohol advertisements, off-sale alcohol outlet density, commerci al alcohol accessibility, and ar ea deprivation) did not load sufficiently with the Neighborhood Stre ngth factor or with each othe r. Therefore, each of these items were included as separate, manifest variables in the structural model. Home and family management practices A two-factor solution best fit the home and family management practice data in 7th grade (CFI = 0.976, TLI = 0.965, RMSEA = 0.059, SRMSR = 0.063). The first factor, Home Alcohol Access, was defined by four items describing the perceived difficulty in getting alcohol from their homes and parents and receiving/taking alc ohol from their parents and homes during their last drinking occasion. While the loadings were low for the items describing receiving/taking alcohol from parents and homes (0.049 and 0.097, respectively), they were included in the model to provide a more comprehensive construct and the model fit when including these items was comparable to that when they were excluded (CFI = 0.974, TLI = 0.958, RMSEA = 0.077, SRMR = 0.042). The second, Family Manage ment, factor was de fined by nine items describing parental monitoring, general pare nt/child communication, and alcohol-specific communication. Alcohol use Once factor comprising all of the alcohol use items adequately fit the data (CFI = 0.995, TLI = 0.989, RMSEA = 0.10). Alcohol Use was defined by the five items assessing alcohol use in the past year, past mont h, past week, heavy episodic use, and having ever been drunk. Structural Model The final structural model is shown in Figure 5-2. All paths were estimated while controlling for treatment group membership. Fit in dices indicated good representation of the data (CFI = 0.974, TLI = 0.978, RMSEA = 0.031). When considering the other neighborhood 114

PAGE 115

constructs, area deprivation did not have any significant direct or indirect effects on alcohol use. Additionally, modeling it s correlations with the other alc ohol-related neighborhood constructs did not improve model fit (CFI = 0.904, TLI = 0.926, RMSEA = 0.061). Therefore, it was excluded from the final model. Significant correlations among the latent and ma nifest factors were observed. At baseline (6th grade): neighborhood stre ngth showed an inverse associat ion with alcohol outlet density ( r = -0.436, p < 0.001) and commercial alcohol access ( r = -0.040, p < 0.01); alcohol outlet density was positively associated with commercial alcohol access ( r = 0.214, p < 0.001) and exposure to alcohol advertisements ( r = 0.036 p < 0.05); and commercial alc ohol access was negatively associated with exposure to alcohol advertisements ( r = -0.080, p < 0.001). In 7th grade, home alcohol access and protective family management practices were inversely associated ( r = 0.462, p < .001). Baseline neighborhood strength was negatively, and exposure to alc ohol advertisements positively, associated with alcohol use in 8th grade ( = -0.082, p < 0.05 and = 0.049, p < 0.05, respectively), after controlling for baseline alcohol use. Alcohol outlet density and commercial alcohol access did not have sta tistically significant direct effects on alcohol use in 8th grade. Neighborhood strength and commercial al cohol access were associated with home alcohol access ( = 0.063, p < 0.05 and = -0.124, p < 0.001, respectively) and family management practices ( = -0.070, p < 0.01 and = 0.073, p < 0.001, respectively) in 7th grade. Alcohol outlet density and exposure to alcohol advertisements did not have a statistically significant effect on home alcohol access or protective family management practices. Home alcohol access showed a positive association with alcohol use ( = 0.399, p < 0.001) in 8th grade, while the association between pr otective family management practices and 115

PAGE 116

alcohol use was not significant when home alcohol access was considered. Tests for indirect effects suggest that home al cohol access may partially mediate the relations between neighborhood strength and alcohol use, although this indirect effect was only marginally significant ( = 0.025, p = 0.062). Discussion This study used SEM to examine the direct a nd indirect relations between alcohol-related neighborhood context, home and family manage ment practices, and alcohol use among a large sample of inner-city, r acial/ethnic minority, young adolescent s. Significant correlations were observed among the alcohol-related neighborhood contextual factor s (i.e., neighborhood strength, alcohol outlet density, commercial alcohol access, and exposure to alcohol advertisements) and among the home and family ma nagement factors (i.e., home alcohol access and protective family management practices). Of particular interest were the large correlations between neighborhood strength and alcohol outlet density, alcohol outlet density and commercial alcohol access, and home alcohol access and protective family management practices. These findings suggest that efforts to minimize alcohol-re lated risk and enhance protective factors (i.e., neighborhood strength, protective family mana gement practices) should be multifaceted, addressing both communityand family-lev el exposure and access to alcohol. Two alcohol-related neighborhood constructs ha d significant, direct relations with alcohol use: increased neighbor hood strength was associated w ith decreased alcohol use and increased exposure to alcohol advertisements was associated with increased alcohol use. Neighborhood strength was positively, and commer cial alcohol access negatively, associated with home alcohol access; while neighborhood strength was negatively, and commercial alcohol access positively, associated with protective fam ily management practices. Increases in home alcohol access were associated w ith increased alcohol use, while increased protective family 116

PAGE 117

management practices was associated with d ecreased alcohol use, albeit not statistically significant. Tests for indirect effects suggested that the pr otective effect of neighborhood strength on alcohol use may be partially reduced if children are exposed to increased alcohol access in the home. The direction of effects for neighborhood strength on protective family management practices and home alcohol access, and commercia l alcohol accessibility on home alcohol access were opposite to those hypothesized, and count erintuitive. The positive relation between commercial alcohol access and protective fam ily management was hypothesized based on previous studies (Beyers et al ., 2003; Brook et al., 1989; Chuang et al., 2005; Rankin & Quane, 2002; Tobler et al., 2007). Together these findi ngs support the hypothesis th at inner-city parents respond to environmental risk, such that as neighborhood risk increa ses (i.e., less neighborhood strength, greater commercial alcohol access), prot ective family management practices increase in addition to decreases in home alcohol access. Thes e findings are consistent with other literature suggesting that parents may bu ffer the effects of risky envi ronments (Beyers et al., 2003; Chuang et al., 2005; Rankin & Quane, 2002), especially during the early adol escent years. Future research should examine whether this bu ffering capacity holds as youth progress through adolescence, becoming increasingly more a part of, and exposed to, their neighborhood environment (Ingoldsby & Shaw, 2002). Given that alcohol use initia tion peaks in early adolescence (Kosterman, Hawkins, Guo, Catalano, & Abbott, 2000; National Center fo r Chronic Disease Prevention and Health Promotion, 2008) and the considerable consequen ces associated with use during this critical developmental period (Grant et al., 2001; Gr uber et al., 1996; Hingson et al., 2002; Hingson, Heeren, Winter et al., 2003; Hingson et al., 2001 ; Hingson, Heeren, Zakocs et al., 2003; Hingson 117

PAGE 118

et al., 2006), preventive efforts targeting young adolescents are important. These findings highlight parent engagement in restricting alcohol access and improving family management practices as key components to preventive e fforts to reduce alcohol use among inner-city, adolescents. Here, the effects of home alcohol access on alcohol use were approximately four times the others considered, consistent with scientific theory regarding more proximal influences on behavior (Flay & Petraitis, 1994) and with other lite rature describing subs tantial increases in risk when alcohol is available or provided at home (Jackson et al., 1999; Resnick et al., 1997; Swahn & Hammig, 2000). Thus, efforts to enga ge and improve parental home and family management practices may be fruitful. Neighborhood strength and exposure to alcohol advertisements in 6th grade were directly and significantly associated with alcohol use in 8th grade, even after controlling for baseline levels of use and considering two prominent, proxi mal predictors of alcohol use. As expected the magnitude of these effects was considerably sma ller for these distal influences; however, they suggest that community characteristics are influential in shap ing alcohol use behaviors among youth. These findings are consistent with othe r studies that have obs erved significant direct effects on alcohol use (Duncan et al., 2002; Pasch et al., 2007; Scheier et al., 2000; Snyder et al., 2006), and suggest that incorporating community -level intervention components that build neighborhood strength and limit exposure to alco hol advertisements ma y enhance intervention effects. This is consistent with extant scientific th eory (e.g., Flay & Petraitis, 1994; Szapocznik & Coatsworth, 1999; Wagenaar & Perry, 1994) acknowledging multiple dimensions that influence adolescent behavior. This study had several limitations. First, the sample for this study comprised only young adolescents, aged 11 to 14 years. The salien cy of context in shaping alcohol use among youth 118

PAGE 119

may vary throughout adolescence (Szapocznik & Coatsworth, 1999). Accordingly, future research should examine associations between alcohol-related context a nd drinking behaviors of youth as they evolve and develop across time. Our data precluded such an examination. Second, the sample for this study were low-income, r acial/ethnic minority, young a dolescents residing in Chicago, Illinois. More studies are needed to examine the consistency of the relations presented here among youth residing in other metropolitan cities as well as rural and suburban areas. Third, given the complexity of the model and sample size, we did not split the sample and conduct independent exploratory and conf irmatory analyses. However, observed effects are similar to other studies among racial/ethnic minority youth examining components of the model identified here, and may be generalizable to other racia l/ethnic minority youth livin g in other metropolitan or rural areas (Beck, Shattuck, Haynie, et al., 1999; Beck & Treiman, 1996; Borawski, et al., 2003; Cleveland et al., 2005; Ke gler et al., 2005; Rankin & Quane, 2002; Sellstrom & Bremberg, 2006). Lastly, measures of al cohol-related neighborhood contex t used do not represent the universe of neighborhood charac teristics which may also influence home and family management and alcohol use among youth. Future research should examine the influence of more broadly defined neighborhood contexts, including additional community measures, such as crime rates, political activism, public polic ies and measures of social structure. Limitations notwithstanding, th is study contributes to a spar se literature describing the etiology of alcohol use among urban, racial/ethni c minority youth, particularly the effects of alcohol-related neighborhood context on home and fa mily management practices and alcohol use. Moreover, multiple dimensions of alcohol-re lated neighborhood context were considered in the analyses, including direct environmental assessments, Census 2000 data, and self-report measures from parents and community leaders. This is a notable strength, as much of the 119

PAGE 120

120 literature describing the influe nce of neighborhood context on drug use and other deleterious health and social outcomes has relied solely on census data (Allison et al., 1999; Chuang et al., 2005; Galea et al., 2007) or self-report measures (Hill & Angel, 2005). Further, the study design allowed for establishment of clear temporal precedence, a great advantage over cross-sectional modeling. The results showed significant direct and indirect associa tions between neighborhood context and alcohol use, and sugge st that inner-city parents re spond to environmental risk and represent a key target for intervention to reduce alcohol use among inner-city adolescents, whether it be through restricting alcohol acce ss in their homes or improving monitoring and communication with their children.

PAGE 121

Table 5-1. Standardized, geomin-rotated factor loadings and fit statistics for measurement models. Model 1 (n = 4170) Model 2 (n = 3778) Model 3 (n = 3801) Item Neighborhood Strength Home Alcohol Access Protective Family Management Alcohol Use Alcohol-related Neighborhood Context Perceived neighborhood strength 0.737 Neighborhood and police preventive action 0.866 Perceived neighborhood problems -0.373 Home and Family Management Last time drank, received alcohol from parent 0.049 -0.010 Last time drank, took alcohol from home 0.097 0.072 Easy to get alcohol from parent 0.793 0.320 Easy to get alcohol from home 0.783 0.381 Parent ask about school 0.206 0.699 Parent praise when do a good job 0.189 0.658 Eat dinner with parent 0.202 0.485 Parent ask who with 0.249 0.571 Parent/child conversations 0.198 0.671 Parent talk about problems alcohol can cause 0.430 0.790 Parent talk about family rules against drinking 0.443 0.542 Parent talk about consequences of drinking 0.455 0.735 Parent talk about influence of ads and commercials 0.362 0.611 Alcohol Use and Intentions Past year alcohol use 0.885 Past month alcohol use 0.972 Past week alcohol use 0.888 Heavy episodic alcohol use 0.857 Ever been drunk 0.777 Fit Indices CFI 1.000 0.976 0.984 TLI 1.000 0.965 0.989 RMSEA < 0.001 0.059 0.102 SRMSR < 0.001 0.063 121

PAGE 122

Figure 5-1. Hypothesize d structural model. 122

PAGE 123

123 Figure 5-2. Structural mode l depicting standardized path s among alcohol-related neighborhood context, home and family management pr actices, and early adol escent alcohol use. (Nonsignificant paths are indi cated with dashed line.)

PAGE 124

CHAPTER 6 DISCUSSION Participants in this study resided in 42 of 77 city-defined communities in Chicago, Illinois. Among these communities, there was co nsiderable racial/ethnic heterogeneity and variability in several census-bas ed indicators of deprivation. Fo r example, the median family income ranged from $14,476 to $133,832 per community, unemployment ranged from 19% to 58% and the proportion of the population below 150% of the poverty threshold ranged from 6% to 69% per community (Table 2-6) Further, nearly half of the communities were relatively racially/ethnically homogeneous; among 14 co mmunities, over 80% of the population was African American and among 6 communities, 80% of the population was Hispanic (Table 2-5). Youth in Chicago, and those of other large, metropolitan cities, face a myriad of distinct challenges related to their environment (Griffin, Botvin, Epstein, Doyle, & Diaz, 2000). For instance, ecological studies have shown that crime, delinquency, drug use, public disorder, and school dropout are significantly clustered in ur ban communities (Coulton et al., 1995; Duncan et al., 2002; Sampson, 1992). In 2006, the rates per 100,000 population in Chicago were higher than those across the United States for murders, ra pes, robberies, aggravat ed assaults, burglaries, and thefts (MDNH, 2008). Moreover, in 2007 youth in Chicago report higher rates of marijuana (44% Chicago, 38% U.S.), heroin (3.7% Chicago, 2.3% U.S.) and ecstasy use than youth across the United States (6.4% Chicago, 5.8%, U.S.; Ce nters for Disease Control and Prevention, 2008). Alcohol use among youth in Chicago is similar to that found nationwide (Centers for Disease Control and Prevention, 2008). Gangs are also problematic. There are approximately 120 street gangs with an estimated 100,000 to 130,000 members in Chicago. Of these, it is estimated that more than 40 percent are under the age of 18, with many joini ng as early as age 11 (Risley, 2004). 124

PAGE 125

Given the social milieu among youth in this study, their alcohol-specific contexts contribute to environments that may place them at elevated risk for deleterious social and behavioral outcomes. This study sought to define the heterogeneous alcohol -related contexts of a large sample of urban, racial/ethnic minority youth and examine how exposure to these contexts related to alcohol use and inten tions. It also examined the role of home and family management practices in mediating these re lationships. These goals were a ccomplished through completion of three distinct empirical investigations. Paper 1 (Chapter 3) examined patterns of alcohol-related neighborhood characteristics of the 42 urban community areas in which the sample resided using multilevel latent class analyses (LCA). Five heterogeneous classes of neighborhoods were identified and characterized by extant social capital and exposure and access to al cohol. Twenty percent of youth resided in communities with low social capital and high access to alcohol; 24% resided in communities with low social capital and low exposure and access to alcohol; 30% resided in communities with moderate social capital and high access to alcohol; 20% reside d in communities with moderate social capital and mode rate exposure to alcohol; and 6% resided in communities with high social capital and moderate exposure a nd high access to alcohol. There was considerable heterogeneity in the racial/ethni c distribution of the classes. For example, African American youth were more likely to live in neighborhoods with high access to alcohol. The majority of both African American and Hispan ic youth in the sample resided in communities characterized by low social capital, highlighting the disadvantage of these inner-city communities and the resource disparity across these racial/ethnic groups. Paper 2 (Chapter 4) used the latent classes identified in Paper 1 to examine the direct effects of alcohol-related nei ghborhood risk class membership on the trajectories of alcohol use 125

PAGE 126

and intentions. None of the ne ighborhood risk classes were signi ficantly associated with the trajectories of alcohol use/intenti ons relative to the othe r classes for either African Americans or Hispanics. Among the Hispanic youth, two contrasts approached marginal significan cerelative to the low social capital/low exposure/low access to alcohol class (Class 2), Hispanic youth in the moderate social capital/low exposure/hi gh access to alcohol and moderate social capital/moderate exposure/low access to alcohol classes were less likely to report alcohol use/intentions over time. However, these eff ects were not statisti cally significant. Given the insignificant effects in Paper 2 when using the more broadly-defined, latent risk class indicator to predict alcohol use/intent ions over time and the co mplex interpretation of effects without a low/no risk clas s to serve as a referent class, Paper 3 (Chapter 5) began with factor analyses to determine the number of la tent constructs among the nine indicators of alcohol-related neighborhood risk. Structural Eq uation Modeling (SEM) was then used to examine the direct and indirect relationship s between alcohol-rela ted neighborhood context, home and family management practices, and alc ohol use. Significant correlations were observed among the alcohol-related neighborhood contextual factors (i.e., neighborhood strength, alcohol outlet density, commercial alcohol access, and exposure to alcohol advertisements) and among the home and family management factors (i.e ., home alcohol access and protective family management practices). Two alcohol-related ne ighborhood constructs had significant, direct relationships with alcohol use: increased neighborhood strength was associated with decreased alcohol use and increased exposur e to alcohol advertisements wa s associated with increased alcohol use. Neighborhood strength was positively and commercial alcohol access was negatively associated with home alcohol access; whereas neighborhood strength was negatively and commercial alcohol access was positively asso ciated with protective family management 126

PAGE 127

practices. Increases in home alcohol access were associated with increased alcohol use, while increased protective family management practices was associated with decreased alcohol use, albeit not statistically significant. Tests for indire ct effects suggested that the protective effect of neighborhood strength on alcohol use may be pa rtially reduced if ch ildren are exposed to increased alcohol access in the home. Collectively, findings from these investigations have important implications for scientific theory and preventive interventions targeting alcohol use among racial/ethnic minority youth in urban cities. First, while ther e was substantive heterogeneity in alcohol-related neighborhood risk, membership in the alcohol-related nei ghborhood risk classes defined using the multilevel LCA did not significantly affect the trajectories of alcohol use/inte ntions. However, the structural equation model suggested that baseline ne ighborhood strength and exposure to alcohol advertisements were significant predictors of alcohol use in 8th grade. These findings, together with a mixed literature on the effects of ne ighborhood context on alcoho l use (Allison et al., 1999; Brook et al., 1989; Chuang et al., 2005; Crum et al., 1996; Duncan et al., 2002; Elliott et al., 1996; Galea et al., 2007; Gibbons et al., 2004 ; Hill & Angel, 2005), suggest that observed effects are particularly sensitive to measurement and unit of analysis. Thus, null effects when examining these and other similar relations should not be construed as unimportant for either theory or preventive efforts. More research is needed to further elucid ate the role of alcoholrelated neighborhood contexts and refine hypothesized direct effects, as we ll as define the best measurement and methodological approaches for examining these complex relations. Findings from this study suggest that it may be more appropriate to individually c onsider the influence of specific neighborhood charac teristics rather than using a more broadly defined, composite indicator of risk. 127

PAGE 128

Second, the observed direct effects among th e alcohol-related neighborhood contextual factors and alcohol use were small in compar ison to those for more proximal factors. For example, the effects of neighborhood strength and exposure to alcohol advertisements in 6th grade on alcohol use in 8th grade were approximately one-fourth those for home alcohol access and protective family management practices. This is consistent with extant scientific theory (e.g., Flay & Petraitis, 1994; Szapocznik & Coatswor th, 1999; Wagenaar & Perry, 1994), as well as other research (Hawkins et al., 1992) and sugge sts that community-level characteristics are important and influential in shaping alcohol use behaviors among youth, albeit with smaller effects than interand intra-personal predictors of use. Therefore, incorporating community-level intervention components that build neighbor hood strength and limit exposure to alcohol advertisements may enhance the effects of pr eventive interventions targeting more proximal influences on alcohol use behavior. Still, given the magnitude of effects, efforts focusing on these two distal factors alone may be insuffici ent to produce sustained, protective effects. Further, given the large correlations between neighborhood strength and alcohol outlet density, alcohol outlet density and commercial alcohol access, and home alcohol access and protective family management practices, efforts to mini mize alcohol-related risk and enhance protective factors (i.e., neighborhood streng th, protective family manage ment practices) should be multifaceted, addressing both communityand fa mily-level exposure and access to alcohol, respectively. Third, in this study the area deprivation s cale did not have any significant direct or indirect effects on alcohol use or protective home and family management practices when considered with the other alcohol-related neighborhood contextual factors (Effects were statistically significant in the structural equati on model when considered alone.). Deprivation has 128

PAGE 129

been significantly associated with a number of deleterious outcomes (Boardman & Saint-Onge, 2005; Scheier et al., 2000). Yet, these findings su ggest that exposure and access to alcohol and neighborhood strength are more prom inent predictors of alcohol use. This is a logical conclusion given these measures were specific to the behavior considered and is an encouraging finding for prevention scientists, as these constructs are more adept to change than poverty, unemployment, housing, and the like (Tur ner, 1998; Wasworthx, 1997). Lastly, findings suggest that i nner-city parents respond to environmental risk, such that as neighborhood risk increases (i.e., less neighbor hood strength, greater commercial alcohol access), protective family management practices increase and home alcohol access decrease. These findings are consistent with other litera ture suggesting that pa rents may buffer the effects of risky environments (Beyers et al ., 2003; Chuang et al., 2005; Rankin & Quane, 2002), especially during the early adolescent years. Thus, parental alcohol-specific and family management behaviors warrant consideration in sc ientific theory and preventive interventions for urban youth. Resnicow and colleagues (2002) describe the need to cultura lly-tailor scientific theory and its applications, as many of our broadl y defined scientific theories may not be robust to variations in racial/ethnic composition and context of the population. However, these results are consistent with a body of literature identif ying parents as key cont ributors to alcohol use behaviors among racial/ethnic majority and mi nority youth alike (Beck et al., 1999; Beck & Treiman, 1996; Borawski et al., 2003; Brody et al., 2006; Komro et al., 2006; Komro et al., 2008; Spoth, Shin, Guyll, Redmond, & Azevedo, 2006). This suggests that the need to engage parents in prevention activities ta rgeting alcohol abuse may be robus t to variations in cultural and geographic context. However, more research is needed to refine scientific theory and determine 129

PAGE 130

how home and family management practices differ and how best to engage parents in prevention strategies across these social and environmental contexts. This study is not without its limitations. Firs t, community-level data included only static measures of neighborhood risk and protective f actors. Ecodevelopmental theory (Szapocznik & Coatsworth, 1999) suggests that youth and their c ontexts evolve across time. Thus, both the number and interpretation of neighborhood latent classes in which the sample resided could vary across time, as could the factors id entified in Paper 3. The data suggest that these alcohol-related risk indicators remained relatively stable as youth progressed from 6th to 8th grade. However, our data precluded a parallel proce ss examination of these contextual changes and alcohol use and intentions. Future research should examine how contexts of youth evolve across time and how this change covaries with adolescent alcohol use. Second, the sample for this study include d low-income, racial/ethnic minority, young adolescents residing in Chicago, Illinois. More studies are needed to examine the associations presented here among youth residing in other metr opolitan cities as well as rural and suburban areas. Effects may not be consistent across diffe ring economic and cultural contexts. However, observed effects are similar to other studies among racial/ethnic minority youth examining components of the model identified here, and may be generalizable to other racial/ethnic minority youth living in other metr opolitan or rural areas (Cleve land et al., 2005; Kegler et al., 2005; Rankin & Quane, 2002; Sellstrom & Bremberg, 2006). Lastly, measures of alcohol-related neighbor hood context used in this study do not represent the universe of neighborhood characteristics which may also influence home and family management and alcohol use among youth. Fu ture research should examine the influence 130

PAGE 131

131 of more comprehensively de fined neighborhood contexts, in cluding additional community measures, such as crime rates, po litical activism, public policies and measures of social structure. Limitations notwithstanding, th is study contributes to a spar se literature describing the contexts and etiology of alcohol use among urban, racial/ethnic minority youth. Moreover, multiple dimensions of alcohol-related neighborhood context were considered, including direct environmental assessments, Census 2000 data, a nd self-report measures from parents and community leaders. This is a notable strength, as much of the literature describing the influence of neighborhood context on drug use a nd other deleterious health and social outcomes has relied on census data (Allison et al., 1999; Chuang et al., 2005; Galea et al., 2007) or self-report measures (Hill & Angel, 2005). Further, the study design allowed for establishment of clear temporal precedence, a great advantage over cros s-sectional modeling. La stly, advanced, stateof-the-science methods were implemented for each investigation. Findin gs showed substantive heterogeneity in the alcohol-related neighborhood contexts in which the sample resided, considerable variation in risk across racial/ethnic subgroup and significant di rect and indirect associations between alcohol-re lated neighborhood context and al cohol use. Further, they suggest that multifaceted, community-level compone nts of alcohol use preventive interventions may be fruitful avenues to reduce alcohol use am ong urban youth in concert with strategies to engage parents in restricting alcohol acce ss in the home or improving monitoring and communication with children.

PAGE 132

APPENDIX A 2002 STUDENT SURVEY 132

PAGE 133

133

PAGE 134

134

PAGE 135

135

PAGE 136

136

PAGE 137

137

PAGE 138

138

PAGE 139

139

PAGE 140

140

PAGE 141

141

PAGE 142

142

PAGE 143

143

PAGE 144

144

PAGE 145

145

PAGE 146

146

PAGE 147

147

PAGE 148

APPENDIX B 2002 PARENT SURVEY 148

PAGE 149

149

PAGE 150

150

PAGE 151

151

PAGE 152

152

PAGE 153

153

PAGE 154

154

PAGE 155

155

PAGE 156

APPENDIX C 2002 COMMUNITY LEADER SURVEY 156

PAGE 157

157

PAGE 158

158

PAGE 159

159

PAGE 160

160

PAGE 161

161

PAGE 162

162

PAGE 163

163

PAGE 164

164

PAGE 165

APPENDIX D ALCOHOL PURCHASE ATTEMPT PROTOCOL 165

PAGE 166

166

PAGE 167

167

PAGE 168

168

PAGE 169

169

PAGE 170

170

PAGE 171

171

PAGE 172

172

PAGE 173

173

PAGE 174

174

PAGE 175

175

PAGE 176

APPENDIX E ALCOHOL ADVERTISEMENT ASSESSM ENT DATA COLLECTION AND CODING PROTOCOL 176

PAGE 177

177

PAGE 178

178

PAGE 179

179

PAGE 180

180

PAGE 181

181

PAGE 182

182

PAGE 183

183

PAGE 184

184

PAGE 185

185

PAGE 186

186

PAGE 187

187

PAGE 188

188

PAGE 189

189

PAGE 190

190

PAGE 191

191

PAGE 192

LIST OF REFERENCES Aas, H., & Klepp, K. (1992). Adolescents alcohol-use related to perceived norms. Scandinavian Journal of Psychology, 33 (4), 315-325. Agresti, A. (2007). An introduction to categorical data analysis, second edition Hoboken, NJ: John Wiley & Sons, Inc. Aizer, A. (2004). Home alone: Supervis ion after school and child behavior. Journal of Public Economics, 88(9/10), 1835-1848. Ajzen, L. (1985). From decisions to actions: A th eory of planned behavior. In J. Kuhl & J. Beckmann (Eds.), Action control: From cognition to behavior Heidelberg, Germany: Springer. Ajzen, L. (1988). Attitudes, personality, and behavior Chicago, IL: Dorsey Press. Akaike, H. (1973). Information theory and an ex tension of the maximum likelihood principle. In B. N. Petrov & F. Csaki (Eds.), Second international symposium on information theory (pp. 267-281). Budapest, Hungary: Akademiai Kaido. Allison, K., Crawford, I., Leone, P., Trickett, E., Perez-Febles, A., Burton, L., et al. (1999). Adolescent substance use: Preliminary exam ining of school and neighborhood context. American Journal of Community Psychology, 27 (2), 111-141. Alvarez, J., Martin, A., Vergeles, M., & Mar tin, A. (2003). Substance use in adolescence: Importance of parental warmth and supervision. Psicothema, 15 (2), 161-166. Arkes, J. (2007). Does the econom y affect teenage substance use? Health Economics, 16 19-36. Ary, D., Tildesley, E., Hops, H., & Andrews, J. (1993). The influence of parent, sibling and peer modeling and the attitudes on adolescent use of alcohol. International Journal of the Addictions, 29(9), 853-880. Asparouhov, T., & Muthn, B. (2006). Multilevel mixture models, version 3 Los Angeles, CA: Muthn & Muthn. Bagnardi, V., Biangiardo, M., La Vecchia, C., & Corrao, G. (2001). Alcohol consumption and the risk of cancer: a meta-analysis. Alcohol Research and Health, 25 (4), 263-270. Barnow, S., Schultz, G., Lucht, M., Ulrich, I ., & Freyberger, H. (2004) The relation between temperament, alcohol expectancies, peer-g roup and alcohol consumption as well as alcohol-related problems in adolescence. Alcohol & Alcoholism, 39 (3), 213-219. Beck, K., & Lockhart, S. (1992). A model of pa rental involvement in adolescent drinking and driving. Journal of Youth and Adolescence, 21 (1), 35-51. 192

PAGE 193

Beck, K., Shattuck, T., Haynie, D., Crump, A ., & Simons-Morton, B. (1999). Associations between parent awareness, monitoring, en forcement and adolescent involvement with alcohol. Health Education Research, 14 (6), 765-775. Beack, K., & Treiman, K. (1996). The relationship of social context of drinking, perceived social norms, and parental influence to various drinking patterns of adolescents. Addictive Behaviors, 21 (5), 633-644. Becker, M. (1974). The health belief model and personal health behavior Thorofare, NJ: Slack. Bergmark, K. H., & Andersson, T. (1999). Development of advanced drinking habits in adolescence: A longitudinal study. Substance Use & Misuse, 34 171-194. Bernstein, K., Galea, S., Ahern, J., Tracy, M ., & Vlahov, D. (2007). The built environment and alcohol consumption in urban neighborhoods. Drug and Alcohol Dependence, 91 244252. Best, D., Manning, V., Gossop, M., Gross, S., & Strang, J. (2006). Excessive drinking and other problem behaviours among 14-16 year old school children. Addictive Behaviors, 31 1424-1435. Beyers, J., Bates, J., Pettit, G., & Dodge, K. (2003). Neighborhood structure, parenting processes, and the development of youths' exte rnalizing behaviors: A multilevel analysis. American Journal of Community Psychology, 31 (1/2), 35-53. Bjarnason, T., Andersson, B., Choquet, M., Elekes, Z., Morgan, M., & Rapinett, G. (2003). Alcohol culture, family structure and adoles cent alcohol use: Mu ltilevel modeling of frequency of heavy drinking among 15 to 16 year old students in 11 European countries. Journal of Studies on Alcohol, 64(2), 200-208. Blount, W., Silverman, I., Sellers, C., & Seese, R. (1994). Alcohol and drug-use among abused women who kill, abused women who don't, and their abusers. Journal of Drug Issues, 24(1-2), 165-177. Boardman, J., & Saint-Onge, J. (2005). Neighborhoods and adolescent development. Children, Youth and Environments, 15 (1), 138-164. Bollen, K. (1989). Structural equations w ith latent variables New York City, NY: John Wiley & Sons. Borawski, E., Ievers-Landis, C., Lovegreen, L ., & Trapl, E. (2003). Parental monitoring, negotiated unsupervised time, and parental trust: The role of perceived parenting practices in adolescent health risk behaviors. Journal of Adolescent Health, 33 (2), 60-70. Borowsky, I., Ireland, M., & Resnick, M. ( 2001). Adolescent suicide attempts: Risks and protectors. Pediatrics, 107 (3), 485. 193

PAGE 194

Bot, S., Engles, R., & Knibbe, R. (2005). The effects of alcohol e xpectancies on drinking behavior in peer groups: Observa tions in a natura listic setting. Addiction, 100 (9), 12701279. Britt, H., Carlin, B., Toomey, T., & Wagenaar, A. (2005). Neighborhood level spatial analysis of the relationship between alcohol outlet density and criminal violence. Environmental and Ecological Statistics, 12 (4), 411-426. Brody, G., Murry, V., Kogan, S., Gerrard, M., Gibbons, F., Molgaard, V., et al. (2006). The Strong African American Families program: A cluster-randomized prevention trial of long-term effects and a mediational model. Journal of Consulting and Clinical Psychology, 74 (2), 356-366. Bromley, R., & Nelson, A. (2002). Alcohol-rela ted crime and disorder across urban space and time: Evidence from a British city. Geoforum, 33(2), 239-254. Brook, J., Nomura, C., & Cohen, P. (1989). A network of influences on adolescent drug involvement: Neighborhood, school, peer and family. Genetic, Social and General Psychology Monographs, 115 (1), 123-145. Brown, S., Tapert, S., Gra nholm, E., & Delis, D. (2000). Neurocognitive functioning of adolescents: Effects of protracted alcohol use. Alcoholism: Clinical and Experimental Research, 24 (2), 164-171. Bryant, A., Schulenberg, J., O'Malley, P., Bachman, J., & Johnston, L. (2003). How academic achievement, attitudes, and behaviors relate to the course of substance use during adolescence: A 6-year, multiwave national longitudinal study. Journal of Research on Adolescence, 13 (3), 361-397. Caetano, R., Clark, C., & Tam, T. (1998). Al cohol consumption among racial/ethnic minorities: Theory and research. Alcohol World: Health & Research, 22 (4), 233-241. Callas, P., Flynn, B., & Worden, J. (2004). Po tentially modifiable psychosocial factors associated with alcohol use during early adolescence. Addictive Behaviors, 29 (8), 15031515. Centers for Disease Contro l and Prevention. (2008). Youth Risk Behavior Survey Atlanta, GA: Centers for Disease Control and Preventi on, National Center for Chronic Disease Prevention and Health Promotion. Christensen, P., Henriksen, L., & Roberts, D. (2000). Substance use in popular prime-time television Washington, D.C.: Office of National Drug Control Policy. Chuang, Y., Ennett, S., Bauman, K., & Fo shee, V. (2005). Neighborhood influences on adolescent cigarette and alcohol use: mediating effects throug h parent and peer behaviors. Journal of Health and Social Behavior, 46 (2), 187-204. 194

PAGE 195

Clark, D., Thatcher, D., & Maisto, S. (2005). Supervisory neglect and adolescent alcohol use disorders: Effects on AUD onset and treatment outcome. Addictive Behaviors, 30(9), 1737-1750. Cleveland, M., Gibbons, F., Gerrard, M., Pmer y, E., & Brody, G. (2005). The impact of parenting on risk cognitions and risk behavior: A study of mediation and moderation in a panel of African American adolescents. Child Development, 76(4), 900-916. Coleman, J. (1994). Foundations of social theory Cambridge, MA: Harvard University Press. Coleman, J. (2007). Social capital in the creation of human capital. The American Journal of Sociology, 94 S95-S120. Coley, R., Morris, J., & Hernandez, D. ( 2004). Out-of-school care and problem behavior trajectories among low-income adolescents : Individual, family and neighborhood characteristics as added risks. Child Development, 75 (3), 948-965. Collins, B., Wileyto, E., Murphy, M., & Munafo, M. (2007). Adolescent environmental tobacco smoke exposure predicts academic achievement test failure. Journal of Adolescent Health, 41(4), 363-370. Collins, L., Fidler, P., Wugalter, S., & Long, J. (1993). Goodness-of-fit test ing for latent class models. Multivariate Behavioral Research, 28 375-389. Collins, R., Ellickson, P., McCaffrey, D., & Hambarsoomians, K. (2007). Early adolescent exposure to alcohol advertising and it s relationship to underage drinking. Journal of Adolescent Health, 40 (6), 527-534. Cook, R., & Clark, D. (2005). Is there an asso ciation between alcohol consumption and sexually transmitted diseases? A systematic review. Sexually Transmitted Diseases, 32 156-164. Corrao, G., Bagnardi, V., Zambon, A., & La V ecchia, C. (2004). A meta-analysis of alcohol consumption and the risk of 15 diseases. Preventive Medicine, 38 (5), 613-619. Corrao, G., Rubbiati, L., Bagnardi, V., Zambon, A., & Poikolainen, K. (2000). Alcohol and coronary heart disease: a meta-analysis. Addiction, 95 (10), 1505-1523. Coulton, C., Korbin, J., Su, M., & Chow, J. (1995). Community level factors and child maltreatment rates. Child Development, 66 1262-1276. Crum, R. M., Lillie-Blanton, M., & Ant hony, J. (1996). Neighborhood environment and opportunity to use cocaine and other drugs in late childhood and early adolescence. Drug and Alcohol Dependence, 43 155-161. Curran, P., Stice, E., & Chassin, L. (1997). The relation between adolescent alcohol use and peer alcohol use: A longitudinal random -effect coefficients model. Journal of Consulting and Clinical Psychology, 65 (1), 130-140. 195

PAGE 196

D'Amico, E., & McCarthy, D. (2006). Escalation and initiation of younger a dolescents' substance use: The impact of perceived peer use. Journal of Adolescent Health, 39 (4), 481-487. Darkes, J., Greenbaum, P., & Goldman, M. (2004). Alcohol expect ancy mediation of biopsychosocial risk: Complex patterns of mediation. Experimental and Clinical Psychopharmacology, 12 (1), 27-38. Dempster, A., & Rubin, D. (1977). Maximum like lihood estimation from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, 39 1-38. Dent, C., Grube, J., & Biglan, A. (2005). Commun ity level alcohol availa bility and enforcement of possession laws as predictors of youth drinking. Preventive Medicine, 40 355-362. DeWit, D., Adlaf, E., Offord, D., & Ogborne, A. (2000). Age at first alcohol use: A risk factor for the development of alcohol disorders. American Journal of Psychiatry, 157 (5), 745750. Dielman, T., Butchart, A., & Shope, J. (1993). Structural equation model tests of patterns of family interaction, peer alcohol use, and in trapersonal predictors of adolescent alcohol use and misuse. Journal of Drug Education, 23 (3), 273-316. Duncan, S., Duncan, T., & Strycker, L. (2002). A multilevel analysis of neighborhood context and youth alcohol and drug problems. Prevention Science, 3 (2), 125-133. Duncan, T., Duncan, S., Strycker, L., Okut, H., Li, F., & Hix-Small, H. (2002). Growth mixture modeling of adolescent alcohol use data Eugene, OR: Oregon Research Institute. Dunn, M., Bartee, R., & Perko, M. (2003). Self -reported alcohol use and sexual behaviors of adolescents. Psychological Reports, 92 (1), 339-348. Eaton, D., Kann, L., Kinchen, S., Ross, J., Hawk ins, J., Harris, W., et al. (2005). Youth risk behavior surveillance --United States, 2005. MMWR, 55 (SS-5), 1-108. Ellickson, P., Collins, R., Hambarsoomians, K., & McCaffrey, D. (2005). Does alcohol advertising promote adolescen t drinking? Results from a longitudinal assessment. Addiction, 100 (8), 1201-1201. Ellickson, P., Tucker, J., & Klein, D. (2003). Ten-year prospective study of public health problems associated with early drinking. Pediatrics, 111(5), 949-955. Elliott, D., Wilson, W., Huizinga, D., Sampson, R., Elliott, A., & Rankin, B. (1996). The effects of neighborhood disadvantage on adolescent development. Journal of Research in Crime and Delinquency, 33, 389-426. English, D., Holman, C., Milne, E., Winter, M., Hulse, G., Codde, G., et al. (1995). The quantification of drug caused morb idity and mortality in Australia Caverra, Australia: Commonwealth Department of Human Services and Health. 196

PAGE 197

ESRI. (2005). ArcGIS ArcMap 9. Redlands, CA: ESRI. Fishbein, M., & Ajzen, L. (1975). Belief, attitude, intention an d behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. Fitzmaurice, G., Laird, N., & Ware, J. (2004). Applied longitudinal analysis Hoboken, NJ: John Wiley & Sons, Inc. Flay, B., & Petraitis, J. (1994). Th e Theory of Triadic Influence: A new theory of health behavior with implications for preventive interventions. Advances in Medical Sociology, 4, 19-44. Fleming, K., Thorson, E., & Atkin, C. (2004) Alcohol advertising exposure and perceptions: Links with alcohol expectancies and intent ions to drink or dr inking in underaged youth and young adults. Journal of Health Communication, 9 (1), 3-29. Foley, K., Altman, D., Durant, R., & Wolfson, M. (2004). Adults' approval and adolescents' alcohol use. Journal of Adolescent Health, 34 345.e317-326. Forrest, R., & Kearns, A. (2001). Social c ohesion, social capital and the neighbourhood. Urban Studies, 38 (12), 2125-2143. Forster, J., McGovern, P., Wagenaar, A., Wolfs on, M., Perry, C., & Anstine, P. (1994). The Ability of young-people to purchase alcohol wi thout age identification in northeastern Minnesota, USA. Addiction, 89 (6), 699-705. Foxcroft, D., & Lowe, G. (1991). Adolescent dr inking behaviour and family socialization factors: A meta-analysis. Journal of Adolescence, 14 255-273. Fulkerson, J., Pasch, K., Perry, C., & Komro, K. (In Press). Relationships between alcoholrelated informal social control, parental monitoring and adolescent problem behaviors among racially diverse urban youth. Journal of Community Health Galea, S., Ahern, J., Tracy, M., & Vlahov, D. (2007). Neighborhood income and income distribution and the use of ciga rettes, alcohol, and marijuana. American Journal of Preventive Medicine, 32 (6S), S195-S202. Garson, G. (2007). Structural equation modeling Raleigh, NC: North Carolina State University. Geckova, A., & van Dijk, J. (2001). Peer impact on smoking, alcohol consumption, drug use, and sports activities in adolescents. Studia Psychologica, 43 (2), 113-123. Gibbons, F., Gerrard, M., Lune, L., Wills, T., Brody, G., & Conger, R. (2004). Context and cognitions: environmental risk, social infl uence, and adolescent substance use. Personality and Social Psychology Bulletin, 30 (8), 1048-1061. Godette, D., Headen, S., & Ford, C. (2006). Wind ows of opportunity: Fundamental concepts for understanding alcohol-related disparities experienced by young Blacks in the United States. Prevention Science, 7, 377-387. 197

PAGE 198

Goldstein, H. (1987). Multilevel models in educ ational and social research New York City, NY: Oxford University Press. Goldstein, H., & Cuttance, P. (1988). A note on national assessment and school comparisons. Journal of Educational Policy, 3 197-202. Goodman, L. (1974). Exploratory latent stru cture analysis using both identifiable and unidentifiable models. Biometrika, 61 215-231. Gorman, D., Labouvie, E., Speer, P., & Subaiya, A. (1998). Alcohol availability and domestic violence. American Journal of Drug and Alcohol Abuse, 24 (4), 661-673. Gorman, D., Speer, P., Labouvie, E., & Subaiya, A. (1998). Risk of assaultive violence and alcohol availability in New Jersey. American Journal of Public Health, 88 (1), 97-100. Grant, B., Stinson, F., & Harford, T. (2001). Ag e at onset of alcohol use and DSM-IV alcohol abuse and dependence: A 12-year follow-up. Journal of Substance Abuse, 13 493-504. Grant, J., Scherrer, J., Lynskey, M., Lyons, M., Eisen, S., Tsuang, M., et al. (2006). Adolescent alcohol use is a risk factor for adult alcohol and drug dependence: ev idence from a twin design. Psychological Medicine, 36 (1), 109-118. Greenfeld, L. A. (1998). Alcohol and crime: An analysis of national data on the prevalence of alcohol in crime Washington, DC: U.S. Department of Justice. Griffin, K., Botvin, G., Epstein, J., Doyle, M ., & Diaz, T. (2000). Psychosocial and behavioral factors in early adolescence as predictors of heavy drinking among high school seniors. Journal of Studies on Alcohol, 61(4), 603-606. Griffin, K., Scheier, L., Botvin, G., & Diaz, T. (2000). Ethnic and gender differences in psychosocial risk, protection, and adolescent alcohol use. Prevention Science, 1 (4), 199212. Grube, J. (1997). Preventing sale s of alcohol to minors: Resu lts from a community trial. Addiction, 93 (Suppl. 2), S251-S260. Gruber, E., DiClemente, R., Anderson, M., & L odico, M. (1996). Early drinking onset and its association with alcohol use and prob lem behavior in late adolescence. Preventive Medicine, 25 (3), 293-300. Gruenewald, P., Johnson, F., & Treno, A. (2002). Outlets, drinking and driving: A multilevel analysis of availability. Journal of Studies on Alcohol, 63 (4), 460-468. Grunbaum, J., Kann, L., Kinchen, S., Ross, J., Hawk ins, J., Lowry, R., et al. (2004). Youth risk behavior surveillance-United States, 2003. MMWR, 53 (SS-2), 1-100. 198

PAGE 199

Guo, J., Chung, I., Hill, K., Hawkins, J., Ca talano, R., & Abbott, R. (2002). Developmental relationships between adoles cent substance use and risky sexual behavior in young adulthood. Journal of Adolescent Health, 31 (4), 354-362. Guo, J., Collins, L., Hill, K., & Hawkins, J. ( 2000). Developmental pathways to alcohol abuse and dependence in young adulthood. Journal of Studies on Alcohol, 61(6), 799-808. Gyimah-Brempong, K. (2001). Alcohol availability and crime: Evidence from census tract data. Southern Economic Journal, 68(1), 2-21. Hackbarth, D., Schnopp-Wyatt, D., Katz, D., Williams, J., Silvestr i, B., & Pfleger, M. (2001). Collaborative research and action to control the geogr aphic placement of outdoor advertising of alcohol and t obacco products in Chicago. Public Health Reports, 116 (6), 558-567. Hagenaars, J., & McCutcheon, A. (2002). Applied latent class analysis models. New York City, NY: Cambridge University Press. Halpern-Felsher, B., Millstein, S., & Ellen, J. (1996). Relationship of alcohol use and risky sexual behavior: A review and analysis of findings. Journal of Adolescent Health, 19 (5), 331-336. Harrison, P., Fulkerson, J., & Park, E. (2000) The relative importance of social versus commercial sources in youth access to tobacco, alcohol and other drugs. Preventive Medicine, 31 39-48. Hawkins, J., Catalano, R., & Miller, J. (1992). Ri sk and protective factor s for alcohol and other drug problems in adolescence and early adult hood implications for substance-abuse prevention. Psychological Bulletin, 112 (1), 64-105. Henry, K., Slater, M., & Oetting, E. (2005). Alco hol use in early adolescence: The effect of changes in risk taking, perceive d harm and friends' alcohol use. Journal of Studies on Alcohol, 66 (2), 275-283. Hill, T., & Angel, R. (2005). Neighborhood disorder psychological distress, and heavy drinking. Social Science & Medicine, 61 965-975. Hingson, R., Heeren, T., Levenson, S., Jamanka, A., & Voas, R. (2002). Age of drinking onset, driving after drinking, and involvement in alcohol related motor-vehicle crashes. Accident Analysis and Prevention, 34 (1), 85-92. Hingson, R., Heeren, T., Winter, M., & Wechsler H. (2003). Early age of first drunkenness as a factor in college students' unplanned and unprotected sex attributable to drinking. Pediatrics, 111 (1), 34-41. Hingson, R., Heeren, T., & Zakocs, R. (2001). Age of drinking onset and involvement in physical fights after drinking. Pediatrics, 108 (4), 872-877. 199

PAGE 200

Hingson, R., Heeren, T., Zakocs, R., Winter, M., & Wechsler, H. (2003). Age of first intoxication, heavy drinking, driving after drinking and ri sk of unintentional injury among US college students. Journal of Studies on Alcohol, 64 (1), 23-31. Hingson, R., Heeren, T., & Winter, M. (2006). Age at drinking onset and alcohol dependence Age at onset, duration, and severity. Archives of Pediatrics & Adolescent Medicine, 160(7), 739-746. Hipwell, A., White, H., Loeber, R., Stouthamer-L oeber, M., Chung, T., & Sembower, M. (2005). Young girls' expectancies about the effects of alcohol, future intentions a nd patterns of use. Journal of Studies on Alcohol, 66 (5), 630-639. Hobbs, F., & Stoops, N. (2002). Demographic trends in the 20th century: Census 2000 special reports Washington, DC: U.S. Census Bureau. Hu, L., & Bentler, P. (1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria vers us new alternatives. Structural Equation Modeling, 6 1-55. Ingoldsby, E., & Shaw, D. (2002). Neighborhood cont extual factors and earl y-starting antisocial pathways. Clinical Child and Family Psychology Review, 5 (1), 21-55. Jackson, C., Henriksen, L., & Dickinson, D. (1999). Alcohol-specific so cialization, parenting behaviors and alcohol use by children. Journal of Studies on Alcohol, 60 (3), 362-367. Jackson, K., Sher, K., Cooper, M., & Wood, P. (2002). Adolescent alco hol and tobacco use: onset, persistence and trajectorie s of use across two samples. Addiction, 97 (5), 517-531. Janz, N., & Bekcer, M. (1984). The h ealth belief model: A decade later. Health Education Quarterly, 11 1-47. Jarrett, R. (1997). Bringing families back in: Ne ighborhood effects on child development. In J. Brooks-Gunn, G. J. Duncan & J. L. Aber (Eds.), Neighborhood poverty: Policy implications in studying neighborhoods New York: Russell Sage Foundation. Johnston, L., O'Malley, P., Bachma n, J., & Schulenberg, J. (2008). Monitoring the Future national results on adolescent drug us e: Overview of key findings, 2007 Bethesda, MD: National Institute on Drug Abuse. Johnston, L., O'Malley, P., Bachma n, J., & Schulenberg, J. (2007). Monitoring the Future: National results on adolescent drug us e, Overview of key findings, 2006 Bethesda, MD: National Institute on Drug Abuse. Johnston, L., O'Malley, P., Bachma n, J., & Schulenberg, J. (2008). Monitoring the Future national survey results on drug use, 1975-2007. Volume I: Secondary school students Bethesda, MD: National Institute on Drug Abuse. 200

PAGE 201

Johnston, L., OMalley, P., Bachma n, J., & Schulenberg, J. (2005). Monitoring the Future National survey results on drug use, 1975-2004. Volume I: Secondary school students Bethesda, MD: National Institute on Drug Abuse. Jones-Webb, R., Toomey, T., Miner, K., Wagen aar, A., Wolfson, M., & Poon, R. (1997). Why and in what context adolescents obtain alcohol from older adults: A pilot study. Substance Use & Misuse, 32 (2), 219-228. Kandel, D., & Jessor, R. (2002). The gateway hypothesis revisited. In Stages and pathways of drug involvement:Examining the gateway hypothesis Cambridge, England: Cambridge University Press. Kandel, D., Yamaguchi, K., & Chen, K. (1992). Stages of progression in drug involvement from adolescence to adulthood: Further evidence for the gateway theory. Journal of Studies on Alcohol, 53 (5), 447-457. Karvonen, S., & Rimpela, A. (1997). Urban small ar ea variation in adolescents' health behaviour. Social Science & Medicine, 45 1089-1098. Kawachi, I., Kenneday, B., Lochner, K., & Prothrow-Stith, D. (1997). Social capital, income inequality, and mortality. American Journal of Public Health, 87 (9), 1491-1498. Kegler, M., Oman, R., Vesely, S., McLeroy, K., Aspy, C., Rodine, S., et al. (2005). Relationships among youth assets and neighborhood and community resources. Health Education & Behavior, 32 380-397. Kelly, K., Comello, M., & Hunn, L. (2002). Pare nt-child communication, perceived sanctions against drug use, and youth drug involvement. Adolescence, 37 (148), 775-787. Klem, L. (2000). Structural eq uation modeling. In L. G. Gr imm & P. R. Yarnold (Eds.), Reading and understanding more multivariate statistics Washington, D.C.: American Psychological Association. Kline, R. (2005). Principles and practice of st ructural equation modeling New York City, NY: The Guilford Press. Komro, K., Maldonado-Molina, M., Tobler, A., B onds, J., & Muller, K. (2007). Effects of home access and availability of alcohol on young adolescents alcohol use. Addiction, 102 1597-1608. Komro, K., Perry, C., Veblen-Mortenson, S., Bosma, L., Dudovitz, B ., Williams, C., et al. (2004). Brief report: the adaptation of Project Northland for urban youth. Journal of Pediatric Psychology, 29 (6), 457-466. Komro, K., Perry, C., Veblen-Mortenson, S., Farba khsh, K., Kugler, K., Al fano, K., et al. (2006). Cross-cultural adaptation of a home-bas ed program for alcohol use prevention among urban youth: The Slick Tracy Home Team program. Journal of Primary Prevention, 27(2), 135-153. 201

PAGE 202

Komro, K., Perry, C., Veblen-Mortenson, S., Farb akhsh, K., Toomey, T., Stigler, M., et al. (2008). Outcomes for a randomized controlled trial of a multi-component alcohol use preventive intervention for urban youth: Project Northland Chicago. Addiction, 103 606618. Komro, K., Perry, C., Williams, C., Stigler, M., Farbakhsh, K ., & Veblen-Mortenson, S. (2001). How did Project Northland reduce alcohol use among young adolesce nts? Analysis of mediating variables. Health Education Research, 16 (1), 59-70. Komro, K., Perry, C., Williams, C ., Veblen-Mortenson, S., Forster, J., Munson, K., et al. (1999). Research and evaluation design of a comm unity-wide program to reduce adolescent alcohol use: Project Northland Ph ase II. In S. Casswell (Ed.), Kettil Bruun Society thematic meeting: Fourth symposium on community action research and the prevention of alcohol and other drug problems (pp. 181-193). Auckland, New Zealand: Alcohol & Public Health Research Unit, University of Auckland. Kosterman, R., Hawkins, J., Guo, J., Catalano, R., & Abbott, R. (2000). The dynamics of alcohol and marijuana initiation: Patterns and predictors of first use in adolescence. American Journal of Public Health, 90 (3), 360-366. Kuha, J. (2004). Comparisons of assumptions and performance. Sociological Methods & Research, 33 (2), 188-229. Kune, G., & Vitetta, L. (1992). Alcohol consump tion and the etiology of colorectal cancer: A review of the sc ientific evidence from 1957 to 1991. Nutrition and Cancer, 18 (2), 97111. Kwate, N., Jernigan, M., & Lee, T. (2007). Preval ence, proximity and predictors of alcohol ads in Central Harlem. Alcohol and Alcoholism, 42(6), 635-640. La Vecchia, C., & Negri, E. (1989). The role of alcohol in oesophageal cancers in non-smokers, and of tobacco in non-drinkers. International Journal of Cancer, 43 (5), 784-785. Lambert, S., Brown, T., Phillips, C., & Ialongo, N. (2004). The relations hip between perceptions of neighborhood characteristics and substa nce use among urban African American adolescents. American Journal of Community Psychology, 34 (3/4), 205-219. Lanza, S., Flaherty, B., & Collins, L. (2003). Latent class and latent transition analysis. In J. A. Schinka & W. F. Velicer (Eds.), Comprehensive handbook of psychology (Vol. 2, pp. 663-685). Hoboken, NJ: Wiley. LaVeist, T., & Wallace, Jr., J. (2000). Health risk and inequitable distribution of liquor stores in African American neighborhood. Social Science & Medicine, 51 (4), 613-617. Ledoux, S., Miller, P., Choquet, M., & Plant, M. (2002). Family stru cture, parent-child relationships, and alcohol and other drug us e among teenagers in France and the United Kingdom. Alcohol and Alcoholism, 37(1), 52-60. 202

PAGE 203

Lo, Y. T., Mendell, N., & Rubin, D. (2001). Te sting the number of co mponents in a normal mixture. Biometrika, 88(3), 767-778. Loveland-Cherry, C., Leech, S., Laetz, V., & Diel man, T. (1996). Correlates of alcohol use and misuse in fourth-grade children: Psychosoc ial, peer, parental, and family factors. Health Education Quarterly, 23(4), 497-511. Maisey, A., & Davies, G. (2003). A case study approach: underage drinking in adolescents. Journal of Research on Social Health, 123 52-24. Mkel K., Osterberg, E., & Sulkunen, P. (1981). Drink in Finla nd: increasing alcohol availability in a monopoly state. In E. Si ngle, P. Morgan & L. de Lint (Eds.), Alcohol, society and the state 2: The social hi story of control in seven countries Toronto, Canada: Addiction Research Foundation. Mkel P. (2002). Who started to drink more? A reanalysis of the change resulting from a new alcohol law in Finland in 1969. In R. Room (Ed.), The effects of Nordic alcohol policies: What happens to drinking and harm when alcohol controls change? Helsinki, Finland: Nordic Council for Alcohol and Drug Research. Marsden, J., Boys, A., Farrell, M., Stillwell, G ., Hutchings, K., Hillebra nd, J., et al. (2005). Personal and social correlates of alc ohol consumption among mid-adolescents. British Journal of Developmental Psychology, 23 427-450. Marsh, H. (1998). Pairwise deletion for missing data in structural equation models: Nonpositive definite matrices, parameter estimates, goodness of fit, and adjusted sample sizes. Structural Equation Modeling, 5 (1), 22-36. McArdle, P., Wiegersma, A., Gilvarry, E., Kolte, B., McCarthy, S., Fitzgerald, M., et al. (2002). European adolescent substance use: The roles of family structure, function and gender. Addiction, 97 (3), 329-339. McGinnis, J., & Foege, W. (1993). Actual causes of death in the United States. JAMA, 270 2207-2212. McGue, M., Iacono, W., Legrand, L., Malone, S. & Elkins, I. (2001). Or igins and consequences of age at first drink. 1. Asso ciations with substance-use diso rders, disinhibitory behavior and psychopathology, and P3 amplitude. Alcoholism-Clinical and Experimental Research, 25 (8), 1156-1165. MDNH. (2008). Chicago crime statistics and crime data Edmonds, WA: Area Connect. Retrieved November 25, 2008, from http://chicago.areaconnect.com/crime1.htm Miller, P. (1997). Family structure, personality, drinking, smoking and il licit drug use: A study of U.K. teenagers. Drug and Alcohol Dependence, 45 (1-2), 121-129. Miller, T., Levy, D., Cohen, M., & Cox, K. ( 2006). Costs of alcohol and drug-involved crime. Prevention Science, 7 333-342. 203

PAGE 204

Mohler-Kuo, M., Dowdall, G., Koss, M., & Wechsler, H. (2004). Correlates of rape while intoxicated in a national sample of college women. Journal of Studies on Alcohol, 65 3745. Mokdad, A., Marks, J., Stroup, D., & Gerberding, J. (2004). Actual causes of death in the United States, 2000. JAMA, 291 (10), 1238-1245. Monti, P., Miranda, R., Nixon, K., Sher, K., Sw artzwelder, H., Tapert S., et al. (2005). Adolescence: booze, brains, and behavior. Alcoholism-Clinical and Experimental Research, 29 (2), 207-220. Muller, K., & Fetterman, B. (2002). Regression and ANOVA: An integrated approach using SAS Software Cary, N.C.: SAS. Murray, D. (1998). Design and analysis of group-randomized trials New York, NY: Oxford University Press. Muthn, B., Brown, C., Masyn, K., Jo, B., Khoo, S., Yang, C., et al. (2002). General growth mixture modeling for randomized preventive interventions. Biostatistics, 3 (4), 459-475. Muthn, B., & Muthn, L. (2000). Integrating pe rson-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism-Clinical and Experimental Research, 24(6), 882-891. Muthn, L., & Muthn, B. (2004). Mplus user's guide. Third Edition. Los Angeles, CA: Muthn & Muthn. Muthn, L., & Muthn, B. (2007). Mplus (Versi on 5.1). Los Angeles, CA: Muthn & Muthn. Nagin, D. (1999). Analyzing developmental trajectories: Semi-parametric, group-based approach. Psychological Methods, 4, 39-177. Nagin, D. (2005). Group-based modeling of development Cambridge, MA: Harvard University Press. Nagin, D., & Tremblay, R. (2001). Analyzing developmental trajectories of distinct but related behaviors: A gr oup-based method. Psychological Methods, 6 (1), 18-34. National Center for Chronic Disease Preventi on and Health Promoti on. (2008). YRBSS Youth online: Comprehensive results. Retrieved September 12, 2008 from http://apps.nccd.cdc.gov/yrbss/. National Institute on Alcohol Abuse and Alcoholism. (2001). Forecast for the future: Strategic plan to address health disparities Rockville: National Institutes of Health. National Research Council, & Institute of Medicine. (2004). Reducing underage drinking: A collective responsibility Washington, D.C.: The National Academies Press. 204

PAGE 205

Nelson, B., Patience, T., & MacDonald, D. (1999). Adolescent risk behavior and the influence of parents and education. Journal of the American Boar d of Family Practice, 12 (6), 436443. National Highway Traffic Safety Administration. (2005). Traffic safety facts 2005 data Washington, D.C.: National Highway Traffic Safety Administration. National Highway Traffic Safety Administration. (2006). Traffic safety facts 2005 data Washington, D.C.: National Highway Traffic Safety Administration. National Institute on Alcohol Abuse and Alcoholism. (2000). Special report to the U.S. Congress on alcohol and health Washington, D.C.: U.S. Department of Health and Human Services. National Institute on Alcohol Abuse and Alcohol ism. (2004). Environmental and contextual considerations. Alcohol Research & Health, 28 (3), 155-162. Nuttall, D., Goldstein, H., Prosser, R., & Rasbash, J. (1989). Differential school effectiveness. International Journal of Educational Research, 13 (769-776). Nylund, K., Asparouhov, T., & Muthn, B. (2007). Deciding on the number of classes in latent class analysis and growth mixture m odeling: A Monte Carlo simulation study Structural Equation Modeling, 14, 535-569. Parker, R. (1993). Alcohol and theories of homicide. In W. L. Adler (Ed.), Advances in Criminological Theory New Brunswick, NJ: Transaction. Parker, R. (1995). Bringing booze back in th e relationship between alcohol and homicide. Journal of Research in Crime and Delinquency, 32 (1), 3-38. Parker, R., & Cartmill, R. (1998). Alcohol and ho micide in the United States 1934-1995 Or one reason why US rates of violence may be going down. Journal of Criminal Law & Criminology, 88(4), 1369-1398. Pasch, K., Komro, K., Perry, C., Hearst, M., & Farbakhsh, K. (In Press). Does outdoor alcohol advertising around elementary schools vary by the ethnicity of stude nts in the school? Ethnicity and Health Pasch, K., Komro, K., Perry, C., Hearst, M., & Farbakhsh, K. (2007). Outdoor alcohol advertising near schools: What does it advertise and how is it re lated to intentions and use of alcohol among young adolescents? Journal of Studies on Alcohol and Drugs, 68 (4), 587-596. Paschall, M., Grube, J., Black, C., & Ringwalt, C. (2007). Is commercial alcohol availability related to adolescent al cohol sources and alcohol use? Fi ndings from a multi-level study. Journal of Adolescent Health, 41 (2), 168-174. 205

PAGE 206

Paschall, M., Grube, J., Black, C., Flewelling, R., Ringwalt, C., & Bigl an, A. (2007). Alcohol outlet characteristics and alcohol sales to youth: Results from alcohol purchase surveys in 45 Oregon communities. Prevention Science, 8 153-159. Perry, C., Komro, K., Veblen-Mortenson, S., Bo sma, L., Farbakhsh, K., Munson, K., et al. (2003). A Randomized controlled trial of th e middle and junior high school DARE and DARE Plus programs. Archives of Pediatrics & Adolescent Medicine, 157 (2), 178-184. Perry, C., Williams, C., Forster, J., Wolfson, M., Wagenaar, A ., Finnegan, J., et al. (1993). Background, conceptualization and design of a community-wide research-program on adolescent alcohol-use Project Northland. Health Education Research, 8 (1), 125-136. Perry, C., Williams, C., Komro, K ., Veblen-Mortenson, S., Stigler, M., Munson, K., et al. (2002). Project Northland: long-term outcomes of co mmunity action to reduc e adolescent alcohol use. Health Education Research, 17 (1), 117-132. Perry, C., Williams, C., Veblen-M ortenson, S., Toomey, T., Komro, K., Anstine, P., et al. (1996). Outcomes of a community-wide alc ohol use prevention pr ogram during early adolescence: Project Northland. American Journal of Public Health, 86 (7), 956-965. Pacific Institute of Resear ch and Evaluation. (1999). Costs of underage drinking Washington, D.C.: United States Department of Justice, O ffice of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Pollack, C., Cubbin, C., Ahn, D., & Winkle by, M. (2005). Neighbourhood deprivation and alcohol consumption: do es the availability of alcohol play a role? International Journal of Epidemiology, 34 (4), 772-780. Poortinga, W. (2006). Do health behaviors mediat e the association between social capital and health? Preventive Medicine, 43 (6), 488-493. Poulin, C., & Graham, L. (2001). The associa tion between substance use, unplanned sexual intercourse and other sexual beha viours among adolescent students. Addiction, 96 (4), 607-621. Preusser, D., & Williams, A. (1992). Sales of alco hol to underage purchasers in three New York counties and Washington, D.C. Journal of Public Health Policy, 13 (3), 306-317. Putnam, R. (1993). The prosperous comm unity. Social capital and public life. The American Prospect, 13 35-42. Raftery, A. (1986). A note on Bayes factors for lo g-linear contingency table models with vague prior information. Journal of the Royal Statistical Society. Series B (Methodological), 48(2), 249-250. Rankin, B., & Quane, J. (2002). Social contexts and urban adolescent out comes: The interrelated effects of neighborhoods, families, a nd peers on African-American youth. Social Problems, 49(1), 79-100. 206

PAGE 207

Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models: Applications and data analysis methods (2nd Edition ed.). Newbury Park, CA: Sage Publications. Rehm, J., Gmel, G., Sempos, C., & Trevisan M. (2003). Alcohol-related morbidity and mortality. Alcohol Research & Health, 27 (1), 39-51. Renna, F. (2007). The economic cost of teen dr inking: Late graduation and lowered earnings. Health Economics, 16 407-419. Resnick, M., Bearman, P., Blum, R., Bauman, K., Harris, K., Jones, J., et al. (1997). Protecting adolescents from harm: findings from th e National Longitudina l Study on Adolescent Health. JAMA, 278 823-832. Resnicow, K., Braithwaite, R., Dilorio, C., & Gl anz, K. (2002). Applying theory to culturally diverse and unique populations. In K. Glanz, B. K. Rimer & F. M. Lewis (Eds.), Health behavior and health education: Theory, research, and practices San Francisco, CA: Jossey-Bass. Richardson, J., Radziszewska, B., Dent, C., & Flay, B. (1993). Relationship between after-school care of adolescents and substance use, ri sk-taking, depressed mood, and academicachievement. Pediatrics, 92 (1), 32-38. Risley, R. (2004). A police officer's perspective on gangs drugs, and guns on the streets of Chicago. Chicago, IL: Illinois Criminal Justice Information Authority. Rogers, R. (1983). Cognitive and psychological proce sses in fear appeals a nd attitude change: A revised theory of protecti on motivation. In J. T. Cncioppo & R. E. Petty (Eds.), Social Psychophysiology. New York City, NY: Guilford. Roizon, J. (1997). Epidemiological issues in alcohol-related violence. Recent Developments in Alcoholism, 13 7-40. Roski, J., Perry, C., McGovern, P., Williams, C., Farbakhsh, K., & Veblen-Mortenson, S. (1997). School and community influences on adolescent alcohol and drug use. Health Education Research, 12 (2), 255-266. Rossow, I., Pape, H., & Storvoll, E. (2005). Sources of intoxication--how do adolescents get hold of alcohol? Tidsskr Nor Laegerforen (Journal of the Norweigian Medical Association), 125, 1160-1162. Sampson, R. (1992). Family management and child development: Insights from social disorganization theory. In J. McCord (Ed.), Advances in Criminological Theory New Brunswick, NJ: Transaction. Santelli, J., Brener, N., Lowry, R., Bhatt, A., & Zabin, L. (1998). Multiple sexual partners among US adolescents and young adults. Family Planning Perspectives, 30 (6), 271-275. 207

PAGE 208

Santelli, J., Robin, L., Brener, N., & Lowry, R. (2001). Timing of alcohol and other drug use and sexual risk behaviors among unmarri ed adolescents and young adults. Family Planning Perspectives, 33 (5), 200. SAS Institute. (2004). SAS/STAT 9.1 user's guide. Cary, NC: SAS Institute, Inc. Schafer, J., & Graham, J. (2002). Missing da ta: Our view of the state of the art. Psychological Methods, 7 (2), 147-177. Scheier, L., & Botvin, G. (1997). Expectancies as mediators of the effects of social influences and alcohol knowledge on adolescent alc ohol use: A prospective analysis. Psychology of Addictive Behaviors, 11 (1), 48-64. Scheier, L., Botvin, G., Diaz, T., & Griffin, K. (1999). Social skills, competence, and drug refusal efficacy as predictors of adolescent alcohol use. Journal of Drug Education, 29(3), 251-278. Scheier, L., Botvin, G., & Miller, N. (2000) Life events, neighborhood stress, psychosocial functioning, and alcohol us e among urban minority youth. Journal of Chile & Adolescent Substance Abuse, 9 (1), 19-50. Schumacker, R., & Lomax, R. (2004). A beginner's guide to stru ctural equation modeling Mahway, NJ: Lawrence Erlbaum Associates. Scribner, R., Cohen, D., Kaplan, S., & Allen, S. (1999). Alcohol availability and homicide in New Orleans: Conceptual considerations for sm all area analysis of the effect of alcohol outlet density. Journal of Studies on Alcohol, 60 (3), 310-316. Scribner, R., Cohen, D., & Fisher, W. (2000). Evid ence of a structural effect for alcohol outlet density: A multilevel analysis. Alcoholism-Clinical and Experimental Research, 24 (2), 188-195. Scribner, R., Mackinnon, D., & Dwyer, J. (1994). Alcohol outlet density and motor-vehicle crashes in Los-Angeles-county cities. Journal of Studies on Alcohol, 55(4), 447-453. Scribner, R., Theall, K., Simonsen, N., Mas on, K., Johnson, S., & DeJong, W. (2007). The contextual role of alcohol outlet density on college drinking. Alcoholism-Clinical and Experimental Research, 31(6), 54A-54A. Seitz, H., & Poschl, G. (1997). Alcohol and ga strointestinal cancer: Pathogenic mechanisms. Addiction Biology, 2 (1), 19-33. Sellstrom, E., & Bremberg, S. (2006). The si gnificance of neighbourhood context to child and adolescent health and well-being: A systematic review of multilevel studies. Scandinavian Journal of Public Health, 34 (5), 544-554. 208

PAGE 209

Shrier, L., Emans, S., Woods, E., & DuRant, R. (1997). The association of sexual risk behaviors and problem drug behaviors in high school students. Journal of Adolescent Health, 20 (5), 377-383. Siahpush, M., Borland, R., Ta ylor, J., Singh, G., Ansari, Z ., & Serraglio, A. (2006). The association of smoking with perception of income inequality, relative material wellbeing, and social capital. Social Science & Medicine, 63 (11), 2801-2812. Singh, G. (2003). Area deprivation and widening inequalities in US mortality, 1969-1998. American Journal of Public Health, 93 (7), 1137-1143. Smart, R., Adlaf, E., & Walsh, G. (1996). Proc urement of alcohol and underage drinking among adolescents in Ontario. Journal of Studies on Alcohol, 57 419-424. Smith, G., Branas, C., & Miller, T. (1999). Fatal nontraffic injuries involving alcohol: A metaanalysis. Annals of Emergency Medicine, 33 (6), 659-668. Snyder, L., Milici, F., Slater, M., Sun, H., & Strizhakova, Y. (2006). Effects of alcohol advertising exposure on drinking among youth. Archives of Pediatrics & Adolescent Medicine, 160 (1), 18-24. Sobel, M. (1982). Asymptotic c onfidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology Washington, D.C.: American Sociological Association. Sorenson, S., & Berk, R. (2001). Handgun sales, beer sales, and youth homicide, California, 1972-1993. Journal of Public Health Policy, 22 (2), 182-197. Speer, P., Gorman, D., Labouvie, E., & Ontkus h, M. (1998). Violent crime and alcohol availability: Relationships in an urban community. Journal of Public Health Policy, 19(3), 303-318. Spoth, R., Shin, C., Guyll, M., Redmond, C., & A zevedo, K. (2006). Universality of effects: An examination of the comparability of long-te rm family intervention effects on substance use across risk-related subgroups. Prevention Science, 7 (2), 209-224. Stacy, A., Zogg, J., Unger, J., & Dent, C. (2004). Exposure to tele vised alcohol ads and subsequent adolesce nt alcohol use. American Journal of Health Behavior, 28 (6), 498509. Stitt, B., & Giacopassi, D. (1992). Alcohol availability and alcohol-related crime. Criminal Justice Reviews, 17, 268-279. Stueve, A., & O'Donnell, L. (2005) Early alcohol ini tiation and subsequent sexual and alcohol risk behaviors among urban youths. American Journal of Public Health, 95 (5), 887-893. 209

PAGE 210

Substance Abuse and Mental Health Services Administration. (2005). The NSDUH report: Alcohol use and delinquent behaviors among youths Rockville, MD: Substance Abuse and Mental Health Administration, Office of Applied Statistics. Substance Abuse and Mental Health Services Administration. (2006). Results from the 2005 National Survey on Drug Use and Health: National findings Rockville, MD: Office of Applied Studies. Swahn, M., & Hammig, B. (2000). Prevalence of youth access to alcohol, guns, illegal drugs, or cigarettes in the home and associa tion with health-risk behaviors. Annals of Epidemiology, 10 (452). Szapocznik, J., & Coatsworth, J. (1999). An ecodevelopmental framework for organizing risk and protection for drug abuse: a developmental model of risk and protection. In M. Glantz & C. R. Hartel (Eds.), Drug Abuse: Origins and Interventions (pp. 331). Washington, D.C.: American Psychological Association. Szreter, S., & Woolcock, M. (2004). Health by associ ation? Social capital, social theory, and the political economy of public health. International Journal of Epidemiology, 33 (4), 650667. Tobler, A., Komro, K., & Maldon ado-Molina, M. (2007, May 29-31). Effects of neighborhood contextual factors on yo ung adolescent alcohol use. Paper presented at the Annual meeting of the Society for Prevention Research, Washington, D.C. Tobler, A., Komro, K., & Ma ldonado-Molina, M. (2008). Patterns of alcohol-related neighborhood risk among early adolescent, multi-ethnic youth residing in urban communities Paper presented at the Annual mee ting of the Society for Prevention Research, San Francisco, CA. Toumbourou, J., Stockwell, T., Neighbors, C., Marlatt, G., Sturge, J., & Rehm, J. (2007). Interventions to reduce harm associat ed with adolescent substance use. The Lancet, 369(9570), 1391-1401. Treno, A., Alaniz, M., & Gruene wald, P. (2000). The use of dri nking places by gender, age and ethnic groups: an analysis of routine drinking activities. Addiction, 95 (4), 537-551. Turner, M. (1998). Moving out of poverty: Expa nding mobility and choice through tenant-based housing assistance. Housing Policy Debate, 9 (2), 373-394. Twisk, J. (2006). Applied multilevel analysis New York City, NY: Cambridge University Press. U.S. Census Bureau. (2000). United States Census 2000 Washington, D.C.: U.S. Department of Commerce. U.S. Census Bureau. (2003). Census Bureau releases population estimates by age, sex, race and Hispanic origin Washington, D.C.: U.S. Census Bureau. Retrieved January 4, 2008 from http://www.census.gov/PressR elease/www/2003/cb03-16.html. 210

PAGE 211

U.S. Census Bureau. (2007). Population estimates Washington, D.C.: U.S. Census Bureau. Retrieved January 4, 2008, from www.census.gov/poptest/estimates.php U.S. Department of Health and Human Services. (2000). 2000 Surgeon General's report-reducing tobacco use: Mi nors' access to tobacco Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. Valdez, A., Kaplan, C., Curtis, R., & Yin, Z. (1995). Illegal drug-use, alcohol and aggressive crime among Mexican-American and White male arrestees in San-Antonio. Journal of Psychoactive Drugs, 27(2), 135-143. van der Vorst, H., Engels, R., Meeus, W., Dekov ic, M., & Van Leeuwe, J. (2005). The role of alcohol-specific socialization in adolescents' drinking behaviour. Addiction, 100 (10), 1464-1476. Vanoers, J., & Garretsen, H. (1993). The geograp hic relationship between alcohol-use, bars, liquor shops and traffic injuries in Rotterdam. Journal of Studies on Alcohol, 54 (6), 739744. Vogt, W. (2005). Dictionary of statistics and methodology Thousand Oaks, CA: Sage Publications. Wagenaar, A., Finnegan, J., Wolfson, M., Anstin e, P., Williams, C., & Perry, C. (1993). Where and how adolescents obtain alcoholic beverages. Public Health Reports, 108(4), 459-464. Wagenaar, A., Murray, D., Wolfson, M., Forster, J., & Finnegan, J. (1994). Communities mobilizing for change on alcohol de sign of a randomized community trial. Journal of Community Psychology 79-101. Wagenaar, A., & Perry, C. (1994). Community st rategies for the reduction of youth drinking: Theory and application. Journal of Research on Adolescence, 4 (2), 319-345. Wagenaar, A., & Streff, F. (1990). Public opinion on alcohol policy. Journal of Public Health Policy, 17 189-205. Wagenaar, A., Toomey, T., & Erickson, D. (2005). Preventing youth access to alcohol: Outcomes from a multi-community time-series trial. Addiction, 100 335-345. Wagenaar, A., Toomey, T., & Lenk, K. ( 2004). Environmental influences on young adult drinking. Alcohol Research & Health, 28 (4), 230-235. Wagenaar, A., Toomey, T., Murray, D., Short, B., Wolfson, M., & Jones-Webb, R. (1996). Sources of alcohol for underage drinkers. Journal of Studies on Alcohol, 57 (3), 325-333. Wagenaar, A., & Wolfson, M. (1994). Enforcem ent of the legal minimum drinking age in the United States. Journal of Public Health Policy, 15 (1), 37-53. 211

PAGE 212

Wagenaar, A., Zobeck, T., Williams, G., & Hings on, R. (1995). Methods used in studies of drink-drive control efforts a metaanal ysis of the literature from 1960 to 1991. Accident Analysis and Prevention, 27(3), 307-316. Wallace, Jr., J. (1999). The social ecology of addiction: Race, risk and resilience. Pediatrics, 103(5), 1122-1127. Warner, L., & White, H. (2003). Longitudinal effects of age at onset and firs t drinking situations on problem drinking. Substance Use & Misuse, 38 (14), 1983-2016. Wasworthx, M. (1997). Changing so cial factors and their long-t erm implications for health. British Medical Bulletin, 53 (1), 198-209. Weitzman, E., & Chen, Y. (2005). Risk modifying e ffects of social capital on measures of heavy alcohol consumption, alcohol abuse, harm s and secondhand effects: national survey findings. Journal of Epidemiology and Community Health, 59 303-309. Williams, S., & Mulhall, P. (2005). Where public school students in Illin ois get cigarettes and alcohol: Characteristics of minor s who use different sources. Prevention Science, 6 4757. Willner, P. (2001). A view through the gateway: expectancies as a possible pathway from alcohol to cannabis. Addiction, 96 (5), 691-703. Wills, T., Gibbons, F., Gerrard, M., Murry, V., & Brody, G. (2003). Family communication and religiosity related to substan ce use and sexual behavior in ear ly adolescence: A test for pathways through self-control and prototype perceptions. Psychology of Addictive Behaviors, 17 (4), 313-323. Wilson, N., Battistich, V., Syme, S., & Boyce, W. (2002). Does elementary school alcohol, tobacco, and marijuana use increase middle school risk? Journal of Adolescent Health, 30(6), 442-447. Wilson, N., Syme, L., Boyce, W., Battistich V., & Selvin, S. (2005). Adolescent alcohol, tobacco, and marijuana use: The influence of neighborhood disorder and hope. American Journal of Health Promotion, 20(1), 11-19. Windle, M., & Davies, P. (1999). Depression and heavy alcohol use among adolescents: Concurrent and prospective relations. Development and Psychopathology, 11 (4), 823844. Yu, J. (2003). The association between parent al alcohol-related be haviors and children's drinking. Drug and Alcohol Dependence, 69 (3), 253-262. Zakrajsek, J., & Shope, J. (2006). Longitudi nal examination of underage drinking and subsequent drinking and risky driving. Journal of Safety Research, 37 (5), 443-451. 212

PAGE 213

213 Zhang, L., Welte, J., Wieczorek, W., & Messner, S. (2000). Alcohol and crime in China. Substance Use & Misuse, 35 (3), 265-279. Zullig, K., Valois, R., Huebner, E., Oeltmann, J., & Drane, J. (2001). Relationship between perceived life satisfaction and adolescents' substance abuse. Journal of Adolescent Health, 29(4), 279-288.

PAGE 214

BIOGRAPHICAL SKETCH Amy Leigh Tobler was born in Ogden, Utah. She is one of four children, having a twin sister and a younger brother and si ster. She grew up mostly in Kaysville, Utah but completed high school at River Ridge High School in New Port Richey, Florida in 1998. She earned her B.H.S. in health science and her M.P.H. in comm unity health education from the University of Florida (UF) in 2002 and 2003, respectively. Upon completion of her masters degree, Amy entered the workforce, first being employed as an Allied Health Department Chair a nd Instructor at City College (2004) and then as a Research Coordinator for the Social and Community Epidemiology Research Program in the College of Medicine, Department of Epidemiol ogy and Health Policy Research at UF. She has worked in this capacity since January 2005. This position allowed her the opportunity to earn her Ph.D. in health and human performance with a specialization in health behavior. Amys work and time as a doctoral student have afforded her many opportunities, including management of 6 externally funded research grants and co-aut horship of several peer-reviewed scientific manuscripts and one book chapter. As a doctoral candidate, Amy received the Health Solutions Graduate Scholarship and Patrick J. Bird Dissertation Research Award. She was also included in Whos Who Among Executives and Professionals in 2008. Amy is a member of the Society for Prevention Research (SPR), SPR Early Career Prevention Network, and Prevention Science Methodology Group. Amy will pursue a faculty position where she can continue her research and teach. She has been married to Jeffrey W. Tobler for 6 years. 214