Research Proposal

bib11.pdf

Home >Social Science homework help >Research Proposal

Paternal incarceration and trajectories of marijuana and other illegal drug use from adolescence into young adulthood: evidence from longitudinal panels of males and females in the United Statesadd_3110 121..132

Michael E. Roettger1, Raymond R. Swisher1,2, Danielle C. Kuhl2 & Jorge Chavez2

National Center for Family and Marriage Research, Bowling Green State University, Bowling Green, OH, USA1 and Department of Sociology, Bowling Green State University, Bowling Green, OH, USA2

ABSTRACT

Aims One-eighth of young adults in the United States report that their biological father has ever been incarcerated (FEI). This study is the first to examine associations between FEI and trajectories of substance use during the transition from adolescence into young adulthood for the US population. Design Using multi-level modeling techniques, trajectories of marijuana and other illegal drug use are examined, with FEI as the primary independent variable. Setting Data are from the first three waves of the National Longitudinal Study of Adolescent Health, a nationally representative sample of US adolescents beginning in 1995. Participants Panels of 7157 males and 7997 females followed from adolescence (7th–12th grades) into early adulthood (ages 18–27 years). Measurements Dependent variables included an ordinal measure of marijuana frequency of use in last thirty days, and a dichotomous measure for whether respondent had any use in the last thirty days of illegal drugs such crystal meth, cocaine, heroin, hallucinogens, PCP, LSD, speed, and ecstasy. Findings Among males and females, respectively, FEI is associated with an increased frequency of marijuana use, and increased odds of any other illegal drug use. Interactions between FEI and age further reveal that FEI is associated with an accentuated trajectory (i.e. a steeper slope) of marijuana use, and an elevated risk (i.e. higher mean level) of other illegal drug use. Conclusions Analysis provides some of the first evidence that paternal incarceration is significantly associated with drug use among U.S. males and females, even after controlling for a number of family background, parental, and individual characteristics.

Keywords Adolescence, collateral consequences of incarceration, drug use trajectories, father incarceration, illegal drug use, marijuana use, young adulthood.

Correspondence to: Michael E. Roettger, National Center for Family and Marriage Research, Bowling Green State University, Bowling Green, OH 43403, USA. E-mail: [email protected] Submitted 3 December 2009; initial review completed 11 January 2010; final version accepted 15 June 2010

INTRODUCTION

With the onset of the ‘War on Drugs’ and ‘War on Crime’, the US incarcerated population increased from 250 000 in 1975 to 2.25 million in 2006 [1,2].

Coinciding with this rise, the number of children with an incarcerated father has increased dramatically. In 2006, nearly 7.5 million children were estimated to have a parent either incarcerated or on probation/parole [3]. Wildeman [4] estimates that by age 14, 4% of white chil- dren and 25% of African American children will have had a parent in prison; moreover, 13% of young adults in

the United States report that their father had spent time in jail or prison [5,6].

Given these trends, researchers are increasingly examining the ‘collateral consequences’ of paternal incarceration for children and families [7–9]. Research has demonstrated associations between parental incarceration and increased levels of delinquency [10–12], mental health problems [13–15], homeless- ness, decreased civic participation [5] and internalizing and externalizing symptoms [13–16]. Similar outcomes for children have also been observed in studies focusing upon parental criminality [9]. Given the wide range of

RESEARCH REPORT doi:10.1111/j.1360-0443.2010.03110.x

these associations, it is surprising that few studies have examined the link between paternal incarceration and youth substance use (see review by Murray & Farrington [9]). To the extent that substance use represents a form of delinquency, or reflects internalizing and externalizing behavior, such an association would be consistent with previous studies.

Using national panels of US male and female respon- dents, we examine associations between biological father’s incarceration and trajectories of marijuana use and other illegal drug use during the transition to adult- hood. To be as conservative as possible, we test if these associations are robust in response to controls for a wide range of individual, family and neighborhood characteristics.

THEORETICAL FRAME

In a review of the literature on the consequences of paren- tal incarceration, Murray & Farrington [9] reported only one previous study with a control group linking paternal incarceration with alcohol and drug use, and no studies examining the issue within a large sample of the general population. Using a matched sample of 258 juveniles institutionalized for mental health treatment, Phillips et al. [14] found that respondents whose parents had been incarcerated were more likely to engage in alcohol and marijuana use. More recently, in a study of 2400 Austra- lian youth, Kinner et al. [17] observed that both maternal and paternal incarceration correlated significantly with alcohol and tobacco use, but that these associations disappeared when controls for socio-economic status, parental and family characteristics were introduced. In a prior study using the same data, Kinner et al. [18] found father’s arrest, but not incarceration, to be correlated with marijuana use in young adulthood [18]. Both studies suggest paternal incarceration may not be associated directly with substance abuse. Perhaps the best quantita- tive evidence of an association between paternal incar- ceration and substance abuse comes from the Cambridge Study of Delinquent Development. Murray & Farrington ([9], p. 161) observed that parental imprisonment in- creased the odds of drug use by a factor of 3.7 between ages 32 and 48 years, suggesting that parental incarcera- tion is associated with use of drugs across the life-course.

As results by Kinner et al. [17] above indicate, it is important to consider the potential role of other variables that might alternatively explain the association between paternal incarceration and youth outcomes. The present study follows the conceptual framework of Murray & Farrington [9], which identifies background factors that place individuals at a greater risk for experiencing paren- tal incarceration, such as low socio-economic status and other disadvantages. Murray & Farrington [9] also

describe a variety of processes through which parental incarceration might be associated with negative out- comes: parental separation, behavioral modeling, economic strain, poor parenting and stigma. Due to uncertainty regarding the timing of father’s incarcera- tion, the present study is unable to strictly differentiate the temporal ordering of background factors, father’s incarceration and potential mediators. Instead, it seeks to produce as conservative an estimate as possible of the association between father’s incarceration and substance abuse by controlling for a wide array of covariates. Thus, we include measures for race and ethnicity, indicators of parental substance use [20–25], family socio-economic status [26], family structure [9,17], parental involvement and monitoring [17,27], history of abuse by a caretaker [24,28], peer drug use [29,30] and low self-control [5,31].

Finally, we will consider the degree to which the asso- ciation between paternal incarceration and substance use is moderated by age and gender [9]. The association between age and problem behavior in adolescence and young adulthood is well established [26,32–34]. Thus, this study uses panel data to examine how age trajecto- ries of substance use vary for youth with and without an incarcerated father. The consequences of paternal incar- ceration may also vary by gender; however, the findings of previous studies have been mixed in this regard. For example, Murray et al. [35] observed a stronger associa- tion between parental incarceration and criminal be- havior for females, whereas Gabel & Shindledecker [13] reported stronger associations between parental incar- ceration and male antisocial behavior.

DATA AND METHODS

Data

Data were obtained from the National Longitudinal Study of Adolescent Health (Add Health). The Add Health in-home sample consists of approximately 20 000 respondents enrolled in grades 7–12 during wave I inter- views in 1995. Follow-up interviews were conducted 1 year later (wave II) and again as young adults (ages 18–27) in 2001–02 (wave III), with approximately 15 000 respondents completing interviews at each wave. Data used in this analysis are from self-reports of respon- dents and parent reports. Answers to sensitive questions, such as regarding paternal incarceration and substance use, were obtained using computer and Audio-CASI interview methods [36], techniques which typically increase reliability. Respondents who completed surveys at both waves I and III are included in the analysis (72% response rate at wave III). All models use the Add Health longitudinal sampling weights, which are designed to

122 Michael E. Roettger et al.

compensate for differential probabilities of sampling and attrition. After removing ~75 cases with substantial missing data, our analysis includes 99.5% of the 15 197 respondents who completed surveys at both waves I and III. Incorporating up to three waves of data for each respondent, we analyze panels containing 19 774 obser- vations based on 7157 males and 22 142 observations from 7997 females.

Data were missing for some variables on a small per- centage of observations. In such instances, multiple impu- tation replicates the error structure, makes optimal use of non-missing data (relative to case-wise deletion or use of dummy variables) and produces unbiased point estimates and standard errors when cases are missing at random [37]. The STATA ‘ice’ procedure was used to estimate missing values, assuming that the data were missing at random [38,39]. To increase randomness in the imputa- tion process, median values for cases with observations were taken from 21 randomly imputed data sets. In supplemental analyses, results based on a subset of respondents with complete data were found to be similar to those reported below based on the imputed data.

Explanatory variables

Paternal incarceration

Paternal incarceration is measured by answers to the wave III question: ‘Has your biological father ever served time in jail or prison?’. Data are unavailable regarding the timing of father’s incarceration. However, given that most males experience first incarcerations by their mid- 30s, and that incarceration rates decline rapidly thereaf- ter, most children in Add Health would have experienced father’s incarceration prior to wave I [40,41]. The stigma associated with criminal histories [42,43] and general findings of reliability for self-reports of delinquency and arrest [44,45] probably minimize false reports. Res- pondents (n = 517 males, 542 females) who refused to answer the question or indicated no knowledge of their father’s incarceration were coded as ‘missing’. Consistent with treatment of other missing data described above, multiple imputation was used for these missing cases, as it did not alter the pattern of findings when compared with case-wise deletion and use of dummy variables.

Controls

A number of additional variables were utilized to control for individual and family characteristics likely to be asso- ciated with both father’s incarceration and youth sub- stance use. These include: age at interview; self-reported race/ethnicity; residing with both biological parents; family socio-economic status based on mother and father’s education and occupation [46]; neighborhood

poverty; low self-control; mother’s binge drinking; prior arrest as a juvenile; peer substance use; prior physical abuse; the respondent’s closeness and involvement to the biological father; parental supervision; and school attachment.

Dependent variables

Frequency of marijuana use

Frequency of marijuana use is measured by respondent’s answer to the question: ‘During the past 30 days, how many times have you used marijuana?’. Responses range from 0 to 120 times. To reduce the influence of outliers, approximately 30 observations with improbably high reports were top-coded to 120. In supplemental analyses, deletion of these cases yielded a similar pattern of results.

Other illegal drug use

A measure of more serious drug use is adapted from Cleve- land & Weibe [47], which accounts for slight variations in the measurement of other illegal drug use across waves. At wave I, respondents were asked if they had ever (i) used inhalants, such as glue or solvents; (ii) used cocaine (defined in lead-in question as ‘including powder, freebase, or crack cocaine’); (iii) used lysergic acid diethylamide (LSD), phencyclidine (PCP), ecstasy, mushrooms, speed, ice, heroin or pills, without a doctor’s prescription; or (iv) ever injected any illegal drug, such as heroin or cocaine. At wave III, they were asked if during the past year they had used: (i) crystal methamphetamine; (ii) any form of cocaine, ‘including powder, freebase, or crack’; or (iii) drugs listed in wave I categories (iii–iv). A dichotomous indicator variable denotes any other illegal drug use at each wave.

STATISTICAL MODELS

Frequency of marijuana use

Frequency of marijuana use is modeled with a Poisson regression for panel data. The distribution for frequency of marijuana use is non-normal, with 80% of males and 85% of females indicating no use. Poisson regression addresses this problem [48]. Additionally, to control for clustering resulting from multiple observations per respondent and a high number of zero-counts (i.e. a ‘zero- inflated Poisson model’), a multi-level model with an individual-level random slope is used [49]. With this mod- eling approach, the resulting observations are assumed to be distributed identically and characterized as:

ln yit it it i( ) = + + +β β β ν0 1Age X (equation 1)

where yit represents the number of times individual i at wave t reports using marijuana; b0 represents the

Father incarceration and illegal drug use 123

intercept; bAgeit represents a row vector of estimated coefficients and the respondent age at wave t; bXit repre- sents a vector of control variables and estimated coeffi- cients; and ni1 is an individual-level random error component that is constant across waves. As a random intercept, ni1 is assumed to vary across individuals with a distribution of ni1~G(1,a). As Rabe-Hesketh & Skrondal [49] note, the measure alpha indicates the degree to which the individual-level intercept corrects for zero- inflated counts where individuals deviate from the restric- tive assumption of a Poisson distribution.

Other illegal drug use

To model other illegal drug usage, we adopt a multi-level logistic regression with an individual-level random effect to control for the repeated measurements of respondents [49]. The probability of other illegal drug use by indi- vidual i at wave t, pit is predicted by:

p e

it yit =

+ − 1

1 (equation 2)

such that yit = b0 + bAgeit + bXit + ni + eij, where b0 represents the estimated slope; bAgeit represents the row vector of estimated coefficients and the respondent’s age at wave t; bXit represents the vector of control vectors and estimated coefficients; ni is the individual-level inter- cept ni|xij~n(0,y); and eij|xij,vi fits a standard logistic distribution.

RESULTS

Table 1 presents means and standard deviations for all variables by sex and father ever incarcerated (FEI) status. Both males and females with an incarcerated father report higher levels of substance use. Mean marijuana frequencies and mean percentage reporting any usage are higher at all three waves for respondents with an incarcerated father. Cumulatively across the three waves, 51.3% of males and 39.3% of females with FEI reported using marijuana, compared to 37.7% and 28.3% of males and females, respectively, without FEI. Similarly, mean levels of other illegal drug use are higher for respondents with a FEI, although differences are smaller for females than males. Cumulatively, 39.4% (versus 27.8%) of males and 28.9% (versus 22.2%) of females with FEI ever report illicit drug usage.

Table 1 also demonstrates that FEI is associated with numerous other negative individual, family and neigh- borhood circumstances. In particular, respondents with a history of father’s incarceration are more likely to report having mothers with a history of binge drinking; arrests as juveniles; friends who engaged in substance use; lower

family and neighborhood socio-economic status; and lower levels of father involvement.

Frequency of marijuana use

Associations between FEI and frequency of marijuana use are provided in Table 2 stratified by sex. Model 1 includes controls for age and race. Model 2 adds a test for the interaction of FEI and age. To test the robustness of these associations, controls are introduced in model 3 for a variety of individual, family and neighborhood factors.

In Fig 1, mean frequencies are predicted using regres- sion coefficients in model 2 of Table 2, for males and females, respectively. Models are fitted using the ‘xtpois- son’ command in STATA version 10.1. The presented mean frequencies are estimated using only the subset of observations in each age by FEI status subgroup.

Males

Among males, having an FEI is associated strongly with increased marijuana use, controlling for age and race. FEI increases the frequency of marijuana use by a factor of e(0.57) = 1.77 (P < 0.001). In model 2, the interaction of FEI with age is highly significant, and suggests that father’s incarceration is associated with an altered trajec- tory (i.e. a higher mean and varying slope associated with age and age-squared) of marijuana use relative to those whose father has not been incarcerated (Fig. 1a). For respondents without an incarcerated father, marijuana use peaks around age 21 at slightly more than five times per month and steadily declines. In contrast, for respon- dents who had a FEI, marijuana use begins at a higher initial frequency, and then increases to a plateau of approximately six times per month. These results suggest that marijuana use follows a typical age–deviance curve among youth without an incarcerated father, with increasing prevalence during adolescence followed by decline in the early 20s. However, those with an incarcer- ated father display continued elevated frequency of usage into young adulthood; those who had a FEI use mari- juana at age 24 at a level equivalent to the peak for those without FEI (age 21). In model 3, despite additional controls, the FEI–marijuana association and FEI–age interactions observed in model 2 remain significant and largely unchanged.

Females

Among females (Table 2), FEI is associated strongly with increased marijuana use. In model 1, FEI increases the frequency of marijuana use by a factor of e(0.56) = 1.75 (P < 0.001) controlling for age and race. The interaction of FEI with age in model 2 is highly significant, suggest- ing that FEI is again associated with a varying trajectory

124 Michael E. Roettger et al.

of marijuana use for females. Plotting the predicted prob- abilities in Fig 1b shows that among females without an incarcerated father marijuana use peaks around age 20, at a frequency just over two times per month before

steadily declining. In contrast, those with an incarcerated father start at an initially higher frequency of use and plateau at an even higher frequency of almost three times per month. Importantly, females with a FEI continue to

Table 1 Mean and standard deviations for respondents by sex and father’s incarceration.

Variable

Male respondents Female respondents

No report that father ever incarcerated

Report that father ever incarcerated

No report that father ever incarcerated

Report that father ever incarcerated

Dependent variables Marijuana usage

Wave I frequency of use 1.93 (7.34) 3.45 (13.52) 0.930 (4.80) 1.74 (7.28) Wave I proportion using 0.159 (0.366) 0.244 (0.429) 0.124 (0.329) 0.190 (0.392) Wave II frequency of use 2.21 (9.67) 3.06 (10.49) 0.997 (4.40) 2.03 (8.18) Wave II proportion using 0.176 (0.381) 0.258 (0.437) 0.141 (0.348) 0.210 (0.407) Wave III frequency of use 4.32 (13.29) 7.25 (18.17) 1.85 (7.96) 2.52 (6.49) Wave III proportion using 0.252 (0.434) 0.348 (0.476) 0.161 (0.367) 0.220 (0.414)

Other illegal drug usage Wave I 0.114 (0.318) 0.18 (0.289) 0.106 (0.236) 0.153 (0.360) Wave II 0.071 (0.257) 0.11 (0.304) 0.075 (0.249) 0.088 (0.283) Wave III 0.199 (0.400) 0.291 (0.454) 0.141 (0.343) 0.169 (0.376)

Independent predictors Age

Wave I 15.66 (1.69) 15.52 (1.72) 15.53 (1.72) 15.33 (1.69) Wave II 16.30 (1.59) 16.23 (1.63) 16.11 (1.63) 15.98 (1.61) Wave III 22.07 (1.73) 21.92 (1.73) 21.89 (1.73) 21.68 (1.70)

Race White [reference] 0.527 (0.499) 0.460 (0.498) 0.531 (0.499) 0.459 (0.498) Black 0.194 (0.39) 0.264 (0.440) 0.220 (0.414) 0.301 (0.458) Hispanic 0.164 (0.362) 0.196 (0.397) 0.151 (0.357) 0.177 (0.381) Asian 0.088 (0.294) 0.028 (0.166) 0.074 (0.262) 0.029 (0.178) Native American 0.016 (0.120) 0.044 (0.204) .0154 (0.123) 0.028 (0.165) Other 0.0105 (0.103) 0.007 (0.084) 0.009 (0.097) 0.006 (0.078) Lives with two biological parentsa 0.581 (0.493) 0.307 (0.461) 0.562 (0.496) 0.260 (0.439) Family SESb 6.43 (2.55) 5.46 (2.51) 6.28 (2.62) 5.29 (2.53) Neighborhood povertyc 0.110 (0.121) 0.144 (0.142) 0.117 (0.130) 0.152 (0.146) Low self-controld 0.271 (0.444) 0.321 (0.461) 0.277 (0.477) 0.345 (0.476) Mother’s binge drinkinge 0.091 (0.288) 0.133 (0.335) 0.094 (0.291) 0.153 (0.360) Respondent arrested as juvenilef 0.067 (0.250) 0.144 (0.351) 0.012 (0.124) 0.025 (0.156) Peer marijuana useg 0.612 (1.00) 0.831 (1.13) 0.540 (0.917) 0.749 (1.04) Prior physical abuseh 0.074 (0.271 ) 0.168 (0.374) 0.073 (0.257) 0.133 (0.340) Closeness to biological fatheri 4.10 (1.11) 3.49 (1.41) 3.74 (1.24) 3.03 (1.37) Father involvementj 1.30 (1.31) 0.934 (1.24) 1.07 (1.20) 0.694 (1.11) Parental supervisionk 1.44 (1.21) 1.41 (1.23) 1.46 (1.18) 1.51 (1.20) School attachmentl 3.96 (0.627) 3.79 (0.720) 3.90 (0.692) 3.76 (0.77) Number of observations 16 964 2744 18 897 3210 Number of respondents 6 146 982 6 830 1152

aRespondent resided with both biological parents at Wave I. bFamily socioeconomic status at wave I. Developed by Ford, Bearman & Moody [54] for use in Add Health. cParent indicates respondent has temperament issues. Used by Hagan & Foster [31] as a measure of low self control. dProportion of families in census tract below poverty level at wave I. eIndicator variable from parent interviews for: (i) biological mother’s self-report of binge drinking; (ii) mother/caregiver’s report that biological brother had history of alcoholism. fRespondent’s self-report of being arrested prior to age 18. gNumber of closest three friends who used marijuana monthly at wave I. hIndicator variable for self-report of physical abuse by parent or caregiver before age 12. iRespondent’s reported closeness to biological father. jScale measuring respondent’s wave I activities with father during the past month for the following activities: (i) gone shopping; (ii) played a sport; (iii) attended church service or activity; (iv) talked about relationship issues; and (v) attended concert, sporting event, movie, play or museum. Coded as: ‘1’ = yes, ‘0’ = no. kWave 1 summary score of whether or not a respondent’s parents set weekend curfews, controlled/limited contact with social circle, set bedtime, set limits on TV viewing and set limits on clothes worn. lWave I school attachment scale used by Hagan & Foster [31], averaging responses to questions regarding agreement with the following statements: (i) you feel close to others at school; (ii) you are happy at school; and (iii) you feel like you are part of your school. Coded responses were: ‘1’ = strongly disagree, ‘2’ = disagree, ‘3’ = neither agree nor disagree, ‘4’ = agree, ‘5’ = strongly agree. SES: socio-economic status.

Father incarceration and illegal drug use 125

T a

bl e

2 P

oi ss

on re

g re

ss io

n of

fr eq

u en

cy of

m ar

ij u

an a

u se

on pr

ed ic

to rs

,b y

se x

(r eg

re ss

io n

co ef

fic ie

n ts

w it

h st

an da

rd er

ro rs

in pa

re n

th es

es ).

V ar

ia bl

M al

es Fe

m al

(1 )

(2 )

(3 )

(1 )

(2 )

(3 )

Fo ca

lI V

s an

d ba

si c

co n

tr ol

s Fa

th er

ev er

in ca

rc er

at ed

(F EI

) 0

.5 7

** *(

0 .1

0 )

5 .4

7 **

*( 0

.4 1

) 5

.3 1

** *(

0 .4

1 )

0 .5

6 **

*( 0

.1 0

) 4

.9 9

** *(

0 .5

5 )

4 .8

6 **

*( 0

.5 5

) A

ge 1

.3 1

** *(

0 .0

2 )

1 .4

3 **

*( 0

.0 2

) 1

.4 3

** *(

0 .0

2 )

1 .1

1 **

*( 0

.0 2

) 1

.2 1

** *(

0 .0

3 )

1 .2

1 **

*( 0

.0 3

) A

ge sq

u ar

ed -0

.0 3

** *(

0 .0

0 )

-0 .0

3 **

*( 0

.0 0

) -0

.0 3

** *(

0 .0

0 )

-0 .0

3 **

*( 0

.0 0

) -0

.0 3

** *(

0 .0

0 )

-0 .0

3 **

*( 0

.0 0

) B

la ck

-0 .2

2 *(

0 .0

9 )

-0 .2

2 *(

0 .0

9 )

-0 .1

2 (0

.0 9

) -0

.3 1

** *(

0 .0

9 )

-0 .3

1 **

*( 0

.0 9

) -0

.5 9

** *(

0 .1

0 )

H is

pa n

ic -0

.1 2

(0 .1

0 )

-0 .1

2 (0

.1 0

) -0

.3 2

** *(

0 .0

9 )

-0 .3

5 **

*( 0

.1 1

) -0

.3 5

** *(

0 .1

1 )

-0 .5

3 **

*( 0

.1 0

) A

si an

-0 .6

4 **

*( 0

.1 3

) -0

.6 4

** *(

0 .1

3 )

-0 .7

5 **

*( 0

.1 2

) -0

.6 9

** *(

0 .1

5 )

-0 .6

9 **

*( 0

.1 5

) -0

.4 7

** *(

0 .1

4 )

N at

iv e

A m

er ic

an 0

.1 5

(0 .2

5 )

0 .1

5 (0

.2 5

) 0

.1 0

(0 .2

3 )

0 .5

5 *(

0 .2

8 )

0 .5

4 **

** (0

.2 8

) 0

.0 8

(0 .2

6 )

O th

er 0

.0 7

(0 .3

5 )

0 .0

8 (0

.3 5

) 0

.5 4

** **

(0 .3

2 )

-0 .0

2 (0

.3 9

) -0

.0 1

(0 .3

9 )

-0 .6

0 **

** (0

.3 6

) In

te ra

ct io

n s

FE I

¥ ag

e –

-0 .5

0 **

*( 0

.0 4

) -0

.5 0

** *(

0 .0

4 )

- -0

.4 2

** *(

0 .0

6 )

-0 .4

3 **

*( 0

.0 6

) FE

I ¥

ag e

sq u

ar ed

– 0

.0 1

** *(

0 .0

0 )

0 .0

1 **

*( 0

.0 0

) -

0 .0

1 **

*( 0

.0 0

) 0

.0 1

** *(

0 .0

0 )

A d

di ti

on al

co n

tr ol

s T

w o

bi ol

og ic

al pa

re n

ts -0

.3 5

** *(

0 .0

8 )

-0 .2

6 **

(0 .0

8 )

Fa m

ily SE

S 0

.0 5

** *(

0 .0

1 )

0 .0

4 **

(0 .0

1 )

N ei

gh bo

rh oo

d po

ve rt

y -1

.0 3

** *(

0 .2

7 )

-0 .2

6 (0

.2 8

) Lo

w se

lf -c

on tr

ol 0

.3 1

** *(

0 .0

7 )

0 .2

8 **

*( 0

.0 8

) M

ot h

er ’s

bi n

ge dr

in ki

n g

0 .3

7 **

*( 0

.1 1

) 0

.3 8

** *(

0 .1

1 )

R es

po n

de n

t ar

re st

as ju

ve n

il e

0 .8

5 **

*( 0

.1 2

) 1

.2 8

** *(

0 .2

8 )

Fr ie

n d

m ar

ij u

an a

u se

0 .5

8 **

*( 0

.0 3

) 0

.7 1

** *(

0 .0

4 )

R ep

ea te

d ph

ys ic

al ab

u se

Fa th

er cl

os en

es s

0 .0

0 (0

.0 3

) -0

.0 4

(0 .0

3 )

Fa th

er in

vo lv

em en

t 0

.0 2

(0 .0

3 )

-0 .0

5 (0

.0 3

) P

ar en

ta ls

u pe

rv is

io n

-0 .0

2 (0

.0 3

) 0

.0 2

(0 .0

3 )

Sc h

oo la

tt ac

h m

en t

-0 .3

5 **

*( 0

.0 5

) -0

.1 8

** *(

0 .0

5 )

In te

rc ep

t -1

2 .2

8 **

*( 0

.1 8

) -1

3 .4

8 **

*( 0

.2 0

) -1

3 .0

6 **

*( 0

.3 2

) -1

0 .6

8 **

*( 0

.2 3

) -1

1 .7

5 **

*( 0

.2 6

) -1

1 .8

3 **

*( 0

.3 5

) A

lp h

a 2

.1 2

** *(

0 .0

2 )

2 .1

2 **

*( 0

.0 2

) 1

.9 4

** *(

0 .0

2 )

2 .3

4 **

*( 0

.0 2

) 2

.3 4

** *(

0 .0

2 )

2 .1

4 **

*( 0

.0 3

) Lo

g- li

ke li

h oo

d -4

9 9

9 1

.6 -4

9 9

3 4

.2 -4

9 5

8 2

.7 -3

2 1

3 8

.5 9

-3 2

0 3

0 .9

-3 1

6 9

4 .5

N u

m be

r of

ob se

rv at

io n

s 1

9 7

7 4

1 9

7 7

4 1

9 7

7 4

2 2

1 4

2 2

2 1

4 2

2 2

1 4

2 N

u m

be r

of re

sp on

de n

ts 7

1 5

7 7

1 5

7 7

1 5

7 7

9 9

7 7

9 9

7 7

9 9

*P <

0 .0

5 (t

w o-

ta il

ed te

st );

** P

< 0

.0 1

;* **

P <

0 .0

0 1

;* **

*P <

0 .0

5 (o

n e-

ta il

ed te

st ).

SE S:

so ci

o- ec

on om

ic st

at u

126 Michael E. Roettger et al.

use marijuana at age 24 at a level higher than the peak level for females without FEI. In model 3, despite addi- tional controls, the association between FEI and mari- juana use and the interaction of FEI with age remain

significant and largely unchanged. These results suggest that both males and females with an incarcerated father are at risk for greater marijuana use and follow a different trajectory of use with age.

A Frequency of Marijuana Usage Among Males

12 14 16 18 20 22 24

Age

P re

d ic

te d

F re

q u

e n

c y o

f M

a ri

ju a n

a U

s a g

Males with FEI

Males without FEI

B Frequency of Marijuana Usage Among Females

0.5

1.5

2.5

3.5

12 14 16 18 20 22 24

Age

P re

d ic

te d

F re

q u

e n

c y o

f M

a ri

ju a n

a U

s a g

Females with FEI

Females without FEI

Figure 1 Trajectories of marijuana use among males and females, by age and whether respondent’s father was ever incarcerated. FEI: father ever incarcerated

Father incarceration and illegal drug use 127

Other illegal drug use

Results for models examining the association between having an incarcerated father and other illegal drug use are presented in Table 3, again stratified by sex. Variables are entered in the same sequence of models as before.

In Fig 2, mean probabilities are predicted using regression coefficients from model 1 of Table 3. Pre- dicted probabilities are estimated using only the subset of panel observations with the corresponding age and FEI status.

Males

Among males, paternal incarceration is strongly associ- ated with increased probability of other illegal drug use, increasing the risk of other illegal drug use in young adulthood by a factor of e(0.78) = 2.20 (P < 0.001), con- trolling for age and race (model 1). However, the interac- tion of FEI with age in model 2 is non-significant, suggesting that FEI is not associated with an altered tra- jectory of other illegal drug use. Plotting the predicted probabilities in Fig 2a shows that, among respondents without an incarcerated father, other illegal drug use rises exponentially and plateaus at a rate of about 14% between ages 21–24. Following a similar curve, risk of other illegal drug use among those with an incarcerated father rises and then stabilizes at a higher rate of about 21–25% at ages 21–24. In model 3, the association of FEI with other illegal drug use remains highly significant (P < 0.001), despite controlling for other factors.

Females

Among females, FEI is associated strongly with increased risk of other illegal drug use. In model 1, father’s incar- ceration increases risk of other illegal drug use over time by a factor of e(1.93) = 6.89 (P < 0.001), controlling for age and race. The interaction of FEI with age in model 2 is non-significant, suggesting that father’s incarceration is not associated with an altered age trajectory for female other illegal drug use. Figure 2b shows the predicted probabilities from model 2 by FEI and age for females. Between the ages of 19 and 24, the percentage of females using other illegal drugs plateaus at approximately 6–7% among those without an incarcerated father, compared to between 8 and 10% among those with a FEI. As was the case for males, the direct association of FEI with other illegal drug use remains statistically significant (P < 0.01) when controlling for a wide range of indi- vidual and family characteristics.

DISCUSSION

In this paper, we have investigated associations of paternal incarceration with substance use during the

transition to adulthood, uncovering two notable find- ings. First, having a father ever incarcerated (FEI) is associated with increased marijuana use and increased risk of other illegal drug use among both males and females. Secondly, in the case of marijuana use, father’s incarceration is associated with altered age-marijuana use trajectories, for both males and females. Further- more, these associations are robust in response to con- trols for background factors and other covariates including parental substance abuse, low self-control, juvenile delinquency, family structure and socio- economic status, peer drug use, prior physical abuse, school attachment and father closeness and attachment. Thus this paper provides some of the first evidence of an association between father’s incarceration and drug use within a large, nationally representative (i.e. non-clinical) sample.

These findings suggest that paternal incarcera- tion is associated with drug use during the transition to adulthood in the United States. Despite the inability of this study to establish causality, it suggests a number of tentative policy implications. Given that paternal incarceration is an increasingly common life-course event [4] and that 13% of young adults in the United States report a biological father having been incar- cerated [5], this study’s findings suggest that a sub- stantial population in the United States is at risk for heightened drug use. Increased drug use associated with paternal incarceration may also lead to substan- tial economic and social costs, including illegal drug market activity, increased crime and incarceration rates, lost work productivity and substance abuse treatment [51]. In a society where drug users make up one-fourth of incarcerated prisoners [52] and 60% of inmates report regular drug use prior to incarceration [53], the early onset and heightened risk of drug use also increases the likelihood of an intergenerational trans- mission of incarceration between fathers and their children.

This study is not without limitations. The school- based sampling design of Add Health may exclude youth who have dropped out of the school system and are most likely to be caught up in the criminal justice system. Further, Add Health has limited information regarding the biological father’s other characteristics, including histories of substance use and criminality. While moth- er’s history of binge drinking or alcoholism helps to capture intergenerational effects from the mother, the intergenerational influence of father’s substance use and criminality remains a potential source of unobserved heterogeneity.

While the measure of other illegal drug use, adapted from Cleveland & Weibe [47], captures a broad range of more serious drugs across three waves of data collection,

128 Michael E. Roettger et al.

T a

bl e

3 Lo

gi st

ic re

g re

ss io

n of

ot h

er il

le ga

ld ru

g u

se on

pr ed

ic to

rs ,b

y se

x (r

eg re

ss io

n co

ef fic

ie n

ts w

it h

st an

da rd

er ro

rs in

pa re

n th

es es

V ar

ia bl

M al

es Fe

m al

(1 )

(2 )

(3 )

(1 )

(2 )

(3 )

Fo ca

lI V

s an

d ba

si c

co n

tr ol

s Fa

th er

ev er

in ca

rc er

at ed

(F EI

) 0

.7 9

** *(

0 .0

9 )

0 .3

0 (2

.2 4

) 0

.4 7

** *(

0 .0

8 )

0 .6

6 **

*( 0

.1 5

) 2

.9 5

(3 .0

1 )

0 .2

5 **

(0 .0

9 )

A ge

0 .8

4 **

*( 0

.0 9

) 0

.8 3

** *(

0 .1

0 )

0 .7

8 **

*( 0

.0 9

) 0

.7 5

** *(

0 .1

2 )

0 .7

9 **

*( 0

.1 3

) 0

.5 4

** *(

0 .0

9 )

A ge

sq u

ar ed

-0 .0

2 **

*( 0

.0 0

) -0

.0 2

** *(

0 .0

0 )

-0 .0

2 **

*( 0

.0 0

) -0

.0 1

** *(

0 .0

0 )

-0 .0

2 **

*( 0

.0 0

) -0

.0 1

** *(

0 .0

0 )

B la

ck -1

.9 4

** *(

0 .1

1 )

-1 .9

4 **

*( 0

.1 1

) -1

.8 7

** *(

0 .1

1 )

-3 .4

4 **

*( 0

.2 3

) -3

.4 4

** *(

0 .2

3 )

-2 .0

0 **

*( 0

.1 2

) H

is pa

n ic

-0 .3

7 **

*( 0

.0 9

) -0

.3 7

** *(

0 .0

9 )

-0 .3

1 **

*( 0

.0 8

) -0

.6 2

** *(

0 .1

5 )

-0 .6

2 **

*( 0

.1 5

) -0

.2 6

** (0

.0 9

) A

si an

-0 .8

3 **

*( 0

.1 3

) -0

.8 3

** *(

0 .1

3 )

-0 .7

6 **

*( 0

.1 2

) -1

.3 4

** *(

0 .2

4 )

-1 .3

4 **

*( 0

.2 4

) -0

.7 2

** *(

0 .1

4 )

N at

iv e

A m

er ic

an 0

.2 9

(0 .2

1 )

0 .2

8 (0

.2 1

) 0

.1 1

(0 .1

9 )

0 .9

5 **

(0 .3

6 )

0 .9

5 **

(0 .3

6 )

0 .2

0 (0

.2 1

) O

th er

-0 .4

3 (0

.3 3

) -0

.4 3

(0 .3

3 )

-0 .4

6 (0

.2 9

) -0

.3 6

(0 .5

5 )

-0 .3

6 (0

.5 5

) In

te ra

ct io

n s

FE I

¥ ag

e –

0 .0

7 (0

.2 4

) –

– -0

.2 6

(0 .3

3 )

– FE

I ¥

ag e

sq u

ar ed

– -0

.0 0

(0 .0

1 )

– –

0 .0

1 (0

.0 1

) –

A d

di ti

on al

co n

tr ol

s T

w o

bi ol

og ic

al pa

re n

ts -0

.1 4

*( 0

.0 7

) -0

.2 1

(0 .3

2 )

Fa m

ily SE

S 0

.0 9

** *(

0 .0

1 )

0 .0

8 **

*( 0

.0 1

) N

ei gh

bo rh

oo d

po ve

rt y

-0 .4

6 (0

.2 8

) -1

.3 1

** *(

0 .3

2 )

Lo w

se lf

-c on

tr ol

0 .1

6 *(

0 .0

6 )

M ot

h er

’s bi

n ge

dr in

ki n

g 0

.0 6

(0 .0

9 )

0 .1

8 **

** (0

.1 0

) R

es po

n de

n t

ar re

st as

ju ve

n il

e 1

.0 5

** *(

0 .0

9 )

1 .3

6 **

*( 0

.2 2

) Fr

ie n

d m

ar ij

u an

a u

se 0

.6 6

** *(

0 .0

3 )

0 .6

6 **

*( 0

.0 3

) R

ep ea

te d

ph ys

ic al

ab u

se 0

.4 2

** *(

0 .1

0 )

0 .4

4 **

*( 0

.1 1

) Fa

th er

cl os

en es

s 0

.0 3

(0 .0

3 )

-0 .0

6 *(

0 .0

3 )

Fa th

er in

vo lv

em en

t -0

.0 2

(0 .0

3 )

-0 .0

1 (0

.0 3

) P

ar en

ta ls

u pe

rv is

io n

0 .0

1 (0

.0 2

) -0

.0 6

*( 0

.0 3

) Sc

h oo

la tt

ac h

m en

t -0

.3 6

** *(

0 .0

4 )

-0 .4

9 **

*( 0

.0 5

) In

te rc

ep t

-1 1

.6 1

** *(

0 .8

9 )

-1 1

.5 5

** *(

0 .9

8 )

-1 0

.5 0

** *(

0 .9

1 )

-9 .4

6 **

*( 0

.8 7

) -9

.6 8

** *(

0 .9

5 )

-6 .5

5 **

*( 0

.9 0

) In

di vi

du al

-l ev

el st

an da

rd er

ro r

0 .1

.5 8

** *(

0 .0

5 6

) 1

.5 8

** *(

0 .0

5 6

) 1

.1 3

** (0

.0 6

) 1

.8 4

** *(

0 .0

6 )

1 .8

4 **

*( 0

.0 6

) 1

.2 2

** *(

0 .0

6 )

Lo g-

li ke

li h

oo d

-7 2

0 5

.5 -7

2 0

5 .2

-6 6

6 8

.3 -7

0 4

2 .8

-7 0

4 2

.4 -6

5 6

9 .6

N u

m be

r of

ob se

rv at

io n

s 1

9 7

7 4

1 9

7 7

4 1

9 7

7 4

2 2

1 4

2 2

2 1

4 2

2 2

1 4

2 N

u m

be r

of re

sp on

de n

ts 7

1 5

7 7

1 5

7 7

1 5

7 7

9 9

7 7

9 9

7 7

9 9

*P <

0 .0

5 (t

w o-

ta il

ed te

st );

** P

< 0

.0 1

;* **

P <

0 .0

0 1

;* **

*P <

0 .0

5 (o

n e-

ta il

ed te

st ).

SE S:

so ci

o- ec

on om

ic st

at u

Father incarceration and illegal drug use 129

it does not allow detailing patterns of use for specific drugs, such as heroin, crystal methamphetamine or cocaine. Recent research suggests that long-term use patterns vary by type of drug [50].

Future research examining type of drug may yield additional insights into influences of paternal incarceration on substance use among youth.

A Other Illegal Drug Use Among Males

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

12 14 16 18 20 22 24

Age

P re

d ic

te d

P ro

b a

b il it

y o

f O

th e

r Il le

g a

l D

ru g

U s

a g

Males with FEI

Males without FEI

B Other Illegal Drug Use Among Females

0.00%

2.00%

4.00%

6.00%

8.00%

10.00%

12.00%

12 14 16 18 20 22 24

Age

P re

d ic

te d

P ro

b a

b il

it y

o f

O th

e r

Il le

g a

l D

ru g

U s

a g

Females with FEI

Females withoutFEI

Figure 2 Trajectories for probability of other illegal drug use among males and females, by age and whether respondent’s father was ever incarcerated. FEI: father ever incarcerated

130 Michael E. Roettger et al.

Finally, while this study substantially expands the existing literature linking paternal incarceration with substance use among youth, causal inference is limited by a lack of information regarding the timing of pater- nal incarceration and concurrent criminality. Future research incorporating these measures would help to establish (i) whether paternal incarceration is linked causally to youth substance abuse and (ii) how mediat- ing mechanisms related to paternal incarceration, such as family instability, poverty and stress, influence adolescents’ future drug use.

Declarations of interest

None.

Acknowledgements

This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman and Kathleen Mullan Harris, and funded by a grant P01- HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgement is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. People interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 West Franklin Street, Chapel Hill, NC 27516-2524, USA ([email protected]). No direct support was received from grant P01-HD31921 for this analysis. This project was supported with a grant to the National Center for Family and Marriage Research from the US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, grant number 1 U01AE000001-01.

References

1. West H., Sabol W. Prison Inmates at Midyear 2008— Statistical Tables. Washington, DC: Bureau of Justice Statistics; 2009.

2. Western B., Wildeman C. The Black family and mass incar- ceration. Ann Am Acad Pol Soc Sci 2009; 621: 221–42.

3. Herman-Stahl M., Kan M., McKay T. Incarceration and the Family: A Review of Research and Promising Approaches for Serving Fathers and Families. Washington, DC: US Depart- ment of Health and Human Services; 2008.

4. Wildeman C. Parental imprisonment, the prison boom, and the concentration of childhood disadvantage. Demography 2009; 46: 265–80.

5. Foster H., Hagan J. Incarceration and intergenerational social exclusion. Soc Probl 2007; 54: 399–433.

6. Foster H., Hagan J. The mass incarceration of parents in America: issues of race/ethnicity, collateral damage to chil- dren, and prisoner reentry. Ann Am Acad Pol Soc ScI 2009; 623: 179–94.

7. Hagan J., Dinovitzer R. Collateral consequences of

imprisonment for children, communities, and prisoners. Crime Soc Justice 1999; 26: 121–62.

8. Johnson E., Waldfogel J. Children of incarcerated parents: multiple risks and children’s living arrangements. In: Pattillo M., Weiman D., Western B., editors. Imprisoning America. Thousand Oaks, CA: Russell Sage Foundation; 2004, p. 97–134.

9. Murray J., Farrington D. Effects of parental imprisonment on children. In: Tonry M., editor. Crime and Justice: A Review of Research. Chicage, IL: University of Chicago Press; 2008, p. 133–206. 37.

10. Murray J., Janson C., Farrington D. Crime in adult offspring of prisoners: a cross-national comparison of two longitudi- nal samples. Crim Justice Behav 2007; 34: 133–49.

11. Gottfredson M., Hirschi T. A General Theory of Crime. Stanford, CA: Stanford University Press; 1990.

12. Thornberry T., Freeman-Gallant A., Lizotte A., Krohn M., Smith C. Linked lives: the intergenerational transmission of antisocial behavior. J Abnorm Child Psychol 2003; 31: 171–84.

13. Gabel S., Shindledecker R. Characteristics of children whose parents have been incarcerated. Hosp Commun Psychiatry 1993; 44: 656–60.

14. Phillips S., Burns B., Wagner H., Kramer T., Robbins J. Parental incarceration among adolescents receiving mental health services. J Child Fam Stud 2002; 11: 385–99.

15. Wilbur M., Marani J., Appugliese D., Woods R., Siegel J., Cabral J. et al. Socioemotional effects of fathers’ incarcera- tion on low-income, urban, school-aged children. Pediatrics 2007; 120: 678–85.

16. Robins L. Deviant Children Grown Up. Baltimore, MD: Williams and Wilkins; 1966.

17. Kinner S., Alati R., Najman J., Williams G. Do paternal arrest and imprisonment lead to child behaviour problems and substance use? A longitudinal analysis. J Child Psychol Psychiatry 2007; 48: 1148–56.

18. Hayatbakhsh M., Kinner S., Jamrozik K., Najman J., Mamun A. ‘Maternal partner criminality and cannabis use in young adulthood: prospective study.’. Aust NZ J Psychiatry 2007; 41: 546–53.

19. Petit B., Western B. Mass imprisonment and the life course: race and class inequality in U.S. incarceration. Am Sociol Rev 2004; 69: 151–69.

20. Chassin L., Curran P., Hussong A., Colder C. The relation of parent alcoholism to adolescent substance use: a longitudi- nal follow-up study. J Abnorm Psychol 1996; 105: 70–80.

21. Dallaire D. Incarcerated mothers and fathers: a comparison of risks for children and families. Fam Relat 2007; 56: 440– 53.

22. Duncan T., Tildesley E., Duncan S., Hops H. The consistency of family and peer influences on the development of sub- stance use in adolescence. Addiction 1995; 90: 1647–60.

23. Johnson J., Leff M. Children of substance abusers: overview of research findings. Pediatrics 1999; 103: 1085–99.

24. Merikangas K., Merikangas K., Stolar M., Stevens D., Goulet J., Preisig M. et al. Familial transmission of substance use disorders. Arch Gen Psychiatry 1998; 55: 973–9.

25. Moss H., Clark D., Kirisci L. Timing of paternal substance use disorder cessation and effects on problem behaviors in sons. Am J Addict 1997; 6: 30–7.

26. Daniel J., Hickman M., Macleod J., Wiles N., Lingford- Hughes A., Farrell M. et al. Is socioeconomic status in early life associated with drug use? A systematic review of the evidence. Drug Alcohol Rev 2009; 28: 142–53.

Father incarceration and illegal drug use 131

27. Gottfredson M. R., Hirschi T. A General Theory of Crime. Stanford, CA: Stanford University Press; 1990.

28. Sheridan M. A proposed intergenerational model of substance-abuse, family functioning, and abuse neglect. Child Abuse Neglect 1995; 19: 519–30.

29. Fergusson D., Horwood L. Early onset cannabis use and psy- chosocial adjustment in young adults. Addiction 1997; 92: 279–96.

30. Dishion T., Xapaldi D., Spracklen K., Li F. Peer ecology of male-adolescent drug-use. Dev Psychopathol 1995; 7: 803–24.

31. Hagan J., Foster H. S/he’s a rebel: toward a sequential stress theory of delinquency and gendered pathways to disadvan- tage in emerging adulthood. Soc Forces 2003; 82: 53–86.

32. Ezell M., Cohen L. Desisting From Crime: Continuity and Change in Long-Term Crime Patterns of Serious Chronic Offend- ers. New York: Oxford University Press; 2005.

33. Moffitt T. Adolescent-limited and life-course persistent anti- social behavior: a developmental taxonomy. Psychol Rev 1993; 100: 647–701.

34. Sampson R., Laub J. Crime in the Making: Pathways and Turning Points Through Life. Cambridge, MA: Harvard University Press; 1993.

35. Murray J., Janson C. -G., Farrington D. P. Crime in adult offspring of prisoners: a cross-national comparison of two longitudinal samples. Crim Justice Behav 2007; 34: 133–49.

36. Harris K., Halpern C., Entzel P., Tabor J., Bearman P., Udry J. The National Longitudinal Study of Adolescent Health: Research Design [WWW document]. Available at: http://www. cpc.unc.edu/projects/addhealth/design (accessed 26 July 2010). (Archived by WebCite® at http://www.webcitation. org/5rWBedFQ4).

37. Allison P. Missing Data.Thousand Oaks, CA: Russell Sage; 2001.

38. Horton N., Kleinman K. Much ado about nothing: a com- parison of missing data methods and software to fit incom- plete data regression models. Am Stat 2007; 61: 79–90.

39. Royston P. Multiple imputation of missing values. Stata Tech J 2005; 5: 527–36.

40. Bonczar T. Prevalence of Imprisonment in the U.S. Population, 1974–2001.Washington, DC: Bureau of Justice Statistics; 2003.

41. Pager D. The mark of a criminal record. Am J Sociol 2003; 108: 937–75.

42. Uggen C., Manza J., Thompson M. Citizenship, democracy, and the civic reintegration of criminal offenders. Ann Am Acad Pol Soc Sci 2006; 605: 281–310.

43. Farrington D., Jolliffe D., Loeber R., Stouthamer-Loeber M., Kalb L. The concentration of offenders in families, and family criminality in the prediction of boys’ delinquency. J Adolesc 2001; 24: 579–96.

44. Hindelang M. Variations in sex–race-age-specific incidence rates of offending. Am Sociol Rev 1981; 46: 461–74.

45. Hindelang M., Hirschi T., Weis J. Correlates of delinquency—illusion of discrepancy between self-report and official measures. Am Sociol Rev 1979; 44: 995–1014.

46. Ford C., Bearman P., Moody J. Foregone health care among adolescents. JAMA 1999; 282: 2227–34.

47. Cleveland H., Wiebe R. Understanding the association between adolescent marijuana use and later serious drug use: gateway effect or developmental trajectory? Dev Psycho- pathol 2008; 20: 615–32.

48. Long S. Regression Models for Categorical and Limited Depen- dent Variables. Thousand Oaks, CA: Sage Publications; 1997.

49. Rabe-Hesketh S., Skrondal A. Multilevel and Longitu- dinal Modeling Using Stata. College Station, TX: StataCorp; 2008.

50. Hser Y., Evans E., Huang D., Brecht M., Li L. Comparing the dynamic course of heroin, cocaine, and methamphetamine use over 10 years. Addict Behav 2008; 33: 1581–9.

51. Cohen M. A., Piquero A. R. New evidence on the monetary value of saving a high risk youth. J Quant Criminol 2009; 25: 25–49.

52. Pastore A., Maguire K. Sourcebook of Criminal Justice Statis- tics, 2010; Table 6.0001.2005. Available at: http://www. albany.edu/sourcebook/pdf/t600012005.pdf (assessed 26 July 2010). (Archived by WebCite® at http://www. webcitation.org/5rW8kqD9U).

53. Karberg J., Mumola C. Drug Use and Dependence, State and Federal Prisoners, 2004.Washington, DC: Bureau of Justice Statistics; 2006.

54. Ford C., Bearman P., Moody J. Foregone health care among adolescents. JAMA 1999; 282: 2227–34.

132 Michael E. Roettger et al.

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.

Users should refer to the original published version of the material.